Compare commits

...

54 Commits

Author SHA1 Message Date
Xavier Roche
5d22b59809 Probe run 4: hunt the full window and count distinct PIDs per set
Run 3 answered the question on the first five spawns and stopped there, which
leaves no collision rate and a sample of five. Runs the full 300s, counts
distinct reissued PIDs per set, and caps only the tree kills.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <xroche@gmail.com>
2026-08-12 17:18:17 +02:00
Xavier Roche
65940365ee Probe run 3: fix the null sets and the console-host confound
Run 2 built every PID set through a helper that returned the HashSet, which
PowerShell unrolls, so each set came back null and matched everything. It also
showed 500 console hosts surviving the orphan-makers, each holding a handle on
its dead creator and reserving that PID -- the likely cause of run 1's null.
Children now inherit this shell's console, and a wscript set adds a child that
attaches to none.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <xroche@gmail.com>
2026-08-12 17:16:11 +02:00
Xavier Roche
95c03624dd Probe run 2: four sets to separate PID reservation from bad luck
Run 1 saw zero collisions in 8095 spawns against 300 dead orphan-parent PIDs,
about 100x below chance, so those PIDs were unavailable rather than unlucky.
Adds baseline, held-handle and dead-child control sets, all freed at the same
instant, so the B result has something to be compared against.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <xroche@gmail.com>
2026-08-12 17:06:57 +02:00
Xavier Roche
bfb9d4611d Scratch probe: does taskkill /F /T follow a stale recorded parent PID?
Branch-scoped windows-2022 job, not wired into any gate. Part 1 dumps the
recorded-parent state of every runner process; part 2 hunts a PID collision
against 300 held orphans and tree-kills the reused PID.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <xroche@gmail.com>
2026-08-12 16:59:18 +02:00
Xavier Roche
de484e0670 Release 3.49.21 (#1226)
* Release 3.49.21

Version bumps in configure.ac, htsglobal.h, version.rc and the AppStream
metainfo, plus the curated release notes.

VERSION_INFO goes to 3:14:0, a revision-only bump: httrackp gained a tail
field (host_alias) and #1202 added an export, so nothing moved and nothing
went away.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Apply the release review: notes and changelog wording

The changelog entry named upstream flags, which the packaging audience does
not want; the notes instantiated example hostnames and quoted a UI label
where their neighbours stay abstract. The copyright bullet cites #1163 as
well, and the new export is named rather than implied.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 16:51:35 +02:00
Xavier Roche
b6d15c7bc8 Pin the host-alias bytes the WinHTTrack validator now depends on (#1224)
hts_host_alias_rule_ok accepts 0x80-0xFF and a leading '-', both by
design, and nothing pinned either. httrack-windows#109 dropped its
mirrored grammar for this exported validator, so a tightened reject set
would break it silently.

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 14:31:07 +00:00
Xavier Roche
db22036173 The Windows watchdog test calls a reaped ping a survivor (#1225)
* The Windows watchdog test calls a reaped ping a survivor

58_watchdog has failed the Win32 leg on every master push since the suite
went parallel. The grandchild check reads tasklist one second after the
watchdog fires and asks whether the ping's PID is still listed; neither half
survives sixteen concurrent tests. Match the image name as well as the PID,
and poll for it to go rather than sleeping once.

On a non-timeout failure the driver re-runs the test traced and appends to
the same log, then reports the last 25 lines: for an intermittent failure
that is the run that passed. Keep the failing run's tail before the trace can
append, and report both.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Control alive(), and pin the report's order and bound

The image-name conjunct is new, so a predicate that never matches would
call every survivor reaped: prove it fires on a live ping first, and fold
case the way proclib.sh's tasklist matchers already do.

172 now pins the two properties the marker alone left open: the failing
tail must come before the label, and both halves must stay tails.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 16:11:55 +02:00
Xavier Roche
50f151e11b freeport hands out a port it has already released (#1221)
`freeport` binds a socket to port 0, records the number and closes it, so the port is free again before its caller does anything with it. Under `make check -j16` two tests can be handed the same one. `start_proxytrack` already retried; the other two callers did not.

The one that matters is `56_local-proxy-noleak`, which wants a port that stays dead for the whole crawl so that connecting to it is refused. A retry cannot help when the exposure lasts the length of the test: if a sibling binds that number, the decoy proxy answers and the test stops testing its own premise while still reporting green. Such a port is now held rather than sampled, bound and never listened on, so connections are still refused. `htsserver_start` retries in the shape `start_proxytrack` already uses, and `freeport`'s comment now says which of the two a caller owes it.

Test 288 asserts the discriminator rather than the symptom, since an unbound port refuses a connection too: a competing bind must fail while the port is held, and succeed once it is released. Extracting the helper also fixed a leak in 276, whose private copy registered its cleanup inside a command substitution and left a python process sleeping 900s behind every run.

Closes #1218
2026-08-12 12:42:29 +00:00
Xavier Roche
729f823db9 A build path containing "assert" or "panic" fails the crawl tests (#1222)
* Read the crawl's own log, not the command line printed above it

hts-log.txt line 2 echoes the engine's whole command line, absolute paths
included, so --log-found and --log-not-found were matching the paths the run
was given. 36_local-bigcrawl searches for "assert", which made it fail for
anyone building under a directory whose name contains that word; TMPDIR, -O and
the source directory reach that line too.

Only that line is skipped. The banner above it names the URL, which
14_local-https does match, and it is left in place if it ever stops looking
like the echo: a "not found" audit hiding a real match is the worse failure.

Closes #1220

Signed-off-by: Xavier Roche <roche@httrack.com>

* Review: pin the banner too, and cut the comments

The pair of cases constrained only "line 2 gone, line 4 present", so the
first-cut mistake this fixes, dropping the banner as well, passed both.
14_local-https catches it but skips where TLS is not built.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 11:42:07 +00:00
Xavier Roche
01274a93f5 The WebHTTrack pages hardcode a copyright year the doc pages now compute (#1217)
* Expand the WebHTTrack footer year instead of hardcoding it (#1165)

The 23 pages under html/server/ carried 1998-2026 as literal markup, which
doc-chrome.py cannot reach, so from next New Year the docs and the GUI would
have disagreed with nothing to catch it.

htsserver already resolves a ${prefix:name} family in the templates it serves,
so the year becomes one more of those. Computed per request, it cannot drift,
and no regeneration or check has to watch it.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Classify the date: token in the template escaping audit (#1165)

185 requires every ${...} in the server templates to carry the escaping prefix
its position needs, and a bare key in element body is a finding. The year the
server computes is not runtime data and cannot carry markup, so it joins the
translations and build-time constants instead.

Paired both ways: a settings value in element body must still be a finding, or
the exemption is a hole rather than a classification.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Fall back to the clock on an unusable SOURCE_DATE_EPOCH, and harden the test

A garbage or over-range override reached time zero, so the footer rendered
1998-1970 rather than ignoring the value. The parse now has to consume the whole
string, and a value gmtime rejects falls back to the real clock.

The test grew what the review found missing: the tree-wide sweep that a single
converted page cannot satisfy, a pinned epoch an hour before New Year UTC read
from 14 hours east of it, so gmtime and localtime give different years, and the
two unusable overrides. Renumbered off 286, which master took meanwhile.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Refuse a SOURCE_DATE_EPOCH whose year no footer can show

glibc's gmtime accepts the whole int range of tm_year, so a parsable
SOURCE_DATE_EPOCH could reach the footer as "-2" or overflow the
tm_year + 1900 add. Both break the digits-only promise test 185's
date: exemption rests on.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 11:33:33 +00:00
Xavier Roche
bd765e4ed6 Give the compile-time rules one spelling, HTS_STATIC_ASSERT (#1219)
Three constants were pinned three ways: an enum dividing by zero in htsparse.c,
and two negative-array typedefs in htsarena.h. htsglobal.h already had
HTS_COMPILE_ASSERT for the expression case, so this adds the declaration case
beside it, on the enum form, and moves all three onto it.

Each of the three was checked by breaking what it guards and confirming the
build stops: a non-power-of-two alignment kills both arena rules, and shrinking
SINGLEFILE_MAX_SPAN kills the parse one.

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 11:24:34 +00:00
Xavier Roche
17e3d241fa Windows CI runs from different branches starve each other's runners (#1214)
* Windows CI runs from different branches starve each other's runners

windows-build keys its concurrency group on github.ref, so it only cancels
superseded runs of the same branch. Runs on different branches overlap freely,
and at two platforms each, up to six Windows jobs can share the pool. #732 and
#733 were two Win32 jobs starting 2 seconds apart that both wedged the 45-minute
step timeout and both passed on a rerun of the identical commits; #1126 wedged
the same way; PR #1203's x64 leg lost communication with the server while three
runs were in flight.

A job-level group keyed on matrix.platform caps that at one x64 and one Win32
job across the whole repository. It needs queue: max, because the default queue
holds a single pending entry and cancels it when the next job arrives, which
would trade starvation for cancelled checks. The workflow-level per-ref cancel
stays, so a second push to a branch still drops the older run and frees its
queue slot.

A Windows job takes about 34 minutes, so a third concurrent PR now waits instead
of racing. That wait shrinks once the suite step stops running its tests
serially.

Closes #1213

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Tighten the comment to two lines

Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 10:23:56 +00:00
Xavier Roche
ee20bfbb2b The Windows suite runs its tests one at a time (#1216)
The Windows suite step takes 23.5 minutes because `ci-windows-suite.sh` runs its 238 tests one at a time, while every `make check` on other platforms uses `-j min(2*nproc,16)`. Measured against the Linux suite test for test, Windows is only 1.61x slower at the same work: the gap is the missing `-j`, not the platform. Scheduling the real per-test times gives 3.3 min at `-j8`.

The pool is the small part. Three assumptions in the shared harness had to go first, because each is only safe while one test runs at a time. `kill_tree` fell back to killing every engine image and every `python.exe` on the host when it could not resolve a Windows PID, which under `-jN` kills the other tests' servers; it now takes a PID captured at fork time, and the host-wide sweep needs an explicit opt-in the driver sets only when it runs serially. `dump_crawl_logs` globbed and deleted every `httrack_local.*` in a shared TMPDIR, so each test now gets its own. The between-test reap taskkills by image name, so it runs once after the pool drains, trading per-test leak attribution for parallelism; a wedge is still bounded by the per-test timeout and its tree kill.

`wait -n` is bash 4.3, and macOS drives this script under 3.2, so the pool falls back to the serial path there. Test 286 runs a stub suite both ways and fails if the parallel leg does not beat the serial one; forcing the driver to one worker makes it red.

Closes #1215
2026-08-12 11:46:51 +02:00
Xavier Roche
bde9eadb05 An early exit from hts_main_internal leaks the rebuilt command line (#1210)
* Release the rebuilt command line on every early exit

hts_main_internal allocated it once and freed it only at the end, so each of
the 88 early returns leaked the token chunks and the slot array. It never
mattered much for the binary, which is exiting anyway, but hts_main2 is an
exported entry point and an embedder pays it per call.

htsmain_free() already runs on every one of those paths, and each caller
returns straight after, so the release belongs there.

Closes #1206

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Review: tighten the comments

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Name the new leak test in the Windows suite's expected skips

It needs a LeakSanitizer build, so it skips there like 114 does, and that job
fails any change to the skip set rather than let coverage drift silently.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 09:43:19 +00:00
Xavier Roche
88f0a0174d The Windows build rebuilds OpenSSL from source on every run (#1209)
The vcpkg binary cache has not served a package since 2026-08-04: both legs log `Restored 0 package(s)` and rebuild openssl from source, 7.7 min on x64 and 5.1 min on Win32. The key named only the manifest, so nothing busted it when a runner image orphaned every archive, and actions/cache skips the save on an exact hit. That is why it stayed broken for eight days rather than one run.

Naming every input vcpkg's ABI hash folds in would only make a list that can still come up short, so the restore keys on the run id and prefix-matches instead, the save runs when the archive set changed, and a cache vcpkg matched nothing in is dropped rather than carried forward. Two runs are needed to show it: the first supersedes the dead entry, the second should restore warm and bring the build step to roughly 2 min.

Closes #1208
2026-08-12 09:00:53 +00:00
Xavier Roche
696553b0f9 Three open-coded chunked pools with the same stable-pointer contract (#1212)
* Give the three chunked pools one implementation

htscore.c grew two of them, for link strings and for lien_url records, and
#1205 added a third for the rebuilt command line. All three exist for the same
reason, stated in one of their own comments: "There are references to item
pointers, so we can not just realloc()". All three picked 32768. The growable
primitives the tree already has are no substitute, since String and TypedArray
move their contents and coucal's pool rebases its keys on purpose.

htsarena.h is that pool, once: chunks that are never resized, doubling to a
ceiling, with the alignment and the overflow guards in one place instead of
three. The 32 KB cap on a single link string goes with it, and so does the
assert that used to abort on a longer one.

Closes #1207

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Charge alignment padding where it is needed, not to every allocation

Rounding each size up cost several bytes on every link string a mirror
records, where the pool it replaces packed them. Only the lien_url records
need the alignment, so pad at the allocation that wants it.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Pin the arena's alignment assumptions at compile time

Review: the padding takes HTS_ARENA_ALIGN for a power of two and the header
for a multiple of it, and nothing said so. A platform where either stops
holding would have miscompiled quietly; now it does not compile.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 08:57:30 +00:00
Xavier Roche
99d494bd39 The resume offer echoes the program path instead of the command line it would run (#1211)
* Quote the command line in the resume offer, instead of the program path

The offer to resume or update a mirror ends with the command line it would
run, but printed argv[0], so it read "OK to Continue httrack /usr/bin/httrack".
It has never worked: the buffer it reads holds the rebuilt command line as
NUL-separated tokens, so the append stopped at the first one.

Printing the rest is not a plain loop over them. HT_PRINT appends through
strcatbuff into a fixed 2 KB buffer and aborts on overflow, which a command
line replayed from doit.log outgrows, so the tokens are clipped to what is
left and closed with an ellipsis.

Closes #1204

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Review: tighten the comments

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Bound the clip mark too, so the append needs nothing of its caller

Review: the ' ...' went in through HT_PRINT, which aborts on overflow, so it
leaned on the preamble the caller had written leaving room. Not reachable
today, since every argv element is bounded, but the function should not
depend on that.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 08:37:37 +00:00
Xavier Roche
53eace3cfe A cleared box in the wizard cannot disable the index, and --purge-old=1 kills the run (#1201)
* A long option drops the value its short form takes

--index=0 dropped its value and read back as the bare --index, which enables
the index: the alias table calls -I a "single" option, and that class throws an
attached value away. So clearing "create a top index" in the WebHTTrack wizard
could not turn it off, and neither could --noindex.

Two new classes carry the value the short form already accepts: "onoff" for the
flags whose parser reads a 0 suffix (-I0), "level" for those that parse a number
(-%v2). Everything still "single" now refuses an attached value rather than
dropping it. The bare form of every alias is unchanged.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Assert the emitted short form against the engine, not the table

Review found the value-taking rows had no test tying them to the parser: the
table loop only checks bare forms, which are identical for all three classes,
so a row wrongly marked onoff would emit -x0 and kill the run unseen. 282 now
sweeps every onoff and level row the engine lists and fails on "invalid
option", and asserts the refusal message and the exact flag set each wizard
box emits rather than a substring.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Three param rows are booleans in disguise, so --purge-old=1 kills the run

purge-old (-X), httpproxy-ftp (-%f) and extended-parsing (-%P) are classified
param, but their parser cases have the onoff shape: set the flag, then consume
a 0 suffix. The class demands a following token, so --purge-old=1 emitted -X1
and died on "invalid option 1", and --purge-old before a URL ate the URL as its
value and mirrored nothing. A profile carrying NoPurgeOldFiles=0 makes the
wizard write exactly that command line.

They were misclassified rather than newly broken, so they move to onoff with
the rest. No fourth row has this shape: every other param row reads a real
value, and every param1 takes a separate string. help() now annotates an
optional value, which regenerates the man page and its rendered HTML.

Closes #1200

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* The users guide still promises a numeric value for the three flags

fcguide.html carries a pasted copy of the option list, where -X, -%f and -%P
read [=N]. Only 0 was ever meaningful, and help() now says [=0].

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* The class sweep reads the table it is checking

Review found it circular: a row that should take a value but is still single
never appears in the list, which is the #1200 bug exactly. step4.html is an
independent source, so every --name=0 it writes must now resolve to a row whose
class takes one. The sweep also tries =1, the value that emitted -X1 and killed
the run, and a refusal must carry a message.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Bound the digit run before indexing it

CodeQL cpp/offset-use-before-range-check: value[i] was read before i < 16
gated it. In bounds either way, but the wrong order for a parser fed user
input. Same boundary: 16 digits accepted, 17 refused.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 10:21:54 +02:00
Xavier Roche
1dcfcf491f The Android guide screenshots crawl the emulator's host-loopback alias (#1203)
10.0.2.2 is the Android emulator's fixed alias for the host machine's own
127.0.0.1, so the URL field and the progress line showed a reader an address
they cannot use. A grep for the literal loopback address reports the set clean,
which is why #1133's purge left it behind on this platform alone.

Regenerated from httrack-works 3988c35, where the walk now maps www.example.com
onto that alias inside the emulator. Two images move; the other 17 captures
reproduce byte for byte.

Closes #1199

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 07:45:41 +00:00
Xavier Roche
14e9108049 A --host-alias rule may carry bytes no host can, and no front end can call the validator (#1202)
* Reject a newline inside a --host-alias rule, and export the validator

hts_host_alias_rule_ok() did not reject '\n', which the engine uses to join
the rules it accepted, so --host-alias $'a.com=b.com\nc.com' validated as one
rule and hts_host_alias_match() then read it as two, dropping the '='-less
fragment without a word.

The same function moves to httrack-library.h behind HTSEXT_API, so a front end
can reject a malformed rule in its own field instead of mirroring the grammar
and drifting from it.

Closes #1192
Closes #1172

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Reject any control byte and an unparsable canonical scheme in a rule

Exporting the validator makes its reject set a supported contract, so close
the two holes review found beside the '\n' one. A control byte anywhere in a
rule is refused: '\v' and '\f' are not on the argv filter's CR/LF/TAB list and
reached the host the fold writes. The canonical side is cut with
jump_protocol_const rather than a literal "://", the way every consumer cuts
it, so "b.com=x://a.com" no longer folds http://b.com to a host of x://a.com.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 07:15:01 +00:00
Xavier Roche
5c068a3c65 A large .httrackrc aborts the engine on the rebuilt command line's token block (#1205)
* Grow the rebuilt command line's token block instead of aborting on it

The block was sized once, from the current argv plus fsize("config") plus
32 KB of slack, and never grew. The engine reads .httrackrc, httrackrc or
/etc/httrack.conf, never a file called "config", so nothing in that sum
accounted for what the expansion would add; past the slack cmdl_ins ran out
of room and strlcpybuff aborted the process on a config file the user wrote
themselves.

The block becomes a chain of chunks, so it grows the way the slot array has
since #1196. A chunk is never resized, which is what makes the growth safe:
every argv[] slot points into it and stays valid, as does argv_firsturl in
htscoremain.c, where a realloc would have had to rebase them all.

Closes #1198

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Name the chunk fields for what they are, and catch a clipped token

Review follow-ups: blk_size/blk_used named a blk field the chunk list replaced,
CMDL_CHUNK_DATA was a macro where its peer cmdl_room is a function, and the
test's two markers would both survive an implementation that clipped a token to
the room left instead of growing.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 07:11:45 +00:00
Xavier Roche
e9336b4d87 Nothing checks that guide.html's hardcoded image sizes match the images (#1193)
* Check guide.html's hardcoded image sizes against the files it ships

Every <img> in the guide carries a width/height pair, re-verified by hand
after each recapture. A screenshot regenerated at another geometry ships
stretched and nothing goes red, which is the failure mode a regeneration is
actually exposed to.

tests/guide-image-check.py parses the tags and compares each declared pair
against PIL's, and catches the two neighbours: an <img> whose file is gone,
and a guide-*.png under html/img/ the guide stopped referencing. 279 drives
it against the tree, then mutates a copy to prove each arm fires.

Closes #1187

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Close the review holes in the guide image check

The distcheck srcdir is read-only and cp -a carried that over, so the first
mutant hit EACCES and leaked its temp tree. Comment stripping, a
case-insensitive tag match and a guard on any src the parse cannot read stop
a defect from being skipped in silence, and a repeated attribute now resolves
the way a browser lays it out.

None of the old mutants touched a PNG, so a checker that never opened a file
passed every one of them. 279 now resizes a shipped image, which is the
defect #1187 is actually about.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Read the PNG geometry from the IHDR, so 279 runs on CI

No workflow installs Pillow and no runner image carries it, so the PIL guard
skipped 279 on every leg: a green suite that never ran the check. png-colors.py
already states the rule this broke, that the suite cannot assume Pillow on a
build host.

Twenty-four bytes of header carry what the check needs, so it reads them
itself, verifying the signature and the IHDR type rather than slicing on
faith. A missing checker is now a hard failure: it is in EXTRA_DIST, so its
absence is a packaging bug rather than an environment to tolerate. Three new
arms cover the reader's own refusals, and the image-side arm swaps in another
shipped PNG instead of resizing one through Pillow.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 02:35:33 +00:00
Xavier Roche
11c357e73a The installed headers are never compiled standalone or pairwise on MSVC (#1190)
* Compile the installed headers standalone and pairwise on MSVC

205, 206, 207 and 269 all reach the installed header set through automake's
install rule and gate on abs_top_builddir, so all four exit 77 on the Windows
job, which runs the test scripts directly. Nothing else compiles
DevIncludes_DATA the way a consumer does: the DLL build only proves the
engine's own fixed include order, with LIBHTTRACK_EXPORTS set.

Move 269's sweep into tests/install-headers-sweep.sh and drive it from both
legs. 269 hands it what automake installed; a new step in the existing
libhttrack job stages the same list out of the source tree and compiles it with
cl.exe, batching every unit into one response file per language because a spawn
on that runner costs more than the compile. 278 audits the wiring, since no
sweep can notice its own step being dropped.

Closes #1153

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Put the bytecode mode back on the command line, widen the wiring audit

Review found the -U half of the sweep could go vacuous: the mode was
emitted into each unit as a #define, so a -DHTS_INTERNAL_BYTECODE
arriving through TEST_CPPFLAGS made both halves the same compile while
the unit count still read right. Both modes are command-line flags
again, with a synthetic header that must fail in one and compile in the
other.

Also: assert five named headers reach the staged set, so a Makefile.am
reformat the awk mis-parses cannot shrink the sweep quietly; resolve a
../ entry against the build tree alone; and teach 278's audit to reject
a conditional or continue-on-error step or job, a --self-test that never
sweeps, a sweep placed before the build, and an invocation left as a
comment, each with its own control.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Find the vcpkg include root instead of spelling it

The MSBuild integration installs under src\vcpkg_installed\<triplet>\<triplet>,
nesting the triplet twice, so the step threw before it reached cl.exe. Search
for it, and name openssl/ssl.h as the thing that has to be found: htsbasenet.h
includes it by default, and it is the only reason the path is needed.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-12 00:28:34 +00:00
Xavier Roche
94d2d79217 A wizard project reload cannot read back a zero or an empty value (#1184)
* Keep a value of zero when a wizard project is reloaded

The winprofile.ini reader turned "0" into an empty value for every key, so a
rate, depth or timeout the user set to zero came back as the wizard's own
default. Only the checkbox keys, which step4.html writes as ${ztest:x:0:1},
need that inverse; everything else now reads its zero back.

Adds tests/274, the first test to reach the reader at all: it drives a real
project reload over HTTP and covers the escape set as well as the coercion.

Closes #1177
Closes #1171

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Let a present winprofile.ini key overwrite the wizard default

An empty value acquired a NULL String, which coucal_readptr refuses, so
${do:copy:} skipped the key and the startup default survived. Also guards the
truncated percent escape both unescape helpers walked past.

Refs #1186

Signed-off-by: Xavier Roche <roche@httrack.com>

* Guard the truncated percent escape in the two remaining copies

htstools.c reads the category out of the same winprofile.ini and proxystrings.h
carries a third copy of the loop, both walking past the terminator on a value
that ends in a bare %.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Default the cache box at startup, not on every step2 render

${do:set:cache:1} ran on each render, so a reloaded profile that cleared the
box had it ticked again by the next visit to step2. Only a fresh session may
take the default, which is what initOn[] is for.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Send the cleared cache and URL-hack boxes to the engine

Both read their state with test:, which emits nothing for an empty value, so
an unchecked box left the engine on its default while the box itself rendered
unticked. ztest: is the form the neighbouring cookies and parse-java lines
already use.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Assert a fresh wizard session still defaults to caching

Nothing covered the startup default itself, so dropping it from initOn[] left
the suite green.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 22:52:21 +00:00
Xavier Roche
52ac9ee142 A doit.log or a long command line overflows the rebuilt argv (#1196)
* Grow the rebuilt command line instead of guessing 1024 spare slots

htscoremain.c allocated argc + 1024 argv slots for the command line it
rebuilds and never re-checked that headroom, so both writers could run
off the end: cmdl_add on a command line whose --opt=value arguments each
expand to two tokens, and cmdl_ins replaying hts-cache/doit.log, which is
bounded only by the 8000-byte line the reader accepts.

The token count is not knowable up front (config files and doit.log each
add an input-defined number), so the slot array now grows on demand and
argc <= size is re-established before every write. cmdl_add, cmdl_ins and
optinclude_file take the array, its capacity and the token block as one
struct, which is what lets a growth reach every writer.

Signed-off-by: Xavier Roche <xroche@gmail.com>
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Cover the config-file writer and refresh the stale macro references

Signed-off-by: Xavier Roche <xroche@gmail.com>
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Address review: cmdl_init/cmdl_free, and a distinct OOM result for config files

The struct was half-initialized between its brace initializer and the two
fields the caller filled in by hand, and three new out-of-memory paths
returned without freeing it. cmdl_init and cmdl_free own both allocations,
so neither state is reachable.

optinclude_file returned "the file was read" on an allocation failure too,
which let the caller proceed with a half-expanded config. It now reports
that case apart, and the caller panics as its sibling writers already do.

Signed-off-by: Xavier Roche <xroche@gmail.com>
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <xroche@gmail.com>
Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 22:44:23 +00:00
Xavier Roche
d432772e1e A rule that begins with a dash kills the whole run (#1183)
* Take a rule that begins with a dash (#1179)

--strip-query and --host-alias take a user-written rule, and a rule may
legitimately start with '-'. The engine read it as the next option name,
so the whole run aborted before the mirror and the valid rules beside it
were lost.

Closes #1179

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Review fixes: narrow the exemption, and prove the rule's bytes

Restore the diagnostic when a rule option is followed by a real option name,
and pin byte fidelity plus the short and '=' spellings in the test.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Review fixes: trim the comments added by this change

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Return the computed boolean rather than a ternary over it

House convention: a comparison or && chain is already an hts_boolean.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Run the no-parameter probes through an absolute httrack path

noparam cds before invoking it, and make check puts the build's src/ on PATH
relatively, so a bare name found nothing there. An installed /usr/bin/httrack
hid that locally; CI has none and every build leg failed.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 22:39:05 +00:00
Xavier Roche
997160c98f The guide's WebHTTrack and Android screenshots predate the fields they describe (#1197)
The WebHTTrack set was shot before #1166 and #1176 landed. Its Spider tab showed neither the Host aliases box nor the four-row Strip query keys that replaced a single-line field, though the surrounding prose already described both. That page is 229 px taller now, and `guide.html`'s hardcoded height moves with it. The Android set predated #97, #114 and #116 the same way.

Captures are httrack-works `1af9fed`, the Android half shot by that session at httrack-android `f3123ad`. Regenerated with `tools/doc-images.py png`.

Seven web captures moved and I checked each. The Spider tab is the real change. Progress and finished carry live crawl counters. The project-name shot picks up a fresh random temp suffix in its base path. Three option tabs differ only by antialiasing on their select widgets, visually identical at a mean channel delta of 39. All 19 Android shots moved because that session recaptured its whole set rather than the four stale tabs, which keeps the version line and clock consistent.

The Windows images belong to #1180, now merged, so the two never overlapped. Every image is re-checked against `guide.html`: 55 of them, one dimension changed, no other drift.

Closes #1194
2026-08-11 22:20:33 +00:00
Xavier Roche
c3a5deef57 freeport reserves the ICP port over TCP, not the UDP it will be bound in (#1185)
* Reserve the ICP port in UDP, and let proxytrack name a bind failure

freeport() reserved both proxytrack ports by binding a plain TCP socket, but
proxytrack opens its ICP socket with SOCK_DGRAM. Windows keeps excluded port
ranges per protocol, so a TCP-derived number can land inside a UDP exclusion
and all three retries lose the same way. It now takes the protocol of each
port it hands out and probes in that one.

The failure was hard to read because proxytrack printed strerror(errno) for a
socket error Winsock reports through WSAGetLastError(), so the one line that
should have named the cause said "No error". It now reports the protocol, the
port and the system message for the call that actually failed.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Report a resolver failure as one, and pin the probe's protocol to its port

Review of the first commit found smallserver_init() reading errno on the
gethost() path, where getaddrinfo() never sets it: an unresolvable host printed
"cannot bind tcp port N ...: Success (0)". It now carries its own code and
proxytrack says "cannot resolve".

Test 276 grew with it. The socket-type spy logs the port beside the kind, so a
probe that opens a UDP socket and hands back a TCP-drawn number no longer
passes; both proxytrack destinations are exercised, and a probe forced to
collide proves the redraw loop reports exhaustion. Renamed smallserver_init's
"family" parameter to socktype, which is what it carries.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 21:11:01 +00:00
Xavier Roche
b33ef757e5 Nothing runs black or flake8, so .flake8 binds nothing (#1182)
* Reflow tools/doc-images.py and tools/doc-chrome.py with black

Pure black output, no edits by hand. Settles the 14 E501 that .flake8
already forbade.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Gate black and flake8 on the tracked Python files in CI

.flake8 has bound nothing since it arrived: no job ran black, flake8 or
any other Python linter, which is how two tools/ scripts drifted. Fold
both into the existing lint job, over git ls-files rather than a glob
list so a .py in a new directory cannot escape.

Closes #1170

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Split the Python lint into per-tool steps

Review follow-up: shellcheck and shfmt each get their own step so a red
names the tool, and the list moves to a shared env var like SHELL_SCRIPTS.
Guard the empty list too, which flake8 would otherwise accept silently.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 21:01:27 +00:00
Xavier Roche
9bb989053d A target that is not a keyword shares one tab, and dark mode prints white on white (#1181)
* html/: target="_new" shares one tab, and dark mode prints white on white

`_new` is not a target keyword, so browsers treat it as a named browsing
context called "new" and every one of the ten hand-written links reuses the
same tab. `_blank` is the keyword they meant. #1157 fixed the generated
footer; these predate it.

doc.css's `@media print` forced a white body but left the palette alone, so
in a browser that keeps matching prefers-color-scheme: dark while printing,
`--ink` stayed #e6e6ee on paper. Reset the tokens instead, which also covers
the panel and anything added later that reads them.

Closes #1164
Closes #1158

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Also fix cache.html's two transposed target="new_"

Same defect one page over, found while reviewing: `new_` is no more a
keyword than `_new`. Test 277 now rejects any target value outside the four
keywords rather than grepping for one misspelling, so a transposition is
caught too, and it reads both quote styles.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 20:56:00 +00:00
Xavier Roche
c27d55c345 The guide's WinHTTrack screenshots predate the address, rate and host-alias changes (#1180)
Four of the guide's WinHTTrack images predate changes that shipped in httrack-windows. The walkthrough now crawls www.example.com:8099 instead of the loopback address (#104), a new project no longer starts capped at 25 KB/s (#103), and the Spider tab gained a Host aliases row (#101). Captures are httrack-works 8f63cbc, regenerated with `tools/doc-images.py png`.

Only those four move. The other fourteen Windows captures are byte-identical, and no web or Android image changed, so nothing rode in from a sibling capture set. Geometry is unchanged on all four, so `guide.html`'s hardcoded width and height still match.

The prose needs no addition: #1175 already flipped the Host aliases row to `data-for="win web droid"` ahead of the field shipping. The new image does surface something I have left alone, though. WinHTTrack draws Host aliases as a single-line box, the same height as Sitemap address, while the guide says "one rule per line". The engine walks newline-separated rules, so that control holds exactly one, which is what #1168 fixed on the WebHTTrack side. Whether it should be multi-line is httrack-windows' call, and I could not confirm the limit from source here, so I have written no caveat.
2026-08-11 20:52:32 +00:00
Xavier Roche
40e94df2cc The wizard can only hold one --strip-query rule (#1176)
* Take several strip-query rules in the wizard, one flag each

The engine accumulates --strip-query into a newline-separated list, one rule
per flag, but the wizard bound the option to a single-line input and emitted a
single flag, so a project built there held one rule. The field is a textarea
now and the arglist: directive #1166 added emits one flag per rule.

Closes #1168

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Review fixes: exact assertions, and a scoped rule that matches

The winprofile and reopen checks matched substrings, so a merged pair passed
both; they compare the whole value now. 26's scoped rule matched no host, which
left a mutant dropping every scoped rule green on all four legs, and the control
strips a key the fixture never carries.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Test the last-match-wins rule the guide now documents, and the escaping

Both legs of the accumulating pair carried the same key, so a union or a
first-match-wins engine passed them; the new leg needs the later rule to
replace the earlier one. 273 posts a rule holding a '<', which raw would close
the box early.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 21:04:30 +02:00
Xavier Roche
c71f37189a Reshoot the Windows progress screenshot on a busy mirror (#1174)
The guide showed a mirror in progress as one connection trickling in a
small file, which is not what the pane looks like in use.
httrack-windows#102 changed the capture walk to serve itself a larger site
over eight connections, so it now shows eight transfers in flight with
their bars part-filled.

Regenerated from httrack-works 82ca70b. guide-win-language moves with it:
its About line carries the engine version, and CI now builds against
3.49-20 where that set was shot at 3.49-19.

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 18:14:17 +00:00
Xavier Roche
cb91ecaedc Show the Host aliases row to WinHTTrack readers (#1175)
httrack-windows#101 merged as 2e65f8f, putting the field on WinHTTrack's
Spider page. The row now names all three front ends.

Closes #1173

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 18:13:55 +00:00
Xavier Roche
faa6964d18 The WebHTTrack wizard cannot reach --host-alias (#1166)
* Expose --host-alias in the WebHTTrack wizard and the guide

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Review fixes: bash 3.2, the reload binding, and the guide wording

mapfile is bash 4 and macOS runs the suite under 3.2, so the argv read becomes
a plain loop. The test now reopens the option page with the rule stored, which
is what a typo in ${attr:hostalias} would break: the field renders empty and
the project loses the rule on every reload.

The guide row takes the CLI help's own wording (hosts fold onto one) and says
what the validator does with a trailing slash.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Take several host-alias rules in the wizard, one flag each

The engine accumulates --host-alias one rule per flag, so a single-line field
could only ever declare one canonical host. The field becomes a textarea and a
new arglist: template directive emits one flag per rule.

Rules separate on any whitespace, the way the URL box separates two URLs,
except beside a ',' or an '=' where the space belongs to the rule the parser
trims it out of: "a.com , b.com = c.com" stays one rule. The profile keeps them
on one ini line with the separator percent-encoded, which unescapeini() turns
back into a line break on reload.

--strip-query has the same shape and is now #1168.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Pin the whitespace shapes the rules box accepts

A table over the runs a user actually types: repeated spaces, a tab, a CRLF
run, leading and trailing whitespace, and the runs beside a ',' or an '=' that
belong to their rule. The empty cases carry the weight, since the engine
refuses an empty rule and a blank line must not reach it as --host-alias "".

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Split the rules box on every whitespace the Android app splits on

The wizard's splitter took " \t\r\n" while the Android field uses Java's \s,
so a pasted "a=b\vc=d" was two rules there and one rejected rule here. Widen
it to isspace()'s C-locale set.

The guide's "Strip query keys" row also claimed win and web only, hiding from
the in-app Android guide a field the app has long shown as editStripQuery.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Show the Host aliases row to Android readers

httrack-android#116 merged as b90e8ec, so the app has the field the row
describes. The row travels with the engine the app pins, so a build that
carries this guide also carries the field.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 16:48:44 +00:00
Xavier Roche
fce40a5511 The Windows guide screenshots lag the rebaselined captures (#1169)
httrack-works 8ae37ab reshot the Windows set from master, moving two of the
images doc-images.py generates. The mirror-progress counters differ because the
demo site grew from 24 pages at 0.35s to 120 at 0.8s, so the animation capture
added by httrack-windows#100 has a crawl long enough to film. The welcome pane
differs only in the runner's D: volume label; no GUI changed.

Regenerating deliberately, because doc-images.py reads the httrack-works working
tree live: left alone these land in whichever unrelated PR next runs the script.

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 15:14:55 +00:00
Xavier Roche
0f3dbcded4 A long URL overflows the dedup key buffer in htshash.c (#1167)
* A long URL overflows the dedup key buffer in htshash.c

The dedup key is the host followed by the path, built with two unbounded
strcpy() into hash->normfil. Both halves come off the wire and each can reach
HTS_URLMAXSIZE * 2, so a hostile URL wrote up to twice what the buffer holds.
The write landed in the next field of the struct, where neither ASan nor
_FORTIFY_SOURCE can see it.

Size the key scratch for what it actually holds, a host plus a path, and clip
rather than abort on anything longer: the key only feeds the hash, equality is
decided by key_adrfil_equals_generic, so a clipped key costs a collision. The
match string hts_query_strip_keys() builds from the same two halves aborted on
the same input; it clips now too.

Closes #1160

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Bound the strip-query match string and let the test own the canary

Review follow-ups on the key fix: use the tree's own strclipbuff() instead of a
hand-rolled clip, size hts_query_strip_keys()'s own host+path buffer so an
in-range URL is never truncated into matching a rule the whole one would not,
and keep an overlong path out of fil_normalized_filtered_ex()'s smaller
scratch.

hash_url_key() is now the key builder the engine itself calls, so the self-test
poisons its own arena and catches an overrun of any size instead of reading a
private field. normfil2 goes back to its old size: it only ever holds one
normalized path.

Signed-off-by: Xavier Roche <xroche@gmail.com>
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Signed-off-by: Xavier Roche <xroche@gmail.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 17:00:14 +02:00
Xavier Roche
04c75c430a A site served under several hostnames is mirrored once per name (#1156)
Some sites answer on several hostnames with the same content. HTTrack treated each name as its own site. Links to the other name were out of scope by default, and once a filter let them in, the whole tree came down again under a second directory.

`--host-alias "alias[,alias...]=canonical"` (`-%C`, repeatable) names the hosts of one site. They then count as the same address for the travel rules, and each link is folded onto the canonical host in `url_savename`, which every link passes before it is type-probed, named, recorded or fetched. So one host reaches the wire and the cache whichever name linked there. Aliases match with the +/- filter glob and the last matching rule wins; the right side is a single literal host, checked when the option is parsed. Either side may name a scheme: on the right it replaces the link's, for a host that answers on one scheme only (`legacy.example.com=https://example.com`), and on the left it narrows the rule to that scheme (`https://www.example.com=ftp://files.example.com`). A rule may name anything an address holds and nothing it does not, so a path is refused with a diagnostic rather than ignored: scheme, credentials, host and port live in `adr`, a path lives in `fil`. Chains are followed to their end, since the fold has to give the same host however often a link is named; rules pointing in a circle name no canonical host, so those are left alone and reported.

The interaction with the url hacks is the subtle part. The mapping is decoupled from `-%u`, as `--strip-query` already is, so `-%u0` cannot silently turn it off. But the host is matched in the form the dedup layer is using: under the `www.` collapse a rule naming `example.com` also covers `www.example.com`, which `-%u` would otherwise key onto the bare name and away from the canonical one, and with the collapse off the rule matches the name as the link spells it, so `--no-www-dedup` still keeps those apart. The collapse applies to the rule as well as to the host, so either spelling of an alias works under `-%u`.

There is a `-#test=hostalias` self-test, and a crawl test that reaches the second name once per engine path that folds it: a shared page, a page nothing else links to, a redirect, a page whose type is known only after a probe, and a sitemap entry. A control run keeps the alias URL absolute and fetches none of them, and an HTTPS leg covers the case that made the canonical-side scheme necessary. A misdeclared direction, where the canonical host redirects back to the alias, was probed by hand: the existing same-file-redirect handling absorbs it and the crawl ends normally.

Closes #16
2026-08-11 13:07:15 +00:00
Xavier Roche
0e0072ec03 The footer copyright year is pinned to 2026 and will disagree with the site (#1163)
* Compute the footer copyright year instead of pinning it to 2026

The year was hardcoded because --check diffs the regenerated chrome against
the shipped bytes, so computing it would have turned every 1st of January
into a red CI leg. That left httrack.com, which prints date('Y'), and the
shipped pages disagreeing from next New Year with nothing to catch it.

Compute it, and let --check ignore the digits: a page shipped in an earlier
year stays in sync until something else rewrites it, and the next
regeneration restamps all 15. The mask is scoped to the footer region
because contact.html carries a second 1998-YYYY in its licence text, which
still has to match byte for byte.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Test the year tolerance, and stop it accepting non-ASCII digits

The mask was written \d, which in a Python str pattern matches every Unicode
decimal digit, so a footer carrying Arabic-Indic or fullwidth digits passed
--check where it failed before. [0-9] is what was meant.

Nothing exercised the new branch either. The tree ships the year it was
generated in, so on CI the two sides agree and the tolerance is never
reached; it would have run for the first time on a New Year, in production.
Honour SOURCE_DATE_EPOCH as man/makeman.sh already does, which pins the
stamp and gives the test a seam, and add 271 to drive it: a lagging year
passes, a drifted footer and non-ASCII digits still fail. Each assertion is
paired with the mutant that kills it.

Also rename undated() to mask_footer_year(): it masks one field in one
region, and the old name promised a page-wide date strip.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 12:26:37 +00:00
Xavier Roche
ceab9c834b Nothing in CI builds our only consumer of the installed headers (#1155)
* Add a CI job building HTTraQt against the branch

HTTraQt is the only consumer of our installed headers, so it is the only
thing exercising the public C++ API surface. Twice an httraqt FTBFS is how
we learned we broke source compat, the second time six days late and by
accident (#1150).

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Review fixes: m4/ in the filter, weekly run, stray-library shapes

m4 macros feed AM_CPPFLAGS for every installed header, so a change there
can break the consumer without touching src/. The weekly run catches Qt
and runner-image drift, since upstream itself is pinned.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Fetch HTTraQt from our mirror, shallow, by SHA

No PR should depend on SourceForge being up. xroche/httraqt mirrors the
tree with its full history; fetching the pinned SHA directly also drops
the whole-history clone and the separate pin check.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Name the mirror for what it is, build infrastructure

xroche/httraqt reads as a claim on a project whose author we have just
asked about restarting it. The -ci-mirror suffix does not.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Compile the installed headers with exceptions and RTTI off too

Qt permits -fno-exceptions/-fno-rtti, so a consumer's translation unit
may see our headers that way and 206 never did. Additive: the
compiler's default mode keeps its own coverage, and the flag pair is
probed like the -std ones so a compiler without it skips.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 11:59:06 +00:00
dependabot[bot]
082d87e284 Bump github/codeql-action from 4.37.4 to 4.37.6 (#1159)
Bumps [github/codeql-action](https://github.com/github/codeql-action) from 4.37.4 to 4.37.6.
- [Release notes](https://github.com/github/codeql-action/releases)
- [Changelog](https://github.com/github/codeql-action/blob/main/CHANGELOG.md)
- [Commits](https://github.com/github/codeql-action/compare/v4.37.4...v4.37.6)

---
updated-dependencies:
- dependency-name: github/codeql-action
  dependency-version: 4.37.6
  dependency-type: direct:production
  update-type: version-update:semver-patch
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>
2026-08-11 11:29:44 +00:00
Xavier Roche
58e8aec88f The doc footer drops the patents chicklet, and two pages escape the check (#1157)
* The doc footer drops the patents chicklet, and two pages escape the check

The shipped doc footer had drifted from httrack.com's: no End Software
Patents chicklet, no <small> around the copyright line so it rendered a
size larger than the site's, and a .5rem top margin holding it off the
panel's black edge. doc.css now takes site.css's .sitefooter shape (flex
row, line-height 1.3, no top margin), with the chicklet at the far end.

The chicklet ships as html/images/esp_chicklet.png and is referenced
relatively rather than by the site's absolute src, since these pages are
read off the install directory where that would be a broken image and an
outbound request from a local page. The link points at the canonical
https://endsoftwarepatents.org/innovating-without-patents/, where the
site's http URL lands after two redirects, and uses target="_blank"
because _new is not a target keyword.

The footer also becomes its own doc-chrome region, as the masthead did in
#1154. It used to sit inside the generated bottom region, which index.html
and guide.html do not have: both build their own sidebar and take the
masthead alone, so their footers were hand-maintained and unchecked.
Corrupting index.html's footer text passes --check on master and fails
after this; a page that loses the markers is reported rather than crashing
the generator.

The year stays hardcoded: --check diffs the regenerated chrome against the
shipped bytes, so computing it would red CI every 1st of January.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Report a malformed chrome region per page instead of aborting the run

A page whose region markers are half-present reached region()'s bare
SystemExit, which named no page and stopped the sweep, so a drift on any
later page went unreported. Catch it per page, print the page, and carry
on. A duplicate marker pair is now rejected too: region() substitutes the
first pair only, so a trailing second one shipped unchecked on any region,
the masthead included.

Also add rel="noopener" to the chicklet link, and reword two comments: the
footer's contrast note no longer claims .8em now that <small> nests inside
it, and doc.css points at site.css's .sitefooter as the shape it mirrors.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 11:28:05 +00:00
Xavier Roche
ce66c2391a The site menu is injected at a guessed </header>, not a marker (#1154)
* The site menu is injected at a guessed </header>, not a marker

httrack.com serves the html/ pages through a wrapper that injects the site
menu under the masthead, and it locates that spot by finding the first
</header>. Only index.html and guide.html carry a literal
<!-- doc-chrome:masthead -->. The other 13 get their masthead emitted inside
the generated top region, so its content is governed while the marker itself
never reaches the shipped bytes.

doc-chrome.py now emits the markers as part of top, so all 15 pages expose the
same injection point and --check keeps them there. Shipped bytes change by
exactly those two comment lines per page and nothing else, which matters
because the site copies these files verbatim. Two checks make the guarantee
direct rather than resting on "it sits in a generated region": a page carrying
the chrome whose file has lost the marker fails --check, and a page loading
doc.css with no markers at all is now an error instead of dropping out of the
check unnoticed.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Reword the three new comments after the review prose pass

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 09:49:17 +00:00
Xavier Roche
49e93b171b Nothing runs the test suite under emulation, so the buildd finds those failures first (#1149)
* Run the suite under qemu-user, where only the buildd does today

cross-arch.yml compiles for the nine ports architectures and stops at make, so
the Debian buildd is the first machine anywhere to execute the suite on one. It
reports by failing the package build, which is how both failures in #1146
reached sid past a green cross leg.

s390x rather than the hppa that reported it: a release architecture, so the
container image and the build-deps come from the main archive instead of
debian-ports, and big-endian at no extra cost. Gated on tests/ changes, since
the harness is where an assumption about the native shape of a process lands.

The script reads /proc/self/cmdline back before building: a runner that stopped
emulating would leave this leg green while covering nothing.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Read the emulation probe off a child, and pin what must pass

qemu answers /proc/self/cmdline from the guest argv it was handed, so a process
asking about itself sees bash and never the interpreter. Only another pid
reaches the host kernel, which is both why the self-check would have failed
forever and why the harness sees the shift at all.

make check exits 0 for an all-SKIP run, and this leg skips the crawl tests by
design, so a container that lost a dependency would report a green covering
nothing. Read the pass count back and pin a floor.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* TEMPORARY: revert #1147 to prove this leg reds on what it targets

Backed out in the next commit. Without it the leg cannot be shown to catch
anything, and the issue asks for exactly that.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Restore #1147, the control having shown the leg reds

The reverted run failed 105_suite-timeout under emulation, which is what the
hppa buildd reported. 151 skipped there rather than failing: its pacer fires on
this runner, where an emulated configure is far cheaper than on a buildd, so the
leg reproduces the argv-shift failure deterministically and the wall-clock one
only on a slow enough host.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 08:56:18 +00:00
Xavier Roche
3ea7a773fa Installed headers break under some include orders (#1152)
A consumer picks its own include order, and five ordered pairs of our installed headers do not compile. Both causes are a guard placed over a block that a later, better-informed include should still have got to run: htsstrings.h kept a partial copy of htsglobal.h's attribute helpers behind the same `#ifndef HTS_UNUSED`, so including it first starved htssafe.h of `HTS_CHECK_RESULT`; htsbasenet.h tested `HTS_USEOPENSSL` before anything had defined it, and its own `HTS_DEFBASENETH` then locked the OpenSSL block out for good, leaving `SSL` undeclared in htsopt.h. Both now include htsglobal.h rather than guess at it.

The first one is not hypothetical. It shipped in 3.49.15 and broke httraqt, our only reverse dependency, as an RC-severity FTBFS ([Debian #1143660](https://bugs.debian.org/1143660)); Debian worked around it downstream by dropping the include and it never reached us, so it is still live in 3.49.20. The second has never been reported and turned up while sweeping for the first.

205 and 206 could not catch either, because they compile each installed header on its own, and on its own every header passes. 269 sweeps all 182 ordered pairs in C and C++ and in both `HTS_INTERNAL_BYTECODE` states, six installed headers gating declarations on that macro. Its control is a synthetic pair that compiles in one order only: it proves the sweep can see the shape it hunts, without asserting that any real header stays broken. Verified by reverting each fix in turn, each failing 269 alone and naming its own pair, and by making the sweep vacuous, which the control catches.

Closes #1150
Closes #1151
2026-08-11 10:32:43 +02:00
Xavier Roche
5175cf986f A caught signal kills every in-flight transfer, and a stopped FTP transfer waits twice (#1145)
`back_wait()` called `select()` and threw the return away. On EINTR Linux leaves the three fd sets exactly as the caller filled them, so every in-flight socket reads back as an exception and the slot dies with a fabricated "Receive Error": any caught signal costs a transfer, and on master a burst of SIGCHLD against a trickling body truncates the mirror. Reading the return is only half of it, because the prompt `^C` people rely on during an HTTP download exists only through that same accident. Nothing on the receive path ever read `opt->state.stop`, so a correct `select()` on its own would leave HTTP waiting out `--timeout` the way FTP does today. `back_wait()` now sweeps every live non-FTP slot when the stop flag is up, sharing the mirror-limit abort loop, and reports the user's stop instead of a socket failure. FTP slots are left alone: the worker thread owns both socket and slot, and finishing one under a live `fwrite()` is a use-after-free. So are the transfers running under a cap, because `back_checkmirror()` raises the same flag the first time `-M` or `--max-time` is reached: while that grace runs the sweep takes only the slots still waiting to connect, and the limit block that owns those transfers ends them with its own status and their WARC record.

On the FTP side a stopped transfer against a silent server waited out `--timeout` twice, once on the data socket and again on the QUIT reply nobody was going to send. Only the second window goes. The data wait keeps its own, since the first `^C` promises to finish pending transfers and clipping that wait truncates transfers still making progress (#1107). With `--timeout=0` a stop stays unbounded, which is what the user asked for, and a second `^C` already exits at once. No thread drain either: teardown is #1051, and a partial drain would hang the exit against that unbounded wait. Tests 262 and 263 cover both halves against a no-signal control arm, and the backstop self-test now pins the wider sweep with the FTP slot as its control.

Closes #1110
Closes #1096

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 08:31:07 +00:00
Xavier Roche
d25dd436ee cleanup_push traps a SIGPIPE the runners never let it see (#1141)
Settles #1136 as (b): PIPE leaves cleanup_push's trap list.

A non-interactive bash cannot trap a signal it inherited as ignored, and that is how the ubuntu runners hand SIGPIPE down, so `trap -p PIPE` reads back `trap -- '' SIGPIPE` and the entry was inert on the very hosts it was meant to cover. `perl -e '$SIG{PIPE}="IGNORE"; system("bash", ...)'` reproduces the CI red exactly: the subject sits out its full 30s sleep and exits 0. The container and macOS legs passed because their process ancestry resets the disposition. Even where it installed the entry bought nothing, since bash runs the EXIT trap for a fatal PIPE and the stack drains either way; only QUIT needs its own trap, because a non-interactive shell ignores it.

256 now reads the trap list back out of a live subject and checks it against the list it knows, so the next signal added without a leg goes red. Putting PIPE back makes that new case fail.

Closes #1136

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 08:18:36 +00:00
Xavier Roche
0fe1e2ba34 An aborted --update run purges against the files it never reached (#1142)
An interrupted run still ran the end-of-mirror purge, comparing old.lst against the truncated new.lst of what it had managed to fetch. Everything it had not reached was deleted, and `--purge-old` is on by default, so no flag has to be set for this.

The gate now reads `opt->state.stop` and `state.exit_xh` once the crawl loop ends, and skips the purge and the `--changes` report when either is set. Reading the cursor is not enough: a stop that lands while a page is in flight makes the parser drop the links that page carries, so a run can reach the end of its list having never queued half the mirror. Every stop route sets `state.stop` (signals, the GUI, and the `--max-time` and `--max-size` caps through `hts_request_stop`), and nothing sets it on a mirror that ran to the end. The report is suppressed for the reason #1061 gives: a run that never walked the list cannot say what is gone.

The rest of the completion sequence stays. `hts_finish_makeindex` and `index_finish` close files the run was writing, `singlefile_process_mirror` only rewrites pages that arrived whole, and `warc_close_opt` commits records for transfers that really happened, where `warc_abort_opt` would throw them away. The summary line is the other change: it says "mirror aborted after" instead of claiming a complete mirror.

Test 261 covers both halves, a SIGTERM during a stalled FTP `--update` and a `--max-time` cap firing while the page carrying the only links to six leaves is still in flight. The complete crawl in between still purges a file it dropped, so a fix that merely silences the purge fails there.

Closes #1109

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 07:47:55 +00:00
Xavier Roche
9bc300296f The Windows suite has no wall-clock bound, so a blown budget reaches the workflow timeout (#1144)
The Windows suite has no wall-clock bound. Its heartbeat kills only a progress log that has gone static, and the suite deadline is read between tests, but every path that follows a blown budget writes a progress line of its own, and each of those re-arms the staticness window. A test starting a second under the deadline can spend the per-test budget, then a traced rerun on a budget of its own, and land on the workflow's 45-minute step timeout. A cancelled step keeps neither its log nor its artifacts, so nobody learns what it was doing.

The heartbeat now takes an absolute cap and kills at 2400s whatever the progress log is doing, and the traced rerun is charged to the suite deadline instead of getting a second budget. 226 asserts the arithmetic against the workflow's own timeout, 171 drives the cap on its virtual clock against a suite that never stops reporting, and both assertions were checked by mutation. I could not reproduce any of this locally, since it needs a Windows runner, so the fix rests on reading the two scripts rather than on a repro.

The artifacts of the runs the issue cites are still live, and they say something else. Both Win32 legs finished the suite on their own terms and kept their step log and artifacts. Run 31397002827 ran 1264s and failed on 79_local-proxytrack-webdav-mime, whose proxytrack bound none of three port pairs; run 31405831680 ran 1060s and failed on 226 tripping over the new name in the driver's comment. 260 ran on both Windows legs of the first run and passed, in 11s on Win32 and 13s on x64, so it goes back into the selection through a crawl-harness category. The one real overrun is the x64 leg of 31405831680, which ran past 44 minutes and was cancelled with nothing surviving. 260 was already renamed out of the globs in that tree, so it was not the cause, and nothing survives to say what was.

Closes #1126

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 07:47:36 +00:00
Xavier Roche
a6315d04a3 Two suite tests fail on Debian's emulated hppa buildd (#1147)
* The hppa buildd kills 151 and hides the engine from 105

Debian's hppa is a qemu-user chroot, and two suite tests fail there in 3.49.20-1.

105 read /proc/<pid>/cmdline's first field as the program name, which binfmt_misc
gives to the interpreter, so its own probe never saw the fake engine start.
proclib already knew that shift; hoist the two regexes it matches on and reach
them from the shell too.

151 was killed at the 600s budget with four cases to go. The pacer reserved 1.5x
the last step for one more step, and its first twelve cases are cheap rejects
while the last four run configure to completion. Reserve for every step left
instead, past the first, which 196 spends warming a shared config.cache.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

* Order 151's expensive cases first instead of projecting the pacer

Reserving every remaining step at the last one's cost skips a run that fits
whenever one step is slower than its neighbours: a 2x outlier in 196 case 2
projects over four remaining cases and skips a run that ends at 562s of 600.
Keep the one-step-ahead reserve and give 151 the ordering it needs instead --
its four full configures first, so no step left can outrun the reserve the one
before it set.

Also fold proc_program_name onto the awk prologue so the interpreter shift has
one implementation, keep qemu-img-static and friends out of the shift (they feed
kill -9), and pin the reserve's half-step margin, which no fixture covered.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

---------

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 07:06:51 +00:00
Xavier Roche
893b542953 getrandom() is called undeclared when configure finds the function but not its header (#1143)
configure probes `sys/random.h` and `getrandom()` separately, so a libc that exports the symbol without shipping the header reached the call site with no prototype in scope. gcc 14, clang 16 and C23 reject an implicit declaration, so that build fails instead of falling back to /dev/urandom. #1138 guarded the include on both macros and left the call on one; a single macro derived at the include now gates both.

Nothing reached that cell before. There is now a CI leg that configures with `ac_cv_header_sys_random_h=no`, checks the combination really came out that way, and builds: red on master, green here. With getrandom gone it is also the only leg that runs the /dev/urandom fallback. The `enum { x = 1/(cond) }` pin at the call site covers the rest, so if the two guards split again it is a build error even where the compiler would only warn.

`hts_random_bytes` now answers a zero-length ask before reaching any source. The getrandom path already did; the fallback opened /dev/urandom to read nothing, and Windows left it to RtlGenRandom's undocumented behavior.

The new selftest (`-#test=random`, test 267) covers `hts_random_bytes` itself, not the compile break. It draws eight times with a different filler byte each round, so a byte the fill skipped is distinguishable from one it wrote, and it checks that nothing past the ask is touched. Mutants that fill half the buffer, run one byte past the end, return a constant, or scribble on a zero-length ask all fail it. A counter stub passes, since what it asserts is extent and liveness, not entropy quality.

Closes #1139

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 06:44:37 +00:00
Xavier Roche
3232f81373 Wizard domain answers match beyond the domain (#1140)
Three of the wizard's answers built their host filter as `<host>*`, with nothing between the host and the wildcard, so `+foo.com*` also authorized `foo.com.evil.org`. Answering "mirror all links located on the same domain" on a page linking to such a host pulled it into the mirror. Answer 5's allow-the-domain branch had the same shape, and answer 2's exclude over-excluded for the same reason. All three now emit `<host>/*`, and nothing is lost at the bare-host end because both forms the wizard matches against insert a `/` after the host.

This is a visible behavior change: someone who answered "same domain" and relied on the loose match will now see fewer hosts accepted.

To make the emission testable without a crawl, it moved out of the answer switch into `wizard_answer_filter()`, which `-#test=wizardfilter` drives from tests/264. A differential against the old inline code over answers, hosts and paths came back with only the three intended changes; answer 1's collapse of a doubled slash is preserved and pinned by the test.

Closes #1119

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-11 06:25:06 +00:00
commitmaniac
4a5854f31d htsrandom: the macOS build fails on an include only getrandom needs (#1138)
macOS's sys/random.h uses u_int without defining it, so a bare include does not compile. Autoconf's header test passes anyway, because it prepends sys/types.h before testing. The only consumer of that header here is getrandom(), which macOS does not have, so the include is dead weight on that platform. Guard it on HAVE_GETRANDOM as well, matching the guard already on the call.

Reported by MacPorts while upgrading to 3.49.19. Tested on macOS 10.7.5 with Xcode 4.6.3.

Signed-off-by: commitmaniac <201806046+commitmaniac@users.noreply.github.com>
2026-08-11 07:08:26 +02:00
Xavier Roche
df0fbd4711 Put the URL pane back as the software-centre screenshot (#1137)
#1132 moved to the scan-rules pane because every wizard pane advertised a
loopback address. #1135 reshot the set with the crawl addressed as
www.example.com, and the site now serves those bytes.

Signed-off-by: Xavier Roche <roche@httrack.com>
Co-authored-by: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-10 21:40:26 +00:00
210 changed files with 7864 additions and 1037 deletions

View File

@@ -700,6 +700,47 @@ jobs:
if: failure()
run: cat tests/test-suite.log 2>/dev/null || true
# The cell of #1139: a libc exporting getrandom() without shipping
# <sys/random.h>. Nothing else reaches it, and it is a hard build break when
# the call is not gated on the header. It also runs the /dev/urandom fallback.
no-getrandom:
name: build (getrandom header hidden)
runs-on: ubuntu-24.04
steps:
- uses: actions/checkout@v7
with:
submodules: recursive
- name: Install build dependencies
run: |
set -euo pipefail
sudo apt-get update || true # a third-party source in the runner image can 403 (#1097)
sudo apt-get install -y --no-install-recommends \
build-essential autoconf automake libtool autoconf-archive \
zlib1g-dev libssl-dev libbrotli-dev libzstd-dev
- name: Configure (sys/random.h hidden from configure)
run: |
set -euo pipefail
autoreconf -fi
./configure ac_cv_header_sys_random_h=no
# A libc without getrandom() at all would pass this job vacuously.
grep -q "define HAVE_GETRANDOM 1" config.h
! grep -q "define HAVE_SYS_RANDOM_H" config.h
- name: Build
run: make -j"$(nproc)"
- name: Test
timeout-minutes: 20
run: |
jobs=$(( $(nproc) * 2 )); [ "$jobs" -le 16 ] || jobs=16
make check -j"$jobs"
- name: Print the test log on failure
if: failure()
run: cat tests/test-suite.log 2>/dev/null || true
# OpenSSL 4.0 removes the API 3.x deprecated. Building with that set hidden
# keeps the engine ready for it: a deprecated call fails here rather than the
# day a distro moves. Compile-only, since the headers are what is under test.
@@ -837,6 +878,7 @@ jobs:
fi
lint:
# The name is a required status check on master; renaming it blocks every PR.
name: lint (shellcheck, shfmt)
runs-on: ubuntu-24.04
# Every tracked shell script; the globs expand at run time. Kept here so the
@@ -865,8 +907,13 @@ jobs:
# version; use it rather than fetching a release binary from github.com.
sudo apt-get install -y --no-install-recommends shellcheck shfmt appstream \
appstream-util
# Pinned to the local dev versions: black's stable style moves across
# majors, so the runner image's build would reformat on its own schedule.
pip install --quiet black==25.1.0 flake8==7.1.1
shfmt --version
appstreamcli --version
black --version
flake8 --version
- name: shellcheck
run: shellcheck $SHELL_SCRIPTS
@@ -874,6 +921,22 @@ jobs:
- name: shfmt
run: shfmt -d -i 4 $SHELL_SCRIPTS
# Read the list rather than hand-keeping one: a .py added in a new
# directory then gets checked without anyone widening a glob.
- name: List the tracked Python files
run: |
set -euo pipefail
mapfile -t py < <(git ls-files '*.py')
test "${#py[@]}" -gt 0 # flake8 given no paths exits 0, checking nothing
echo "${#py[@]} tracked Python files"
echo "PYTHON_FILES=${py[*]}" >> "$GITHUB_ENV"
- name: black
run: black --check --diff $PYTHON_FILES
- name: flake8
run: flake8 $PYTHON_FILES
# A misnamed test falls outside the glob "make check" runs, so it goes
# missing without failing anything.
- name: test file names

View File

@@ -38,7 +38,7 @@ jobs:
build-essential autoconf automake libtool autoconf-archive \
zlib1g-dev libssl-dev
- uses: github/codeql-action/init@v4.37.4
- uses: github/codeql-action/init@v4.37.6
with:
languages: c-cpp
build-mode: manual
@@ -67,6 +67,6 @@ jobs:
./configure
make -j"$(nproc)"
- uses: github/codeql-action/analyze@v4.37.4
- uses: github/codeql-action/analyze@v4.37.6
with:
category: "/language:c-cpp"

78
.github/workflows/emulated-suite.yml vendored Normal file
View File

@@ -0,0 +1,78 @@
# cross-arch.yml stops at compile for the ports architectures, so the Debian
# buildd is the first machine to run the suite on one, and it reports by failing
# the package build (#1146, #1148). This runs it under qemu-user, where a process
# is not the shape a native one is and every step costs far more.
name: Emulated suite
on:
# The harness is where a native-shape assumption lands, so gate on it rather
# than on every PR: this leg is slow and is not a required check.
pull_request:
paths:
- "tests/**"
- "tools/emulated-suite.sh"
- ".github/workflows/emulated-suite.yml"
push:
branches: [master]
schedule:
- cron: "41 5 * * 1"
workflow_dispatch:
# Least privilege: the workflow only needs to read the repo.
permissions:
contents: read
concurrency:
group: emulated-suite-${{ github.ref }}
cancel-in-progress: true
jobs:
emulated:
name: emulated suite (${{ matrix.arch }})
runs-on: ubuntu-24.04
# The emulated apt, configure and build take the bulk of it, well under the
# 6h ceiling: a wedge should not run that long.
timeout-minutes: 90
strategy:
fail-fast: false
matrix:
# Emulation is what is under test, not parisc: the hppa that reported
# #1146 lives in debian-ports, with no container image and a far more
# fragile archive. s390x is in main, and big-endian, which nothing else
# here covers.
arch: [s390x]
steps:
- name: Register the binfmt handlers
run: |
set -euo pipefail
sudo apt-get update
# Registered with the F flag, so the kernel holds the interpreter open
# and the container needs no copy of it.
sudo apt-get install -y --no-install-recommends qemu-user-static binfmt-support
test -e "/proc/sys/fs/binfmt_misc/qemu-${{ matrix.arch }}" ||
{ ls /proc/sys/fs/binfmt_misc; exit 1; }
- uses: actions/checkout@v7
with:
submodules: recursive
- name: Build and run the suite emulated
run: |
set -euo pipefail
# The build dir is a mounted volume, so the suite log outlives the
# container: a failure here has no detail otherwise.
mkdir -p "$RUNNER_TEMP/emul-bld"
docker run --rm --platform "linux/${{ matrix.arch }}" \
-v "$GITHUB_WORKSPACE:$GITHUB_WORKSPACE" -w "$GITHUB_WORKSPACE" \
-v "$RUNNER_TEMP/emul-bld:/bld" \
-e "GITHUB_WORKSPACE=$GITHUB_WORKSPACE" \
-e "EMULATED_ARCH=${{ matrix.arch }}" \
debian:sid bash ./tools/emulated-suite.sh
- name: Collect the suite log
if: failure()
uses: actions/upload-artifact@v7
with:
name: emulated-suite-${{ matrix.arch }}-log
path: ${{ runner.temp }}/emul-bld/tests/test-suite.log
if-no-files-found: warn

106
.github/workflows/httraqt.yml vendored Normal file
View File

@@ -0,0 +1,106 @@
# Build HTTraQt, the only consumer of our installed headers, against this branch.
name: HTTraQt
# Filtered, so it must never join the required set: a check that a path filter
# skips stays "Expected" forever and blocks the merge.
on:
push:
branches: [master]
paths: ['src/**', 'm4/**', 'configure.ac', '.github/workflows/httraqt.yml']
pull_request:
paths: ['src/**', 'm4/**', 'configure.ac', '.github/workflows/httraqt.yml']
schedule:
- cron: '37 05 * * 1' # upstream is pinned; this catches drift in Qt and the runner image
workflow_dispatch:
permissions:
contents: read
concurrency:
group: httraqt-${{ github.ref }}
cancel-in-progress: true
env:
# Unofficial mirror of the SourceForge tree, so no PR depends on a third-party host.
HTTRAQT_REPO: https://github.com/xroche/httraqt-ci-mirror.git
HTTRAQT_REF: a1175565c61ab9afd7bda2dc96ed1182c2d44ec3 # 1.4.11 + cmake patch, upstream HEAD since 2023-01-06
PREFIX: /usr/local
jobs:
build:
name: build HTTraQt against this branch
runs-on: ubuntu-24.04
steps:
- uses: actions/checkout@v7
with:
submodules: recursive
- name: Install build dependencies
run: |
set -euo pipefail
sudo apt-get update || true # a third-party source in the runner image can 403 (#1097)
sudo apt-get install -y --no-install-recommends \
build-essential autoconf automake libtool autoconf-archive \
zlib1g-dev libssl-dev libbrotli-dev libzstd-dev \
cmake qt6-base-dev qt6-multimedia-dev
- name: Build and install libhttrack
run: |
set -euo pipefail
autoreconf -fi
bld="$RUNNER_TEMP/bld"
mkdir -p "$bld"
(cd "$bld" && "$GITHUB_WORKSPACE/configure" --prefix="$PREFIX" --disable-static)
make -C "$bld" -j"$(nproc)"
sudo make -C "$bld" install
sudo ldconfig
- name: Assert no distribution libhttrack can shadow ours
run: |
set -euo pipefail
# FindHttrack.cmake hardcodes /usr/include/httrack first, so a stray libhttrack-dev would test Debian's headers.
for stray in /usr/include/httrack /usr/lib/libhttrack.so /usr/lib/*/libhttrack.so; do
if [ -e "$stray" ]; then
echo "::error::$stray exists; HTTraQt would build against it, not this branch"
exit 1
fi
done
test -f "$PREFIX/include/httrack/htsglobal.h"
- name: Fetch HTTraQt
run: |
set -euo pipefail
# Fetching the SHA itself needs no pin check afterwards, and keeps the
# tree out of the workspace so our own build cannot pick it up.
rm -rf "$RUNNER_TEMP/httraqt"
git init -q "$RUNNER_TEMP/httraqt"
ok=
for attempt in 1 2 3; do
if git -C "$RUNNER_TEMP/httraqt" fetch -q --depth 1 "$HTTRAQT_REPO" "$HTTRAQT_REF"; then ok=1; break; fi
echo "fetch failed (attempt $attempt)"
[ "$attempt" -eq 3 ] || sleep $((attempt * 15))
done
[ -n "$ok" ] || { echo "::error::cannot reach $HTTRAQT_REPO"; exit 1; }
git -C "$RUNNER_TEMP/httraqt" checkout -q --detach FETCH_HEAD
- name: Configure HTTraQt
run: |
set -euo pipefail
# CMAKE_MINIMUM_REQUIRED is 2.8 upstream, which CMake 4 rejects outright.
# The redirect is load-bearing: a bare MESSAGE() is NOTICE mode, which goes to stderr.
cmake -S "$RUNNER_TEMP/httraqt" -B "$RUNNER_TEMP/httraqt-bld" \
-DCMAKE_POLICY_VERSION_MINIMUM=3.5 \
2>&1 | tee "$RUNNER_TEMP/cmake.log"
# A configure that resolved elsewhere would compile someone else's headers.
grep -q "httrack header directory found: $PREFIX/include/httrack/" "$RUNNER_TEMP/cmake.log"
- name: Build HTTraQt
run: cmake --build "$RUNNER_TEMP/httraqt-bld" -j"$(nproc)"
- name: Assert it linked against the libhttrack we just built
run: |
set -euo pipefail
# Not bin/: CMAKE_RUNTIME_OUTPUT_DIRECTORY is set after ADD_EXECUTABLE, too late.
bin="$RUNNER_TEMP/httraqt-bld/httraqt"
ldd "$bin" | grep -E "libhttrack\.so.* => $PREFIX/lib/" \
|| { ldd "$bin"; exit 1; }

View File

@@ -31,6 +31,11 @@ jobs:
matrix:
platform: [x64, Win32]
configuration: [Release]
# Windows runs from other branches starve this one's runner (#1213), so cap
# them repo-wide at one per platform. queue: max: the default cancels a waiter.
concurrency:
group: windows-build-${{ matrix.platform }}
queue: max
# Redirect vcpkg's default `files` binary cache into the workspace so
# actions/cache can persist it. vcpkg builds openssl/brotli/zlib/zstd from
# source otherwise, several minutes every run.
@@ -67,17 +72,30 @@ jobs:
run: New-Item -ItemType Directory -Force -Path $env:VCPKG_DEFAULT_BINARY_CACHE | Out-Null
# x-gha is gone (vcpkg-tool #1662 dropped it after GitHub changed the cache
# API), so cache the binary archives directly. Keyed on the manifest, which
# carries the builtin-baseline, so a Dependabot bump busts it; restore-keys
# still seeds the unchanged ports' archives, so only the bumped one rebuilds.
- name: Cache vcpkg binary archives
uses: actions/cache@v6
# API), so cache the binary archives directly. An exact key hit skips the save,
# so a cache orphaned by a toolchain bump pins itself, and naming every input
# vcpkg's ABI hash folds in is a list that can come up short: key on the run
# id instead, and restore by prefix.
- name: Restore the vcpkg binary archives
uses: actions/cache/restore@v6
with:
path: ${{ github.workspace }}\vcpkg_cache
key: vcpkg-${{ matrix.platform }}-${{ hashFiles('src/vcpkg.json') }}
key: vcpkg-${{ matrix.platform }}-${{ hashFiles('src/vcpkg.json') }}-${{ github.run_id }}-${{ github.run_attempt }}
restore-keys: |
vcpkg-${{ matrix.platform }}-${{ hashFiles('src/vcpkg.json') }}-
vcpkg-${{ matrix.platform }}-
# Counted before the build, or vcpkg's own writes during it make a cold cache
# look warm.
- name: Count the restored vcpkg archives
shell: pwsh
run: |
$a = @(Get-ChildItem -Recurse -File -Filter *.zip `
$env:VCPKG_DEFAULT_BINARY_CACHE -EA SilentlyContinue)
Write-Host "restored $($a.Count) archive(s)"
$a.FullName | Out-File -FilePath vcpkg-restored.txt
"VCPKG_ARCHIVES_BEFORE=$($a.Count)" | Out-File -FilePath $env:GITHUB_ENV -Append
# The runner image's vcpkg checkout is pinned to some commit; our manifest's
# builtin-baseline is usually newer, so `git show <baseline>:versions/...`
# fails until that commit is local. Fetch exactly the pinned baseline (read
@@ -107,6 +125,50 @@ jobs:
if ($LASTEXITCODE -ne 0) { throw "$proj failed" }
}
# A stale cache and a warm one both restore and rebuild; only archives
# restored and unmatched prove staleness.
- name: Check whether vcpkg could use the restored archives
id: vcpkg
if: ${{ !cancelled() }}
shell: pwsh
run: |
$before = [int]$env:VCPKG_ARCHIVES_BEFORE
$stale = $false
# The log is missing when the build never reached MSBuild.
if (Test-Path msbuild-libhttrack.log) {
$m = Select-String -Path msbuild-libhttrack.log -Pattern 'Restored (\d+) package' |
Select-Object -First 1
if ($m) {
$restored = [int]$m.Matches[0].Groups[1].Value
Write-Host "vcpkg restored $restored of $before archive(s)"
$stale = ($restored -eq 0 -and $before -gt 0)
} else { Write-Host "no vcpkg restore line in the build log" }
} else { Write-Host "no build log" }
if ($stale) {
Write-Host "::warning::stale vcpkg cache: $before archives, none matching this toolchain; every port rebuilt from source"
# Provably dead: drop them rather than carry every past toolchain
# generation in the saved entry forever. -LiteralPath because a piped
# string binds -Path, which globs on a name carrying [ ].
Get-Content vcpkg-restored.txt -EA SilentlyContinue | Where-Object { $_ } |
ForEach-Object { Remove-Item -LiteralPath $_ -Force -EA SilentlyContinue }
}
$after = @(Get-ChildItem -Recurse -File -Filter *.zip `
$env:VCPKG_DEFAULT_BINARY_CACHE -EA SilentlyContinue).Count
Write-Host "$after archive(s) to cache"
# Pruning can leave the count unmoved while replacing every archive, so the
# stale case saves on its own account rather than on growth.
"save=$($stale -or ($after -gt $before))".ToLower() |
Out-File -FilePath $env:GITHUB_OUTPUT -Append
# Only when the archive set changed, so a warm run adds no entry and the
# eviction budget stays for the runs that rebuilt something.
- name: Save the vcpkg binary archives
if: ${{ !cancelled() && steps.vcpkg.outputs.save == 'true' }}
uses: actions/cache/save@v6
with:
path: ${{ github.workspace }}\vcpkg_cache
key: vcpkg-${{ matrix.platform }}-${{ hashFiles('src/vcpkg.json') }}-${{ github.run_id }}-${{ github.run_attempt }}
# WinHTTrack links src\$(Platform)\$(Configuration)\libhttrack.lib, so the
# import lib and the DLL both have to land there under that exact name.
- name: Check the binaries landed where they are expected
@@ -167,6 +229,33 @@ jobs:
throw "CRT mismatch across the boundary: libhttrack.dll=$a httrack.exe=$b"
}
# The DLL build only proves the engine's own fixed include order, the one
# order a consumer need not use, and 205/206/207/269 all gate on an automake
# environment this job has not (#1153).
- name: Compile the installed headers standalone and pairwise
shell: pwsh
timeout-minutes: 20
run: |
$ErrorActionPreference = "Stop"
$vswhere = "${env:ProgramFiles(x86)}\Microsoft Visual Studio\Installer\vswhere.exe"
$vs = & $vswhere -latest -products * -property installationPath | Select-Object -First 1
if (-not $vs) { throw "Visual Studio not found" }
Import-Module "$vs\Common7\Tools\Microsoft.VisualStudio.DevShell.dll"
$arch = if ("${{ matrix.platform }}" -eq "x64") { "x64" } else { "x86" }
Enter-VsDevShell -VsInstallPath $vs -SkipAutomaticLocation `
-DevCmdArguments "-arch=$arch -host_arch=x64"
# htsbasenet.h reaches for <openssl/ssl.h> unless a consumer turns it
# off, so the sweep needs what vcpkg installed for the build. Searched,
# not spelled: the install root nests the triplet twice.
$dirs = @(Get-ChildItem -Path "src\vcpkg_installed" -Recurse -Depth 3 `
-Directory -Filter include -ErrorAction SilentlyContinue)
if (-not ($dirs | Where-Object { Test-Path (Join-Path $_.FullName "openssl\ssl.h") })) {
throw "no vcpkg include directory under src\vcpkg_installed holds openssl/ssl.h"
}
$inc = @($dirs | ForEach-Object { "-I" + $_.FullName })
bash tests/install-headers-sweep.sh --backend cl --srcdir . -- @inc
if ($LASTEXITCODE -ne 0) { throw "the installed headers do not compile standalone and pairwise" }
# The engine ships ~90 tests and none had ever run on Windows: "make check"
# is Linux/macOS only. These are the offline ones, driven from Git Bash
# against the native httrack.exe. They subsume the self-tests this step

364
.github/workflows/zz-pidreuse-probe.yml vendored Normal file
View File

@@ -0,0 +1,364 @@
# Scratch probe: does taskkill /F /T follow a recorded parent PID whose owner is
# long dead? Not part of the build gate; branch-scoped, delete when answered.
name: zz pid-reuse probe
on:
push:
branches: [zz-pidreuse-probe]
permissions:
contents: read
jobs:
probe:
runs-on: windows-2022
timeout-minutes: 15
steps:
- name: Probe
shell: pwsh
run: |
$ErrorActionPreference = 'Continue'
$ProgressPreference = 'SilentlyContinue'
$sum = $env:GITHUB_STEP_SUMMARY
function Say([string]$m) {
Write-Host $m
if ($sum) { try { Add-Content -LiteralPath $sum -Value $m -ErrorAction SilentlyContinue } catch { } }
}
Say "# PID-reuse probe on $env:COMPUTERNAME $(Get-Date -Format o)"
Say ""
Say '```'
# ---------------------------------------------------------------
# Part 1: is the hazard even present?
# ---------------------------------------------------------------
Say "== PART 1: recorded-parent state of the runner's own processes =="
$snap = @(Get-CimInstance Win32_Process)
$byPid = @{}
foreach ($p in $snap) { $byPid[[int]$p.ProcessId] = $p }
Say ("snapshot: {0} live processes" -f $snap.Count)
function Stamp($d) { if ($null -eq $d) { return '?' } else { return ([datetime]$d).ToString('HH:mm:ss.fff') } }
function Classify([int]$q) {
$c = $byPid[$q]
if ($null -eq $c) { return 'NOPROC' }
$par = $byPid[[int]$c.ParentProcessId]
if ($null -eq $par) { return 'DEAD-PARENT' }
if ($null -eq $par.CreationDate -or $null -eq $c.CreationDate) { return 'OK' }
if ($par.CreationDate -gt $c.CreationDate) { return 'RECYCLED-PARENT' }
return 'OK'
}
function Chain([int]$q) {
$out = @()
$seen = @{}
$cur = $q
for ($i = 0; $i -lt 24; $i++) {
$c = $byPid[$cur]
if ($null -eq $c) { $out += ("{0}=<no such process>" -f $cur); break }
$out += ("{0}/{1}@{2}" -f $c.ProcessId, $c.Name, (Stamp $c.CreationDate))
if ($seen.ContainsKey($cur)) { $out += '<cycle>'; break }
$seen[$cur] = $true
$st = Classify $cur
if ($st -ne 'OK') { $out += ("<<{0}>>" -f $st); break }
$cur = [int]$c.ParentProcessId
if ($cur -eq 0) { break }
}
return ($out -join ' <- ')
}
$interesting = @($snap | Where-Object {
$_.Name -like 'Runner.*' -or $_.Name -like 'Agent.*' -or $_.Name -like 'hosted-*' -or
$_.Name -eq 'bash.exe' -or $_.Name -eq 'pwsh.exe' -or
$_.Name -eq 'powershell.exe' -or $_.Name -eq 'dotnet.exe' -or
$_.Name -eq 'node.exe' -or $_.Name -eq 'cmd.exe' -or
$_.Name -eq 'conhost.exe' -or $_.Name -eq 'sh.exe'
} | Sort-Object Name, ProcessId)
$armed = 0
foreach ($p in $interesting) {
$st = Classify ([int]$p.ProcessId)
Say (" {0,-24} pid={1,-7} ppid={2,-7} {3,-16} {4}" -f $p.Name, $p.ProcessId, $p.ParentProcessId, $st, (Chain ([int]$p.ProcessId)))
if ($st -ne 'OK' -and ($p.Name -like 'Runner.*' -or $p.Name -like 'Agent.*' -or $p.Name -like 'hosted-*')) { $armed++ }
}
$dead = 0; $rec = 0; $ok = 0
$boxOrphan = @{}
foreach ($p in $snap) {
$st = Classify ([int]$p.ProcessId)
if ($st -eq 'DEAD-PARENT') {
$dead++
$k = [int]$p.ParentProcessId
if (-not $boxOrphan.ContainsKey($k)) { $boxOrphan[$k] = @() }
$boxOrphan[$k] += ("{0}/{1}" -f $p.ProcessId, $p.Name)
} elseif ($st -eq 'RECYCLED-PARENT') { $rec++ } else { $ok++ }
}
Say ""
Say ("box-wide: {0} processes, {1} with a DEAD recorded parent, {2} with a RECYCLED one, {3} sound" -f $snap.Count, $dead, $rec, $ok)
foreach ($k in ($boxOrphan.Keys | Sort-Object)) {
Say (" dead ppid {0,-7} still claimed as parent by: {1}" -f $k, ($boxOrphan[$k] -join ', '))
}
Say ""
if ($armed -gt 0) {
Say "PART 1 VERDICT: HAZARD PRESENT - $armed runner-agent process(es) record a parent PID that is dead or already recycled."
} elseif ($dead -gt 0) {
Say "PART 1 VERDICT: no Runner.*/Agent.* process has a stale recorded parent, but $dead other live processes do."
} else {
Say "PART 1 VERDICT: NOT ARMED - every live process's recorded parent is alive and older."
}
Say ""
# ---------------------------------------------------------------
# Part 2. Five sets of freed PIDs, released at the same instant, so
# that a null on the set the theory needs can be told apart from a
# probe that never recycles anything:
# A plain ping, killed, .NET handle disposed (baseline: does reuse happen?)
# C plain ping, killed, handle deliberately HELD (does a handle pin the PID?)
# D cmd orphan-maker, child killed (is the construction sound?)
# B cmd orphan-maker, ping child left ALIVE (the question)
# E cmd orphan-maker, wscript child left ALIVE (same, with a child that
# attaches to no console)
# Every process inherits this shell's console, so no console host survives
# to hold a handle on the dead cmd. Run 2 created 500 of those.
# ---------------------------------------------------------------
Say "== PART 2: can a dead parent's PID be handed out again, and does taskkill /T follow it? =="
$vbs = Join-Path $env:TEMP 'zzsleep.vbs'
Set-Content -LiteralPath $vbs -Value 'WScript.Sleep 600000' -Encoding ascii
function New-Ping {
$psi = New-Object System.Diagnostics.ProcessStartInfo
$psi.FileName = "$env:SystemRoot\System32\PING.EXE"
$psi.Arguments = '-n 600 127.0.0.1'
$psi.UseShellExecute = $false
$psi.CreateNoWindow = $false
$psi.RedirectStandardOutput = $true
$psi.RedirectStandardError = $true
return [System.Diagnostics.Process]::Start($psi)
}
function New-Orphan([string]$childArgs) {
# cmd exits at once, leaving the child alive with cmd's now-free PID recorded as parent.
$psi = New-Object System.Diagnostics.ProcessStartInfo
$psi.FileName = "$env:ComSpec"
$psi.Arguments = $childArgs
$psi.UseShellExecute = $false
$psi.CreateNoWindow = $false
return [System.Diagnostics.Process]::Start($psi)
}
function Get-Proc([int]$q) {
return (Get-CimInstance Win32_Process -Filter "ProcessId = $q" -ErrorAction SilentlyContinue | Select-Object -First 1)
}
# --- control: a tree kill of a LIVE parent must kill its child, or the probe is blind.
Say ""
Say "-- control: taskkill /F /T on a live cmd whose child ping is genuine"
$ctl = New-Orphan '/c ping -n 600 127.0.0.1 > NUL'
Start-Sleep -Milliseconds 1500
$ctlKid = @(Get-CimInstance Win32_Process -Filter "ParentProcessId = $($ctl.Id)")
if ($ctlKid.Count -eq 0) {
Say " control INCONCLUSIVE: no child found under cmd $($ctl.Id)"
try { $ctl.Kill() } catch { }
} else {
$kp = [int]$ctlKid[0].ProcessId
Say (" cmd {0} -> child {1}/{2}" -f $ctl.Id, $kp, $ctlKid[0].Name)
$co = (& taskkill.exe /F /T /PID $ctl.Id 2>&1 | Out-String).Trim()
Say (" taskkill: " + ($co -replace "`r?`n", ' | '))
Start-Sleep -Milliseconds 800
$still = Get-Proc $kp
Say (" control result: child {0} {1}" -f $kp, $(if ($null -eq $still) { 'GONE - the probe can observe a tree kill' } else { 'STILL ALIVE - probe is blind, part 2 is inconclusive' }))
}
try { $ctl.Dispose() } catch { }
$conBefore = @(Get-Process conhost -ErrorAction SilentlyContinue).Count
# --- build the sets.
$NB = 300; $ND = 200; $NE = 200; $NA = 200; $NC = 200
Say ""
Say "-- building sets: B=$NB (ping child alive) E=$NE (wscript child alive) D=$ND (child killed) A=$NA (baseline) C=$NC (handle held)"
$bCmd = New-Object 'System.Collections.Generic.List[System.Diagnostics.Process]'
for ($i = 0; $i -lt $NB; $i++) { try { $bCmd.Add((New-Orphan '/c start /b ping -n 600 127.0.0.1 > NUL')) } catch { } }
$eCmd = New-Object 'System.Collections.Generic.List[System.Diagnostics.Process]'
for ($i = 0; $i -lt $NE; $i++) { try { $eCmd.Add((New-Orphan ('/c start /b wscript.exe //B //Nologo "' + $vbs + '"'))) } catch { } }
$dCmd = New-Object 'System.Collections.Generic.List[System.Diagnostics.Process]'
for ($i = 0; $i -lt $ND; $i++) { try { $dCmd.Add((New-Orphan '/c start /b ping -n 600 127.0.0.1 > NUL')) } catch { } }
$aProc = New-Object 'System.Collections.Generic.List[System.Diagnostics.Process]'
for ($i = 0; $i -lt $NA; $i++) { try { $aProc.Add((New-Ping)) } catch { } }
$pinned = New-Object 'System.Collections.Generic.List[System.Diagnostics.Process]'
for ($i = 0; $i -lt $NC; $i++) { try { $pinned.Add((New-Ping)) } catch { } }
$bPid = New-Object 'System.Collections.Generic.HashSet[int]'
foreach ($p in $bCmd) { [void]$bPid.Add([int]$p.Id) }
$ePid = New-Object 'System.Collections.Generic.HashSet[int]'
foreach ($p in $eCmd) { [void]$ePid.Add([int]$p.Id) }
$dPid = New-Object 'System.Collections.Generic.HashSet[int]'
foreach ($p in $dCmd) { [void]$dPid.Add([int]$p.Id) }
$aPid = New-Object 'System.Collections.Generic.HashSet[int]'
foreach ($p in $aProc) { [void]$aPid.Add([int]$p.Id) }
$cPid = New-Object 'System.Collections.Generic.HashSet[int]'
foreach ($p in $pinned) { [void]$cPid.Add([int]$p.Id) }
Say (" distinct PIDs held: B={0} E={1} D={2} A={3} C={4}" -f $bPid.Count, $ePid.Count, $dPid.Count, $aPid.Count, $cPid.Count)
if ($bPid.Count -eq 0 -or $aPid.Count -eq 0) { Say " set construction failed, aborting"; Say '```'; exit 1 }
Start-Sleep -Seconds 3
# Map the surviving children back to the cmd that made them.
$orphan = @{}
$dKids = New-Object 'System.Collections.Generic.List[int]'
$nB = 0; $nE = 0
foreach ($p in @(Get-CimInstance Win32_Process | Where-Object { $_.Name -eq 'ping.exe' -or $_.Name -eq 'wscript.exe' })) {
$pp = [int]$p.ParentProcessId
if ($bPid.Contains($pp) -or $ePid.Contains($pp)) {
if (-not $orphan.ContainsKey($pp)) { $orphan[$pp] = @() }
$orphan[$pp] += [pscustomobject]@{ Pid = [int]$p.ProcessId; Created = $p.CreationDate }
if ($bPid.Contains($pp)) { $nB++ } else { $nE++ }
} elseif ($dPid.Contains($pp)) {
$dKids.Add([int]$p.ProcessId)
}
}
Say (" surviving orphan children: {0} under B parents, {1} under E parents, {2} under D parents (to be killed)" -f $nB, $nE, $dKids.Count)
# Free every set at the same instant. Only C keeps its .NET handle open.
foreach ($p in $bCmd) { try { [void]$p.WaitForExit(2000) } catch { }; try { $p.Dispose() } catch { } }
foreach ($p in $eCmd) { try { [void]$p.WaitForExit(2000) } catch { }; try { $p.Dispose() } catch { } }
foreach ($p in $dCmd) { try { [void]$p.WaitForExit(2000) } catch { }; try { $p.Dispose() } catch { } }
foreach ($p in $aProc) { try { $p.Kill() } catch { } }
foreach ($p in $pinned) { try { $p.Kill() } catch { } }
Start-Sleep -Milliseconds 1000
foreach ($p in $aProc) { try { $p.Dispose() } catch { } }
$aProc.Clear(); $bCmd.Clear(); $eCmd.Clear(); $dCmd.Clear()
if ($dKids.Count -gt 0) { Stop-Process -Id $dKids.ToArray() -Force -ErrorAction SilentlyContinue }
[GC]::Collect(); [GC]::WaitForPendingFinalizers(); [GC]::Collect()
Start-Sleep -Seconds 2
$liveNow = New-Object 'System.Collections.Generic.HashSet[int]'
foreach ($p in @(Get-CimInstance Win32_Process)) { [void]$liveNow.Add([int]$p.ProcessId) }
function CountLive($s) { $n = 0; foreach ($x in $s) { if ($liveNow.Contains($x)) { $n++ } }; return $n }
Say (" set members still live (must be 0): B={0} E={1} D={2} A={3} C={4}" -f (CountLive $bPid), (CountLive $ePid), (CountLive $dPid), (CountLive $aPid), (CountLive $cPid))
$conAfter = @(Get-Process conhost -ErrorAction SilentlyContinue).Count
Say (" conhost.exe: {0} before, {1} after (a surviving console host would hold a handle on its dead creator and pin its PID)" -f $conBefore, $conAfter)
$kidsAlive = 0
foreach ($k in $orphan.Keys) { foreach ($o in $orphan[$k]) { if ($liveNow.Contains($o.Pid)) { $kidsAlive++ } } }
Say (" orphan children still alive under a dead B/E parent: {0}" -f $kidsAlive)
# --- hunt.
$hitA = New-Object 'System.Collections.Generic.HashSet[int]'
$hitB = New-Object 'System.Collections.Generic.HashSet[int]'
$hitC = New-Object 'System.Collections.Generic.HashSet[int]'
$hitD = New-Object 'System.Collections.Generic.HashSet[int]'
$hitE = New-Object 'System.Collections.Generic.HashSet[int]'
$boxHits = 0
$spawns = 0; $tested = 0; $followed = 0; $survived = 0
$KILLCAP = 25
$distinct = New-Object 'System.Collections.Generic.HashSet[int]'
$window = New-Object 'System.Collections.Generic.Queue[System.Diagnostics.Process]'
$minPid = [int]::MaxValue; $maxPid = 0
function Test-Hit([int]$np) {
if ($script:aPid.Contains($np)) { [void]$script:hitA.Add($np) }
if ($script:dPid.Contains($np)) { [void]$script:hitD.Add($np) }
if ($script:cPid.Contains($np)) {
if ($script:hitC.Add($np) -and $script:hitC.Count -le 5) { Say ("!! set C hit: PID $np was handed out although an open process handle on it is still held") }
}
if ($script:boxOrphan.ContainsKey($np)) {
$script:boxHits++
Say ""
Say ("!! REAL COLLISION: our spawn got PID $np, which live box process(es) still record as their parent: $($script:boxOrphan[$np] -join ', ')")
Say " (not tree-killing this one: it would reap a process we did not start)"
Say ""
}
$isB = $script:bPid.Contains($np)
$isE = $script:ePid.Contains($np)
if (-not ($isB -or $isE)) { return $false }
if ($isB) { [void]$script:hitB.Add($np) } else { [void]$script:hitE.Add($np) }
if ($script:tested -ge $script:KILLCAP) { return $false }
$live = @()
foreach ($o in $script:orphan[$np]) {
$cur = Get-Proc $o.Pid
if ($null -ne $cur -and $cur.CreationDate -eq $o.Created) { $live += $o }
}
if ($live.Count -eq 0) { return $false }
$script:tested++
Say ""
Say ("### HIT #$($script:tested) on set $(if ($isB) { 'B' } else { 'E' }): spawn #$($script:spawns) at t=$([int]$script:sw.Elapsed.TotalSeconds)s was handed PID $np, the PID of a dead parent whose child is still running")
foreach ($k in @(Get-CimInstance Win32_Process -Filter "ParentProcessId = $np")) {
Say (" the process table calls this a child of $np : {0}/{1} created {2}" -f $k.ProcessId, $k.Name, $k.CreationDate)
}
foreach ($o in $live) { Say (" stale-PPID orphan {0} is alive before the kill" -f $o.Pid) }
Say (" issuing: taskkill /F /T /PID $np")
$out = (& taskkill.exe /F /T /PID $np 2>&1 | Out-String).Trim()
Say (" taskkill said: " + ($out -replace "`r?`n", ' | '))
Start-Sleep -Milliseconds 1000
$anyDied = $false
foreach ($o in $live) {
$cur = Get-Proc $o.Pid
$alive = ($null -ne $cur -and $cur.CreationDate -eq $o.Created)
if (-not $alive) { $anyDied = $true }
Say (" stale-PPID orphan {0}: alive after the kill = {1}" -f $o.Pid, $alive)
}
if (-not $anyDied) { $script:survived++ }
if ($anyDied) {
$script:followed++
Say " >>> THE TREE WALK FOLLOWED A STALE PPID. taskkill /T does not validate that the recorded parent is the real one."
} else {
Say " >>> the orphan survived: taskkill /T did not follow the stale PPID."
}
Say ""
return $false
}
$sw = [Diagnostics.Stopwatch]::StartNew()
Say ""
Say "-- hunting: rolling window of 200 live spawns, 300s cap"
$done = $false
while (-not $done -and $sw.Elapsed.TotalSeconds -lt 300) {
try { $np = (New-Ping) } catch { Start-Sleep -Milliseconds 50; continue }
$window.Enqueue($np)
$spawns++
$id = [int]$np.Id
[void]$distinct.Add($id)
if ($id -lt $minPid) { $minPid = $id }
if ($id -gt $maxPid) { $maxPid = $id }
try { $done = Test-Hit $id } catch { Say ("Test-Hit failed on $id : " + $_.Exception.Message) }
while ($window.Count -gt 200) {
$old = $window.Dequeue()
try { $old.Kill() } catch { }
try { $old.Dispose() } catch { }
}
if (($spawns % 250) -eq 0) { Write-Host (" ... spawn $spawns, distinct $($distinct.Count), A=$($hitA.Count) B=$($hitB.Count) C=$($hitC.Count) D=$($hitD.Count) E=$($hitE.Count), t=$([int]$sw.Elapsed.TotalSeconds)s") }
}
Say ""
Say "== RESULT =="
Say ("spawns {0}, distinct PIDs {1}, range {2}..{3}, {4:N0}s" -f $spawns, $distinct.Count, $minPid, $maxPid, $sw.Elapsed.TotalSeconds)
Say ("distinct set PIDs reissued, by set:")
Say (" A killed, handle disposed, no child : {0} of {1}" -f $hitA.Count, $aPid.Count)
Say (" C killed, handle still OPEN : {0} of {1}" -f $hitC.Count, $cPid.Count)
Say (" D dead orphan-maker, child killed : {0} of {1}" -f $hitD.Count, $dPid.Count)
Say (" B dead orphan-maker, ping child ALIVE : {0} of {1}" -f $hitB.Count, $bPid.Count)
Say (" E dead orphan-maker, wscript ALIVE : {0} of {1}" -f $hitE.Count, $ePid.Count)
Say ("tree kills issued on a reissued stale-parent PID: {0}; reached the orphan: {1}; orphan survived: {2}" -f $tested, $followed, $survived)
Say ("collisions with a dead PPID recorded by a process we did NOT start: {0}" -f $boxHits)
if ($tested -gt 0 -and $followed -eq 0) {
Say "INTERPRETATION: a dead parent's PID IS reissued while a live process still records it as its parent, in $tested measured cases, and taskkill /F /T left that process alone every time. The tree walk does not follow a stale PPID: it skips a claimed child that predates the process now holding the PID."
} elseif ($hitA.Count -eq 0 -and $hitB.Count -eq 0 -and $hitE.Count -eq 0) {
Say "INTERPRETATION: nothing came back, so this run recycled nothing and says nothing either way."
} elseif ($tested -eq 0) {
Say "INTERPRETATION: PIDs recycle (A>0) but no PID with a live process recording it as parent ever came back. On this kernel such a PID stays reserved, so a tree kill cannot land on one and the stale-PPID mechanism cannot fire."
} elseif ($followed -gt 0) {
Say "INTERPRETATION: a dead parent's PID IS reissued while a live process records it, and taskkill /F /T reaped that process. The mechanism is real on this image."
} else {
Say "INTERPRETATION: the PID was reissued, but taskkill /F /T left the stale-PPID orphan alone: the tree walk validates parentage."
}
Say '```'
while ($window.Count -gt 0) { $o = $window.Dequeue(); try { $o.Kill() } catch { } }
try { Get-Process ping, wscript -ErrorAction SilentlyContinue | Stop-Process -Force -ErrorAction SilentlyContinue } catch { }
Write-Host "probe done"

View File

@@ -1,6 +1,6 @@
AC_PREREQ([2.71])
AC_INIT([httrack], [3.49.20], [roche+packaging@httrack.com], [httrack], [http://www.httrack.com/])
AC_INIT([httrack], [3.49.21], [roche+packaging@httrack.com], [httrack], [http://www.httrack.com/])
AC_COPYRIGHT([
HTTrack Website Copier, Offline Browser for Windows and Unix
Copyright (C) 1998-2015 Xavier Roche and other contributors
@@ -29,6 +29,8 @@ AC_CONFIG_SRCDIR(src/httrack.c)
AC_CONFIG_MACRO_DIR([m4])
AC_CONFIG_HEADERS(config.h)
AM_INIT_AUTOMAKE([subdir-objects])
# 3:14:0: revision-only bump. httrackp gained a tail field (host_alias) and #1202
# added an export (hts_host_alias_rule_ok); no layout moved, nothing went away.
# 3:13:0: revision-only bump. htsstrings.h's StringRoomTotal grew saturation in
# its macro body; no installed struct moved and no export came or went.
# 3:12:0: revision-only bump. httrackp gained a tail field (singlefile_state) and
@@ -44,7 +46,7 @@ AM_INIT_AUTOMAKE([subdir-objects])
# moves nothing). Soname stays .so.3: HTTrackQt is the only consumer of the installed
# headers, so a libhttrack4 rename isn't worth it.
# (3:0:0 was the htsblk mime-buffer widening, the ABI break that moved .so.2 -> .so.3.)
VERSION_INFO="3:13:0"
VERSION_INFO="3:14:0"
AM_MAINTAINER_MODE
AC_USE_SYSTEM_EXTENSIONS

9
debian/changelog vendored
View File

@@ -1,3 +1,12 @@
httrack (3.49.21-1) unstable; urgency=medium
* New upstream release: a site answering to several hostnames can now be
mirrored once rather than once per name, plus fixes to mirror
interruption, to an update pass that purged files it had never reached,
and to several buffer overflows; full list in history.txt.
-- Xavier Roche <xavier@debian.org> Wed, 12 Aug 2026 15:05:15 +0200
httrack (3.49.20-1) unstable; urgency=medium
* New upstream release: an off-site link that looked like a file could pull

View File

@@ -4,6 +4,24 @@ HTTrack Website Copier release history:
This file lists all changes and fixes that have been made for HTTrack
3.49-21
+ New: --host-alias (-%C) folds the several hostnames of one site onto a single canonical host, so a site that answers under more than one hostname is fetched once instead of once per name (#16)
+ New: the WebHTTrack wizard can set the host-alias rules and the guide documents them for every front end; a malformed rule is refused where it can still be corrected, through a newly exported hts_host_alias_rule_ok() (#1161, #1172, #1192)
+ Fixed: any signal caught during a mirror killed every transfer in flight, so a burst of them silently truncated the mirror; a stop now ends an HTTP transfer as the user's stop rather than a fabricated receive error (#1110)
+ Fixed: stopping an FTP transfer against a silent server waited out --timeout twice, once on the data socket and again on a QUIT reply nobody was going to send (#1096)
+ Fixed: a crawl that ended early ran --update's purge against the files it had never reached, deleting good ones (#1109)
+ Fixed: a long URL overflowed the duplicate-detection key buffer (#1160)
+ Fixed: a long command line, a large doit.log or a large .httrackrc aborted the engine while it rebuilt its arguments (#1189, #1198)
+ Fixed: a scan rule beginning with a dash ended the whole run (#1179)
+ Fixed: --index=0 and --noindex enabled the index rather than disabling it, and --purge-old=1 killed the run; an option that takes no value now refuses one instead of silently dropping it (#1195, #1200)
+ Fixed: reloading a WebHTTrack project lost every zero and every cleared field, so a rate limit of 0 came back as 100 KB/s and a cleared user-agent refilled itself (#1177, #1171, #1186)
+ Fixed: the wizard kept only the last of several --strip-query rules, and its domain answers matched beyond the domain they named (#1168, #1119)
+ Fixed: the offer to resume an interrupted mirror printed the program path instead of the command line it would rerun (#1204)
+ Fixed: the build failed where getrandom() has no declaration, and the installed headers did not compile under some include orders (#1139, #1150, #1151)
+ Changed: the documentation and the WebHTTrack pages compute their copyright year instead of carrying a fixed one (#1163, #1165)
+ Changed: refreshed the guide screenshots for all three interfaces, and repaired the documentation's dark mode and its search-target tabs (#1164, #1158)
+ Changed: multiple internal hardening, build, test and CI improvements
3.49-20
+ Fixed: a link to another host whose name looked like a file was scanned as a page, so one link could mirror the whole foreign site on the referring page's depth budget (#121)
+ Fixed: a size-bounded filter built both of its filter URLs wrong when the path had no leading slash (#1095)

View File

@@ -13,12 +13,14 @@
<body>
<!-- doc-chrome:top -->
<!-- doc-chrome:masthead -->
<a class="skip" href="#main">Skip to content</a>
<header class="masthead">
<img src="images/wordmark.svg" width="400" height="36" alt="HTTrack Website Copier">
<div class="tagline">Free software offline browser</div>
</header>
<!-- /doc-chrome:masthead -->
<div class="wrap">
@@ -534,9 +536,14 @@ And then, put the email address in your pages through:
</div>
<dialog id="zoom" aria-label="Enlarged image"><img src="" alt=""></dialog>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- /doc-chrome:bottom -->
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -13,12 +13,14 @@
<body>
<!-- doc-chrome:top -->
<!-- doc-chrome:masthead -->
<a class="skip" href="#main">Skip to content</a>
<header class="masthead">
<img src="images/wordmark.svg" width="400" height="36" alt="HTTrack Website Copier">
<div class="tagline">Free software offline browser</div>
</header>
<!-- /doc-chrome:masthead -->
<div class="wrap">
@@ -75,8 +77,8 @@ The main advantages of this cache are:
<ul>
<li>One single file for a complete website cache archive</li>
<li>Standard <a href="http://www.pkware.com/products/enterprise/white_papers/appnote.txt" target="_new">ZIP</a> format, that can be easily reused on most platforms and languages</li>
<li>Compressed data with the efficient and opened <a href="http://www.gzip.org/zlib/" target="_new">zlib</a> format</li>
<li>Standard <a href="http://www.pkware.com/products/enterprise/white_papers/appnote.txt" target="_blank">ZIP</a> format, that can be easily reused on most platforms and languages</li>
<li>Compressed data with the efficient and opened <a href="http://www.gzip.org/zlib/" target="_blank">zlib</a> format</li>
</ul>
The cache is made of ZIP files entries ; with one ZIP file entry per fetched URL (successfully or not - errors are also stored).<br />
@@ -84,7 +86,7 @@ For each entry:
<ul>
<li>The ZIP file name is the original URL [<small><a href="#orig">see notes below</a></small>]</li>
<li>The ZIP file contents, <b>if available</b>, is the original (compressed, using the deflate algorythm) data</li>
<li>The ZIP file extra field (in the local file header) contains a list of meta-fields, very similar to the <a href="http://www.ietf.org/rfc/rfc2616.txt?number=2616" target="new_">HTTP</a> headers fields. See also <a href="http://www.ietf.org/rfc/rfc2396.txt?number=2396" target="new_">RFC</a>.</li><br />
<li>The ZIP file extra field (in the local file header) contains a list of meta-fields, very similar to the <a href="http://www.ietf.org/rfc/rfc2616.txt?number=2616" target="_blank">HTTP</a> headers fields. See also <a href="http://www.ietf.org/rfc/rfc2396.txt?number=2396" target="_blank">RFC</a>.</li><br />
<li>The ZIP file timestamp follows the "Last-Modified-Since" field given for this URL, if any</li>
</ul>
@@ -230,9 +232,14 @@ Libraries should generally handle this peculiar format, however.
</div>
<dialog id="zoom" aria-label="Enlarged image"><img src="" alt=""></dialog>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- /doc-chrome:bottom -->
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -13,12 +13,14 @@
<body>
<!-- doc-chrome:top -->
<!-- doc-chrome:masthead -->
<a class="skip" href="#main">Skip to content</a>
<header class="masthead">
<img src="images/wordmark.svg" width="400" height="36" alt="HTTrack Website Copier">
<div class="tagline">Free software offline browser</div>
</header>
<!-- /doc-chrome:masthead -->
<div class="wrap">
@@ -211,9 +213,14 @@ scope: both change what a mirror directory contains.</li>
</div>
<dialog id="zoom" aria-label="Enlarged image"><img src="" alt=""></dialog>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- /doc-chrome:bottom -->
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -13,12 +13,14 @@
<body>
<!-- doc-chrome:top -->
<!-- doc-chrome:masthead -->
<a class="skip" href="#main">Skip to content</a>
<header class="masthead">
<img src="images/wordmark.svg" width="400" height="36" alt="HTTrack Website Copier">
<div class="tagline">Free software offline browser</div>
</header>
<!-- /doc-chrome:masthead -->
<div class="wrap">
@@ -117,6 +119,7 @@ the index" problems disappear.</p>
<tr class="head"><td><b>Option</b></td><td><b>What it controls</b></td></tr>
<tr><td><tt>--depth (-r)</tt></td><td>Maximum link depth. <b>The start page is level 1</b>, so one level of links out is <tt>-r2</tt>, not <tt>-r1</tt>.</td></tr>
<tr><td><tt>--stay-on-same-address (-a), --stay-on-same-domain (-d), --stay-on-same-tld (-l), --go-everywhere (-e)</tt></td><td>How far off the starting host the crawl may travel: same address (host), same principal domain, same top-level domain (for example .com), or everywhere. The default keeps you on the starting host.</td></tr>
<tr><td><tt>--host-alias (-%C)</tt></td><td>Declares that several hostnames serve the same site, for example <tt>--host-alias "www2.example.com,m.example.com=example.com"</tt>. Those names then count as the starting host: their links travel under the same-address rules, are fetched from the canonical host, and are saved once rather than once per name. The left side takes filter-style globs and comma-separated names, the right side one literal host, port included. Either side may name a scheme (<tt>legacy.example.com=https://example.com</tt> reaches a host that only answers over HTTPS, and <tt>https://www.example.com=ftp://files.example.com</tt> matches only the HTTPS form). A path is refused: this maps hosts, not URLs.</td></tr>
<tr><td><tt>--can-go-down (-D), --can-go-up (-U), --stay-on-same-dir (-S), --can-go-up-and-down (-B)</tt></td><td>Directory travel: down into subdirectories only, up to parent directories only, stay in the same directory, or both up and down.</td></tr>
<tr><td><tt>--near (-n)</tt></td><td>Also fetch non-HTML files "near" a followed link, such as an image linked from a page you kept but hosted elsewhere.</td></tr>
<tr><td><tt>--ext-depth (-%e)</tt></td><td>How many levels of external links to follow once the crawl leaves your scope (default 0).</td></tr>
@@ -146,7 +149,9 @@ usually not a scope option at all: it is an off-host redirect. A start URL of
<tt>http://example.com/</tt> that redirects to <tt>https://www.example.com/</tt>
lands you on a different host, and same-host scope stops the crawl there. Start
from the final URL, or add a filter that re-admits the real host (see
<a href="#filters">Filters</a>). The log will show the redirect.</p>
<a href="#filters">Filters</a>). The log will show the redirect. When both names
really are one site, <tt>--host-alias</tt> says so once and the mirror stops
depending on which one you started from.</p>
<p><tt>-n</tt> is the fix for pages that render locally without their images or
stylesheets: it lets HTTrack pull in requisites that sit just outside scope. Note
@@ -503,9 +508,14 @@ all disabled.</small></p>
</div>
<dialog id="zoom" aria-label="Enlarged image"><img src="" alt=""></dialog>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- /doc-chrome:bottom -->
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -23,12 +23,14 @@
<body>
<!-- doc-chrome:top -->
<!-- doc-chrome:masthead -->
<a class="skip" href="#main">Skip to content</a>
<header class="masthead">
<img src="images/wordmark.svg" width="400" height="36" alt="HTTrack Website Copier">
<div class="tagline">Free software offline browser</div>
</header>
<!-- /doc-chrome:masthead -->
<div class="wrap">
@@ -206,9 +208,14 @@ roche at httrack dot com (Xavier ROCHE)<br>
</div>
<dialog id="zoom" aria-label="Enlarged image"><img src="" alt=""></dialog>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- /doc-chrome:bottom -->
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -13,12 +13,14 @@
<body>
<!-- doc-chrome:top -->
<!-- doc-chrome:masthead -->
<a class="skip" href="#main">Skip to content</a>
<header class="masthead">
<img src="images/wordmark.svg" width="400" height="36" alt="HTTrack Website Copier">
<div class="tagline">Free software offline browser</div>
</header>
<!-- /doc-chrome:masthead -->
<div class="wrap">
@@ -103,9 +105,14 @@ compared to the previous mirror. This page describes that file.
</div>
<dialog id="zoom" aria-label="Enlarged image"><img src="" alt=""></dialog>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- /doc-chrome:bottom -->
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -121,11 +121,17 @@ body {
}
/* Colour inherited: nothing lighter than --ink reaches 4.5:1 on the --field
this sits on, and at .8em it is small text. */
this sits on, and at this size it is small text. */
/* Mirrors site.css's .sitefooter: flush on the panel's black edge, chicklet at the far end. */
footer {
max-width: 68rem;
margin: .5rem auto 0;
margin: 0 auto;
font-size: .8em;
line-height: 1.3;
display: flex;
justify-content: space-between;
align-items: center;
gap: 1rem;
}
@media (prefers-color-scheme: dark) {
@@ -360,8 +366,26 @@ dialog#zoom::backdrop { background: rgba(0, 0, 0, .8); }
z-index: 10;
}
/* Some browsers still apply prefers-color-scheme: dark when printing, so the ink
comes out #e6e6ee on white paper. Reset the palette, not each rule reading it. */
@media print {
:root {
--field: #77b;
--panel: #ccd;
--accent: #99c;
--ink: #111;
--ink-soft: #444;
--rule: #aab;
--link: #00c;
--link-hover: #c00;
--code-bg: #e8e8f2;
--chip: #dde;
--shadow: rgba(0, 0, 0, .25);
}
body { background: #fff; }
.toc, .filter { display: none; }
.wrap { display: block; border: 0; }
/* Not a token, same paper: the dark scheme inverts the wordmark to white. */
.masthead img { filter: none; }
}

View File

@@ -13,12 +13,14 @@
<body>
<!-- doc-chrome:top -->
<!-- doc-chrome:masthead -->
<a class="skip" href="#main">Skip to content</a>
<header class="masthead">
<img src="images/wordmark.svg" width="400" height="36" alt="HTTrack Website Copier">
<div class="tagline">Free software offline browser</div>
</header>
<!-- /doc-chrome:masthead -->
<div class="wrap">
@@ -318,7 +320,7 @@ However, be sure to always download HTTrack from a <b>trusted source</b> (prefer
<br><br><a NAME="QG0c">Q: <strong>This software is 'free', but I bought it from an authorized reseller . What's going on?</strong></a><br>
A: <em>
HTTrack is free (free as in 'freedom') as it is covered by the <a href="http://www.gnu.org/licenses/gpl.txt" target="_new">GNU General Public License (GPL)</a>.
HTTrack is free (free as in 'freedom') as it is covered by the <a href="http://www.gnu.org/licenses/gpl.txt" target="_blank">GNU General Public License (GPL)</a>.
You can freely download it, without paying any fees, copy it to your friends, and modify it if you respect the license.
There are NO official/authorized resellers, because HTTrack is <b>NOT</b> a commercial product.
But you can be charged for duplication fees, or any other services (example: software CDroms or shareware collections, or fees for maintenance),
@@ -881,9 +883,14 @@ A: <em>Feel free to <a href="contact.html">contact us</a>!
</div>
<dialog id="zoom" aria-label="Enlarged image"><img src="" alt=""></dialog>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- /doc-chrome:bottom -->
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -13,12 +13,14 @@
<body>
<!-- doc-chrome:top -->
<!-- doc-chrome:masthead -->
<a class="skip" href="#main">Skip to content</a>
<header class="masthead">
<img src="images/wordmark.svg" width="400" height="36" alt="HTTrack Website Copier">
<div class="tagline">Free software offline browser</div>
</header>
<!-- /doc-chrome:masthead -->
<div class="wrap">
@@ -174,7 +176,7 @@ Action options:
Proxy options:
P proxy use (-P proxy:port or -P user:pass@proxy:port) (--proxy &lt;param&gt;)
%f *use proxy for ftp (f0 don't use) (--httpproxy-ftp[=N])
%f *use proxy for ftp (f0 don't use) (--httpproxy-ftp[=0])
Limits options:
rN set the mirror depth to N (* r9999) (--depth[=N])
@@ -195,7 +197,7 @@ Flow control:
HN host is abandoned if: 0=never, 1=timeout, 2=slow, 3=timeout or slow (--host-control[=N])
Links options:
%P *extended parsing, attempt to parse all links, even in unknown tags or Javascript (%P0 don't use) (--extended-parsing[=N])
%P *extended parsing, attempt to parse all links, even in unknown tags or Javascript (%P0 don't use) (--extended-parsing[=0])
n get non-html files 'near' an html file (ex: an image located outside) (--near)
t test all URLs (even forbidden ones) (--test)
%L &lt;file&gt; add all URL located in this text file (one URL per line) (--list &lt;param&gt;)
@@ -209,7 +211,7 @@ Build options:
%x do not include any password for external password protected websites (%x0 include) (--no-passwords)
%q *include query string for local files (useless, for information purpose only) (%q0 don't include) (--include-query-string)
o *save the server's error pages (404..) (o0 discard them) (--generate-errors)
X *purge old files after update (X0 keep delete) (--purge-old[=N])
X *purge old files after update (X0 keep delete) (--purge-old[=0])
Spider options:
bN accept cookies in cookies.txt (0=do not accept,* 1=accept) (--cookies[=N])
@@ -2669,9 +2671,14 @@ for only 2 simultaneous sesions.
</div>
<dialog id="zoom" aria-label="Enlarged image"><img src="" alt=""></dialog>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- /doc-chrome:bottom -->
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -13,12 +13,14 @@
<body>
<!-- doc-chrome:top -->
<!-- doc-chrome:masthead -->
<a class="skip" href="#main">Skip to content</a>
<header class="masthead">
<img src="images/wordmark.svg" width="400" height="36" alt="HTTrack Website Copier">
<div class="tagline">Free software offline browser</div>
</header>
<!-- /doc-chrome:masthead -->
<div class="wrap">
@@ -416,9 +418,14 @@ See also: The <a href="faq.html#VF1">FAQ</a><br>
</div>
<dialog id="zoom" aria-label="Enlarged image"><img src="" alt=""></dialog>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- /doc-chrome:bottom -->
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -629,7 +629,7 @@ bottom are the ones that keep you welcome.</p>
<figure>
<img data-for="win" src="img/guide-win-opt-spider.png" alt="WinHTTrack Spider tab" width="524" height="448">
<img data-for="web" src="img/guide-web-opt-spider.png" alt="WebHTTrack Spider tab" width="1024" height="1541">
<img data-for="web" src="img/guide-web-opt-spider.png" alt="WebHTTrack Spider tab" width="1024" height="1770">
<img data-for="droid" src="img/guide-droid-opt-spider.png" alt="Android Spider tab" width="600" height="1300">
</figure>
@@ -703,11 +703,23 @@ bottom are the ones that keep you welcome.</p>
mirror something that needs you logged in, by exporting the session from your browser.</p>
</div>
<div class="opt" data-for="win web">
<div class="opt" data-for="win web droid">
<div class="opt-head"><span class="opt-name">Strip query keys</span><span class="cli">-%g, --strip-query</span></div>
<p>Comma-separated query keys to drop when naming saved files, such as
<code>sid,utm_source</code>. Keeps one page from being saved several times under tracking
parameters that change nothing.</p>
parameters that change nothing. You can scope a rule to the addresses it covers, for example
<code>shop.example.com/*=sid</code>.</p>
<p data-for="web">The box takes one rule per line. Only the last matching rule applies, so keep
the keys for one scope together on a line rather than on lines of their own.</p>
</div>
<div class="opt" data-for="win web droid">
<div class="opt-head"><span class="opt-name">Host aliases</span><span class="cli">-%C, --host-alias</span></div>
<p>Names the other hostnames one site answers to, one rule per line, written
<code>www2.example.com,m.example.com=example.com</code>. They fold onto that host: their links
travel under the same rules and each page is saved once instead of once per name. Either side may
carry a scheme, as in <code>legacy.example.com=https://example.com</code>. A trailing slash is
ignored, but a path is refused: this maps hosts, not addresses.</p>
</div>
</section>
@@ -915,7 +927,12 @@ are how you widen a crawl deliberately.</p>
<dialog id="zoom" aria-label="Enlarged screenshot"><img src="" alt=""></dialog>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -76,7 +76,7 @@ offline browser : copy websites to a local directory</p>
[ <b>-w, --mirror</b> ] [ <b>-W, --mirror-wizard</b> ] [
<b>-g, --get-files</b> ] [ <b>-i, --continue</b> ] [ <b>-Y,
--mirrorlinks</b> ] [ <b>-P, --proxy</b> ] [ <b>-%f,
--httpproxy-ftp[=N]</b> ] [ <b>-%b, --bind</b> ] [ <b>-rN,
--httpproxy-ftp[=0]</b> ] [ <b>-%b, --bind</b> ] [ <b>-rN,
--depth[=N]</b> ] [ <b>-%eN, --ext-depth[=N]</b> ] [ <b>-mN,
--max-files[=N]</b> ] [ <b>-MN, --max-size[=N]</b> ] [
<b>-EN, --max-time[=N]</b> ] [ <b>-AN, --max-rate[=N]</b> ]
@@ -85,48 +85,50 @@ offline browser : copy websites to a local directory</p>
--sockets[=N]</b> ] [ <b>-TN, --timeout[=N]</b> ] [ <b>-RN,
--retries[=N]</b> ] [ <b>-JN, --min-rate[=N]</b> ] [ <b>-HN,
--host-control[=N]</b> ] [ <b>-%P,
--extended-parsing[=N]</b> ] [ <b>-n, --near</b> ] [ <b>-t,
--extended-parsing[=0]</b> ] [ <b>-n, --near</b> ] [ <b>-t,
--test</b> ] [ <b>-%L, --list</b> ] [ <b>-%S, --urllist</b>
] [ <b>-%m, --sitemap</b> ] [ <b>-NN, --structure[=N]</b> ]
[ <b>-%N, --delayed-type-check</b> ] [ <b>-%D,
--cached-delayed-type-check</b> ] [ <b>-%M, --mime-html</b>
] [ <b>-%Z, --single-file</b> ] [ <b>-LN,
--long-names[=N]</b> ] [ <b>-KN, --keep-links[=N]</b> ] [
<b>-x, --replace-external</b> ] [ <b>-%x,
--disable-passwords</b> ] [ <b>-%q,
--include-query-string</b> ] [ <b>-%g, --strip-query</b> ] [
<b>-o, --generate-errors</b> ] [ <b>-X, --purge-old[=N]</b>
] [ <b>-%p, --preserve</b> ] [ <b>-%T, --utf8-conversion</b>
[ <b>-%N, --delayed-type-check[=N]</b> ] [ <b>-%D,
--cached-delayed-type-check[=0]</b> ] [ <b>-%M,
--mime-html[=0]</b> ] [ <b>-%Z, --single-file[=0]</b> ] [
<b>-LN, --long-names[=N]</b> ] [ <b>-KN,
--keep-links[=N]</b> ] [ <b>-x, --replace-external</b> ] [
<b>-%x, --disable-passwords[=0]</b> ] [ <b>-%q,
--include-query-string[=0]</b> ] [ <b>-%g, --strip-query</b>
] [ <b>-%C, --host-alias</b> ] [ <b>-o,
--generate-errors[=N]</b> ] [ <b>-X, --purge-old[=0]</b> ] [
<b>-%p, --preserve</b> ] [ <b>-%T, --utf8-conversion[=0]</b>
] [ <b>-bN, --cookies[=N]</b> ] [ <b>-%K, --cookies-file</b>
] [ <b>-%Y, --why</b> ] [ <b>-u, --check-type[=N]</b> ] [
<b>-j, --parse-java[=N]</b> ] [ <b>-sN, --robots[=N]</b> ] [
<b>-%h, --http-10</b> ] [ <b>-%k, --keep-alive</b> ] [
<b>-%z, --disable-compression</b> ] [ <b>-%B, --tolerant</b>
] [ <b>-%s, --updatehack</b> ] [ <b>-%u, --urlhack</b> ] [
<b>-%A, --assume</b> ] [ <b>-@iN, --protocol[=N]</b> ] [
<b>-%w, --disable-module</b> ] [ <b>-F, --user-agent</b> ] [
<b>-%R, --referer</b> ] [ <b>-%E, --from</b> ] [ <b>-%F,
--footer</b> ] [ <b>-%l, --language</b> ] [ <b>-%a,
--accept</b> ] [ <b>-%X, --headers</b> ] [ <b>-C,
--cache[=N]</b> ] [ <b>-k, --store-all-in-cache</b> ] [
<b>-%r, --warc</b> ] [ <b>-%d, --changes</b> ] [ <b>-%n,
--do-not-recatch</b> ] [ <b>-%v, --display</b> ] [ <b>-Q,
--do-not-log</b> ] [ <b>-q, --quiet</b> ] [ <b>-z,
--extra-log</b> ] [ <b>-Z, --debug-log</b> ] [ <b>-v,
--verbose</b> ] [ <b>-f, --file-log</b> ] [ <b>-f2,
--single-log</b> ] [ <b>-I, --index</b> ] [ <b>-%i,
--build-top-index</b> ] [ <b>-%I, --search-index</b> ] [
<b>-pN, --priority[=N]</b> ] [ <b>-S, --stay-on-same-dir</b>
] [ <b>-D, --can-go-down</b> ] [ <b>-U, --can-go-up</b> ] [
<b>-%h, --http-10[=0]</b> ] [ <b>-%k, --keep-alive[=0]</b> ]
[ <b>-%z, --disable-compression[=0]</b> ] [ <b>-%B,
--tolerant[=0]</b> ] [ <b>-%s, --updatehack[=0]</b> ] [
<b>-%u, --urlhack[=0]</b> ] [ <b>-%A, --assume</b> ] [
<b>-@iN, --protocol[=N]</b> ] [ <b>-%w, --disable-module</b>
] [ <b>-F, --user-agent</b> ] [ <b>-%R, --referer</b> ] [
<b>-%E, --from</b> ] [ <b>-%F, --footer</b> ] [ <b>-%l,
--language</b> ] [ <b>-%a, --accept</b> ] [ <b>-%X,
--headers</b> ] [ <b>-C, --cache[=N]</b> ] [ <b>-k,
--store-all-in-cache</b> ] [ <b>-%r, --warc</b> ] [ <b>-%d,
--changes[=0]</b> ] [ <b>-%n, --do-not-recatch[=0]</b> ] [
<b>-%v, --display[=N]</b> ] [ <b>-Q, --do-not-log</b> ] [
<b>-q, --quiet</b> ] [ <b>-z, --extra-log</b> ] [ <b>-Z,
--debug-log</b> ] [ <b>-v, --verbose</b> ] [ <b>-f,
--file-log</b> ] [ <b>-f2, --single-log</b> ] [ <b>-I,
--index[=0]</b> ] [ <b>-%i, --build-top-index[=0]</b> ] [
<b>-%I, --search-index[=N]</b> ] [ <b>-pN,
--priority[=N]</b> ] [ <b>-S, --stay-on-same-dir</b> ] [
<b>-D, --can-go-down</b> ] [ <b>-U, --can-go-up</b> ] [
<b>-B, --can-go-up-and-down</b> ] [ <b>-a,
--stay-on-same-address</b> ] [ <b>-d,
--stay-on-same-domain</b> ] [ <b>-l, --stay-on-same-tld</b>
] [ <b>-e, --go-everywhere</b> ] [ <b>-%H,
--debug-headers</b> ] [ <b>-%!,
--disable-security-limits</b> ] [ <b>-V, --userdef-cmd</b> ]
[ <b>-%W, --callback</b> ] [ <b>-y,
--background-on-suspend</b> ] [ <b>-K, --keep-links[=N]</b>
]</p>
--disable-security-limits[=0]</b> ] [ <b>-V,
--userdef-cmd</b> ] [ <b>-%W, --callback</b> ] [ <b>-y,
--background-on-suspend[=0]</b> ] [ <b>-K,
--keep-links[=N]</b> ]</p>
<h2>DESCRIPTION
<a name="DESCRIPTION"></a>
@@ -297,7 +299,7 @@ httrack --continue</b></p>
<p>*use proxy for ftp (f0 don&rsquo;t use)
(--httpproxy-ftp[=N])</p> </td></tr>
(--httpproxy-ftp[=0])</p> </td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -529,7 +531,7 @@ errors (*R1) (--retries[=N])</p></td></tr>
<p style="margin-top: 1em">*extended parsing, attempt to
parse all links, even in unknown tags or Javascript (%P0
don&rsquo;t use) (--extended-parsing[=N])</p></td></tr>
don&rsquo;t use) (--extended-parsing[=0])</p></td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -635,7 +637,7 @@ structure, 1+: see below) (--structure[=N])</p></td></tr>
make any link test but wait for files download to start
instead (experimental) (%N0 don&rsquo;t use, %N1 use for
unknown extensions, * %N2 always use)
(--delayed-type-check)</p> </td></tr>
(--delayed-type-check[=N])</p> </td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -648,7 +650,7 @@ unknown extensions, * %N2 always use)
<p>cached delayed type check, don&rsquo;t wait for remote
type during updates, to speedup them (%D0 wait, * %D1
don&rsquo;t wait) (--cached-delayed-type-check)</p></td></tr>
don&rsquo;t wait) (--cached-delayed-type-check[=0])</p></td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -660,7 +662,7 @@ don&rsquo;t wait) (--cached-delayed-type-check)</p></td></tr>
<p>generate a RFC MIME-encapsulated full-archive (.mht)
(--mime-html)</p> </td></tr>
(--mime-html[=0])</p> </td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -678,7 +680,7 @@ between pages stay relative; audio and video stay links);
--single-file-max-size N caps each asset (default 10485760
bytes). %M is the better container where a Chromium-family
browser is a given: one archive, no base64 tax on text, a
shared asset stored once (--single-file)</p></td></tr>
shared asset stored once (--single-file[=0])</p></td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -740,7 +742,8 @@ absolute URI links, K5 transparent proxy link)
<p>do not include any password for external password
protected websites (%x0 include) (--disable-passwords)</p></td></tr>
protected websites (%x0 include)
(--disable-passwords[=0])</p> </td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -753,7 +756,7 @@ protected websites (%x0 include) (--disable-passwords)</p></td></tr>
<p>*include query string for local files (useless, for
information purpose only) (%q0 don&rsquo;t include)
(--include-query-string)</p> </td></tr>
(--include-query-string[=0])</p> </td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -772,13 +775,26 @@ information purpose only) (%q0 don&rsquo;t include)
<td width="4%">
<p>-%C</p></td>
<td width="5%"></td>
<td width="82%">
<p>fold other hostnames of one site onto it
([scheme://]alias[,...]=[scheme://]host) (--host-alias
&lt;param&gt;)</p> </td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
<p>-o</p></td>
<td width="5%"></td>
<td width="82%">
<p>*save the server&rsquo;s error pages (404..) (o0 discard
them) (--generate-errors)</p></td></tr>
them) (--generate-errors[=N])</p></td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -790,7 +806,7 @@ them) (--generate-errors)</p></td></tr>
<p>*purge old files after update (X0 keep delete)
(--purge-old[=N])</p> </td></tr>
(--purge-old[=0])</p> </td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -813,7 +829,7 @@ them) (--generate-errors)</p></td></tr>
<td width="82%">
<p>links conversion to UTF-8 (--utf8-conversion)</p></td></tr>
<p>links conversion to UTF-8 (--utf8-conversion[=0])</p></td></tr>
</table>
<h3>Spider options:
@@ -909,7 +925,7 @@ rules)) (--robots[=N])</p></td></tr>
<p>force HTTP/1.0 requests (reduce update features, only
for old servers or proxies) (--http-10)</p></td></tr>
for old servers or proxies) (--http-10[=0])</p></td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -922,7 +938,7 @@ for old servers or proxies) (--http-10)</p></td></tr>
<p>use keep-alive if possible, greately reducing latency
for small files and test requests (%k0 don&rsquo;t use)
(--keep-alive)</p> </td></tr>
(--keep-alive[=0])</p> </td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -934,7 +950,7 @@ for small files and test requests (%k0 don&rsquo;t use)
<p>do not request compressed content (%z0 request)
(--disable-compression)</p> </td></tr>
(--disable-compression[=0])</p> </td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -946,7 +962,7 @@ for small files and test requests (%k0 don&rsquo;t use)
<p>tolerant requests (accept bogus responses on some
servers, but not standard!) (--tolerant)</p></td></tr>
servers, but not standard!) (--tolerant[=0])</p></td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -959,7 +975,7 @@ servers, but not standard!) (--tolerant)</p></td></tr>
<p>update hacks: various hacks to limit re-transfers when
updating (identical size, bogus response..)
(--updatehack)</p> </td></tr>
(--updatehack[=0])</p> </td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -971,7 +987,7 @@ updating (identical size, bogus response..)
<p>url hacks: various hacks to limit duplicate URLs (strip
//, www.foo.com==foo.com..) (--urlhack)</p></td></tr>
//, www.foo.com==foo.com..) (--urlhack[=0])</p></td></tr>
</table>
<p style="margin-left:18%;">opt out of one url-hack part:
@@ -1186,7 +1202,7 @@ past N bytes, --warc-cdx also writes a sorted CDXJ index,
<p>write hts-changes.json listing what this crawl left new,
changed, unchanged and gone compared to the previous mirror
(--changes)</p> </td></tr>
(--changes[=0])</p> </td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -1198,7 +1214,7 @@ changed, unchanged and gone compared to the previous mirror
<p>do not re-download locally erased files
(--do-not-recatch)</p> </td></tr>
(--do-not-recatch[=0])</p> </td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -1210,7 +1226,7 @@ changed, unchanged and gone compared to the previous mirror
<p>display on screen filenames downloaded (in realtime) - *
%v1 short version - %v2 full animation (--display)</p></td></tr>
%v1 short version - %v2 full animation (--display[=N])</p></td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -1298,7 +1314,7 @@ changed, unchanged and gone compared to the previous mirror
<td width="82%">
<p>*make an index (I0 don&rsquo;t make) (--index)</p></td></tr>
<p>*make an index (I0 don&rsquo;t make) (--index[=0])</p></td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -1310,7 +1326,7 @@ changed, unchanged and gone compared to the previous mirror
<p>make a top index for a project folder (* %i0 don&rsquo;t
make) (--build-top-index)</p></td></tr>
make) (--build-top-index[=0])</p></td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="4%">
@@ -1322,7 +1338,7 @@ make) (--build-top-index)</p></td></tr>
<p>make an searchable index for this mirror (* %I0
don&rsquo;t make) (--search-index)</p></td></tr>
don&rsquo;t make) (--search-index[=N])</p></td></tr>
</table>
<h3>Expert options:
@@ -1627,7 +1643,7 @@ with -#test=NAME [args])</p></td></tr>
<td width="82%">
<p>wait time (--advanced-wait)</p></td></tr>
<p>wait time (--advanced-wait[=N])</p></td></tr>
<tr valign="top" align="left">
<td width="9%"></td>
<td width="8%">
@@ -1661,7 +1677,7 @@ with -#test=NAME [args])</p></td></tr>
<p style="margin-top: 1em">bypass built-in security limits
aimed to avoid bandwidth abuses (bandwidth, simultaneous
connections) (--disable-security-limits)</p></td></tr>
connections) (--disable-security-limits[=0])</p></td></tr>
</table>
<p style="margin-left:18%;">IMPORTANT NOTE: DANGEROUS
@@ -1711,7 +1727,7 @@ myfoo.so[,myparameters]) (--callback &lt;param&gt;)</p></td></tr>
<p>go to background when suspended (y0 don&rsquo;t)
(--background-on-suspend)</p> </td></tr>
(--background-on-suspend[=0])</p> </td></tr>
</table>
<h3>Details: Option N

Binary file not shown.

After

Width:  |  Height:  |  Size: 842 B

Binary file not shown.

Before

Width:  |  Height:  |  Size: 34 KiB

After

Width:  |  Height:  |  Size: 34 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 49 KiB

After

Width:  |  Height:  |  Size: 50 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 51 KiB

After

Width:  |  Height:  |  Size: 51 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 110 KiB

After

Width:  |  Height:  |  Size: 108 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 94 KiB

After

Width:  |  Height:  |  Size: 103 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 44 KiB

After

Width:  |  Height:  |  Size: 43 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 40 KiB

After

Width:  |  Height:  |  Size: 40 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 44 KiB

After

Width:  |  Height:  |  Size: 44 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 44 KiB

After

Width:  |  Height:  |  Size: 51 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 54 KiB

After

Width:  |  Height:  |  Size: 54 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 32 KiB

After

Width:  |  Height:  |  Size: 31 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 42 KiB

After

Width:  |  Height:  |  Size: 42 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 62 KiB

After

Width:  |  Height:  |  Size: 66 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 31 KiB

After

Width:  |  Height:  |  Size: 31 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 51 KiB

After

Width:  |  Height:  |  Size: 52 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 36 KiB

After

Width:  |  Height:  |  Size: 36 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 32 KiB

After

Width:  |  Height:  |  Size: 32 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 24 KiB

After

Width:  |  Height:  |  Size: 25 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 36 KiB

After

Width:  |  Height:  |  Size: 37 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 43 KiB

After

Width:  |  Height:  |  Size: 43 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 82 KiB

After

Width:  |  Height:  |  Size: 82 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 93 KiB

After

Width:  |  Height:  |  Size: 93 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 61 KiB

After

Width:  |  Height:  |  Size: 61 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 113 KiB

After

Width:  |  Height:  |  Size: 114 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 44 KiB

After

Width:  |  Height:  |  Size: 44 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 28 KiB

After

Width:  |  Height:  |  Size: 28 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 8.9 KiB

After

Width:  |  Height:  |  Size: 8.9 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 7.9 KiB

After

Width:  |  Height:  |  Size: 7.7 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 8.6 KiB

After

Width:  |  Height:  |  Size: 9.0 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 19 KiB

After

Width:  |  Height:  |  Size: 20 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 18 KiB

After

Width:  |  Height:  |  Size: 18 KiB

Binary file not shown.

Before

Width:  |  Height:  |  Size: 13 KiB

After

Width:  |  Height:  |  Size: 13 KiB

View File

@@ -110,7 +110,12 @@ not own.</p>
</main>
</div>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -13,12 +13,14 @@
<body>
<!-- doc-chrome:top -->
<!-- doc-chrome:masthead -->
<a class="skip" href="#main">Skip to content</a>
<header class="masthead">
<img src="images/wordmark.svg" width="400" height="36" alt="HTTrack Website Copier">
<div class="tagline">Free software offline browser</div>
</header>
<!-- /doc-chrome:masthead -->
<div class="wrap">
@@ -73,9 +75,14 @@ You may also want to check the <tt>httrack.c</tt> and <tt>httrack.h<tt> files to
</div>
<dialog id="zoom" aria-label="Enlarged image"><img src="" alt=""></dialog>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- /doc-chrome:bottom -->
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -13,12 +13,14 @@
<body>
<!-- doc-chrome:top -->
<!-- doc-chrome:masthead -->
<a class="skip" href="#main">Skip to content</a>
<header class="masthead">
<img src="images/wordmark.svg" width="400" height="36" alt="HTTrack Website Copier">
<div class="tagline">Free software offline browser</div>
</header>
<!-- /doc-chrome:masthead -->
<div class="wrap">
@@ -279,9 +281,14 @@ httrack --wrapper mylibrary,myparameter-string http://www.example.com
</div>
<dialog id="zoom" aria-label="Enlarged image"><img src="" alt=""></dialog>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- /doc-chrome:bottom -->
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -13,12 +13,14 @@
<body>
<!-- doc-chrome:top -->
<!-- doc-chrome:masthead -->
<a class="skip" href="#main">Skip to content</a>
<header class="masthead">
<img src="images/wordmark.svg" width="400" height="36" alt="HTTrack Website Copier">
<div class="tagline">Free software offline browser</div>
</header>
<!-- /doc-chrome:masthead -->
<div class="wrap">
@@ -155,9 +157,14 @@ Below additional function names that can be defined inside the optional libhttra
</div>
<dialog id="zoom" aria-label="Enlarged image"><img src="" alt=""></dialog>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- /doc-chrome:bottom -->
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -13,12 +13,14 @@
<body>
<!-- doc-chrome:top -->
<!-- doc-chrome:masthead -->
<a class="skip" href="#main">Skip to content</a>
<header class="masthead">
<img src="images/wordmark.svg" width="400" height="36" alt="HTTrack Website Copier">
<div class="tagline">Free software offline browser</div>
</header>
<!-- /doc-chrome:masthead -->
<div class="wrap">
@@ -198,9 +200,14 @@ Script example:
</div>
<dialog id="zoom" aria-label="Enlarged image"><img src="" alt=""></dialog>
<footer>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</footer>
<!-- /doc-chrome:bottom -->
<!-- doc-chrome:footer -->
<footer>
<small>&copy; 1998-2026 Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small>
<a href="https://endsoftwarepatents.org/innovating-without-patents/" target="_blank" rel="noopener" title="This site is innovating without patents!"><img src="images/esp_chicklet.png" width="91" height="17" alt="End Software Patents"></a>
</footer>
<!-- /doc-chrome:footer -->
</body>
</html>

View File

@@ -92,7 +92,7 @@ ${LANG_K3} : ${HTTRACK_WEB}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -159,7 +159,7 @@ ${do:end-if}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -43,17 +43,29 @@
</developer>
<screenshots>
<screenshot type="default">
<caption>Choosing which files a mirror includes</caption>
<caption>Choosing the addresses and options for a new mirror</caption>
<!-- A guide asset, so every screenshot reshoot refreshes this too.
appstream-network.yml resolves the URL weekly if the name moves.
Not a wizard pane: the walk crawls a loopback server, so every pane
showing the address advertises 127.0.0.1 and an ephemeral port. -->
<image>https://www.httrack.com/html/img/guide-web-opt-scan-rules.png</image>
appstream-network.yml resolves the URL weekly if the name moves; it
cannot see a stale deploy, since the name survives a reshoot. -->
<image>https://www.httrack.com/html/img/guide-web-project-setup.png</image>
</screenshot>
</screenshots>
<content_rating type="oars-1.1"/>
<!-- Newest first; tests/01_engine-version-macros.test enforces it. -->
<releases>
<release version="3.49.21" date="2026-08-12">
<description>
<ul>
<li>A site that answers to several names is mirrored once, instead of once per name</li>
<li>Stopping a mirror ends the transfers under way instead of reporting them as network errors</li>
<li>A mirror that ends early no longer deletes files the update pass never reached</li>
<li>Reloading a project keeps the boxes you cleared and the values you set to zero</li>
<li>Clearing the option to build an index page now turns it off</li>
<li>A very long address, or a very large settings file, no longer stops the mirror</li>
<li>Refreshed screenshots throughout the documentation</li>
</ul>
</description>
</release>
<release version="3.49.20" date="2026-08-10">
<description>
<ul>

View File

@@ -72,7 +72,7 @@ ${html:error}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -61,7 +61,7 @@ function info(str) {
${do:output-mode:html-urlescaped}
href="file://${path}/"
${do:output-mode:}
target="_new">
target="_blank">
${LANG_D8}
</a>
</td></tr>
@@ -96,7 +96,7 @@ ${do:loadhash}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -99,12 +99,12 @@ ${do:end-if}
${LANG_G8} :
${/* an http: page cannot navigate to file:, so the mirror is reached through the server */}
<a href="/website/index.html" target="_new">
<a href="/website/index.html" target="_blank">
${html:path}/${html:projname}
</a></li>
<ul>
<li><a href="/website/index.html" target="_new">${LANG_D8}</a></li>
<li><a href="/website/hts-log.txt" target="_new">${LANG_D4}</a></li>
<li><a href="/website/index.html" target="_blank">${LANG_D8}</a></li>
<li><a href="/website/hts-log.txt" target="_blank">${LANG_D4}</a></li>
</ul>
<form method="POST" action="exit.html" name="form">
@@ -143,7 +143,7 @@ ${html:path}/${html:projname}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -57,7 +57,7 @@ function info(str) {
<table class="tableWidth" border="0" width="100%" cellspacing="0">
<tr><td class="tabCtrl" align="left">
<a style="background:black;color: white" href="about.html" target="_new"
<a style="background:black;color: white" href="about.html" target="_blank"
title='${html:LANG_G21}' onMouseOver="info('${js:LANG_G21}'); return true" onMouseOut="info('&nbsp;'); return true"
>
${LANG_O16}...
@@ -66,7 +66,7 @@ ${LANG_O16}...
<tr><td class="tabCtrl" align="left">
<a style="background:black;color: white"
href="http://www.httrack.com/update.php3?Product=HTTrack&Version=${attr:HTTRACK_VERSIONID}&VersionStr=${attr:HTTRACK_VERSION}&Platform=${attr:HTS_PLATFORM}&LanguageId=${attr:lang}" target="_new"
href="http://www.httrack.com/update.php3?Product=HTTrack&Version=${attr:HTTRACK_VERSIONID}&VersionStr=${attr:HTTRACK_VERSION}&Platform=${attr:HTS_PLATFORM}&LanguageId=${attr:lang}" target="_blank"
title='${html:LANG_O17}' onMouseOver="info('${js:LANG_O17}'); return true" onMouseOut="info('&nbsp;'); return true"
>
${LANG_O17}...
@@ -77,7 +77,7 @@ ${/* show help only if available */}
${do:if-file-exists:html/index.html}
<tr><td class="tabCtrl" align="left">
<a style="background:black;color: white"
href="/index.html" target="_new"
href="/index.html" target="_blank"
title='${html:LANG_TIPHELP}' onMouseOver="info('${js:LANG_TIPHELP}'); return true" onMouseOut="info('&nbsp;'); return true"
>
${LANG_P16}
@@ -103,7 +103,7 @@ ${do:end-if}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -132,7 +132,7 @@ ${LANG_THANKYOU}!
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -158,7 +158,7 @@ ${do:end-if}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -163,7 +163,7 @@ ${LANG_IOPT10}:
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -250,7 +250,7 @@ ${LANG_W3}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -191,7 +191,7 @@ ${listid:build:LISTDEF_3}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -140,7 +140,7 @@ ${do:output-mode:}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -194,7 +194,7 @@ ${listid:travel3:LISTDEF_11}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -189,7 +189,7 @@ ${LANG_I46}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -228,7 +228,7 @@ ${LANG_I64b}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -156,7 +156,7 @@ ${LANG_I43b}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -148,7 +148,7 @@ ${LANG_B13}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -181,9 +181,17 @@ ${LANG_SITEMAPURL}
<br><br>
${LANG_STRIPQUERY}
<input name="stripquery" value="${attr:stripquery}" size="40"
<br>
<textarea name="stripquery" cols="60" rows="4"
title='${html:LANG_STRIPQUERYTIP}' onMouseOver="info('${js:LANG_STRIPQUERYTIP}'); return true" onMouseOut="info('&nbsp;'); return true"
>
>${do:output-mode:html}${stripquery}${do:output-mode:}</textarea>
<br><br>
${LANG_HOSTALIAS}
<br>
<textarea name="hostalias" cols="60" rows="4"
title='${html:LANG_HOSTALIASTIP}' onMouseOver="info('${js:LANG_HOSTALIASTIP}'); return true" onMouseOut="info('&nbsp;'); return true"
>${do:output-mode:html}${hostalias}${do:output-mode:}</textarea>
<br><br>
<input type="checkbox" name="toler" ${checked:toler}
@@ -235,7 +243,7 @@ ${LANG_STRIPQUERY}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -208,7 +208,7 @@ ${listid:logtype:LISTDEF_9}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -204,7 +204,7 @@ ${LANG_H20} ${html:info.currentjob}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -99,9 +99,6 @@ ${do:if-file-exists:html/index.html}
${do:end-if}
</tr></table>
${/* Default values for empty settings */}
${do:set:cache:1}
${/* Convert winprofile.ini into internal variables */}
${do:copy:CurrentUrl:urls}
${do:copy:Category:projcateg}
@@ -140,6 +137,7 @@ ${do:copy:KeepWww:keepwww}
${do:copy:KeepSlashes:keepslashes}
${do:copy:KeepQueryOrder:keepqueryorder}
${do:copy:StripQuery:stripquery}
${do:copy:HostAlias:hostalias}
${do:copy:StoreAllInCache:cache2}
${do:copy:Sitemap:sitemap}
${do:copy:SitemapUrl:sitemapurl}
@@ -293,7 +291,7 @@ ${do:end-if:}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -196,7 +196,7 @@ ${do:output-mode:}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -144,7 +144,7 @@ ${do:end-if}${do:output-mode:html}
${test:hidequery:--include-query-string=1:--include-query-string=0}
${test:nopurge:--purge-old=1:--purge-old=0}
\
${test:cache:--cache=0:}
${ztest:cache:--cache=0:}
${test:filter::--priority=0:--priority=1:--priority=2:--priority=7}
${test:travel::--stay-on-same-dir:--can-go-down:--can-go-up:--can-go-up-and-down}
${test:travel2::--stay-on-same-address:--stay-on-same-domain:--stay-on-same-tld:--go-everywhere}
@@ -179,13 +179,14 @@ ${/* -m<n> resets the html limit, so the bare form must precede the -m,<n> one *
${ztest:cookies:--cookies=0:}
${ztest:parsejava:--parse-java=0:}
${test:updhack:--updatehack}
${test:urlhack:--urlhack=0:--urlhack}
${ztest:urlhack:--urlhack=0:--urlhack}
${test:keepwww:--keep-www-prefix}
${test:keepslashes:--keep-double-slashes}
${test:keepqueryorder:--keep-query-order}
${test:cookiesfile:--cookies-file "}${arg:cookiesfile}${test:cookiesfile:"}
${test:pausefiles:--pause "}${arg:pausefiles}${test:pausefiles:"}
${test:stripquery:--strip-query "}${arg:stripquery}${test:stripquery:"}
${arglist:stripquery:--strip-query}
${arglist:hostalias:--host-alias}
${test:toler:--tolerant}
${test:http10:--http-10}
${test:cache2:--store-all-in-cache}
@@ -202,8 +203,8 @@ ${/* -m<n> resets the html limit, so the bare form must precede the -m,<n> one *
${test:norecatch:--do-not-recatch}
${test:logf:--single-log}
${test:logtype:::--extra-log:--debug-log}
${test:index:--index=0:}
${test:index2:--search-index=0:--search-index}
${ztest:index:--index=0:}
${ztest:index2:--search-index=0:--search-index}
${test:prox:--proxy "}${do:if-not-empty:prox}${test:proxytype::socks5:connect}${test:proxytype:\3A//}${do:end-if}${do:output-mode:html}${arg:prox}${test:prox:\3A}${arg:portprox}${test:prox:"}
${test:ftpprox:--httpproxy-ftp=0:--httpproxy-ftp}
</textarea>
@@ -248,6 +249,7 @@ KeepWww=${ztest:keepwww:0:1}
KeepSlashes=${ztest:keepslashes:0:1}
KeepQueryOrder=${ztest:keepqueryorder:0:1}
StripQuery=${stripquery}
HostAlias=${hostalias}
StoreAllInCache=${ztest:cache2:0:1}
Sitemap=${ztest:sitemap:0:1}
SitemapUrl=${sitemapurl}
@@ -346,7 +348,7 @@ ${do:output-mode:}
<table width="76%" border="0" align="center" valign="bottom" cellspacing="0" cellpadding="0">
<tr>
<td id="footer"><small><small>&copy; 1998-2026 Xavier Roche & other contributors - Web Design: Leto Kauler.</small></small></td>
<td id="footer"><small><small>&copy; 1998-${date:year} Xavier Roche &amp; other contributors - Web Design: Leto Kauler.</small></small></td>
</tr>
</table>

View File

@@ -1034,6 +1034,10 @@ LANG_STRIPQUERY
Strip query keys:
LANG_STRIPQUERYTIP
Comma-separated query keys to drop from the saved-file naming (e.g. sid,utm_source).
LANG_HOSTALIAS
Host aliases:
LANG_HOSTALIASTIP
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
LANG_WARC
Write a WARC archive of the crawl
LANG_WARCTIP

View File

@@ -1014,3 +1014,7 @@ WARC segment size limit:
Íàé-ãîëÿì ðàçìåð íà WARC ñåãìåíò:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
Çàïî÷âàíå íà íîâ WARC ñåãìåíò, êîãàòî òåêóùèÿò íàäõâúðëè òîçè áðîé áàéòîâå; îñòàâåòå ïðàçíî èëè 0 çà åäèí ôàéë.
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1014,3 +1014,7 @@ WARC segment size limit:
Tamaño máximo de segmento WARC:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
Empezar un nuevo segmento WARC cuando el actual supere este número de bytes; déjelo en blanco o 0 para un solo archivo.
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1014,3 +1014,7 @@ WARC segment size limit:
Nejvìtší velikost segmentu WARC:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
Zaèít nový segment WARC, jakmile aktuální pøekroèí tento poèet bajtù; ponechte prázdné nebo 0 pro jeden soubor.
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1014,3 +1014,7 @@ WARC segment size limit:
WARC 分段大小上限:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
目前分段超過此位元組數時開始新的 WARC 分段;留空或填 0 則保持單一檔案。
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1014,3 +1014,7 @@ WARC segment size limit:
WARC 分段大小上限:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
当前分段超过该字节数时开始新的 WARC 分段;留空或填 0 则保持单个文件。
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1014,3 +1014,7 @@ WARC segment size limit:
Najveæa velièina WARC segmenta:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
Zapoèni novi WARC segment kada trenutni prijeðe ovaj broj bajtova; ostavite prazno ili 0 za jednu datoteku.
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1044,3 +1044,7 @@ WARC segment size limit:
Største størrelse på WARC-segment:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
Begynd et nyt WARC-segment, når det aktuelle overstiger dette antal byte; lad feltet stå tomt eller skriv 0 for én fil.
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1014,3 +1014,7 @@ WARC segment size limit:
Maximale Größe eines WARC-Segments:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
Ein neues WARC-Segment beginnen, sobald das aktuelle diese Anzahl Bytes überschreitet; leer lassen oder 0 für eine einzige Datei.
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1014,3 +1014,7 @@ WARC segment size limit:
WARC-segmendi suurim maht:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
Alusta uut WARC-segmenti, kui praegune ületab selle baitide arvu; jäta tühjaks või 0, et hoida üks fail.
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1044,3 +1044,7 @@ WARC segment size limit:
WARC segment size limit:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1014,3 +1014,7 @@ WARC segment size limit:
WARC-osan enimmäiskoko:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
Aloita uusi WARC-osa, kun nykyinen ylittää tämän tavumäärän; jätä tyhjäksi tai 0, jolloin tiedosto pysyy yhtenä.
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1044,3 +1044,7 @@ WARC segment size limit:
Taille maximale d'un segment WARC :
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
Commencer un nouveau segment WARC dès que le courant dépasse ce nombre d'octets ; laissez vide ou 0 pour un fichier unique.
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1014,3 +1014,7 @@ WARC segment size limit:
Μέγιστο μέγεθος τμήματος WARC:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
Έναρξη νέου τμήματος WARC μόλις το τρέχον ξεπεράσει αυτόν τον αριθμό byte. Αφήστε κενό ή 0 για ένα ενιαίο αρχείο.
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1014,3 +1014,7 @@ WARC segment size limit:
Dimensione massima del segmento WARC:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
Inizia un nuovo segmento WARC quando quello corrente supera questo numero di byte; lascia vuoto o 0 per un solo file.
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1014,3 +1014,7 @@ WARC segment size limit:
WARC セグメントの最大サイズ:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
現在のセグメントがこのバイト数を超えたら新しい WARC セグメントを開始します。空欄または 0 で単一ファイルのままにします。
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

View File

@@ -1014,3 +1014,7 @@ WARC segment size limit:
Íà¼ãîëåìà ãîëåìèíà íà WARC ñåãìåíò:
Start a new WARC segment once the current one grows past this many bytes; leave blank or 0 to keep a single file.
Çàïî÷íè íîâ WARC ñåãìåíò êîãà òåêîâíèîò <20>å ãî íàäìèíå îâî¼ áðî¼ áà¼òè; îñòàâåòå ïðàçíî èëè 0 çà åäíà äàòîòåêà.
Host aliases:
Host aliases:
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).
Other hostnames serving this same site, folded onto one, one rule per line (e.g. www2.example.com,m.example.com=example.com).

Some files were not shown because too many files have changed in this diff Show More