Split -%u URL Hacks into independent www/slash/query toggles (#271)

-%u (--urlhack) bundled three dedup normalizations under one switch:
www.host == host, redundant // collapse, and query-argument reordering.
A mirror that needed one but not another (e.g. keep www. distinct) had to
turn the whole umbrella off. Add three opt-out sub-options, defaulting to
the umbrella so existing -%u/-%u0 behavior is unchanged:

  --keep-www-prefix      keep www.foo.com distinct from foo.com   (-%j)
  --keep-double-slashes  keep redundant // in the path            (-%o)
  --keep-query-order     keep query-argument order significant    (-%y)

The split is resolved once in hash_init() into norm_host/norm_slash/
norm_query and threaded through the dedup hash (htshash.c), the savename
lookup key (htsname.c) and the redirect-loop diagnostic (htsparse.c) so
all three stay consistent. fil_normalized() gains an internal
fil_normalized_ex(do_slash, do_query) core; the public
fil_normalized()/fil_normalized_filtered() keep their signatures.

Normalization (slash/query) now follows urlhack and its sub-flags
uniformly, while --strip-query stays orthogonal. So with urlhack off,
strip-query strips keys without sorting the remainder; the url_savename
urlhack-off branch is moved to the same do_slash=0/do_query=0 normalizer
the hash uses, so a URL is always looked up under the key it was stored
with (a self-lookup mismatch this otherwise introduced).

http/https are always merged in the dedup key (the scheme is stripped
regardless of -%u), so that part of the request needs no toggle.

The opt-outs are spelled positively (--keep-*) because httrack's generic
--no<opt> prefix only appends the disabling "0" for parametered options,
not "single" booleans, so --nowww-dedup would silently no-op.

opt grows three hts_boolean fields appended at the struct tail (offsets
stable, no soname bump, matching the strip_query addition in #112).

Tested by a -#test=urlhack engine self-test (hash_url_equals over each
flag combination) plus a -%u0 + --strip-query crawl case exercising the
urlhack-off savename branch.

Closes #271

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

.github/workflows/ci.yml

@@ -188,6 +188,51 @@ jobs:
         if: failure()
         run: cat tests/test-suite.log 2>/dev/null || true
 
+  # MemorySanitizer catches reads of uninitialized memory (#143's stack-garbage
+  # size filter) that ASan/UBSan miss. It flags any byte an uninstrumented lib
+  # wrote, so the job stays in our own code: offline self-tests only, no openssl
+  # (--disable-https), no zlib cache tests, static (the runtime is not in .so's).
+  msan:
+    name: msan (MemorySanitizer, clang)
+    runs-on: ubuntu-24.04
+    steps:
+      - uses: actions/checkout@v6
+        with:
+          submodules: recursive
+
+      - name: Install build dependencies
+        run: |
+          set -euo pipefail
+          sudo apt-get update
+          sudo apt-get install -y --no-install-recommends \
+            build-essential clang autoconf automake libtool autoconf-archive \
+            zlib1g-dev
+
+      - name: Configure (MSan, static, no https)
+        run: |
+          set -euo pipefail
+          autoreconf -fi
+          ./configure CC=clang \
+            CFLAGS="-fsanitize=memory -fsanitize-memory-track-origins=2 -fno-sanitize-recover=all -g -O1 -fno-omit-frame-pointer" \
+            LDFLAGS="-fsanitize=memory" \
+            --disable-https --disable-shared --enable-static
+
+      - name: Build
+        run: make -j"$(nproc)"
+
+      - name: Test (offline self-tests under MSan)
+        env:
+          MSAN_OPTIONS: abort_on_error=1:halt_on_error=1
+        run: |
+          set -euo pipefail
+          # Engine self-tests only; the cache trio pulls in uninstrumented zlib.
+          tests="$(cd tests && ls 01_engine-*.test | grep -v -- '-cache' | tr '\n' ' ')"
+          make check TESTS="$tests"
+
+      - name: Print the test log on failure
+        if: failure()
+        run: cat tests/test-suite.log 2>/dev/null || true
+
   # Optional-dependency build: compile and test with HTTPS/OpenSSL disabled --
   # the configuration users on minimal systems build, and one libssl is not even
   # installed here so configure cannot silently re-enable it. The matrix above
@@ -232,30 +277,42 @@ jobs:
   deb:
     name: deb package (lintian)
     runs-on: ubuntu-24.04
+    # Build and gate inside Debian sid, the upload target. A Debian dpkg-deb
+    # produces archive-legal xz members (an Ubuntu host defaults to zstd, which
+    # the archive's lintian rejects), and sid's lintian carries the same
+    # data-driven checks (embedded-lib fingerprints and the like) the buildds and
+    # UDD apply -- so issues surface here instead of after upload.
+    container: debian:sid
     steps:
-      - uses: actions/checkout@v6
-        with:
-          submodules: recursive
-
       - name: Install packaging toolchain
         run: |
           set -euo pipefail
-          sudo apt-get update
+          apt-get update
-          sudo apt-get install -y --no-install-recommends \
+          apt-get install -y --no-install-recommends \
+            ca-certificates git \
             build-essential autoconf automake libtool autoconf-archive \
             zlib1g-dev libssl-dev \
             debhelper devscripts lintian fakeroot
 
+      - uses: actions/checkout@v6
+        with:
+          submodules: recursive
+
       # --unsigned: CI has no GPG key (also skips the release sig/checksums).
-      # debuild builds every package, then lintian gates on errors.
+      # mkdeb builds every package then runs the lintian gate (--fail-on=error,
+      # warning); debuild runs the packaged test pass.
       #
       # DEB_BUILD_OPTIONS trims work CI does not need (release builds via
       # mkdeb.sh are untouched): noautodbgsym drops the -dbgsym packages whose
       # LTO payloads are slow to compress and that CI never ships; parallel uses
-      # every core. We let debuild run its test pass -- the only one now that
+      # every core.
-      # mkdeb no longer runs its own -- so CI exercises the packaged tests.
+      - name: Build and lint Debian packages
-      - name: Build Debian packages
         run: |
+          set -euo pipefail
+          # The workspace volume is owned by the host runner uid, but the
+          # container runs as root, so mkdeb's git calls (superproject and the
+          # coucal submodule) trip "dubious ownership"; mark them all safe.
+          git config --global --add safe.directory "*"
           export DEB_BUILD_OPTIONS="noautodbgsym parallel=$(nproc)"
           bash tools/mkdeb.sh --unsigned --no-release-artifacts
 

AGENTS.md

@@ -33,8 +33,9 @@ the operational checklist: toolchain, invariants, and how to ship a change.
 - Be terse. Comment the why, in English; translate French comments you touch.
 - Strip AI tells from prose (em-dash overuse, rule-of-three, filler, vague
   attributions). Ref: Wikipedia "Signs of AI writing". Claude Code: `/humanizer`.
-- Behavior change → add a test. Fast path: a hidden `httrack -#N` debug
+- Behavior change → add a test. Fast path: a hidden `httrack -#test=NAME` engine
-  subcommand (`htscoremain.c`) driven by a `tests/NN_*.test`, over a slow crawl.
+  self-test (registry in `htsselftest.c`; `-#test` lists them) driven by a
+  `tests/NN_*.test`, over a slow crawl.
 
 ## Review your change adversarially (strongly suggested)
 Before pushing, and when reviewing others, don't skim for bugs:

configure.ac

@@ -215,9 +215,12 @@ AC_SUBST(OPENSSL_LIBS)
 fi
 
 ### Support IPv6
+V6_SUPPORT=no
 AC_CHECK_LIB(c, getaddrinfo, [V6_FLAG="-DINET6"
+V6_SUPPORT=yes
 AC_DEFINE(HTS_INET6, 1, [Check for IPv6])], AC_MSG_WARN([*** IPv6 not found IPv6 compatibility disabled]))
 AC_SUBST(V6_FLAG)
+AC_SUBST(V6_SUPPORT)
 
 ### Check for LFS
 AC_CHECK_LIB(c, fopen64, [LFS_FLAG="-D_FILE_OFFSET_BITS=64 -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE"

debian/libhttrack3.lintian-overrides

@@ -1,3 +1,8 @@
 # The shared libraries ship without a versioned symbols control file (ABI is
 # tracked via the SONAME plus a >= upstream-version dependency, see debian/rules).
 libhttrack3: no-symbols-control-file usr/lib/*
+
+# Bundled, locally patched minizip (src/minizip): it adds a zipFlush() API the
+# system libminizip lacks (htscache.c flushes the cache .zip so an interrupted
+# crawl leaves a valid archive), plus Android/old-zlib portability fixes.
+libhttrack3: embedded-library *libminizip*

debian/proxytrack.lintian-overrides

@@ -0,0 +1,3 @@
+# Statically linked against httrack's bundled, patched minizip (see src/minizip
+# and libhttrack3's override): the zipFlush() API is absent from the system one.
+proxytrack: embedded-library *libminizip*

man/httrack.1

@@ -3,7 +3,7 @@
 .\"
 .\" This file is generated by man/makeman.sh; do not edit by hand.
 .\" SPDX-License-Identifier: GPL-3.0-or-later
-.TH httrack 1 "13 June 2026" "httrack website copier"
+.TH httrack 1 "27 June 2026" "httrack website copier"
 .SH NAME
 httrack \- offline browser : copy websites to a local directory
 .SH SYNOPSIS
@@ -43,6 +43,7 @@ httrack \- offline browser : copy websites to a local directory
 [ \fB\-x, \-\-replace\-external\fR ]
 [ \fB\-%x, \-\-disable\-passwords\fR ]
 [ \fB\-%q, \-\-include\-query\-string\fR ]
+[ \fB\-%g, \-\-strip\-query\fR ]
 [ \fB\-o, \-\-generate\-errors\fR ]
 [ \fB\-X, \-\-purge\-old[=N]\fR ]
 [ \fB\-%p, \-\-preserve\fR ]
@@ -198,6 +199,8 @@ replace external html links by error pages (\-\-replace\-external)
 do not include any password for external password protected websites (%x0 include) (\-\-disable\-passwords)
 .IP \-%q
 *include query string for local files (useless, for information purpose only) (%q0 don't include) (\-\-include\-query\-string)
+.IP \-%g
+strip query keys for dedup ([host/pattern=]key1,key2,...) (\-\-strip\-query <param>)
 .IP \-o
 *generate output html file in case of error (404..) (o0 don't generate) (\-\-generate\-errors)
 .IP \-X
@@ -225,6 +228,8 @@ tolerant requests (accept bogus responses on some servers, but not standard!) (\
 update hacks: various hacks to limit re\-transfers when updating (identical size, bogus response..) (\-\-updatehack)
 .IP \-%u
 url hacks: various hacks to limit duplicate URLs (strip //, www.foo.com==foo.com..) (\-\-urlhack)
+.br
+opt out of one url\-hack part: \-\-keep\-www\-prefix (www.foo.com<>foo.com), \-\-keep\-double\-slashes (//), \-\-keep\-query\-order (?b&a)
 .IP \-%A
 assume that a type (cgi,asp..) is always linked with a mime type (\-%A php3,cgi=text/html;dat,bin=application/x\-zip) (\-\-assume <param>)
 .br
@@ -313,12 +318,8 @@ debug HTTP headers in logfile (\-\-debug\-headers)
 .SS Guru options: (do NOT use if possible)
 .IP \-#X
 *use optimized engine (limited memory boundary checks) (\-\-fast\-engine)
-.IP \-#0
+.IP \-#test
-filter test (\-#0 '*.gif' 'www.bar.com/foo.gif') (\-\-debug\-testfilters <param>)
+list engine self\-tests (run one with \-#test=NAME [args])
-.IP \-#1
-simplify test (\-#1 ./foo/bar/../foobar)
-.IP \-#2
-type test (\-#2 /foo/bar.php)
 .IP \-#C
 cache list (\-#C '*.com/spider*.gif' (\-\-debug\-cache <param>)
 .IP \-#R

src/Makefile.am

@@ -56,7 +56,7 @@ whttrackrundir = $(bindir)
 whttrackrun_SCRIPTS = webhttrack
 
 libhttrack_la_SOURCES =  htscore.c htsparse.c htsback.c htscache.c \
-	htscache_selftest.c htsdns_selftest.c \
+	htscache_selftest.c htsdns_selftest.c htsselftest.c \
 	htscatchurl.c htsfilters.c htsftp.c htshash.c coucal/coucal.c \
 	htshelp.c htslib.c htscoremain.c \
 	htsname.c htsrobots.c htstools.c htswizard.c \
@@ -66,7 +66,7 @@ libhttrack_la_SOURCES =  htscore.c htsparse.c htsback.c htscache.c \
 	md5.c \
 	minizip/ioapi.c minizip/mztools.c minizip/unzip.c minizip/zip.c \
 	hts-indextmpl.h htsalias.h htsback.h htsbase.h htssafe.h \
-	htsbasenet.h htsbauth.h htscache.h htscache_selftest.h htsdns_selftest.h htscatchurl.h  \
+	htsbasenet.h htsbauth.h htscache.h htscache_selftest.h htsdns_selftest.h htsselftest.h htscatchurl.h  \
 	htsconfig.h htscore.h htsparse.h htscoremain.h htsdefines.h  \
 	htsfilters.h htsftp.h htsglobal.h htshash.h coucal/coucal.h \
 	htshelp.h htsindex.h htslib.h htsmd5.h \

src/htsalias.c

@@ -60,6 +60,9 @@ Please visit our Website: http://www.httrack.com
   param1 : this option must be alone, and needs one distinct parameter (-P <path>)
   param0 : this option must be alone, but the parameter should be put together (+*.gif)
 */
+/* clang-format off: hand-aligned table; clang-format reflows the whole
+   initializer (2->4 space) on any edit, churning every untouched row. */
+/* clang-format off */
 const char *hts_optalias[][4] = {
   /*   {"","","",""}, */
   {"path", "-O", "param1", "output path"},
@@ -107,6 +110,8 @@ const char *hts_optalias[][4] = {
   {"disable-passwords", "-%x", "single", ""}, {"disable-password", "-%x",
                                                "single", ""},
   {"include-query-string", "-%q", "single", ""},
+  {"strip-query", "-%g", "param1",
+   "strip [host/pattern=]key1,key2,... from URLs"},
   {"generate-errors", "-o", "single", ""},
   {"do-not-generate-errors", "-o0", "single", ""},
   {"purge-old", "-X", "param", ""},
@@ -123,6 +128,9 @@ const char *hts_optalias[][4] = {
   {"tolerant", "-%B", "single", ""},
   {"updatehack", "-%s", "single", ""}, {"sizehack", "-%s", "single", ""},
   {"urlhack", "-%u", "single", ""},
+  {"keep-www-prefix", "-%j", "single", ""},
+  {"keep-double-slashes", "-%o", "single", ""},
+  {"keep-query-order", "-%y", "single", ""},
   {"user-agent", "-F", "param1", "user-agent identity"},
   {"referer", "-%R", "param1", "default referer URL"},
   {"from", "-%E", "param1", "from email address"},
@@ -241,6 +249,7 @@ const char *hts_optalias[][4] = {
 
   {"", "", "", ""}
 };
+/* clang-format on */
 
 /* 
   Check for alias in command-line 

src/htsback.c

@@ -57,7 +57,10 @@ Please visit our Website: http://www.httrack.com
 // DOS
 #include <process.h>            /* _beginthread, _endthread */
 #endif
+#include <io.h> /* _chsize_s */
+#define HTS_FTRUNCATE(fp, sz) _chsize_s(_fileno(fp), (sz))
 #else
+#define HTS_FTRUNCATE(fp, sz) ftruncate(fileno(fp), (sz))
 #endif
 
 #define VT_CLREOL       "\33[K"
@@ -73,6 +76,8 @@ struct_back *back_new(httrackp *opt, int back_max) {
 
   sback->count = back_max;
   sback->lnk = (lien_back *) calloct((back_max + 1), sizeof(lien_back));
+  sback->connect_fallback = (hts_connect_fallback *) calloct(
+      (back_max + 1), sizeof(hts_connect_fallback));
   sback->ready = coucal_new(0);
   hts_set_hash_handler(sback->ready, opt);
   coucal_set_name(sback->ready, "back_new");
@@ -83,6 +88,7 @@ struct_back *back_new(httrackp *opt, int back_max) {
     sback->lnk[i].r.location = sback->lnk[i].location_buffer;
     sback->lnk[i].status = STATUS_FREE;
     sback->lnk[i].r.soc = INVALID_SOCKET;
+    sback->connect_fallback[i].addr_count = -1; // not yet probed
   }
   return sback;
 }
@@ -93,6 +99,7 @@ void back_free(struct_back ** sback) {
       freet((*sback)->lnk);
       (*sback)->lnk = NULL;
     }
+    freet((*sback)->connect_fallback);
     if ((*sback)->ready != NULL) {
       coucal_delete(&(*sback)->ready);
       (*sback)->ready_size_bytes = 0;
@@ -102,6 +109,72 @@ void back_free(struct_back ** sback) {
   }
 }
 
+/* Per-candidate connect deadline cap (seconds): a connecting slot with another
+   address to try waits at most this long before falling back, instead of the
+   full (default 120s) slot timeout. Caps the dead-IPv6 stall while staying well
+   above a normal handshake. The last candidate still gets the full timeout. */
+#define HTS_CONNECT_FALLBACK_TIMEOUT 10
+
+int back_connect_fallback_due(int addr_index, int addr_count, int elapsed,
+                              int timeout) {
+  int deadline;
+
+  if (addr_index + 1 >= addr_count) // last (or only) candidate: no fallback
+    return 0;
+  if (timeout <= 0) // no timeout management: never force it
+    return 0;
+  deadline = (timeout < HTS_CONNECT_FALLBACK_TIMEOUT)
+                 ? timeout
+                 : HTS_CONNECT_FALLBACK_TIMEOUT;
+  return elapsed >= deadline;
+}
+
+/* Pending-connect result for a non-blocking socket reported ready by select():
+   0 = connected, >0 = the connect errno (refused, unreachable, ...), -1 if the
+   probe itself failed. A failed connect is reported writable too, so this is
+   how success is told from failure without blocking. */
+static int connect_socket_error(T_SOC soc) {
+  int soerr = 0;
+  socklen_t len = (socklen_t) sizeof(soerr);
+
+  if (getsockopt(soc, SOL_SOCKET, SO_ERROR, (char *) &soerr, &len) != 0)
+    return -1;
+  return soerr;
+}
+
+/* Retry a stuck/failed connecting slot against its next resolved address.
+   Closes the current socket and starts a non-blocking connect to the next
+   candidate, leaving the slot in STATUS_CONNECTING. Returns 1 if a new connect
+   was started, 0 if no fallback address remains (caller fails the slot). */
+static int back_connect_next(httrackp *opt, struct_back *sback, int i) {
+  hts_connect_fallback *const cf = &sback->connect_fallback[i];
+  lien_back *const back = sback->lnk;
+  const int next = cf->addr_index + 1;
+  T_SOC soc;
+
+  if (next >= cf->addr_count)
+    return 0;
+
+  if (back[i].r.soc != INVALID_SOCKET) {
+    deletehttp(&back[i].r);
+    back[i].r.soc = INVALID_SOCKET;
+  }
+  soc = newhttp_addr(opt, back[i].url_adr, &back[i].r, -1, 0, next, NULL);
+  if (soc == INVALID_SOCKET)
+    return 0;
+
+  back[i].r.soc = soc;
+  cf->addr_index = next;
+  cf->connect_start = time_local();
+  if (back[i].timeout > 0)
+    back[i].timeout_refresh = cf->connect_start;
+  back[i].status = STATUS_CONNECTING;
+  hts_log_print(opt, LOG_DEBUG,
+                "connect failed, trying next address (%d/%d) for %s", next + 1,
+                cf->addr_count, back[i].url_adr);
+  return 1;
+}
+
 void back_delete_all(httrackp * opt, cache_back * cache, struct_back * sback) {
   if (sback != NULL) {
     int i;
@@ -1911,8 +1984,11 @@ int back_add(struct_back * sback, httrackp * opt, cache_back * cache, const char
       // ouvrir liaison, envoyer requète
       // ne pas traiter ou recevoir l'en tête immédiatement
       hts_init_htsblk(&back[p].r);
-      //memset(&(back[p].r), 0, sizeof(htsblk)); 
+      // memset(&(back[p].r), 0, sizeof(htsblk));
       back[p].r.location = back[p].location_buffer;
+      // fresh connect: address list not yet probed, start at the first
+      sback->connect_fallback[p].addr_index = 0;
+      sback->connect_fallback[p].addr_count = -1;
       // recopier proxy
       if ((back[p].r.req.proxy.active = opt->proxy.active)) {
         if (StringBuff(opt->proxy.bindhost) != NULL)
@@ -2369,21 +2445,25 @@ void back_wait(struct_back * sback, httrackp * opt, cache_back * cache,
       // en cas de gestion du connect préemptif
 #if HTS_XCONN
       if (back[i].status == STATUS_CONNECTING) {        // connexion
-        do_wait = 1;
+        // a connecting slot always carries a live socket; guard anyway so a
+        // stray INVALID_SOCKET can never reach FD_SET (mirrors the recv branch)
+        if (back[i].r.soc != INVALID_SOCKET) {
+          do_wait = 1;
 
-        // noter socket write
+          // noter socket write
-        FD_SET(back[i].r.soc, &fds_c);
+          FD_SET(back[i].r.soc, &fds_c);
 
-        // noter socket erreur
+          // noter socket erreur
-        FD_SET(back[i].r.soc, &fds_e);
+          FD_SET(back[i].r.soc, &fds_e);
 
-        // calculer max
+          // calculer max
-        if (max_c) {
+          if (max_c) {
-          max_c = 0;
+            max_c = 0;
-          nfds = back[i].r.soc;
+            nfds = back[i].r.soc;
-        } else if (back[i].r.soc > nfds) {
+          } else if (back[i].r.soc > nfds) {
-          // ID socket la plus élevée
+            // ID socket la plus élevée
-          nfds = back[i].r.soc;
+            nfds = back[i].r.soc;
+          }
         }
 
       } else
@@ -2517,8 +2597,20 @@ void back_wait(struct_back * sback, httrackp * opt, cache_back * cache,
       }
       // ---- FLAG WRITE MIS A UN?: POUR LE CONNECT
       if (back[i].status == STATUS_CONNECTING) {        // attendre connect
+        hts_connect_fallback *const cf = &sback->connect_fallback[i];
         int dispo = 0;
 
+        // probe the resolved address list once per fresh connect (cache hit:
+        // the host was resolved when this connect was opened)
+        if (cf->addr_count < 0 && back[i].r.soc != INVALID_SOCKET &&
+            !back[i].r.is_file) {
+          SOCaddr scratch[HTS_MAXADDRNUM];
+
+          cf->addr_count = hts_dns_resolve_all(opt, back[i].url_adr, scratch,
+                                               HTS_MAXADDRNUM, NULL);
+          cf->connect_start = time_local();
+        }
+
         // vérifier l'existance de timeout-check
         if (!gestion_timeout)
           if (back[i].timeout > 0)
@@ -2526,7 +2618,20 @@ void back_wait(struct_back * sback, httrackp * opt, cache_back * cache,
 
         // connecté?
         dispo = FD_ISSET(back[i].r.soc, &fds_c);
-        if (dispo) {            // ok connected!!
+        if (dispo) { // socket ready: connect() finished (ok or failed)
+          // a refused/failed connect is reported writable too; probe SO_ERROR
+          // and, on failure, fall back to the next address (or fail the slot)
+          if (connect_socket_error(back[i].r.soc) != 0) {
+            if (!back_connect_next(opt, sback, i)) {
+              deletehttp(&back[i].r);
+              back[i].r.soc = INVALID_SOCKET;
+              back[i].r.statuscode = STATUSCODE_CONNERROR;
+              strcpybuff(back[i].r.msg, "Connect Error");
+              back[i].status = STATUS_READY;
+              back_set_finished(sback, i);
+            }
+            continue; // reconnected (stay connecting) or failed
+          }
           busy_state = 1;
 
 #if HTS_USEOPENSSL
@@ -3661,7 +3766,27 @@ void back_wait(struct_back * sback, httrackp * opt, cache_back * cache,
                     }
 #endif
 /********** **************************** ********** */
-                  } else {      // il faut aller le chercher
+                  }
+                  // MIME type excluded by a -mime: filter: abort, don't fetch
+                  // the body (#58)
+                  else if (HTTP_IS_OK(back[i].r.statuscode) &&
+                           !back[i].testmode &&
+                           strnotempty(back[i].r.contenttype) &&
+                           hts_acceptmime(opt, 0, back[i].url_adr,
+                                          back[i].url_fil,
+                                          back[i].r.contenttype) == 1) {
+                    deletehttp(&back[i].r);
+                    back[i].r.soc = INVALID_SOCKET;
+                    back[i].status = STATUS_READY;
+                    back_set_finished(sback, i);
+                    back[i].r.statuscode = STATUSCODE_EXCLUDED;
+                    strcpybuff(back[i].r.msg, "Excluded by MIME type filter");
+                    hts_log_print(
+                        opt, LOG_NOTICE,
+                        "File excluded by MIME type filter (%s): %s%s",
+                        back[i].r.contenttype, back[i].url_adr,
+                        back[i].url_fil);
+                  } else { // il faut aller le chercher
 
                     // effacer buffer (requète)
                     if (!noFreebuff) {
@@ -3672,35 +3797,70 @@ void back_wait(struct_back * sback, httrackp * opt, cache_back * cache,
                     // xxc SI CHUNK VERIFIER QUE CA MARCHE??
                     if (back[i].r.statuscode == 206) {  // on nous envoie un morceau (la fin) coz une partie sur disque!
                       off_t sz = fsize_utf8(back[i].url_sav);
+                      /* RFC 7233: resume at the server's Content-Range start,
+                         not the offset we requested; a server may resume
+                         earlier and appending the overlap duplicates bytes
+                         (#198). */
+                      const LLint resume = back[i].r.crange_start;
+                      const hts_boolean range_ok =
+                          back[i].r.crange > 0 && resume >= 0 &&
+                          resume <= (LLint) sz &&
+                          back[i].r.crange_end + 1 == back[i].r.crange &&
+                          (back[i].r.totalsize < 0 ||
+                           back[i].r.totalsize ==
+                               back[i].r.crange_end - resume + 1);
 
 #if HDEBUG
                       printf("partial content: " LLintP " on disk..\n",
                              (LLint) sz);
 #endif
-                      if (sz >= 0) {
+                      if (sz >= 0 && range_ok) {
                         if (!is_hypertext_mime(opt, back[i].r.contenttype, back[i].url_sav)) {  // pas HTML
                           if (opt->getmode & HTS_GETMODE_NONHTML) {
                             filenote(&opt->state.strc, back[i].url_sav, NULL);  // noter fichier comme connu
                             file_notify(opt, back[i].url_adr, back[i].url_fil,
                                         back[i].url_sav, 0, 1,
                                         back[i].r.notmodified);
-                            back[i].r.out = FOPEN(fconv(catbuff, sizeof(catbuff), back[i].url_sav), "ab");       // append
+                            back[i].r.out =
+                                FOPEN(fconv(catbuff, sizeof(catbuff),
+                                            back[i].url_sav),
+                                      "r+b"); // resume in place
                             if (back[i].r.out && opt->cache != 0) {
-                              back[i].r.is_write = 1;   // écrire
+                              back[i].r.is_write = 1;
-                              back[i].r.size = sz;      // déja écrit
+                              back[i].r.size = resume; // bytes already on disk
-                              back[i].r.statuscode = HTTP_OK;   // Forcer 'OK'
+                              back[i].r.statuscode = HTTP_OK; // force 'OK'
                               if (back[i].r.totalsize >= 0)
-                                back[i].r.totalsize += sz;      // plus en fait
+                                back[i].r.totalsize += resume; // -> full size
-                              fseek(back[i].r.out, 0, SEEK_END);        // à la fin
+                              // drop bytes past the resume point; a silent
-                              /* create a temporary reference file in case of broken mirror */
+                              // failure could leave a stale tail, so on error
-                              if (back_serialize_ref(opt, &back[i]) != 0) {
+                              // drop the partial and refetch the whole file
-                                hts_log_print(opt, LOG_WARNING,
+                              if (HTS_FTRUNCATE(back[i].r.out,
-                                              "Could not create temporary reference file for %s%s",
+                                                (off_t) resume) != 0) {
-                                              back[i].url_adr, back[i].url_fil);
+                                fclose(back[i].r.out);
-                              }
+                                back[i].r.out = NULL;
+                                url_savename_refname_remove(
+                                    opt, back[i].url_adr, back[i].url_fil);
+                                UNLINK(back[i].url_sav);
+                                back[i].status = STATUS_READY;
+                                back_set_finished(sback, i);
+                                strcpybuff(back[i].r.msg,
+                                           "Can not truncate partial file, "
+                                           "restarting");
+                              } else {
+                                fseeko(back[i].r.out, (off_t) resume, SEEK_SET);
+                                /* create a temporary reference file in case of
+                                 * broken mirror */
+                                if (back_serialize_ref(opt, &back[i]) != 0) {
+                                  hts_log_print(opt, LOG_WARNING,
+                                                "Could not create temporary "
+                                                "reference file for %s%s",
+                                                back[i].url_adr,
+                                                back[i].url_fil);
+                                }
 #if HDEBUG
-                              printf("continue interrupted file\n");
+                                printf("continue interrupted file\n");
 #endif
+                              }
                             } else {    // On est dans la m**
                               back[i].status = STATUS_READY;    // terminé (voir plus loin)
                               back_set_finished(sback, i);
@@ -3712,17 +3872,18 @@ void back_wait(struct_back * sback, httrackp * opt, cache_back * cache,
                           FILE *fp =
                             FOPEN(fconv(catbuff, sizeof(catbuff), back[i].url_sav), "rb");
                           if (fp) {
-                            LLint alloc_mem = sz + 1;
+                            LLint alloc_mem = resume + 1;
 
                             if (back[i].r.totalsize >= 0)
                               alloc_mem += back[i].r.totalsize; // AJOUTER RESTANT!
                             if (deleteaddr(&back[i].r)
                                 && (back[i].r.adr =
                                     (char *) malloct((size_t) alloc_mem))) {
-                              back[i].r.size = sz;
+                              back[i].r.size = resume;
                               if (back[i].r.totalsize >= 0)
-                                back[i].r.totalsize += sz;      // plus en fait
+                                back[i].r.totalsize += resume; // -> full size
-                              if ((fread(back[i].r.adr, 1, sz, fp)) != sz) {
+                              if ((fread(back[i].r.adr, 1, (size_t) resume,
+                                         fp)) != (size_t) resume) {
                                 back[i].status = STATUS_READY;  // terminé (voir plus loin)
                                 back_set_finished(sback, i);
                                 strcpybuff(back[i].r.msg,
@@ -3740,14 +3901,30 @@ void back_wait(struct_back * sback, httrackp * opt, cache_back * cache,
                                          "No memory for partial file");
                             }
                             fclose(fp);
-                          } else {      // Argh.. 
+                          } else {                              // open failed
                             back[i].status = STATUS_READY;      // terminé (voir plus loin)
                             back_set_finished(sback, i);
                             strcpybuff(back[i].r.msg,
                                        "Can not open partial file");
                           }
                         }
-                      } else {  // Non trouvé??
+                      } else if (sz >=
+                                 0) { // unusable range -> restart whole file
+                        hts_log_print(opt, LOG_WARNING,
+                                      "Unusable partial-content range for %s%s "
+                                      "(have " LLintP " bytes, got " LLintP
+                                      "-" LLintP "/" LLintP "), restarting",
+                                      back[i].url_adr, back[i].url_fil,
+                                      (LLint) sz, back[i].r.crange_start,
+                                      back[i].r.crange_end, back[i].r.crange);
+                        url_savename_refname_remove(opt, back[i].url_adr,
+                                                    back[i].url_fil);
+                        UNLINK(back[i].url_sav);
+                        back[i].status = STATUS_READY;
+                        back_set_finished(sback, i);
+                        strcpybuff(back[i].r.msg,
+                                   "Unusable partial content, restarting");
+                      } else {                          // partial not found
                         back[i].status = STATUS_READY;  // terminé (voir plus loin)
                         back_set_finished(sback, i);
                         strcpybuff(back[i].r.msg, "Can not find partial file");
@@ -3828,7 +4005,6 @@ void back_wait(struct_back * sback, httrackp * opt, cache_back * cache,
 
                       }
                     }
-
                   }
 
                   /*} */
@@ -3884,6 +4060,29 @@ void back_wait(struct_back * sback, httrackp * opt, cache_back * cache,
 
         if (back[i].status > 0) {       // réception/connexion/..
           if (back[i].timeout > 0) {
+            // a stuck connect with a fallback address: retry the next one well
+            // before the full timeout (dead IPv6 on a dual-stack host, ...)
+            if (back[i].status == STATUS_CONNECTING) {
+              const hts_connect_fallback *const cf =
+                  &sback->connect_fallback[i];
+
+              if (back_connect_fallback_due(cf->addr_index, cf->addr_count,
+                                            (int) (act - cf->connect_start),
+                                            back[i].timeout)) {
+                if (back_connect_next(opt, sback, i)) {
+                  continue; // reconnected to the next candidate
+                }
+                // fallback was due but no socket could be opened
+                // (back_connect_next closed the dead one): stop now rather than
+                // spin on an invalid fd
+                back[i].r.soc = INVALID_SOCKET;
+                back[i].r.statuscode = STATUSCODE_CONNERROR;
+                strcpybuff(back[i].r.msg, "Connect Error");
+                back[i].status = STATUS_READY;
+                back_set_finished(sback, i);
+                continue;
+              }
+            }
             //printf("time check %d\n",((int) (act-back[i].timeout_refresh))-back[i].timeout);
             if (((int) (act - back[i].timeout_refresh)) >= back[i].timeout) {
               hts_log_print(opt, LOG_DEBUG, "connection timed out for %s%s", back[i].url_adr,

src/htsbasenet.h

@@ -146,7 +146,8 @@ typedef enum BackStatusCode {
   STATUSCODE_NON_FATAL = -5,
   STATUSCODE_SSL_HANDSHAKE = -6,
   STATUSCODE_TOO_BIG = -7,
-  STATUSCODE_TEST_OK = -10
+  STATUSCODE_TEST_OK = -10,
+  STATUSCODE_EXCLUDED = -11 /* aborted: MIME excluded by a -mime: filter */
 } BackStatusCode;
 
 /** HTTrack status ('status' member of of 'lien_back') **/

src/htscache.c

@@ -220,6 +220,25 @@ struct cache_back_zip_entry {
 	} \
 } while(0)
 
+/* A cache (new.zip) write failed: storage is gone (disk full / dropped share),
+   so the mirror is doomed too. Abort it via exit_xh, don't crash as assertf
+   did. */
+static void cache_zip_write_failed(httrackp *opt, cache_back *cache,
+                                   const char *what, int zErr) {
+  if (!cache->zipWriteFailed) {
+    cache->zipWriteFailed = HTS_TRUE;
+    if (check_fatal_io_errno()) {
+      hts_log_print(opt, LOG_ERROR,
+                    "Mirror aborted: disk full or filesystem problems");
+    } else {
+      hts_log_print(opt, LOG_ERROR,
+                    "Mirror aborted: cache write failed (%s): %s", what,
+                    hts_get_zerror(zErr));
+    }
+  }
+  opt->state.exit_xh = -1; /* fatal: stop the mirror, exit non-zero */
+}
+
 /* Ajout d'un fichier en cache */
 void cache_add(httrackp * opt, cache_back * cache, const htsblk * r,
                const char *url_adr, const char *url_fil, const char *url_save,
@@ -236,6 +255,10 @@ void cache_add(httrackp * opt, cache_back * cache, const htsblk * r,
   const char *url_save_suffix = url_save;
   int zErr;
 
+  /* already failed and aborting; don't touch the broken stream again */
+  if (cache->zipWriteFailed)
+    return;
+
   // robots.txt hack
   if (url_save == NULL) {
     dataincache = 0;            // testing links
@@ -346,9 +369,8 @@ void cache_add(httrackp * opt, cache_back * cache, const htsblk * r,
                                    */
                                   headers, (uInt) strlen(headers), NULL, 0, NULL,       /* comment */
                                   Z_DEFLATED, Z_DEFAULT_COMPRESSION)) != Z_OK) {
-    int zip_zipOpenNewFileInZip_failed = 0;
+    cache_zip_write_failed(opt, cache, "opening a cache entry", zErr);
-
+    return;
-    assertf(zip_zipOpenNewFileInZip_failed);
   }
 
   /* Write data in cache */
@@ -358,9 +380,8 @@ void cache_add(httrackp * opt, cache_back * cache, const htsblk * r,
         if ((zErr =
              zipWriteInFileInZip((zipFile) cache->zipOutput, r->adr,
                                  (int) r->size)) != Z_OK) {
-          int zip_zipWriteInFileInZip_failed = 0;
+          cache_zip_write_failed(opt, cache, "writing to the cache", zErr);
-
+          return;
-          assertf(zip_zipWriteInFileInZip_failed);
         }
       }
     } else {
@@ -381,9 +402,10 @@ void cache_add(httrackp * opt, cache_back * cache, const htsblk * r,
               if ((zErr =
                    zipWriteInFileInZip((zipFile) cache->zipOutput, buff,
                                        (int) nl)) != Z_OK) {
-                int zip_zipWriteInFileInZip_failed = 0;
+                cache_zip_write_failed(opt, cache, "writing to the cache",
-
+                                       zErr);
-                assertf(zip_zipWriteInFileInZip_failed);
+                fclose(fp);
+                return;
               }
             }
           } while(nl > 0);
@@ -397,16 +419,14 @@ void cache_add(httrackp * opt, cache_back * cache, const htsblk * r,
 
   /* Close */
   if ((zErr = zipCloseFileInZip((zipFile) cache->zipOutput)) != Z_OK) {
-    int zip_zipCloseFileInZip_failed = 0;
+    cache_zip_write_failed(opt, cache, "closing a cache entry", zErr);
-
+    return;
-    assertf(zip_zipCloseFileInZip_failed);
   }
 
   /* Flush */
   if ((zErr = zipFlush((zipFile) cache->zipOutput)) != 0) {
-    int zip_zipFlush_failed = 0;
+    cache_zip_write_failed(opt, cache, "flushing the cache", zErr);
-
+    return;
-    assertf(zip_zipFlush_failed);
   }
 }
 

src/htscache_selftest.c

@@ -47,6 +47,7 @@ Please visit our Website: http://www.httrack.com
 #include "htslib.h"
 #include "htszlib.h"
 
+#include <errno.h>
 #include <stdio.h>
 #include <string.h>
 
@@ -316,6 +317,136 @@ static int disk_fallback_selftest(httrackp *opt) {
   return fail;
 }
 
+typedef struct {
+  size_t budget;  /**< bytes allowed through before writes start failing */
+  int fail_errno; /**< errno set on the failing write (ENOSPC, EIO, ...) */
+  int writes;     /**< zwrite call count, to detect re-entry into the stream */
+} writefail_inject;
+
+/* zwrite that copies until the budget runs out, then fails with inj->fail_errno
+   (the #174/#219 condition). Counts calls so the test can prove a flagged cache
+   never re-enters the stream. */
+static uLong selftest_failing_zwrite(voidpf opaque, voidpf stream,
+                                     const void *buf, uLong size) {
+  writefail_inject *inj = (writefail_inject *) opaque;
+
+  inj->writes++;
+  if (inj->budget >= (size_t) size) {
+    inj->budget -= (size_t) size;
+    return (uLong) fwrite(buf, 1, (size_t) size, (FILE *) stream);
+  }
+  errno = inj->fail_errno;
+  return 0; /* short write -> the minizip op returns an error */
+}
+
+/* Open a ZIP whose writes fail past inj->budget, so cache_add() hits an error.
+ */
+static zipFile selftest_open_failing_zip(const char *path,
+                                         writefail_inject *inj) {
+  zlib_filefunc_def ff;
+
+  fill_fopen_filefunc(&ff); /* real fopen/read/seek/close; ignores opaque */
+  ff.zwrite_file = selftest_failing_zwrite;
+  ff.opaque = inj;
+  return zipOpen2(path, APPEND_STATUS_CREATE, NULL, &ff);
+}
+
+/* Store one octet-stream body into `cache` (all-in-cache, body in the ZIP). */
+static void writefail_store(httrackp *opt, cache_back *cache, const char *fil,
+                            const char *body, size_t body_len) {
+  htsblk r;
+  char locbuf[4];
+  char *bodycopy = malloct(body_len);
+
+  hts_init_htsblk(&r);
+  r.statuscode = 200;
+  r.size = (LLint) body_len;
+  strcpybuff(r.msg, "OK");
+  strcpybuff(r.contenttype, "application/octet-stream");
+  locbuf[0] = '\0';
+  r.location = locbuf;
+  r.is_write = 0;
+  memcpy(bodycopy, body, body_len);
+  r.adr = bodycopy;
+  cache_add(opt, cache, &r, "example.com", fil, "example.com/blob.bin", 1,
+            NULL);
+  freet(bodycopy);
+}
+
+/* #174/#219: a failing cache write used to crash via assertf(); it must instead
+   stop the mirror (exit_xh = -1) without crashing. Assert that, plus the cache
+   is flagged and a sibling write doesn't re-enter the broken stream. */
+int cache_write_failure_selftest(httrackp *opt, const char *dir) {
+  int fail = 0;
+  char path[HTS_URLMAXSIZE];
+  /* incompressible + big, so deflate flushes (and fails) mid-write, before
+   * close */
+  static const size_t body_len = 256 * 1024;
+  char *body = malloct(body_len);
+  int phase;
+
+  gen_body(body, body_len, 1 /* incompressible */);
+  fconcat(path, sizeof(path), dir, "/wfail.zip");
+
+  /* phase 0: fail on the body write, fatal errno (ENOSPC, the disk-full
+     branch). phase 1: fail on the open, non-fatal errno (EIO, dropped-share
+     branch). Both must abort the mirror. */
+  for (phase = 0; phase < 2; phase++) {
+    cache_back cache;
+    writefail_inject inj;
+    int writes_after_fail;
+
+    inj.budget = (phase == 0) ? 4096 : 0;
+    inj.fail_errno = (phase == 0) ? ENOSPC : EIO;
+    inj.writes = 0;
+    memset(&cache, 0, sizeof(cache));
+    cache.type = 1;
+    cache.log = stderr;
+    cache.errlog = stderr;
+    cache.hashtable = coucal_new(0);
+    cache.zipOutput = selftest_open_failing_zip(path, &inj);
+    if (cache.zipOutput == NULL) {
+      fprintf(stderr, "cache-writefail: could not open injected ZIP\n");
+      fail++;
+      continue;
+    }
+
+    opt->state.exit_xh = 0; /* clear; the failing write must set it to -1 */
+    writefail_store(opt, &cache, "/blob.bin", body, body_len);
+    if (!cache.zipWriteFailed) {
+      fprintf(stderr, "cache-writefail: phase %d: write error not caught\n",
+              phase);
+      fail++;
+    }
+    if (opt->state.exit_xh != -1) {
+      fprintf(stderr,
+              "cache-writefail: phase %d: mirror not aborted (exit_xh=%d)\n",
+              phase, opt->state.exit_xh);
+      fail++;
+    }
+
+    /* a flagged cache must no-op a sibling write: no further backend write */
+    writes_after_fail = inj.writes;
+    writefail_store(opt, &cache, "/blob2.bin", body, 16);
+    if (inj.writes != writes_after_fail) {
+      fprintf(stderr,
+              "cache-writefail: phase %d: sibling write re-entered the broken "
+              "stream (%d extra backend writes)\n",
+              phase, inj.writes - writes_after_fail);
+      fail++;
+    }
+
+    if (cache.zipOutput != NULL) {
+      zipClose(cache.zipOutput,
+               NULL); /* best-effort; may fail on the backend */
+      cache.zipOutput = NULL;
+    }
+  }
+
+  freet(body);
+  return fail;
+}
+
 int cache_selftests(httrackp *opt, const char *dir) {
   int failures = 0;
   cache_back cache;

src/htscache_selftest.h

@@ -52,6 +52,10 @@ int cache_selftests(httrackp *opt, const char *dir);
    committed file, never by the test). Returns the failed-check count. */
 int cache_golden_selftest(httrackp *opt, const char *dir, int regen);
 
+/* #174/#219: assert a failing cache write aborts the mirror cleanly instead of
+   crashing. Returns the failed-check count. */
+int cache_write_failure_selftest(httrackp *opt, const char *dir);
+
 #endif
 
 #endif

src/htscore.c

@@ -736,26 +736,39 @@ int httpmirror(char *url1, httrackp * opt) {
     /* OPTIMIZED for fast load */
     if (StringNotEmpty(opt->filelist)) {
       char *filelist_buff = NULL;
-      const size_t filelist_sz = off_t_to_size_t(fsize(StringBuff(opt->filelist)));
+      size_t filelist_sz = 0;
+      const char *filelist_err = NULL; /* failure reason, NULL on success */
+      const off_t fs = fsize(StringBuff(opt->filelist));
 
-      if (filelist_sz != (size_t) -1) {
+      if (fs < 0) {
+        /* fsize() hides the cause; redo stat() for a precise errno (#49) */
+        struct stat st;
+        filelist_err = stat(StringBuff(opt->filelist), &st) != 0
+                           ? strerror(errno)
+                           : "not a regular file";
+      } else if ((filelist_sz = off_t_to_size_t(fs)) == (size_t) -1) {
+        filelist_err = "file too large";
+        filelist_sz = 0;
+      } else {
         FILE *fp = fopen(StringBuff(opt->filelist), "rb");
 
-        if (fp) {
+        if (fp == NULL) {
+          filelist_err = strerror(errno);
+        } else {
           filelist_buff = malloct(filelist_sz + 1);
-          if (filelist_buff) {
+          if (filelist_buff == NULL) {
-            if (fread(filelist_buff, 1, filelist_sz, fp) != filelist_sz) {
+            filelist_err = "out of memory";
-              freet(filelist_buff);
+          } else if (fread(filelist_buff, 1, filelist_sz, fp) != filelist_sz) {
-              filelist_buff = NULL;
+            freet(filelist_buff);
-            } else {
+            filelist_err = "read error";
-              *(filelist_buff + filelist_sz) = '\0';
+          } else {
-            }
+            filelist_buff[filelist_sz] = '\0';
           }
           fclose(fp);
         }
       }
 
-      if (filelist_buff) {
+      if (filelist_buff != NULL) {
         int filelist_ptr = 0;
         int n = 0;
         char BIGSTK line[HTS_URLMAXSIZE * 2];
@@ -780,8 +793,8 @@ int httpmirror(char *url1, httrackp * opt) {
         // Free buffer
         freet(filelist_buff);
       } else {
-        hts_log_print(opt, LOG_ERROR, "Could not include URL list: %s",
+        hts_log_print(opt, LOG_ERROR, "Could not include URL list \"%s\": %s",
-                      StringBuff(opt->filelist));
+                      StringBuff(opt->filelist), filelist_err);
       }
     }
 
@@ -3726,6 +3739,9 @@ HTSEXT_API int copy_htsopt(const httrackp * from, httrackp * to) {
   if (StringNotEmpty(from->user_agent))
     StringCopyS(to->user_agent, from->user_agent);
 
+  if (StringNotEmpty(from->strip_query))
+    StringCopyS(to->strip_query, from->strip_query);
+
   if (from->retry > -1)
     to->retry = from->retry;
 

src/htscore.h

@@ -152,6 +152,15 @@ struct lien_adrfilsave {
   char save[HTS_URLMAXSIZE * 2]; /**< local save path (with directory) */
 };
 
+/** Per-slot connect-fallback bookkeeping (parallel to struct_back.lnk).
+    Tracks which resolved address the slot is currently connecting to so a
+    stuck connect can be retried against the next one. */
+typedef struct hts_connect_fallback {
+  int addr_index;       /**< candidate being connected (0-based) */
+  int addr_count;       /**< resolved addresses; -1 = not yet probed */
+  TStamp connect_start; /**< when the current candidate's connect began */
+} hts_connect_fallback;
+
 /** The download-slot ring: the set of concurrent transfers in flight.
     Allocated/owned by the engine; consumers (status callbacks, the loop)
     read it but do not resize or free it. */
@@ -168,6 +177,7 @@ struct struct_back {
   int count;              /**< number of usable slots (back_max) */
   coucal ready;           /**< index of slots whose transfer completed */
   LLint ready_size_bytes; /**< total bytes buffered in completed slots */
+  hts_connect_fallback *connect_fallback; /**< per-slot, count+1 entries */
 };
 
 typedef struct cache_back_zip_entry cache_back_zip_entry;
@@ -204,6 +214,8 @@ struct cache_back {
   cache_back_zip_entry *zipEntries;
   int zipEntriesOffs;
   int zipEntriesCapa;
+  hts_boolean
+      zipWriteFailed; /**< a cache write failed; stop touching the stream */
 };
 
 #ifndef HTS_DEF_FWSTRUCT_hash_struct
@@ -222,8 +234,12 @@ struct hash_struct {
   coucal adrfil;
   /* former address+path -> link index (renamed/moved entries) */
   coucal former_adrfil;
-  /* scratch buffers reused across lookups (not reentrant) */
+  /* effective urlhack sub-flags: www.==host / // collapse / query-arg sort */
-  int normalized;
+  hts_boolean norm_host;
+  hts_boolean norm_slash;
+  hts_boolean norm_query;
+  /* query-strip keys (not owned); set from opt->strip_query at hash_init */
+  const char *strip_query;
   char normfil[HTS_URLMAXSIZE * 2];
   char normfil2[HTS_URLMAXSIZE * 2];
   char catbuff[CATBUFF_SIZE];
@@ -352,6 +368,22 @@ int fspc(httrackp * opt, FILE * fp, const char *type);
 
 char *next_token(char *p, int flag);
 
+/* Like fil_normalized(), but first drops query keys in STRIP (comma-separated,
+   "*" = all); STRIP NULL/empty behaves exactly like fil_normalized(). */
+char *fil_normalized_filtered(const char *source, char *dest,
+                              const char *strip);
+
+/* As fil_normalized_filtered(), but DO_SLASH/DO_QUERY gate the // collapse and
+   the query-argument sort independently (the urlhack sub-flags). */
+char *fil_normalized_filtered_ex(const char *source, char *dest,
+                                 const char *strip, int do_slash, int do_query);
+
+/* For URL ADR/FIL, return (in DEST) the comma keylist to strip from the
+   '\n'-separated "[pattern=]keys" RULES (patterns matched on host/path via
+   strjoker, last wins); NULL if none match. Feeds fil_normalized_filtered(). */
+const char *hts_query_strip_keys(const char *rules, const char *adr,
+                                 const char *fil, char *dest, size_t destsize);
+
 /* Read a whole file into a freshly malloc'd, NUL-terminated buffer; the caller
    owns it and must release it with freet(). Return NULL on missing/unreadable
    file (readfile_or substitutes defaultdata instead). The byte content is NOT
@@ -372,6 +404,13 @@ void check_rate(TStamp stat_timestart, int maxrate);
 /* Backing (download-slot) scheduler. Operate on the back[] ring (struct_back).
    Not thread-safe; call from the single crawl loop. */
 
+/* True if a connecting slot should give up on the current address and try the
+   next one: a fallback address remains (addr_index+1 < addr_count) and the
+   candidate has been connecting for at least its deadline, min(timeout, an
+   internal cap). elapsed/timeout in seconds. Exposed for the -#D self-test. */
+int back_connect_fallback_due(int addr_index, int addr_count, int elapsed,
+                              int timeout);
+
 /* How many new sockets may be opened now, honoring maxsoc and the maxconn rate
    limit (>=0). _strict ignores reserved-slot headroom; the plain form leaves
    room for naming tests and stops at 0 when the stack is nearly full. */

src/htsdns_selftest.c

@@ -62,7 +62,7 @@ typedef struct mock_host {
   const char *name;
   int gai_err; /* non-zero: getaddrinfo returns this */
   int naddr;
-  mock_addr addr[3];
+  mock_addr addr[6];
   int calls; /* times the backend resolved this host */
 } mock_host;
 
@@ -83,6 +83,17 @@ static mock_host mock_hosts[] = {
      {{AF_INET, {9, 10, 11, 12}},
       {AF_INET6, {0x20, 0x01, 0x0d, 0xb8, [15] = 3}}},
      0},
+    /* more addresses than HTS_MAXADDRNUM: the list must clamp to the cap. */
+    {"many.test",
+     0,
+     6,
+     {{AF_INET, {10, 0, 0, 1}},
+      {AF_INET, {10, 0, 0, 2}},
+      {AF_INET, {10, 0, 0, 3}},
+      {AF_INET, {10, 0, 0, 4}},
+      {AF_INET, {10, 0, 0, 5}},
+      {AF_INET, {10, 0, 0, 6}}},
+     0},
     {"nodns.test", EAI_NONAME, 0, {{0}}, 0},
 };
 
@@ -197,9 +208,9 @@ int dns_selftests(httrackp *opt) {
   CHECK(resolve_family_nocache("v4only.test") == AF_INET);
   CHECK(resolve_family_nocache("v6only.test") == AF_INET6);
 
-  /* Dual-stack: the current resolver keeps only the *first* address. Both
+  /* Dual-stack: the single-address API returns the *first* resolved address.
-     orderings pin that (not a family preference); PR2 (multi-address) widens
+     Both orderings pin selection by position, not a family preference. The
-     it. */
+     multi-address API (resolve_all, below) exposes the whole list. */
   CHECK(resolve_family_nocache("dual.test") == AF_INET6); /* v6 listed first */
   CHECK(resolve_family_nocache("dual4.test") == AF_INET); /* v4 listed first */
 
@@ -240,6 +251,100 @@ int dns_selftests(httrackp *opt) {
     CHECK(mock_find("nodns.test")->calls == 1); /* resolved once, then cached */
   }
 
+  /* Multi-address resolution: count and order are the connect-fallback
+     contract. A dead first address is retried against the next, so both must be
+     exact. */
+  mock_reset_calls();
+  {
+    SOCaddr addrs[HTS_MAXADDRNUM];
+    char ip[64];
+    const char *err = NULL;
+
+    /* dual-stack, in resolver order: [0]=v6, [1]=v4 */
+    CHECK(hts_dns_resolve_all(opt, "dual.test", addrs, HTS_MAXADDRNUM, &err) ==
+          2);
+    CHECK(SOCaddr_sinfamily(addrs[0]) == AF_INET6);
+    CHECK(SOCaddr_sinfamily(addrs[1]) == AF_INET);
+    SOCaddr_inetntoa(ip, sizeof(ip), addrs[1]);
+    CHECK(strcmp(ip, "5.6.7.8") == 0);
+    CHECK(mock_find("dual.test")->calls ==
+          1); /* one backend hit for the list */
+
+    /* single-address host: count 1 */
+    CHECK(hts_dns_resolve_all(opt, "v4only.test", addrs, HTS_MAXADDRNUM,
+                              &err) == 1);
+    SOCaddr_inetntoa(ip, sizeof(ip), addrs[0]);
+    CHECK(strcmp(ip, "1.2.3.4") == 0);
+
+    /* does-not-resolve: count 0 (negative), no addresses */
+    CHECK(hts_dns_resolve_all(opt, "nodns.test", addrs, HTS_MAXADDRNUM, &err) ==
+          0);
+
+    /* more than the cap: the kept list is clamped to HTS_MAXADDRNUM, keeping
+       the FIRST addresses in resolver order (not some other window) */
+    CHECK(hts_dns_resolve_all(opt, "many.test", addrs, HTS_MAXADDRNUM, &err) ==
+          HTS_MAXADDRNUM);
+    SOCaddr_inetntoa(ip, sizeof(ip), addrs[0]);
+    CHECK(strcmp(ip, "10.0.0.1") == 0);
+    SOCaddr_inetntoa(ip, sizeof(ip), addrs[HTS_MAXADDRNUM - 1]);
+    CHECK(strcmp(ip, "10.0.0.4") == 0);
+
+    /* family filter still applies through the list path */
+    IPV6_resolver = 1;
+    CHECK(hts_dns_resolve_all(opt, "dual4.test", addrs, HTS_MAXADDRNUM, &err) ==
+          1);
+    CHECK(SOCaddr_sinfamily(addrs[0]) == AF_INET);
+    IPV6_resolver = 0;
+  }
+
+  /* newhttp_addr() must connect to the addr_index-th address, not always the
+     first: this is what back_connect_next relies on to reach the fallback. */
+  {
+    htsblk r;
+    int count = -1;
+    T_SOC s;
+
+    hts_init_htsblk(&r);
+    s = newhttp_addr(opt, "dual.test", &r, 80, 0, 0, &count);
+    CHECK(count == 2);
+    CHECK(SOCaddr_sinfamily(r.address) == AF_INET6); /* index 0 = v6 */
+    if (s != INVALID_SOCKET)
+      deletesoc(s);
+
+    hts_init_htsblk(&r);
+    count = -1;
+    s = newhttp_addr(opt, "dual.test", &r, 80, 0, 1, &count);
+    CHECK(count == 2);
+    CHECK(SOCaddr_sinfamily(r.address) == AF_INET); /* index 1 = v4 */
+    if (s != INVALID_SOCKET)
+      deletesoc(s);
+
+    /* out-of-range index: no address selected (address stays unset) */
+    hts_init_htsblk(&r);
+    s = newhttp_addr(opt, "dual.test", &r, 80, 0, 2, NULL);
+    CHECK(s == INVALID_SOCKET);
+    if (s != INVALID_SOCKET)
+      deletesoc(s);
+  }
+
+  /* Connect-fallback decision (consumer of the multi-address list): when a
+     stuck connect should abandon the current address for the next one. */
+  {
+    /* no fallback for the last/only candidate, whatever the elapsed time */
+    CHECK(back_connect_fallback_due(0, 1, 9999, 120) == 0);
+    CHECK(back_connect_fallback_due(1, 2, 9999, 120) == 0);
+    CHECK(back_connect_fallback_due(3, 4, 9999, 120) == 0);
+    /* fallback available: wait the per-candidate deadline (cap 10s here) */
+    CHECK(back_connect_fallback_due(0, 2, 9, 120) == 0);
+    CHECK(back_connect_fallback_due(0, 2, 10, 120) == 1);
+    CHECK(back_connect_fallback_due(2, 4, 10, 120) == 1);
+    /* a shorter slot timeout shortens the deadline (min(timeout, cap)) */
+    CHECK(back_connect_fallback_due(0, 2, 4, 5) == 0);
+    CHECK(back_connect_fallback_due(0, 2, 5, 5) == 1);
+    /* no timeout management: never force a fallback */
+    CHECK(back_connect_fallback_due(0, 2, 9999, 0) == 0);
+  }
+
   hts_dns_set_resolver_backend(NULL);
   return failures;
 }

src/htsfilters.c

@@ -76,7 +76,8 @@ int fa_strjoker(int type, char **filters, int nfil, const char *nom, LLint * siz
     }
     if (size)
       sz = *size;
-    if (strjoker(nom, filters[i] + filteroffs, &sz, size_flag)) {       // reconnu
+    /* size unknown (scan time): no size pointer => size tests stay neutral */
+    if (strjoker(nom, filters[i] + filteroffs, size ? &sz : NULL, size_flag)) {
       if (size)
         if (sz != *size)
           sizelimit = sz;

src/htsglobal.h

@@ -408,6 +408,10 @@ typedef int T_SOC;
 /* Buffer size for a printed network address (IPv4 or IPv6, NUL included). */
 #define HTS_MAXADDRLEN 64
 
+/* Max resolved addresses kept per host for connect fallback (dead IPv6 etc.).
+ */
+#define HTS_MAXADDRNUM 4
+
 #ifdef _WIN32
 #else
 #define __cdecl

src/htshash.c

@@ -106,10 +106,10 @@ static coucal_hashkeys key_adrfil_hashes_generic(void *arg,
   const lien_url*const lien = (const lien_url*) value;
   const char *const adr = !former ? lien->adr : lien->former_adr;
   const char *const fil = !former ? lien->fil : lien->former_fil;
-  const char *const adr_norm = adr != NULL ? 
+  const char *const adr_norm =
-    ( hash->normalized  ? jump_normalized_const(adr)
+      adr != NULL ? (hash->norm_host ? jump_normalized_const(adr)
-                        : jump_identification_const(adr) )
+                                     : jump_identification_const(adr))
-    : NULL;
+                  : NULL;
 
   // copy address
   assertf(adr_norm != NULL);
@@ -117,10 +117,18 @@ static coucal_hashkeys key_adrfil_hashes_generic(void *arg,
 
   // copy link
   assertf(fil != NULL);
-  if (hash->normalized) {
+  {
-    fil_normalized(fil, &hash->normfil[strlen(hash->normfil)]);
+    /* resolve the per-URL strip keys; strip applies even when urlhack is off */
-  } else {
+    char BIGSTK keybuf[HTS_URLMAXSIZE];
-    strcpy(&hash->normfil[strlen(hash->normfil)], fil);
+    const char *const keys = hts_query_strip_keys(hash->strip_query, adr, fil,
+                                                  keybuf, sizeof(keybuf));
+
+    if (hash->norm_slash || hash->norm_query || keys != NULL) {
+      fil_normalized_filtered_ex(fil, &hash->normfil[strlen(hash->normfil)],
+                                 keys, hash->norm_slash, hash->norm_query);
+    } else {
+      strcpy(&hash->normfil[strlen(hash->normfil)], fil);
+    }
   }
 
   // hash
@@ -132,8 +140,7 @@ static int key_adrfil_equals_generic(void *arg,
                                      coucal_key_const a_,
                                      coucal_key_const b_, 
                                      const int former) {
-  hash_struct *const hash = (hash_struct*) arg;
+  hash_struct *const hash = (hash_struct *) arg;
-  const int normalized = hash->normalized;
   const lien_url*const a = (const lien_url*) a_;
   const lien_url*const b = (const lien_url*) b_;
   const char *const a_adr = !former ? a->adr : a->former_adr;
@@ -150,10 +157,10 @@ static int key_adrfil_equals_generic(void *arg,
   assertf(b_fil != NULL);
 
   // skip scheme and authentication to the domain (possibly without www.)
-  ja = normalized
+  ja = hash->norm_host ? jump_normalized_const(a_adr)
-    ? jump_normalized_const(a_adr) : jump_identification_const(a_adr);
+                       : jump_identification_const(a_adr);
-  jb = normalized
+  jb = hash->norm_host ? jump_normalized_const(b_adr)
-    ? jump_normalized_const(b_adr) : jump_identification_const(b_adr);
+                       : jump_identification_const(b_adr);
   assertf(ja != NULL);
   assertf(jb != NULL);
   if (strcasecmp(ja, jb) != 0) {
@@ -161,12 +168,23 @@ static int key_adrfil_equals_generic(void *arg,
   }
 
   // now compare pathes
-  if (normalized) {
+  {
-    fil_normalized(a_fil, hash->normfil);
+    char BIGSTK ka[HTS_URLMAXSIZE], kb[HTS_URLMAXSIZE];
-    fil_normalized(b_fil, hash->normfil2);
+    const char *const keysa =
-    return strcmp(hash->normfil, hash->normfil2) == 0;
+        hts_query_strip_keys(hash->strip_query, a_adr, a_fil, ka, sizeof(ka));
-  } else {
+    const char *const keysb =
-    return strcmp(a_fil, b_fil) == 0;
+        hts_query_strip_keys(hash->strip_query, b_adr, b_fil, kb, sizeof(kb));
+
+    if (hash->norm_slash || hash->norm_query || keysa != NULL ||
+        keysb != NULL) {
+      fil_normalized_filtered_ex(a_fil, hash->normfil, keysa, hash->norm_slash,
+                                 hash->norm_query);
+      fil_normalized_filtered_ex(b_fil, hash->normfil2, keysb, hash->norm_slash,
+                                 hash->norm_query);
+      return strcmp(hash->normfil, hash->normfil2) == 0;
+    } else {
+      return strcmp(a_fil, b_fil) == 0;
+    }
   }
 }
 
@@ -222,11 +240,17 @@ static int key_former_adrfil_equals(void *arg,
   return key_adrfil_equals_generic(arg, a, b, 1);
 }
 
-void hash_init(httrackp *opt, hash_struct * hash, int normalized) {
+void hash_init(httrackp *opt, hash_struct *hash, hts_boolean normalized) {
   hash->sav = coucal_new(0);
   hash->adrfil = coucal_new(0);
   hash->former_adrfil = coucal_new(0);
-  hash->normalized = normalized;
+  /* urlhack is the umbrella; per-feature negatives opt out of each part */
+  hash->norm_host = normalized && !opt->no_www_dedup;
+  hash->norm_slash = normalized && !opt->no_slash_dedup;
+  hash->norm_query = normalized && !opt->no_query_dedup;
+  /* snapshot the query-strip list (not owned; valid for the hash lifetime) */
+  hash->strip_query =
+      StringNotEmpty(opt->strip_query) ? StringBuff(opt->strip_query) : NULL;
 
   hts_set_hash_handler(hash->sav, opt);
   hts_set_hash_handler(hash->adrfil, opt);
@@ -282,6 +306,26 @@ void hash_free(hash_struct *hash) {
   }
 }
 
+/* Test helper: do the two URLs dedupe to the same key under opt's urlhack
+   flags? Exercises the live hash compare (norm_host/slash/query resolution). */
+hts_boolean hash_url_equals(httrackp *opt, const char *adra, const char *fila,
+                            const char *adrb, const char *filb) {
+  hash_struct hash;
+  lien_url la, lb;
+  hts_boolean eq;
+
+  memset(&la, 0, sizeof(la));
+  memset(&lb, 0, sizeof(lb));
+  la.adr = key_duphandler(NULL, adra);
+  la.fil = key_duphandler(NULL, fila);
+  lb.adr = key_duphandler(NULL, adrb);
+  lb.fil = key_duphandler(NULL, filb);
+  hash_init(opt, &hash, opt->urlhack);
+  eq = key_adrfil_equals(&hash, &la, &lb);
+  hash_free(&hash);
+  return eq;
+}
+
 // retour: position ou -1 si non trouvé
 int hash_read(const hash_struct * hash, const char *nom1, const char *nom2,
               hash_struct_type type) {

src/htshash.h

@@ -51,8 +51,12 @@ typedef enum hash_struct_type {
 } hash_struct_type;
 
 // tables de hachage
-void hash_init(httrackp *opt, hash_struct *hash, int normalized);
+void hash_init(httrackp *opt, hash_struct *hash, hts_boolean normalized);
 void hash_free(hash_struct *hash);
+/* Test helper: HTS_TRUE if the two URLs dedupe together under opt's urlhack
+   flags. */
+hts_boolean hash_url_equals(httrackp *opt, const char *adra, const char *fila,
+                            const char *adrb, const char *filb);
 int hash_read(const hash_struct * hash, const char *nom1, const char *nom2,
               hash_struct_type type);
 void hash_write(hash_struct * hash, size_t lpos);

src/htshelp.c

@@ -563,6 +563,7 @@ void help(const char *app, int more) {
     (" %x  do not include any password for external password protected websites (%x0 include)");
   infomsg
     (" %q *include query string for local files (useless, for information purpose only) (%q0 don't include)");
+  infomsg(" %g  strip query keys for dedup ([host/pattern=]key1,key2,...)");
   infomsg
     ("  o *generate output html file in case of error (404..) (o0 don't generate)");
   infomsg("  X *purge old files after update (X0 keep delete)");
@@ -587,6 +588,9 @@ void help(const char *app, int more) {
     (" %s  update hacks: various hacks to limit re-transfers when updating (identical size, bogus response..)");
   infomsg
     (" %u  url hacks: various hacks to limit duplicate URLs (strip //, www.foo.com==foo.com..)");
+  infomsg("     opt out of one url-hack part: --keep-www-prefix "
+          "(www.foo.com<>foo.com), --keep-double-slashes (//), "
+          "--keep-query-order (?b&a)");
   infomsg
     (" %A  assume that a type (cgi,asp..) is always linked with a mime type (-%A php3,cgi=text/html;dat,bin=application/x-zip)");
   infomsg("     shortcut: '--assume standard' is equivalent to -%A "
@@ -646,9 +650,7 @@ void help(const char *app, int more) {
   infomsg("");
   infomsg("Guru options: (do NOT use if possible)");
   infomsg(" #X *use optimized engine (limited memory boundary checks)");
-  infomsg(" #0  filter test (-#0 '*.gif' 'www.bar.com/foo.gif')");
+  infomsg(" #test  list engine self-tests (run one with -#test=NAME [args])");
-  infomsg(" #1  simplify test (-#1 ./foo/bar/../foobar)");
-  infomsg(" #2  type test (-#2 /foo/bar.php)");
   infomsg(" #C  cache list (-#C '*.com/spider*.gif'");
   infomsg(" #R  cache repair (damaged cache)");
   infomsg(" #d  debug parser");

src/htslib.c

@@ -2297,14 +2297,27 @@ htsblk http_test(httrackp * opt, const char *adr, const char *fil, char *loc) {
 // peut ouvrir avec des connect() non bloquants: waitconnect=0/1
 T_SOC newhttp(httrackp * opt, const char *_iadr, htsblk * retour, int port,
               int waitconnect) {
+  return newhttp_addr(opt, _iadr, retour, port, waitconnect, 0, NULL);
+}
+
+T_SOC newhttp_addr(httrackp *opt, const char *_iadr, htsblk *retour, int port,
+                   int waitconnect, int addr_index, int *addr_count) {
   T_SOC soc;                    // descipteur de la socket
 
+  if (addr_count != NULL) {
+    *addr_count = 0;
+  }
+
   if (strcmp(_iadr, "file://") != 0) {  /* non fichier */
     SOCaddr server;
+    SOCaddr addrs[HTS_MAXADDRNUM];
+    int naddr;
     const char *error = "unknown error";
 
     // tester un éventuel id:pass et virer id:pass@ si détecté
     const char *const iadr = jump_identification_const(_iadr);
+    const char *resolve_host = iadr;
+    char BIGSTK iadr2[HTS_URLMAXSIZE * 2];
 
     SOCaddr_clear(server);
 
@@ -2326,7 +2339,6 @@ T_SOC newhttp(httrackp * opt, const char *_iadr, htsblk * retour, int port,
 #endif
 
       if (a != NULL) {
-        char BIGSTK iadr2[HTS_URLMAXSIZE * 2];
         int i = -1;
 
         iadr2[0] = '\0';
@@ -2337,18 +2349,19 @@ T_SOC newhttp(httrackp * opt, const char *_iadr, htsblk * retour, int port,
 
         // adresse véritable (sans :xx)
         strncatbuff(iadr2, iadr, (int) (a - iadr));
-
+        resolve_host = iadr2;
-        // adresse sans le :xx
-        hts_dns_resolve2(opt, iadr2, &server, &error);
-
-      } else {
-
-        // adresse normale (port par défaut par la suite)
-        hts_dns_resolve2(opt, iadr, &server, &error);
       }
+    }
 
-    } else {                    // port défini
+    // resolve the full address list and pick the requested candidate; the
-      hts_dns_resolve2(opt, iadr, &server, &error);
+    // scheduler retries the next index when a connect fails (dead IPv6 etc.)
+    naddr =
+        hts_dns_resolve_all(opt, resolve_host, addrs, HTS_MAXADDRNUM, &error);
+    if (addr_count != NULL) {
+      *addr_count = naddr;
+    }
+    if (addr_index >= 0 && addr_index < naddr) {
+      SOCaddr_copy_SOCaddr(server, addrs[addr_index]);
     }
 
     if (!SOCaddr_is_valid(server)) {
@@ -3597,7 +3610,10 @@ static int sortNormFnc(const void *a_, const void *b_) {
   return strcmp(*a + 1, *b + 1);
 }
 
-HTSEXT_API char *fil_normalized(const char *source, char *dest) {
+/* Path normalizer core: optionally collapse redundant '//' (DO_SLASH) and/or
+   sort query arguments (DO_QUERY) so equivalent URLs dedupe. */
+static char *fil_normalized_ex(const char *source, char *dest, int do_slash,
+                               int do_query) {
   char lastc = 0;
   int gotquery = 0;
   int ampargs = 0;
@@ -3607,8 +3623,8 @@ HTSEXT_API char *fil_normalized(const char *source, char *dest) {
   for(i = j = 0; source[i] != '\0'; i++) {
     if (!gotquery && source[i] == '?')
       gotquery = ampargs = 1;
-    if ((!gotquery && lastc == '/' && source[i] == '/') // foo//bar -> foo/bar
+    if (do_slash && !gotquery && lastc == '/' && source[i] == '/') {
-      ) {
+      // foo//bar -> foo/bar
     } else {
       if (gotquery && source[i] == '&') {
         ampargs++;
@@ -3620,7 +3636,7 @@ HTSEXT_API char *fil_normalized(const char *source, char *dest) {
   dest[j++] = '\0';
 
   /* Sort arguments (&foo=1&bar=2 == &bar=2&foo=1) */
-  if (ampargs > 1) {
+  if (do_query && ampargs > 1) {
     char **amps = malloct(ampargs * sizeof(char *));
     char *copyBuff = NULL;
     size_t qLen = 0;
@@ -3668,6 +3684,153 @@ HTSEXT_API char *fil_normalized(const char *source, char *dest) {
   return dest;
 }
 
+HTSEXT_API char *fil_normalized(const char *source, char *dest) {
+  return fil_normalized_ex(source, dest, 1, 1);
+}
+
+/* Is query key ARG[0..keylen) in the comma-separated STRIP list? "*" = all;
+   case-sensitive, space-trimmed tokens. */
+static int hts_query_key_stripped(const char *arg, size_t keylen,
+                                  const char *strip) {
+  const char *p = strip;
+
+  while (*p != '\0') {
+    const char *start = p;
+    size_t toklen;
+
+    while (*p != '\0' && *p != ',')
+      p++;
+    toklen = (size_t) (p - start);
+    while (toklen > 0 && *start == ' ') {
+      start++;
+      toklen--;
+    }
+    while (toklen > 0 && start[toklen - 1] == ' ')
+      toklen--;
+    if (toklen == 1 && start[0] == '*')
+      return 1;
+    if (toklen == keylen && strncmp(start, arg, keylen) == 0)
+      return 1;
+    if (*p == ',')
+      p++;
+  }
+  return 0;
+}
+
+/* see htscore.h */
+char *fil_normalized_filtered_ex(const char *source, char *dest,
+                                 const char *strip, int do_slash,
+                                 int do_query) {
+  const char *query;
+  char BIGSTK tmp[HTS_URLMAXSIZE * 2];
+  htsbuff cb;
+  int wrote = 0;
+
+  /* No strip list, or no query: plain normalization. */
+  if (strip == NULL || *strip == '\0' ||
+      (query = strchr(source, '?')) == NULL) {
+    return fil_normalized_ex(source, dest, do_slash, do_query);
+  }
+
+  /* Copy the path, re-emit kept query args, let fil_normalized() sort. Walk
+     every field incl. empty/trailing ("a&","?&&") so the result is a fixpoint
+     (the read re-normalizes it; a dropped empty arg would miss dedup). */
+  cb = htsbuff_ptr(tmp, sizeof(tmp));
+  htsbuff_catn(&cb, source, (size_t) (query - source));
+  for (query++;;) {
+    const char *const arg = query;
+    const char *eq = NULL;
+    size_t keylen, arglen;
+
+    while (*query != '\0' && *query != '&') {
+      if (eq == NULL && *query == '=')
+        eq = query;
+      query++;
+    }
+    arglen = (size_t) (query - arg);
+    keylen = eq != NULL ? (size_t) (eq - arg) : arglen;
+    if (!hts_query_key_stripped(arg, keylen, strip)) {
+      htsbuff_catc(&cb, wrote ? '&' : '?');
+      htsbuff_catn(&cb, arg, arglen);
+      wrote = 1;
+    }
+    if (*query == '\0')
+      break;
+    query++;
+  }
+  return fil_normalized_ex(tmp, dest, do_slash, do_query);
+}
+
+/* see htscore.h */
+char *fil_normalized_filtered(const char *source, char *dest,
+                              const char *strip) {
+  return fil_normalized_filtered_ex(source, dest, strip, 1, 1);
+}
+
+/* see htscore.h */
+const char *hts_query_strip_keys(const char *rules, const char *adr,
+                                 const char *fil, char *dest, size_t destsize) {
+  const char *p, *q;
+  const char *result = NULL;
+  char BIGSTK url[HTS_URLMAXSIZE * 2];
+
+  if (rules == NULL || *rules == '\0' || destsize == 0)
+    return NULL;
+
+  /* Match string = normalized host/path, query removed. jump_normalized_const
+     collapses www+scheme/auth so read and write (double-normalized) agree;
+     query excluded keeps the decision on host/path only. */
+  url[0] = '\0';
+  strcatbuff(url, jump_normalized_const(adr));
+  if (fil[0] != '/')
+    strcatbuff(url, "/");
+  q = strchr(fil, '?');
+  if (q != NULL)
+    strncatbuff(url, fil, (int) (q - fil));
+  else
+    strcatbuff(url, fil);
+
+  /* Walk the '\n' entries; last match wins (like the +/- filter eval). Each is
+     "pattern=keys"; no '=' is the bare form, pattern "*". */
+  for (p = rules; *p != '\0';) {
+    const char *const line = p;
+    const char *eol, *eq, *keys;
+    char BIGSTK pat[HTS_URLMAXSIZE * 2];
+
+    while (*p != '\0' && *p != '\n')
+      p++;
+    eol = p;
+    if (*p == '\n')
+      p++;
+    if (eol == line)
+      continue;
+    eq = memchr(line, '=', (size_t) (eol - line));
+    if (eq != NULL) {
+      size_t patlen = (size_t) (eq - line);
+
+      if (patlen >= sizeof(pat))
+        patlen = sizeof(pat) - 1;
+      memcpy(pat, line, patlen);
+      pat[patlen] = '\0';
+      keys = eq + 1;
+    } else {
+      pat[0] = '*';
+      pat[1] = '\0';
+      keys = line;
+    }
+    if (strjoker(url, pat, NULL, NULL) != NULL) {
+      size_t klen = (size_t) (eol - keys);
+
+      if (klen >= destsize)
+        klen = destsize - 1;
+      memcpy(dest, keys, klen);
+      dest[klen] = '\0';
+      result = dest;
+    }
+  }
+  return result;
+}
+
 #define endwith(a) ( (len >= (sizeof(a)-1)) ? ( strncmp(dest, a+len-(sizeof(a)-1), sizeof(a)-1) == 0 ) : 0 );
 HTSEXT_API char *adr_normalized_sized(const char *source, char *dest,
                                       size_t destsize) {
@@ -4164,9 +4327,10 @@ HTSEXT_API hts_boolean get_httptype_sized(httrackp *opt, char *s, size_t ssize,
     /* Check html -> text/html */
     const char *a = fil + strlen(fil) - 1;
 
-    while((*a != '.') && (*a != '/') && (a > fil))
+    /* a < fil when fil is empty: bound before dereferencing */
+    while ((a > fil) && (*a != '.') && (*a != '/'))
       a--;
-    if (*a == '.' && strlen(a) < 32) {
+    if (a >= fil && *a == '.' && strlen(a) < 32) {
       int j = 0;
 
       a++;
@@ -4753,66 +4917,53 @@ int hts_read(htsblk * r, char *buff, int size) {
 // -- Gestion cache DNS --
 // 'RX98
 
-// 'capsule' contenant uniquement le cache
+// Free a DNS cache record (coucal value handler).
-t_dnscache *hts_cache(httrackp * opt) {
+static void hts_cache_value_free(coucal_opaque arg, coucal_value value) {
+  void *record = value.ptr;
+
+  (void) arg;
+  freet(record);
+}
+
+// opt's DNS cache hashtable, created on first use. Records (t_dnscache*) are
+// owned by the table and freed by hts_cache_value_free on coucal_delete.
+coucal hts_cache(httrackp *opt) {
   assertf(opt != NULL);
   if (opt->state.dns_cache == NULL) {
-    opt->state.dns_cache = (t_dnscache *) malloct(sizeof(t_dnscache));
+    coucal cache = coucal_new(0);
-    memset(opt->state.dns_cache, 0, sizeof(t_dnscache));
+
+    coucal_set_name(cache, "dns_cache");
+    coucal_value_set_value_handler(cache, hts_cache_value_free, NULL);
+    opt->state.dns_cache = cache;
   }
   assertf(opt->state.dns_cache != NULL);
-  /* first entry is NULL */
-  assertf(opt->state.dns_cache->iadr == NULL);
   return opt->state.dns_cache;
 }
 
-// Free DNS cache.
+// MUST BE LOCKED (coucal is not internally serialized vs FTP/web threads)
-void hts_cache_free(t_dnscache *const root) {
+// Look up iadr in the DNS cache, filling out[0..min(count,max)-1].
-  if (root != NULL) {
+// Returns: -1 not yet tested; 0 negative-cached (not in DNS); >0 address count.
-    t_dnscache *cache;
+static int hts_ghbn_all(coucal cache, const char *const iadr,
-    for(cache = root; cache != NULL; ) {
+                        SOCaddr *const out, const int max) {
-      t_dnscache *const next = cache->next;
+  void *ptr;
-      cache->next = NULL;
-      freet(cache);
-      cache = next;
-    }
-  }
-}
 
-// lock le cache dns pour tout opération d'ajout
+  assertf(out != NULL);
-// plus prudent quand plusieurs threads peuvent écrire dedans..
-// -1: status? 0: libérer 1:locker
-
-// MUST BE LOCKED
-// routine pour le cache - retour optionnel à donner à chaque fois
-// NULL: nom non encore testé dans le cache
-// si h_length==0 alors le nom n'existe pas dans le dns
-static SOCaddr* hts_ghbn(const t_dnscache *cache, const char *const iadr, SOCaddr *const addr) {
-  assertf(addr != NULL);
   assertf(iadr != NULL);
   if (*iadr == '\0') {
-    return NULL;
+    return -1;
   }
-  /* first entry is empty */
+  if (coucal_read_pvoid(cache, iadr, &ptr)) { // ok trouvé
-  if (cache->iadr == NULL) {
+    const t_dnscache *const record = (const t_dnscache *) ptr;
-    cache = cache->next;
+    int i;
-  }
+
-  for(; cache != NULL; cache = cache->next) {
+    assertf(record->host_count <= HTS_MAXADDRNUM);
-    assertf(cache != NULL);
+    for (i = 0; i < record->host_count && i < max; i++) {
-    assertf(cache->iadr != NULL);
+      assertf(record->host_length[i] <= sizeof(record->host_addr[i]));
-    assertf(cache->iadr == (const char*) cache + sizeof(t_dnscache));
+      SOCaddr_copyaddr2(out[i], record->host_addr[i], record->host_length[i]);
-    if (strcmp(cache->iadr, iadr) == 0) {       // ok trouvé
-      if (cache->host_length != 0) {     // entrée valide
-        assertf(cache->host_length <= sizeof(cache->host_addr));
-        SOCaddr_copyaddr2(*addr, cache->host_addr, cache->host_length);
-        return addr;
-      } else {                  // erreur dans le dns, déja vérifié
-        SOCaddr_clear(*addr);
-        return addr;
-      }
     }
+    return record->host_count;
   }
-  return NULL;
+  return -1;
 }
 
 #if HTS_INET6 != 0
@@ -4826,84 +4977,239 @@ static const hts_resolver_backend *hts_resolver = &hts_resolver_libc;
 void hts_dns_set_resolver_backend(const hts_resolver_backend *backend) {
   hts_resolver = (backend != NULL) ? backend : &hts_resolver_libc;
 }
-#endif
 
-static SOCaddr *hts_dns_resolve_nocache2_(const char *const hostname,
+/* Debug/test hook: HTTRACK_DEBUG_RESOLVE="host:ip[,ip...]" pins the resolution
-                                          SOCaddr *const addr,
+   of `host` to the listed addresses (curl --resolve style), so the connect
-                                          const char **error) {
+   fallback can be exercised deterministically (a dead address first, a live one
-  {
+   next). Any other host resolves normally. Below: an addrinfo backend that owns
-#if HTS_INET6==0
+   its chain (its own freeaddrinfo), so a synthesized and a delegated result
-    /* IPv4 resolver */
+   free the same way. */
-    struct hostent *const hp = gethostbyname(hostname);
 
-    if (hp != NULL) {
+/* Deep-copy a libc addrinfo chain into our own allocations. */
-      SOCaddr_copyaddr2(addr, hp->h_addr_list[0], hp->h_length);
+static struct addrinfo *resolver_dup_chain(const struct addrinfo *src) {
-      return SOCaddr_is_valid(addr) ? &addr : NULL;
+  struct addrinfo *head = NULL, *tail = NULL;
-    } else {
-      SOCaddr_clear(*addr);
-    }
-#else
-    /* IPv6 resolver */
-    struct addrinfo *res = NULL;
-    struct addrinfo hints;
-    int gerr;
 
-    SOCaddr_clear(*addr);
+  for (; src != NULL; src = src->ai_next) {
-    memset(&hints, 0, sizeof(hints));
+    struct addrinfo *const ai = calloct(1, sizeof(*ai));
-    if (IPV6_resolver == 1)     // V4 only (for bogus V6 entries)
+
-      hints.ai_family = PF_INET;
+    ai->ai_family = src->ai_family;
-    else if (IPV6_resolver == 2)        // V6 only (for testing V6 only)
+    ai->ai_socktype = src->ai_socktype;
-      hints.ai_family = PF_INET6;
+    ai->ai_protocol = src->ai_protocol;
-    else                        // V4 + V6
+    ai->ai_addrlen = src->ai_addrlen;
-      hints.ai_family = PF_UNSPEC;
+    ai->ai_addr = malloct(src->ai_addrlen);
-    hints.ai_socktype = SOCK_STREAM;
+    memcpy(ai->ai_addr, src->ai_addr, src->ai_addrlen);
-    hints.ai_protocol = IPPROTO_TCP;
+    if (head == NULL)
-    if ((gerr = hts_resolver->getaddrinfo(hostname, NULL, &hints, &res)) == 0) {
+      head = ai;
-      if (res != NULL) {
+    else
-        if (res->ai_addr != NULL && res->ai_addrlen != 0) {
+      tail->ai_next = ai;
-          SOCaddr_copyaddr2(*addr, res->ai_addr, res->ai_addrlen);
+    tail = ai;
-        }
-      }
-    } else {
-      if (error != NULL) {
-        *error = gai_strerror(gerr);
-      }
-    }
-    if (res) {
-      hts_resolver->freeaddrinfo(res);
-    }
-#endif
   }
-
+  return head;
-  return SOCaddr_is_valid(*addr) ? addr : NULL;
 }
 
-HTSEXT_API SOCaddr* hts_dns_resolve_nocache2(const char *const hostname, 
+/* Build one addrinfo node from an IPv4/IPv6 literal, or NULL if it does not
-                                     SOCaddr *const addr, const char **error) {
+   parse or is filtered out by want_family (AF_INET/AF_INET6/PF_UNSPEC). */
-  /* Protection */
+static struct addrinfo *resolver_make_ai(const char *ip, int want_family) {
-  if (!strnotempty(hostname)) {
+  struct addrinfo *ai;
-    return NULL;
+
+  if (strchr(ip, ':') != NULL) { // IPv6 literal
+    struct sockaddr_in6 sa6;
+
+    if (want_family != PF_UNSPEC && want_family != AF_INET6)
+      return NULL;
+    memset(&sa6, 0, sizeof(sa6));
+    if (inet_pton(AF_INET6, ip, &sa6.sin6_addr) != 1)
+      return NULL;
+    sa6.sin6_family = AF_INET6;
+    ai = calloct(1, sizeof(*ai));
+    ai->ai_family = AF_INET6;
+    ai->ai_addrlen = sizeof(sa6);
+    ai->ai_addr = malloct(sizeof(sa6));
+    memcpy(ai->ai_addr, &sa6, sizeof(sa6));
+  } else { // IPv4 literal
+    struct sockaddr_in sa;
+
+    if (want_family != PF_UNSPEC && want_family != AF_INET)
+      return NULL;
+    memset(&sa, 0, sizeof(sa));
+    if (inet_pton(AF_INET, ip, &sa.sin_addr) != 1)
+      return NULL;
+    sa.sin_family = AF_INET;
+    ai = calloct(1, sizeof(*ai));
+    ai->ai_family = AF_INET;
+    ai->ai_addrlen = sizeof(sa);
+    ai->ai_addr = malloct(sizeof(sa));
+    memcpy(ai->ai_addr, &sa, sizeof(sa));
+  }
+  return ai;
+}
+
+static void override_freeaddrinfo(struct addrinfo *res) {
+  while (res != NULL) {
+    struct addrinfo *const next = res->ai_next;
+
+    freet(res->ai_addr);
+    freet(res);
+    res = next;
+  }
+}
+
+static int override_getaddrinfo(const char *node, const char *service,
+                                const struct addrinfo *hints,
+                                struct addrinfo **res) {
+  const char *const spec = getenv("HTTRACK_DEBUG_RESOLVE");
+  const int want = (hints != NULL) ? hints->ai_family : PF_UNSPEC;
+  const char *colon;
+
+  *res = NULL;
+  if (spec != NULL && node != NULL && (colon = strchr(spec, ':')) != NULL &&
+      (size_t) (colon - spec) == strlen(node) &&
+      strncmp(spec, node, colon - spec) == 0) {
+    struct addrinfo *head = NULL, *tail = NULL;
+    char buf[256];
+    char *p;
+
+    buf[0] = '\0';
+    strncatbuff(buf, colon + 1, sizeof(buf) - 1);
+    for (p = strtok(buf, ","); p != NULL; p = strtok(NULL, ",")) {
+      struct addrinfo *const ai = resolver_make_ai(p, want);
+
+      if (ai != NULL) {
+        if (head == NULL)
+          head = ai;
+        else
+          tail->ai_next = ai;
+        tail = ai;
+      }
+    }
+    if (head == NULL)
+      return EAI_NONAME;
+    *res = head;
+    return 0;
   }
 
-  /*
+  /* not overridden: delegate to libc, copying into our owned format */
-     Strip [] if any : [3ffe:b80:1234:1::1] 
+  {
-     The resolver doesn't seem to handle IP6 addresses in brackets
+    struct addrinfo *sys = NULL;
-   */
+    int gerr = getaddrinfo(node, service, hints, &sys);
+
+    if (gerr != 0)
+      return gerr;
+    *res = resolver_dup_chain(sys);
+    freeaddrinfo(sys);
+    return 0;
+  }
+}
+
+static const hts_resolver_backend hts_resolver_override = {
+    override_getaddrinfo, override_freeaddrinfo};
+
+/* Install the env override once, unless a backend was already set (self-test).
+ */
+static void hts_resolver_check_env(void) {
+  static int checked = 0;
+
+  if (!checked) {
+    checked = 1;
+    if (hts_resolver == &hts_resolver_libc &&
+        getenv("HTTRACK_DEBUG_RESOLVE") != NULL) {
+      hts_resolver = &hts_resolver_override;
+    }
+  }
+}
+#endif
+
+// Resolve hostname into up to max addresses (resolver/RFC 6724 order), no
+// cache. Returns the count copied into out[0..count-1]; 0 = does not resolve.
+static int hts_dns_resolve_nocache_list_(const char *const hostname,
+                                         SOCaddr *const out, const int max,
+                                         const char **error) {
+  int count = 0;
+
+#if HTS_INET6==0
+  /* IPv4 resolver */
+  struct hostent *const hp = gethostbyname(hostname);
+
+  if (hp != NULL) {
+    char **h;
+
+    for (h = hp->h_addr_list; count < max && h != NULL && *h != NULL; h++) {
+      SOCaddr_clear(out[count]);
+      SOCaddr_copyaddr2(out[count], *h, hp->h_length);
+      if (SOCaddr_is_valid(out[count]))
+        count++;
+    }
+  }
+#else
+  /* IPv6 resolver */
+  struct addrinfo *res = NULL, *cur;
+  struct addrinfo hints;
+  int gerr;
+
+  hts_resolver_check_env();
+  memset(&hints, 0, sizeof(hints));
+  if (IPV6_resolver == 1) // V4 only (for bogus V6 entries)
+    hints.ai_family = PF_INET;
+  else if (IPV6_resolver == 2) // V6 only (for testing V6 only)
+    hints.ai_family = PF_INET6;
+  else // V4 + V6
+    hints.ai_family = PF_UNSPEC;
+  hints.ai_socktype = SOCK_STREAM;
+  hints.ai_protocol = IPPROTO_TCP;
+  if ((gerr = hts_resolver->getaddrinfo(hostname, NULL, &hints, &res)) == 0) {
+    for (cur = res; cur != NULL && count < max; cur = cur->ai_next) {
+      if (cur->ai_addr != NULL && cur->ai_addrlen != 0) {
+        SOCaddr_clear(out[count]);
+        SOCaddr_copyaddr2(out[count], cur->ai_addr, cur->ai_addrlen);
+        if (SOCaddr_is_valid(out[count]))
+          count++;
+      }
+    }
+  } else if (error != NULL) {
+    *error = gai_strerror(gerr);
+  }
+  if (res) {
+    hts_resolver->freeaddrinfo(res);
+  }
+#endif
+
+  return count;
+}
+
+// Strip [] around a literal IPv6 ([3ffe:b80:1234:1::1]) the resolver won't
+// take, then resolve into a list. Returns the count.
+static int hts_dns_resolve_nocache_list(const char *const hostname,
+                                        SOCaddr *const out, const int max,
+                                        const char **error) {
+  if (!strnotempty(hostname) || max <= 0) {
+    return 0;
+  }
   if ((hostname[0] == '[') && (hostname[strlen(hostname) - 1] == ']')) {
-    SOCaddr *ret;
     size_t size = strlen(hostname);
     char *copy = malloct(size + 1);
+    int count;
+
     assertf(copy != NULL);
     copy[0] = '\0';
     strncat(copy, hostname + 1, size - 2);
-    ret =  hts_dns_resolve_nocache2_(copy, addr, error);
+    count = hts_dns_resolve_nocache_list_(copy, out, max, error);
     freet(copy);
-    return ret;
+    return count;
   } else {
-    return hts_dns_resolve_nocache2_(hostname, addr, error);
+    return hts_dns_resolve_nocache_list_(hostname, out, max, error);
   }
 }
 
+HTSEXT_API SOCaddr *hts_dns_resolve_nocache2(const char *const hostname,
+                                             SOCaddr *const addr,
+                                             const char **error) {
+  SOCaddr_clear(*addr);
+  if (hts_dns_resolve_nocache_list(hostname, addr, 1, error) > 0) {
+    return SOCaddr_is_valid(*addr) ? addr : NULL;
+  }
+  return NULL;
+}
+
 HTSEXT_API SOCaddr* hts_dns_resolve_nocache(const char *const hostname, SOCaddr *const addr) {
   return hts_dns_resolve_nocache2(hostname, addr, NULL);
 }
@@ -4914,16 +5220,18 @@ HTSEXT_API int check_hostname_dns(const char *const hostname) {
 }
 
 // Needs locking
-// cache dns interne à HTS // ** FREE A FAIRE sur la chaine
+// Internal DNS cache. Fill out[0..count-1] with up to max addresses for _iadr,
-static SOCaddr* hts_dns_resolve_(httrackp * opt, const char *_iadr,
+// resolving (and caching the full list) on a miss. Returns the count.
-                                 SOCaddr *const addr, const char **error) {
+static int hts_dns_resolve_list_(httrackp *opt, const char *_iadr,
+                                 SOCaddr *const out, const int max,
+                                 const char **error) {
   char BIGSTK iadr[HTS_URLMAXSIZE * 2];
-  t_dnscache *cache = hts_cache(opt);  // adresse du cache
+  coucal cache = hts_cache(opt); // le cache dns
-  SOCaddr *sa;
+  int count;
 
   assertf(opt != NULL);
   assertf(_iadr != NULL);
-  assertf(addr != NULL);
+  assertf(out != NULL);
 
   strcpybuff(iadr, jump_identification_const(_iadr));
   // couper éventuel :
@@ -4935,55 +5243,67 @@ static SOCaddr* hts_dns_resolve_(httrackp * opt, const char *_iadr,
   }
 
   /* get IP from the dns cache */
-  sa = hts_ghbn(cache, iadr, addr);
+  count = hts_ghbn_all(cache, iadr, out, max);
-  if (sa != NULL) {
+  if (count >= 0) { // cache hit (0 == negative-cached)
-    return SOCaddr_is_valid(*sa) ? sa : NULL;
+    return count;
-  } else {                      // non présent dans le cache dns, tester
+  } else { // non présent dans le cache dns, tester
-    const size_t iadr_len = strlen(iadr) + 1;
+    SOCaddr resolved[HTS_MAXADDRNUM];
-
+    t_dnscache *record;
-    // find queue
+    int i;
-    for(; cache->next != NULL; cache = cache->next) ;
 
 #if DEBUGDNS
     printf("resolving (not cached) %s\n", iadr);
 #endif
 
-    sa = hts_dns_resolve_nocache2(iadr, addr, error);     // calculer IP host
+    count = hts_dns_resolve_nocache_list(iadr, resolved, HTS_MAXADDRNUM, error);
 
 #if HTS_WIDE_DEBUG
     DEBUG_W("gethostbyname done\n");
 #endif
 
-    /* attempt to store new entry */
+    /* attempt to store new entry (coucal owns it and dups the host key) */
-    cache->next = malloct(sizeof(t_dnscache) + iadr_len);
+    record = malloct(sizeof(t_dnscache));
-    if (cache->next != NULL) {
+    if (record != NULL) {
-      t_dnscache *const next = cache->next;
+      memset(record, 0, sizeof(*record));
-      char *const block = (char*) cache->next;
+      record->host_count = count;
-      char *const str = block + sizeof(t_dnscache);
+      for (i = 0; i < count; i++) {
-      memcpy(str, iadr, iadr_len);
+        record->host_length[i] = SOCaddr_size(resolved[i]);
-      next->iadr = str;
+        assertf(record->host_length[i] <= sizeof(record->host_addr[i]));
-      if (sa != NULL) {
+        memcpy(record->host_addr[i], &SOCaddr_sockaddr(resolved[i]),
-        next->host_length = SOCaddr_size(*sa);
+               record->host_length[i]);
-        assertf(next->host_length <= sizeof(next->host_addr));
-        memcpy(next->host_addr, &SOCaddr_sockaddr(*sa), next->host_length);
-      } else {
-        next->host_length = 0;      // non existant dans le dns
       }
-      next->next = NULL;
+      coucal_add_pvoid(cache, iadr, record);
-      return sa;
     }
 
-    /* return result if any */
+    /* copy result to caller (cache store may have failed; result still valid)
-    return sa;
+     */
-  }                             // retour hp du cache
+    for (i = 0; i < count && i < max; i++) {
+      SOCaddr_copy_SOCaddr(out[i], resolved[i]);
+    }
+    return count;
+  } // retour hp du cache
 }
 
-SOCaddr* hts_dns_resolve2(httrackp * opt, const char *_iadr, SOCaddr *const addr, const char **error) {
+int hts_dns_resolve_all(httrackp *opt, const char *iadr, SOCaddr *out, int max,
-  SOCaddr *ret;
+                        const char **error) {
+  int count;
+
+  if (!strnotempty(iadr) || max <= 0) {
+    return 0;
+  }
   hts_mutexlock(&opt->state.lock);
-  ret = hts_dns_resolve_(opt, _iadr, addr, error);
+  count = hts_dns_resolve_list_(opt, iadr, out, max, error);
   hts_mutexrelease(&opt->state.lock);
-  return ret;
+  return count;
+}
+
+SOCaddr *hts_dns_resolve2(httrackp *opt, const char *_iadr, SOCaddr *const addr,
+                          const char **error) {
+  SOCaddr_clear(*addr);
+  if (hts_dns_resolve_all(opt, _iadr, addr, 1, error) > 0) {
+    return SOCaddr_is_valid(*addr) ? addr : NULL;
+  }
+  return NULL;
 }
 
 SOCaddr* hts_dns_resolve(httrackp * opt, const char *_iadr, SOCaddr *const addr) {
@@ -5720,7 +6040,11 @@ HTSEXT_API httrackp *hts_create_opt(void) {
   opt->verbosedisplay = HTS_VERBOSE_NONE; // no text animation
   opt->sizehack = HTS_FALSE;
   opt->urlhack = HTS_TRUE;
+  opt->no_www_dedup = HTS_FALSE;
+  opt->no_slash_dedup = HTS_FALSE;
+  opt->no_query_dedup = HTS_FALSE;
   StringCopy(opt->footer, HTS_DEFAULT_FOOTER);
+  StringCopy(opt->strip_query, "");
   opt->ftp_proxy = HTS_TRUE;
   opt->convert_utf8 = HTS_TRUE;
   StringCopy(opt->filelist, "");
@@ -5823,14 +6147,14 @@ HTSEXT_API void hts_free_opt(httrackp * opt) {
 
     /* Cache */
     if (opt->state.dns_cache != NULL) {
-      t_dnscache *root;
+      coucal root;
 
       hts_mutexlock(&opt->state.lock);
       root = opt->state.dns_cache;
       opt->state.dns_cache = NULL;
       hts_mutexrelease(&opt->state.lock);
 
-      hts_cache_free(root);
+      coucal_delete(&root); // frees records via hts_cache_value_free
     }
 
     /* Cancel chain */
@@ -5865,6 +6189,7 @@ HTSEXT_API void hts_free_opt(httrackp * opt) {
     StringFree(opt->urllist);
     StringFree(opt->footer);
     StringFree(opt->mod_blacklist);
+    StringFree(opt->strip_query);
 
     StringFree(opt->path_html);
     StringFree(opt->path_html_utf8);

src/htslib.h

@@ -147,11 +147,13 @@ struct OLD_htsblk {
 #define HTS_DEF_FWSTRUCT_t_dnscache
 typedef struct t_dnscache t_dnscache;
 #endif
+// One DNS cache record, stored as a coucal value keyed by hostname.
 struct t_dnscache {
-  struct t_dnscache *next;
+  // resolved addresses, in resolver (RFC 6724) order; host_count==0 means the
-  const char *iadr;
+  // name does not resolve (negative cache). host_count<=HTS_MAXADDRNUM.
-  size_t host_length;                   // length ; (4 or 16) ; 0 for error
+  int host_count;
-  char host_addr[HTS_MAXADDRLEN];
+  size_t host_length[HTS_MAXADDRNUM]; // sockaddr length of each (16 or 28)
+  char host_addr[HTS_MAXADDRNUM][HTS_MAXADDRLEN];
 };
 
 /* Library internal definictions */
@@ -191,6 +193,13 @@ int http_cookie_header_selftest(t_cookie *cookie, const char *domain,
 //int newhttp(char* iadr,char* err=NULL);
 T_SOC newhttp(httrackp * opt, const char *iadr, htsblk * retour, int port,
               int waitconnect);
+/* Like newhttp(), but connect to the addr_index-th resolved address of the host
+   (0-based) instead of always the first; *addr_count, if non-NULL, is set to
+   the total resolved addresses. newhttp() == newhttp_addr(...,0,NULL). Used by
+   the slot scheduler to try the next address when a connect fails (dead IPv6
+   etc.). */
+T_SOC newhttp_addr(httrackp *opt, const char *iadr, htsblk *retour, int port,
+                   int waitconnect, int addr_index, int *addr_count);
 HTS_INLINE void deletehttp(htsblk * r);
 HTS_INLINE int deleteaddr(htsblk * r);
 HTS_INLINE void deletesoc(T_SOC soc);
@@ -215,9 +224,14 @@ void treatfirstline(htsblk * retour, const char *rcvd);
 
 // sous-fonctions
 LLint http_xfread1(htsblk * r, int bufl);
-HTS_INLINE SOCaddr* hts_dns_resolve2(httrackp * opt, const char *iadr,
+/* Cached resolver: fill out[0..count-1] with up to max addresses for iadr (in
-                                     SOCaddr *const addr, 
+   resolver order), returning the count (0 = does not resolve, negative-cached).
-                                     const char **error);
+   Resolves once per host; later calls read the DNS cache. Must hold no lock
+   (brackets opt->state.lock itself). */
+int hts_dns_resolve_all(httrackp *opt, const char *iadr, SOCaddr *out, int max,
+                        const char **error);
+HTS_INLINE SOCaddr *hts_dns_resolve2(httrackp *opt, const char *iadr,
+                                     SOCaddr *const addr, const char **error);
 HTS_INLINE SOCaddr* hts_dns_resolve(httrackp * opt, const char *iadr,
                                     SOCaddr *const addr);
 HTSEXT_API SOCaddr* hts_dns_resolve_nocache2(const char *const hostname, 
@@ -230,8 +244,9 @@ HTSEXT_API int check_hostname_dns(const char *const hostname);
 int ftp_available(void);
 
 #if HTS_DNSCACHE
-void hts_cache_free(t_dnscache *const cache);
+/* Return opt's DNS cache hashtable (hostname -> t_dnscache record), creating it
-t_dnscache *hts_cache(httrackp * opt);
+   on first use. Records are owned by the table and freed on coucal_delete. */
+coucal hts_cache(httrackp *opt);
 #endif
 
 // outils divers

src/htsname.c

@@ -198,6 +198,13 @@ int url_savename(lien_adrfilsave *const afs,
   // copy of fil, used for lookups (see urlhack)
   const char *normadr = adr;
   const char *normfil = fil_complete;
+  /* query keys to strip for this URL (NULL = none); decoupled from urlhack */
+  char BIGSTK stripkeys[HTS_URLMAXSIZE];
+  const char *const strip =
+      StringNotEmpty(opt->strip_query)
+          ? hts_query_strip_keys(StringBuff(opt->strip_query), adr,
+                                 fil_complete, stripkeys, sizeof(stripkeys))
+          : NULL;
   const char *const print_adr = jump_protocol_const(adr);
   const char *start_pos = NULL, *nom_pos = NULL, *dot_pos = NULL;     // Position nom et point
 
@@ -230,9 +237,13 @@ int url_savename(lien_adrfilsave *const afs,
   // www-42.foo.com -> foo.com
   // foo.com/bar//foobar -> foo.com/bar/foobar
   if (opt->urlhack) {
-    // copy of adr (without protocol), used for lookups (see urlhack)
+    // dedup-lookup key; honor the per-feature negatives like htshash.c so
-    normadr = adr_normalized_sized(adr, normadr_, sizeof(normadr_));
+    // distinct URLs keep distinct savenames (else keep normadr = adr)
-    normfil = fil_normalized(fil_complete, normfil_);
+    if (!opt->no_www_dedup)
+      normadr = adr_normalized_sized(adr, normadr_, sizeof(normadr_));
+    normfil =
+        fil_normalized_filtered_ex(fil_complete, normfil_, strip,
+                                   !opt->no_slash_dedup, !opt->no_query_dedup);
   } else {
     if (link_has_authority(adr_complete)) {     // https or other protocols : in "http/" subfolder
       char *pos = strchr(adr_complete, ':');
@@ -245,6 +256,11 @@ int url_savename(lien_adrfilsave *const afs,
         normadr = normadr_;
       }
     }
+    // strip still applies with urlhack off (host left untouched); no // or
+    // query-sort here, to match the hash key (norm_slash/norm_query are 0 when
+    // urlhack is off) so a URL is looked up under the key it was stored with
+    if (strip != NULL)
+      normfil = fil_normalized_filtered_ex(fil_complete, normfil_, strip, 0, 0);
   }
 
   // à afficher sans ftp://
@@ -760,9 +776,9 @@ int url_savename(lien_adrfilsave *const afs,
         strcatbuff(fil, DEFAULT_HTML);  // nommer page par défaut (à priori ici html depuis un proxy http)
     }
   }
-  // Changer extension?
+  // Change the extension? e.g. php3 saved as html, cgi as html or gif/xbm
-  // par exemple, php3 sera sauvé en html, cgi en html ou gif, xbm etc.. selon les cas
+  // depending on the resolved type.
-  if (ext_chg && !opt->no_type_change) {                // changer ext
+  if (ext_chg && !opt->no_type_change) {
     char *a = fil + strlen(fil) - 1;
 
     if ((opt->debug > 1) && (opt->log != NULL)) {
@@ -774,11 +790,19 @@ int url_savename(lien_adrfilsave *const afs,
                       adr_complete, fil_complete, ext);
     }
     if (ext_chg == 1) {
+      // Cut the old extension only when it is empty (a bare trailing dot), the
+      // new one, or a recognized one; an unknown trailing ".token" (e.g.
+      // /article-1.884291, #115) is part of the name, not an extension.
+      const char *const old_ext = get_ext(catbuff, sizeof(catbuff), fil);
+      const int known_ext = !*old_ext || strfield2(old_ext, ext) ||
+                            is_knowntype(opt, fil) || is_dyntype(old_ext) ||
+                            ishtml_ext(old_ext) != -1;
+
       while((a > fil) && (*a != '.') && (*a != '/'))
         a--;
-      if (*a == '.')
+      if (*a == '.' && known_ext)
-        *a = '\0';              // couper
+        *a = '\0';          // cut
-      strcatbuff(fil, ".");     // recopier point
+      strcatbuff(fil, "."); // re-add the dot
     } else {
       while((a > fil) && (*a != '/'))
         a--;
@@ -786,7 +810,7 @@ int url_savename(lien_adrfilsave *const afs,
         a++;
       *a = '\0';
     }
-    strcatbuff(fil, ext);       // copier ext/nom
+    strcatbuff(fil, ext); // append ext/name
   }
   // Rechercher premier / et dernier .
   {
@@ -1721,10 +1745,10 @@ char *url_savename_refname_fullpath(httrackp * opt, const char *adr,
     StringBuff(opt->path_log), digest_filename);
 }
 
-/* remove refname if any */
+/* remove refname if any; HTS_TRUE if it was removed */
-void url_savename_refname_remove(httrackp * opt, const char *adr,
+hts_boolean url_savename_refname_remove(httrackp *opt, const char *adr,
-                                 const char *fil) {
+                                        const char *fil) {
   char *filename = url_savename_refname_fullpath(opt, adr, fil);
 
-  (void) UNLINK(filename);
+  return UNLINK(filename) == 0 ? HTS_TRUE : HTS_FALSE;
 }

src/htsname.h

@@ -104,8 +104,9 @@ char *url_md5(char *digest_buffer, const char *fil_complete);
 void url_savename_refname(const char *adr, const char *fil, char *filename);
 char *url_savename_refname_fullpath(httrackp * opt, const char *adr,
                                     const char *fil);
-void url_savename_refname_remove(httrackp * opt, const char *adr,
+/* Remove the temp-ref for (adr,fil); HTS_TRUE if it was removed. */
-                                 const char *fil);
+hts_boolean url_savename_refname_remove(httrackp *opt, const char *adr,
+                                        const char *fil);
 #endif
 
 #endif

src/htsopt.h

@@ -241,7 +241,7 @@ struct htsoptstate {
   char *userhttptype;
   int verif_backblue_done; /**< backblue.gif/fade.gif already emitted */
   int verif_external_status;
-  t_dnscache *dns_cache;  /**< DNS resolution cache */
+  coucal dns_cache; /**< DNS resolution cache: hostname -> t_dnscache record */
   int dns_cache_nthreads; /**< number of in-flight DNS resolver threads */
   /* HTML parsing state */
   char _hts_errmsg[HTS_CDLMAXSIZE + 256]; /**< last engine error message */
@@ -529,6 +529,12 @@ struct httrackp {
   htslibhandles libHandles; /**< loaded external module handles */
   //
   htsoptstate state; /**< embedded live engine state */
+  String strip_query; /**< query keys to drop when deduping URLs (-strip-query);
+                           appended at the tail to keep field offsets stable */
+  hts_boolean
+      no_www_dedup; /**< with urlhack, keep www.host distinct from host */
+  hts_boolean no_slash_dedup; /**< with urlhack, keep redundant // in paths */
+  hts_boolean no_query_dedup; /**< with urlhack, keep query-argument order */
 };
 
 /* Running statistics for a mirror. */

src/htsparse.c

@@ -3602,16 +3602,28 @@ int hts_mirror_check_moved(htsmoduleStruct * str,
              ident_url_relatif(mov_url, urladr(), urlfil(), moved)) >= 0) {
           int set_prio_to = 0;  // pas de priotité fixéd par wizard
 
-          // check whether URLHack is harmless or not
+          // check whether URLHack is harmless or not (per the effective
-          if (opt->urlhack) {
+          // sub-flags)
+          if (opt->urlhack && (!opt->no_www_dedup || !opt->no_slash_dedup ||
+                               !opt->no_query_dedup)) {
+            const int norm_host = !opt->no_www_dedup;
+            const int norm_slash = !opt->no_slash_dedup;
+            const int norm_query = !opt->no_query_dedup;
             char BIGSTK n_adr[HTS_URLMAXSIZE * 2], n_fil[HTS_URLMAXSIZE * 2];
             char BIGSTK pn_adr[HTS_URLMAXSIZE * 2], pn_fil[HTS_URLMAXSIZE * 2];
 
-            n_adr[0] = n_fil[0] = '\0';
+            strlcpybuff(n_adr,
-            (void) adr_normalized_sized(moved->adr, n_adr, sizeof(n_adr));
+                        norm_host ? jump_normalized_const(moved->adr)
-            (void) fil_normalized(moved->fil, n_fil);
+                                  : jump_identification_const(moved->adr),
-            (void) adr_normalized_sized(urladr(), pn_adr, sizeof(pn_adr));
+                        sizeof(n_adr));
-            (void) fil_normalized(urlfil(), pn_fil);
+            strlcpybuff(pn_adr,
+                        norm_host ? jump_normalized_const(urladr())
+                                  : jump_identification_const(urladr()),
+                        sizeof(pn_adr));
+            fil_normalized_filtered_ex(moved->fil, n_fil, NULL, norm_slash,
+                                       norm_query);
+            fil_normalized_filtered_ex(urlfil(), pn_fil, NULL, norm_slash,
+                                       norm_query);
             if (strcasecmp(n_adr, pn_adr) == 0
                 && strcasecmp(n_fil, pn_fil) == 0) {
               hts_log_print(opt, LOG_WARNING,
@@ -3749,44 +3761,60 @@ int hts_mirror_check_moved(htsmoduleStruct * str,
 
       }                         // bloc
       // erreur HTTP (ex: 404, not found)
-    } else if ((r->statuscode == HTTP_PRECONDITION_FAILED)
+    } else if ((r->statuscode == HTTP_PRECONDITION_FAILED) ||
-               || (r->statuscode == HTTP_REQUESTED_RANGE_NOT_SATISFIABLE)
+               (r->statuscode == HTTP_REQUESTED_RANGE_NOT_SATISFIABLE)) {
-      ) {                       // Precondition Failed, c'est à dire pour nous redemander TOUT le fichier
+      // 412/416: the resume partial is stale; re-get the whole file (#206)
-      if (fexist_utf8(heap(ptr)->sav)) {
+      lien_back *itemback = NULL;
-        remove(heap(ptr)->sav);        // Eliminer
+      int had_partial = 0;
-      } else {
+      int ref_existed = 0;
-        hts_log_print(opt, LOG_WARNING,
+      int ref_gone;
-                      "Unexpected 412/416 error (%s) for %s%s, '%s' could not be found on disk",
+
-                      r->msg, urladr(), urlfil(),
+      // Drop the temp-ref, its partial, and heap->sav so the re-get carries no
-                      heap(ptr)->sav != NULL ? heap(ptr)->sav : "");
+      // Range; else back_add rebuilds the same Range and loops.
+      if (back_unserialize_ref(opt, heap(ptr)->adr, heap(ptr)->fil,
+                               &itemback) == 0) {
+        had_partial = 1;
+        ref_existed = 1;
+        // best-effort: an orphaned partial cannot re-Range once the ref is gone
+        if (fexist_utf8(itemback->url_sav))
+          (void) UNLINK(fconv(OPT_GET_BUFF(opt), OPT_GET_BUFF_SIZE(opt),
+                              itemback->url_sav));
+        back_clear_entry(itemback);
+        freet(itemback);
       }
-      if (!fexist_utf8(heap(ptr)->sav)) {    // Bien éliminé? (sinon on boucle..)
+      // don't re-record if the ref survived (it would re-Range and loop)
-#if HDEBUG
+      ref_gone =
-        printf("Partial content NOT up-to-date, reget all file for %s\n",
+          url_savename_refname_remove(opt, heap(ptr)->adr, heap(ptr)->fil) ||
-               heap(ptr)->sav);
+          !ref_existed;
-#endif
+      if (fexist_utf8(heap(ptr)->sav)) {
+        had_partial = 1;
+        remove(heap(ptr)->sav);
+      }
+
+      // Re-get once, only if a partial existed and both Range triggers are
+      // gone; a failed removal gives up rather than looping. range_used is
+      // unreliable (it does not survive the delayed-type two-pass).
+      if (had_partial && ref_gone && !fexist_utf8(heap(ptr)->sav)) {
         hts_log_print(opt, LOG_DEBUG, "Partial file reget (%s) for %s%s",
                       r->msg, urladr(), urlfil());
-        // enregistrer le MEME lien
         if (hts_record_link(opt, heap(ptr)->adr, heap(ptr)->fil, heap(ptr)->sav, "", "", NULL)) {
-          heap_top()->testmode = heap(ptr)->testmode;   // mode test?
+          heap_top()->testmode = heap(ptr)->testmode;
-          heap_top()->link_import = 0;   // pas mode import
+          heap_top()->link_import = 0;
           heap_top()->depth = heap(ptr)->depth;
           heap_top()->pass2 = max(heap(ptr)->pass2, numero_passe);
           heap_top()->retry = heap(ptr)->retry;
           heap_top()->premier = heap(ptr)->premier;
           heap_top()->precedent = ptr;
-          //
-          // canceller lien actuel
           error = 1;
-          hts_invalidate_link(opt, ptr);  // invalidate hashtable entry
+          hts_invalidate_link(opt, ptr); // invalidate hashtable entry
-          //
+        } else {                         // out of memory
-        } else {              // oups erreur, plus de mémoire!!
+          XH_uninit;
-          XH_uninit;          // désallocation mémoire & buffers
           return 0;
         }
       } else {
-        hts_log_print(opt, LOG_ERROR, "Can not remove old file %s", urlfil());
+        hts_log_print(opt, LOG_WARNING,
+                      "Giving up on partial reget (%s) for %s%s", r->msg,
+                      urladr(), urlfil());
         error = 1;
       }
 

src/htsselftest.h

@@ -0,0 +1,52 @@
+/* ------------------------------------------------------------ */
+/*
+HTTrack Website Copier, Offline Browser for Windows and Unix
+Copyright (C) 2026 Xavier Roche and other contributors
+
+SPDX-License-Identifier: GPL-3.0-or-later
+
+This program is free software: you can redistribute it and/or modify
+it under the terms of the GNU General Public License as published by
+the Free Software Foundation, either version 3 of the License, or
+(at your option) any later version.
+
+This program is distributed in the hope that it will be useful,
+but WITHOUT ANY WARRANTY; without even the implied warranty of
+MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+GNU General Public License for more details.
+
+You should have received a copy of the GNU General Public License
+along with this program. If not, see <http://www.gnu.org/licenses/>.
+
+Ethical use: we kindly ask that you NOT use this software to harvest email
+addresses or to collect any other private information about people. Doing so
+would dishonor our work and waste the many hours we have spent on it.
+
+Please visit our Website: http://www.httrack.com
+*/
+
+/* ------------------------------------------------------------ */
+/* File: htsselftest.h                                          */
+/*       named dispatch for the hidden engine self-tests        */
+/* Author: Xavier Roche                                         */
+/* ------------------------------------------------------------ */
+
+#ifndef HTSSELFTEST_DEFH
+#define HTSSELFTEST_DEFH
+
+#ifdef HTS_INTERNAL_BYTECODE
+
+#ifndef HTS_DEF_FWSTRUCT_httrackp
+#define HTS_DEF_FWSTRUCT_httrackp
+typedef struct httrackp httrackp;
+#endif
+
+/* Run engine self-test `name` over the positional args argv[0..argc-1], or list
+   the available tests when name is NULL, empty, or "list". Prints the result;
+   returns the process exit code (0 == success). The caller owns option cleanup.
+   Reached through the hidden `httrack -#test[=NAME ...]` subcommand. */
+int hts_selftest(httrackp *opt, const char *name, int argc, char **argv);
+
+#endif
+
+#endif

tests/01_engine-cache-golden.test

@@ -4,7 +4,7 @@
 # POSIX /bin/sh on some platforms (e.g. macOS), so avoid bashisms and GNU-only
 # tool flags despite the #!/bin/bash above.
 
-# Golden cache-format regression test (driven by 'httrack -#B <dir>').
+# Golden cache-format regression test (driven by 'httrack -#test=cache-golden <dir>').
 #
 # 01_engine-cache.test writes the cache with the same build it reads back (a
 # round-trip), so it cannot catch a read-path or ZIP-format regression where
@@ -13,7 +13,7 @@
 # byte-exact.
 #
 # Regenerate the fixture after a deliberate format change with
-# 'httrack -#B <dir> regen', then copy <dir>/hts-cache/new.zip over the
+# 'httrack -#test=cache-golden <dir> regen', then copy <dir>/hts-cache/new.zip over the
 # committed file.
 
 set -eu
@@ -37,11 +37,11 @@ trap 'rm -rf "$dir"' EXIT
 mkdir -p "$dir/hts-cache"
 cp "$fixture/hts-cache/new.zip" "$dir/hts-cache/new.zip"
 
-out=$(httrack -#B "$dir")
+out=$(httrack -#test=cache-golden "$dir")
 
 # Match the exact success line: the read must have found and verified every
-# entry, not merely failed to enter the mode (a bad -#B falls through to the
+# entry, not merely failed to enter the mode (a renamed/removed test prints the
-# usage screen, which also exits non-zero but never prints this).
+# registry to stderr, which also exits non-zero but never prints this).
 test "$out" = "cache-golden: OK" || {
     echo "expected 'cache-golden: OK', got: $out" >&2
     exit 1

tests/01_engine-cache-writefail.test

@@ -0,0 +1,24 @@
+#!/bin/bash
+#
+# Keep this POSIX-portable: the harness runs it via $(BASH), which is a plain
+# POSIX /bin/sh on some platforms (e.g. macOS), so avoid bashisms and GNU-only
+# tool flags despite the #!/bin/bash above.
+
+# Cache write-failure handling (httrack -#test=cache-writefail <dir>). #174/#219.
+# A failing new.zip write (disk full) used to crash the process via assertf; it
+# must instead stop the mirror with a fatal error (exit_xh=-1), no crash. The
+# self-test asserts that; reverting the fix makes -#test=cache-writefail abort (SIGABRT) and fail.
+
+set -eu
+
+dir=$(mktemp -d)
+trap 'rm -rf "$dir"' EXIT
+
+out=$(httrack -#test=cache-writefail "$dir")
+
+# Match the exact success line (error logs also go to stdout); a renamed/removed
+# test prints the registry to stderr, which exits non-zero but never prints this.
+printf '%s\n' "$out" | grep -qx "cache-writefail: OK" || {
+    echo "expected 'cache-writefail: OK', got: $out" >&2
+    exit 1
+}

tests/01_engine-cache.test

@@ -4,7 +4,7 @@
 # POSIX /bin/sh on some platforms (e.g. macOS), so avoid bashisms and GNU-only
 # tool flags despite the #!/bin/bash above.
 
-# Cache create/read/update logic (driven by 'httrack -#A <dir>').
+# Cache create/read/update logic (driven by 'httrack -#test=cache <dir>').
 #
 # The in-process self-test stores several hand-crafted edge entries (normal
 # HTML, an empty redirect with a near-limit location, a non-HTML body kept via
@@ -20,13 +20,13 @@ set -eu
 dir=$(mktemp -d)
 trap 'rm -rf "$dir"' EXIT
 
-# Like the other -# debug modes, a trailing token (the working directory) is
+# The working directory is a required argument; without it the test prints a
-# required; a bare '-#A' falls through to the usage screen.
+# usage line to stderr and returns non-zero.
-out=$(httrack -#A "$dir")
+out=$(httrack -#test=cache "$dir")
 
 # Match the exact success line, so the test cannot pass for an unrelated reason
-# (e.g. the -#A mode being gone and falling through to the usage screen, which
+# (e.g. the cache test being gone, which prints the registry to stderr but
-# also exits non-zero but never prints this).
+# never prints this line).
 test "$out" = "cache-selftest: OK" || {
     echo "expected 'cache-selftest: OK', got: $out" >&2
     exit 1

tests/01_engine-charset.test

@@ -4,13 +4,13 @@
 set -euo pipefail
 
 # charset -> UTF-8 conversion (hts_convertStringToUTF8).
-# -#3 <charset> <string> prints the string re-decoded from <charset> as UTF-8.
+# -#test=charset <charset> <string> prints the string re-decoded from <charset> as UTF-8.
 conv() {
-    test "$(httrack -O /dev/null -#3 "$1" "$2")" == "$3" || exit 1
+    test "$(httrack -O /dev/null -#test=charset "$1" "$2")" == "$3" || exit 1
 }
 # crash probe: malformed input must exit cleanly, not abort.
 runs() {
-    httrack -O /dev/null -#3 "$1" "$2" >/dev/null 2>&1 || exit 1
+    httrack -O /dev/null -#test=charset "$1" "$2" >/dev/null 2>&1 || exit 1
 }
 
 # the source bytes below are UTF-8 (this file is UTF-8); "café" is 0x63 61 66 C3 A9.
@@ -31,7 +31,7 @@ conv 'us-ascii' 'hello' 'hello'
 # unknown charset: ASCII passes through unchanged, but non-ASCII input cannot be
 # decoded and yields empty output (an error is printed to stderr).
 conv 'no-such-charset-xyz' 'abc' 'abc'
-test "$(httrack -O /dev/null -#3 'no-such-charset-xyz' 'café' 2>/dev/null)" == "" || exit 1
+test "$(httrack -O /dev/null -#test=charset 'no-such-charset-xyz' 'café' 2>/dev/null)" == "" || exit 1
 
 # malformed UTF-8 (lone continuation byte, truncated lead byte) must not crash
 runs 'utf-8' $'\x80'

tests/01_engine-cookies.test

@@ -1,14 +1,15 @@
 #!/bin/bash
 #
 # Issue #151 guard: the request Cookie header must be bare RFC 6265 name=value
-# pairs, no $Version/$Path attributes. Driven by the 'httrack -#Q' selftest.
+# pairs, no $Version/$Path attributes. Driven by the 'httrack -#test=cookies' selftest.
 
 set -eu
 
-# A trailing token is required; a bare '-#Q' falls through to the usage screen.
+# 'run' is an ignored placeholder argument.
-out=$(httrack -#Q run)
+out=$(httrack -#test=cookies run)
 
-# Exact-match the success line so a fall-through to usage can't pass the test.
+# Exact-match the success line so a renamed/removed test (it prints the registry
+# to stderr) can't pass.
 test "$out" = "cookie-header: OK" || {
     echo "expected 'cookie-header: OK', got: $out" >&2
     exit 1

tests/01_engine-copyopt.test

@@ -2,15 +2,16 @@
 #
 # Regression guard for the unsigned-enum sentinel trap: copy_htsopt's
 # `if (from->X > -1)` guard is always false for unsigned hts_boolean fields, so
-# they silently stop being copied. Driven by the in-process 'httrack -#9' test.
+# they silently stop being copied. Driven by the in-process 'httrack -#test=copyopt' test.
 # Keep POSIX-portable (harness runs it via $(BASH), a plain /bin/sh on macOS).
 
 set -eu
 
-# A trailing token is required; a bare '-#9' falls through to the usage screen.
+# 'run' is an ignored placeholder argument.
-out=$(httrack -#9 run)
+out=$(httrack -#test=copyopt run)
 
-# Exact-match the success line so a fall-through to usage can't pass the test.
+# Exact-match the success line so a renamed/removed test (it prints the registry
+# to stderr) can't pass.
 test "$out" = "copy-htsopt: OK" || {
     echo "expected 'copy-htsopt: OK', got: $out" >&2
     exit 1

tests/01_engine-dns.test

@@ -5,9 +5,8 @@ set -euo pipefail
 
 # DNS resolver/cache self-test: a mock getaddrinfo (no network) checks address
 # family, single-address selection, the -@i4/-@i6 family filter, and cache reuse.
-# The trailing token is required, like the other -# selftests, so a bare command
+# 'run' is an ignored placeholder argument.
-# line isn't treated as "no arguments" and routed to the usage screen.
+out=$(httrack -#test=dns run)
-out=$(httrack -#D run)
 
 test "$out" = "dns-selftest: OK" || {
     echo "expected 'dns-selftest: OK', got: $out" >&2

tests/01_engine-entities.test

@@ -4,13 +4,13 @@
 set -euo pipefail
 
 # HTML entity unescaping (hts_unescapeEntitiesWithCharset).
-# -#6 <string> prints the string with entities decoded (UTF-8 output).
+# -#test=entities <string> prints the string with entities decoded (UTF-8 output).
 ent() {
-    test "$(httrack -O /dev/null -#6 "$1")" == "$2" || exit 1
+    test "$(httrack -O /dev/null -#test=entities "$1")" == "$2" || exit 1
 }
 # crash probe: malformed input must exit cleanly, not abort.
 runs() {
-    httrack -O /dev/null -#6 "$1" >/dev/null 2>&1 || exit 1
+    httrack -O /dev/null -#test=entities "$1" >/dev/null 2>&1 || exit 1
 }
 
 # named entities

tests/01_engine-filelist.test

@@ -0,0 +1,65 @@
+#!/bin/bash
+#
+# -%L URL-list loading (#49): a readable list is honored; an unusable one fails
+# with the reason (errno / not-a-regular-file), not a bare "Could not include
+# URL list". Offline: file:// fixture, no server. Asserts on httrack's own
+# strings and the message shape, so it is locale-independent.
+
+set -euo pipefail
+
+tmp=$(mktemp -d "${TMPDIR:-/tmp}/httrack_filelist.XXXXXX") || exit 1
+trap 'rm -rf "$tmp"' EXIT HUP INT QUIT PIPE TERM
+
+echo '<html><body>hi</body></html>' >"$tmp/index.html"
+
+# run httrack with the given -%L target; structured log lands in $out/hts-log.txt
+run() {
+    local out="$1" list="$2"
+    rm -rf "$out"
+    mkdir -p "$out"
+    httrack -O "$out" --quiet -n "-%L" "$list" >"$out/.stdout" 2>&1 || true
+    LOG="$out/hts-log.txt"
+}
+
+fail() {
+    echo "FAIL: $1"
+    cat "$LOG"
+    exit 1
+}
+loghas() {
+    grep -Eq "$1" "$LOG" || fail "expected /$1/ in $LOG"
+}
+lognot() {
+    if grep -Eq "$1" "$LOG"; then fail "unexpected /$1/ in $LOG"; fi
+}
+
+# readable list: its one URL is loaded and counted (count must be non-zero)
+printf 'file://%s/index.html\n' "$tmp" >"$tmp/urls.txt"
+run "$tmp/ok" "$tmp/urls.txt"
+loghas '[1-9][0-9]* links added from'
+
+# missing file: quoted name + a non-empty reason, never the old reasonless
+# "Could not include URL list: <name>". The reason is the stat() errno, not the
+# directory fallback literal (guards against dropping the errno lookup).
+run "$tmp/miss" "$tmp/nope.txt"
+loghas 'Could not include URL list "[^"]+": .+'
+lognot 'Could not include URL list: '
+lognot 'not a regular file'
+
+# a directory is rejected with our own reason (locale-independent)
+mkdir -p "$tmp/adir"
+run "$tmp/dir" "$tmp/adir"
+loghas 'Could not include URL list "[^"]+": not a regular file'
+
+# unreadable regular file: the fopen() errno arm fires, distinct from the
+# directory branch. Root bypasses mode 000, so skip it there.
+if test "$(id -u)" -ne 0; then
+    : >"$tmp/noperm.txt"
+    chmod 000 "$tmp/noperm.txt"
+    run "$tmp/perm" "$tmp/noperm.txt"
+    chmod 644 "$tmp/noperm.txt"
+    loghas 'Could not include URL list "[^"]+": .+'
+    lognot 'not a regular file'
+fi
+
+exit 0

tests/01_engine-filter.test

@@ -4,13 +4,13 @@
 set -euo pipefail
 
 # wildcard filter engine (strjoker), the core of +/- include/exclude rules.
-# -#0 <filter> <string> prints "<string> does match <filter>" or "... does NOT match ...".
+# -#test=filter <filter> <string> prints "<string> does match <filter>" or "... does NOT match ...".
 
 match() {
-    test "$(httrack -O /dev/null -#0 "$1" "$2")" == "$2 does match $1" || exit 1
+    test "$(httrack -O /dev/null -#test=filter "$1" "$2")" == "$2 does match $1" || exit 1
 }
 nomatch() {
-    test "$(httrack -O /dev/null -#0 "$1" "$2")" == "$2 does NOT match $1" || exit 1
+    test "$(httrack -O /dev/null -#test=filter "$1" "$2")" == "$2 does NOT match $1" || exit 1
 }
 
 # bare star matches everything
@@ -71,3 +71,27 @@ nomatch '*[\[\]]' '[' # not matched, despite the docs
 match '*[\[\]]' ']'   # only via the empty class-match + trailing ']'
 match '*[\[\]]' '[]'  # one of {'[','\'} then the trailing ']'
 nomatch '*[\[\]]' '[]x'
+
+# Size-based rules (-#test=filtersize <size> <string> <filter...>): a negative size
+# means the size is still unknown (scan time). A size exclusion must stay neutral
+# then, so the file is fetched and only cancelled once its size is known (#143).
+fsize() {
+    local want="$1"
+    shift
+    test "$(httrack -O /dev/null -#test=filtersize "$@")" == "$want" || exit 1
+}
+fsize 'verdict=allowed size_flag=0' -1 foo.jpg -* '+*.jpg' '-*.jpg*[<10]'   # scan time: keep
+fsize 'verdict=forbidden size_flag=1' 5 foo.jpg -* '+*.jpg' '-*.jpg*[<10]'  # <10KB: cancel
+fsize 'verdict=allowed size_flag=1' 20 foo.jpg -* '+*.jpg' '-*.jpg*[<10]'   # >=10KB: keep
+fsize 'verdict=forbidden size_flag=0' -1 foo.txt -* '+*.jpg' '-*.jpg*[<10]' # not a jpg
+# the '>' operator is just as neutral at scan time, and fires once size is known
+fsize 'verdict=allowed size_flag=0' -1 foo.jpg -* '+*.jpg' '-*.jpg*[>10]'   # scan time: keep
+fsize 'verdict=forbidden size_flag=1' 20 foo.jpg -* '+*.jpg' '-*.jpg*[>10]' # >10KB: cancel
+
+# [name]/[file]/[path] never span '?' mid-string; a trailing query is still
+# tolerated by the global '?' rule (same as plain *.aspx), not the class (#144).
+nomatch '*[path]/end' 'a?b/end'
+nomatch '*[file]end' 'foo?xend'
+nomatch '*[name]X' 'abc?X'
+match '*[file]' 'foo?x=1' # trailing query: tolerated, as for *.aspx
+match '*.aspx' 'page.aspx?y=2'

tests/01_engine-hashtable.test

@@ -3,5 +3,7 @@
 
 set -euo pipefail
 
-# httrack internal hashtable autotest on 100K keys
+# httrack internal hashtable autotest on 100K keys. Assert the success line (on
-httrack -#7 100000
+# stderr) so a misrouted registry entry can't pass on exit code alone.
+out=$(httrack -#test=hashtable 100000 2>&1)
+printf '%s\n' "$out" | grep -q "all hashtable tests were successful!" || exit 1

tests/01_engine-idna.test

@@ -3,13 +3,13 @@
 
 set -euo pipefail
 
-# IDNA / punycode encode (-#4) and decode (-#5). This code has a CVE history,
+# IDNA / punycode encode (-#test=idna-encode) and decode (-#test=idna-decode). This code has a CVE history,
 # so the edge cases below cover passthrough, round-trips, and malformed input.
 
-enc() { test "$(httrack -O /dev/null -#4 "$1")" == "$2" || exit 1; }
+enc() { test "$(httrack -O /dev/null -#test=idna-encode "$1")" == "$2" || exit 1; }
-dec() { test "$(httrack -O /dev/null -#5 "$1")" == "$2" || exit 1; }
+dec() { test "$(httrack -O /dev/null -#test=idna-decode "$1")" == "$2" || exit 1; }
 # crash probe: malformed ACE input must exit cleanly, not abort.
-runs() { httrack -O /dev/null -#5 "$1" >/dev/null 2>&1 || exit 1; }
+runs() { httrack -O /dev/null -#test=idna-decode "$1" >/dev/null 2>&1 || exit 1; }
 
 # encode
 enc 'www.café.com' 'www.xn--caf-dma.com'

tests/01_engine-mime.test

@@ -4,13 +4,13 @@
 set -euo pipefail
 
 # MIME type guessing from extension (get_httptype / give_mimext).
-# -#2 <path> prints "<path> is '<mime>'" then "and its local type is '.<ext>'".
+# -#test=mime <path> prints "<path> is '<mime>'" then "and its local type is '.<ext>'".
 
 mime() {
-    test "$(httrack -O /dev/null -#2 "$1" | head -1)" == "$1 is '$2'" || exit 1
+    test "$(httrack -O /dev/null -#test=mime "$1" | head -1)" == "$1 is '$2'" || exit 1
 }
 unknown() {
-    test "$(httrack -O /dev/null -#2 "$1" | head -1)" == "$1 is of an unknown MIME type" || exit 1
+    test "$(httrack -O /dev/null -#test=mime "$1" | head -1)" == "$1 is of an unknown MIME type" || exit 1
 }
 
 mime '/a/b.html' 'text/html'

tests/01_engine-relative.test

@@ -8,7 +8,7 @@ set -euo pipefail
 # relative path from <curr>'s directory to <link>
 rel() {
     local got
-    got=$(httrack -O /dev/null -#l "$1" "$2")
+    got=$(httrack -O /dev/null -#test=relative "$1" "$2")
     test "$got" == "relative=$3" ||
         {
             echo "FAIL rel($1, $2): got '$got' want 'relative=$3'"
@@ -19,7 +19,7 @@ rel() {
 # resolve <link> against origin <adr>/<fil> -> adr=.. fil=..
 ident() {
     local got
-    got=$(httrack -O /dev/null -#i "$1" "$2" "$3")
+    got=$(httrack -O /dev/null -#test=resolve "$1" "$2" "$3")
     test "$got" == "$4" ||
         {
             echo "FAIL ident($1, $2, $3): got '$got' want '$4'"

tests/01_engine-savename.test

@@ -0,0 +1,41 @@
+#!/bin/bash
+#
+
+set -euo pipefail
+
+# Local save-name extension resolution (url_savename via -#test=savename <fil> <content-type>).
+# Asserts on the basename of "savename: <path>".
+
+name() {
+    out="$(httrack -O /dev/null -#test=savename "$1" "$2" | sed -n 's/^savename: //p')"
+    test "${out##*/}" == "$3" || {
+        echo "FAIL: '$1' '$2' -> '$out' (want '$3')"
+        exit 1
+    }
+}
+
+# #115: an unknown trailing ".token" is part of the name, keep it and append the type.
+name '/article-1.884291' 'text/html' 'article-1.884291.html'
+name '/news/story-12345.987654' 'text/html' 'story-12345.987654.html'
+
+# Recognized extensions still collapse to the resolved type.
+name '/page.php' 'text/html' 'page.html'
+name '/page.asp' 'text/html' 'page.html'
+name '/foo' 'text/html' 'foo.html'
+
+# A bare trailing dot is not a tail to keep.
+name '/page.' 'text/html' 'page.html'
+
+# Soft-404 (#267/#408): a binary URL served as HTML is named .html.
+name '/x.pdf' 'text/html' 'x.html'
+name '/x.gif' 'text/html' 'x.html'
+
+# Type agrees with the extension: keep it, no churn, no double extension.
+name '/x.pdf' 'application/pdf' 'x.pdf'
+name '/x.jpg' 'image/jpeg' 'x.jpg'
+name '/x.html' 'text/html' 'x.html'
+name '/x.js' 'application/x-javascript' 'x.js'
+name '/types/data.json' 'application/json' 'data.json'
+
+# Agreeing type must not rewrite the extension's casing (no strip-and-reappend).
+name '/x.JPG' 'image/jpeg' 'x.JPG'

tests/01_engine-selftest-dispatch.test

@@ -0,0 +1,17 @@
+#!/bin/bash
+#
+# The -#test dispatch itself: a bare -#test lists the registry, and an unknown
+# name errors (non-zero, diagnostic) instead of silently passing.
+
+set -eu
+
+# Bare -#test lists known tests (printed to stderr).
+list=$(httrack -#test 2>&1)
+printf '%s\n' "$list" | grep -q "filter" || exit 1
+printf '%s\n' "$list" | grep -q "cache-writefail" || exit 1
+
+# Unknown name: non-zero exit + diagnostic, and no test result line.
+rc=0
+err=$(httrack -#test=bogus 2>&1) || rc=$?
+test "$rc" -ne 0 || exit 1
+printf '%s\n' "$err" | grep -q "Unknown self-test" || exit 1

tests/01_engine-simplify.test

@@ -5,7 +5,7 @@ set -euo pipefail
 
 # path simplify engine (fil_simplifie): collapses ./ and ../ segments.
 simp() {
-    test "$(httrack -O /dev/null -#1 "$1")" == "simplified=$2" || exit 1
+    test "$(httrack -O /dev/null -#test=simplify "$1")" == "simplified=$2" || exit 1
 }
 
 simp './foo/bar/' 'foo/bar/'

tests/01_engine-stripquery.test

@@ -0,0 +1,8 @@
+#!/bin/bash
+#
+
+set -euo pipefail
+
+# --strip-query: pattern-scoped query-key stripping for dedup. All assertions
+# live in the engine self-test (hts_query_strip_keys + fil_normalized_filtered).
+httrack -O /dev/null -#test=stripquery | grep -q "strip-query self-test OK"

tests/01_engine-strsafe.test

@@ -3,23 +3,22 @@
 
 set -euo pipefail
 
-# htssafe.h bounded string operations (driven by 'httrack -#8').
+# htssafe.h bounded string operations (driven by 'httrack -#test=strsafe').
 
 # Success path: every bounded op (strcpybuff/strcatbuff/strncatbuff/strlcpybuff)
-# must behave correctly. Like the other -# debug modes, a trailing token is
+# must behave correctly. 'run' selects the success path (vs the overflow modes).
-# required (a bare '-#8' falls through to the usage screen).
 rc=0
-out=$(httrack -#8 run) || rc=$?
+out=$(httrack -#test=strsafe run) || rc=$?
 test "$rc" -eq 0 || exit 1
 test "$out" == "strsafe: OK" || exit 1
 
 # Overflow path: an over-capacity write into a sized buffer must be caught by
 # the bounded macro and abort the process, not be silently truncated/completed.
 # Assert the htssafe abort signature specifically, so the test cannot pass for
-# an unrelated reason (e.g. the -#8 mode being gone and falling through to the
+# an unrelated reason (e.g. the strsafe test being gone, which prints the
-# usage screen, which also exits non-zero).
+# registry to stderr and also exits non-zero).
 # the bounded macro aborts (non-zero exit), so don't let set -e trip on it
-err=$(httrack -#8 overflow "this string is far too long for the buffer" 2>&1) || true
+err=$(httrack -#test=strsafe overflow "this string is far too long for the buffer" 2>&1) || true
 case "$err" in
 *"strsafe: NOT aborted"*)
     echo "over-capacity write was NOT caught" >&2
@@ -36,7 +35,7 @@ esac
 # capacity (4 bytes into a 4-byte buffer), so this also pins the boundary: a
 # '<=' off-by-one in the capacity check would let it through (and print "NOT
 # aborted"). Match the specific htsbuff abort message, not just any assert.
-err=$(httrack -#8 overflow-buff "abcd" 2>&1) || true
+err=$(httrack -#test=strsafe overflow-buff "abcd" 2>&1) || true
 case "$err" in
 *"strsafe: NOT aborted"*)
     echo "htsbuff over-capacity write was NOT caught" >&2

tests/01_engine-urlhack.test

@@ -0,0 +1,8 @@
+#!/bin/bash
+#
+
+set -euo pipefail
+
+# -%u url-hack split (#271): www / // / query-order dedup toggle independently.
+# All assertions live in the engine self-test (hash compare flag resolution).
+httrack -O /dev/null -#test=urlhack run | grep -q "urlhack self-test OK"

tests/19_local-connect-fallback.test

@@ -0,0 +1,110 @@
+#!/bin/bash
+#
+# A host that resolves to several addresses must fall back to the next one when
+# a connect fails, instead of giving up on the first (dead IPv6 on a dual-stack
+# host, ...). HTTRACK_DEBUG_RESOLVE pins "deadhost" to a refused address first
+# (127.0.0.2, nothing listening) then the live server (127.0.0.1): the crawl
+# only succeeds if httrack retries the second address. A second case pins every
+# address to a refused one, so the slot must exhaust the list and error out
+# (rather than hang or loop).
+
+set -euo pipefail
+
+: "${top_srcdir:=..}"
+
+if test "${V6_SUPPORT:-}" == "no"; then
+    echo "no IPv6 support (resolver list/override is IPv6-only), skipping"
+    exit 77
+fi
+if ! command -v python3 >/dev/null 2>&1; then
+    echo "python3 missing, skipping"
+    exit 77
+fi
+
+server="$top_srcdir/tests/local-server.py"
+root="$top_srcdir/tests/server-root"
+tmpdir=$(mktemp -d)
+serverpid=
+
+cleanup() {
+    if test -n "$serverpid"; then
+        kill "$serverpid" 2>/dev/null || true
+        wait "$serverpid" 2>/dev/null || true
+    fi
+    rm -rf "$tmpdir"
+    return 0
+}
+trap cleanup EXIT
+
+# bind the live server to 127.0.0.1 only, so 127.0.0.2 refuses the connect
+python3 "$server" --root "$root" --bind 127.0.0.1 >"$tmpdir/srv.out" 2>"$tmpdir/srv.err" &
+serverpid=$!
+port=
+for _ in $(seq 1 50); do
+    line=$(head -n1 "$tmpdir/srv.out" 2>/dev/null || true)
+    if test "${line%% *}" == "PORT"; then
+        port="${line#PORT }"
+        break
+    fi
+    kill -0 "$serverpid" 2>/dev/null || {
+        echo "server exited early: $(cat "$tmpdir/srv.err")"
+        exit 1
+    }
+    sleep 0.1
+done
+test -n "$port" || {
+    echo "could not discover server port"
+    exit 1
+}
+
+out="$tmpdir/crawl"
+HTTRACK_DEBUG_RESOLVE="deadhost:127.0.0.2,127.0.0.1" \
+    httrack "http://deadhost:$port/simple/basic.html" -O "$out" \
+    -c1 --robots=0 --timeout=30 --quiet -Z >"$tmpdir/log" 2>&1
+
+log="$out/hts-log.txt"
+
+# the dead address was tried, then the next one (proves the fallback ran)
+if ! grep -q "trying next address" "$log"; then
+    echo "FAIL: no connect fallback happened"
+    cat "$log"
+    exit 1
+fi
+
+# 0 errors and the file was actually fetched (over the live address)
+errs=$(grep -iEc "^[0-9:]*[[:space:]]Error:" "$log" || true)
+test "$errs" == "0" || {
+    echo "FAIL: $errs error(s) reported"
+    grep -iE "Error:" "$log"
+    exit 1
+}
+test -f "$out/deadhost_$port/simple/basic.html" || {
+    echo "FAIL: basic.html not downloaded via fallback"
+    find "$out" -type f
+    exit 1
+}
+
+# every address refused: the slot exhausts the list, then errors out (the
+# harness timeout would catch a hang/loop; refused connects are instant)
+out2="$tmpdir/crawl2"
+HTTRACK_DEBUG_RESOLVE="alldead:127.0.0.2,127.0.0.3" \
+    httrack "http://alldead:$port/simple/basic.html" -O "$out2" \
+    -c1 --robots=0 --timeout=30 --quiet -Z >"$tmpdir/log2" 2>&1
+log2="$out2/hts-log.txt"
+
+grep -q "trying next address" "$log2" || {
+    echo "FAIL: exhaustion path never tried the fallback address"
+    cat "$log2"
+    exit 1
+}
+grep -iqE "^[0-9:]*[[:space:]]Error:" "$log2" || {
+    echo "FAIL: all addresses failing did not report an error"
+    cat "$log2"
+    exit 1
+}
+test ! -f "$out2/alldead_$port/simple/basic.html" || {
+    echo "FAIL: file downloaded despite every address failing"
+    exit 1
+}
+
+echo "OK: connect fallback succeeds, and exhausting all addresses errors out"

tests/20_local-resume-loop.test

@@ -0,0 +1,113 @@
+#!/bin/bash
+# Issue #206: a continue/update crawl looped forever when the resume Range got a
+# 416. Pass 1 leaves a partial + temp-ref; pass 2 must terminate and not loop.
+set -u
+
+: "${top_srcdir:=..}"
+testdir=$(cd "$(dirname "$0")" && pwd)
+server="${testdir}/local-server.py"
+
+command -v python3 >/dev/null || ! echo "python3 not found; skipping" || exit 77
+
+tmpdir=$(mktemp -d "${TMPDIR:-/tmp}/httrack_206.XXXXXX") || exit 1
+serverpid=
+crawlpid=
+cleanup() {
+    test -n "$crawlpid" && kill -9 "$crawlpid" 2>/dev/null
+    if test -n "$serverpid"; then
+        kill "$serverpid" 2>/dev/null
+        wait "$serverpid" 2>/dev/null
+    fi
+    rm -rf "$tmpdir"
+}
+trap cleanup EXIT HUP INT QUIT PIPE TERM
+
+# --- start the server, discover its ephemeral port --------------------------
+# RESUME_COUNTER gets a byte per /resume/blob.txt request (pass-2 delta bounds re-gets).
+serverlog="${tmpdir}/server.log"
+counter="${tmpdir}/blobcount"
+RESUME_COUNTER="$counter" python3 "$server" --root "${testdir}/server-root" >"$serverlog" 2>&1 &
+serverpid=$!
+port=
+for _ in $(seq 1 50); do
+    line=$(head -n1 "$serverlog" 2>/dev/null)
+    if test "${line%% *}" == "PORT"; then
+        port="${line#PORT }"
+        break
+    fi
+    kill -0 "$serverpid" 2>/dev/null || {
+        echo "server exited early: $(cat "$serverlog")"
+        exit 1
+    }
+    sleep 0.1
+done
+test -n "$port" || {
+    echo "could not discover server port"
+    exit 1
+}
+base="http://127.0.0.1:${port}"
+
+which httrack >/dev/null || {
+    echo "could not find httrack"
+    exit 1
+}
+out="${tmpdir}/crawl"
+mkdir "$out"
+common=(-O "$out" --quiet --disable-security-limits --robots=0 --timeout=30 --retries=0)
+refdir="${out}/hts-cache/ref"
+
+# --- pass 1: crawl, interrupt once the blob download is underway -------------
+printf '[pass 1: interrupt mid-download] ..\t'
+httrack "${common[@]}" "${base}/resume/index.html" >"${tmpdir}/log1" 2>&1 &
+crawlpid=$!
+# Wait until blob.txt is requested, then SIGTERM so httrack's exit handler
+# finalizes the cache and serializes the temp-ref.
+for _ in $(seq 1 300); do
+    test -s "$counter" && break
+    kill -0 "$crawlpid" 2>/dev/null || break
+    sleep 0.1
+done
+sleep 0.5
+kill -TERM "$crawlpid" 2>/dev/null
+wait "$crawlpid" 2>/dev/null
+crawlpid=
+test -n "$(find "$refdir" -name '*.ref' 2>/dev/null)" || {
+    echo "FAIL: no temp-ref survived pass 1; cannot drive #206"
+    exit 1
+}
+echo "OK (temp-ref present)"
+before=$(wc -c <"$counter" 2>/dev/null || echo 0)
+
+# --- pass 2: --continue -> resume Range -> 416, bounded against the #206 loop -
+# Kill pass 2 after a deadline (portable stand-in for `timeout`, absent on macOS).
+printf '[pass 2: resume must terminate] ..\t'
+HANG_RC=137 # 128 + SIGKILL
+httrack "${common[@]}" --continue "${base}/resume/index.html" >"${tmpdir}/log2" 2>&1 &
+crawlpid=$!
+(sleep 30 && kill -9 "$crawlpid" 2>/dev/null) &
+guard=$!
+rc=0
+wait "$crawlpid" 2>/dev/null || rc=$?
+crawlpid=
+kill "$guard" 2>/dev/null || true
+wait "$guard" 2>/dev/null || true
+if test "$rc" -eq "$HANG_RC"; then
+    echo "FAIL: pass 2 did not terminate (#206 resume->416 loop)"
+    exit 1
+fi
+echo "OK (terminated, rc=$rc)"
+
+# The fix re-gets once (resume Range + range-less re-get = 2): the lower bound
+# rejects a drop-the-link non-fix (1), the upper bound rejects the loop (many).
+after=$(wc -c <"$counter" 2>/dev/null || echo 0)
+hits=$((after - before))
+printf '[bounded re-get count] ..\t'
+if test "$hits" -lt 2; then
+    echo "FAIL: only ${hits} pass-2 request(s); the stale partial was not re-got"
+    exit 1
+fi
+if test "$hits" -gt 8; then
+    echo "FAIL: ${hits} pass-2 requests for blob.txt (resume is looping)"
+    exit 1
+fi
+echo "OK (${hits} requests)"

tests/21_local-intl-update.test

@@ -0,0 +1,11 @@
+#!/bin/bash
+#
+# #157: a dotless, accented URL named .html on the first crawl must keep .html
+# across an update -- not revert to the extensionless name.
+
+: "${top_srcdir:=..}"
+
+bash "$top_srcdir/tests/local-crawl.sh" --errors 0 --rerun \
+    --found 'intl/Instalação_CVS_no_Ubuntu.html' \
+    --not-found 'intl/Instalação_CVS_no_Ubuntu' \
+    httrack 'BASEURL/intl/index.html'

tests/22_local-broken-size.test

@@ -0,0 +1,17 @@
+#!/bin/bash
+# Issues #32/#41: a Content-Length that disagrees with the body warns "bogus
+# state (broken size)" and skips the cache; -%B (tolerant) accepts it.
+
+: "${top_srcdir:=..}"
+
+# Default: warn, but the file is still written.
+bash "$top_srcdir/tests/local-crawl.sh" --errors 0 \
+    --found 'size/oversize.bin' \
+    --log-found 'bogus state \(broken size' \
+    httrack 'BASEURL/size/index.html'
+
+# -%B (tolerant): no warning, file written.
+bash "$top_srcdir/tests/local-crawl.sh" --errors 0 \
+    --found 'size/oversize.bin' \
+    --log-not-found 'bogus state' \
+    httrack 'BASEURL/size/index.html' '-%B'

tests/23_local-errpage.test

@@ -0,0 +1,19 @@
+#!/bin/bash
+# Issue #17: with "no error pages" (-o0), 4xx/5xx bodies must not be written;
+# a genuine 0-byte 200 stays. Default (-o1) writes the error page. (#17's purge
+# half also does not reproduce; the purge path is not exercised here.)
+set -e
+
+: "${top_srcdir:=..}"
+
+# -o0: 404 suppressed, good page and the legit 0-byte 200 kept.
+bash "$top_srcdir/tests/local-crawl.sh" --errors 1 \
+    --found 'errpage/good.html' \
+    --found 'errpage/empty.html' \
+    --not-found 'errpage/missing.html' \
+    httrack 'BASEURL/errpage/index.html' '-o0'
+
+# Control -o1 (default): the 404 error page is written.
+bash "$top_srcdir/tests/local-crawl.sh" --errors 1 \
+    --found 'errpage/missing.html' \
+    httrack 'BASEURL/errpage/index.html' '-o1'

tests/24_local-resume-overlap.test

@@ -0,0 +1,109 @@
+#!/bin/bash
+# Issue #198: on a resumed download the server may answer the Range with a 206
+# that starts *before* the offset we asked for (block-aligned ranges). httrack
+# must honor the returned Content-Range, not blindly append, or the overlap
+# bytes get duplicated and the file grows (corrupt PDFs). Pass 1 interrupts
+# flaky.bin mid-body (partial + temp-ref); pass 2 resumes against a 206 that
+# backs up 8 bytes. The result must equal the same bytes fetched whole (full.bin).
+set -eu
+
+: "${top_srcdir:=..}"
+testdir=$(cd "$(dirname "$0")" && pwd)
+server="${testdir}/local-server.py"
+
+command -v python3 >/dev/null || ! echo "python3 not found; skipping" || exit 77
+
+tmpdir=$(mktemp -d "${TMPDIR:-/tmp}/httrack_198.XXXXXX") || exit 1
+serverpid=
+crawlpid=
+cleanup() {
+    if test -n "$crawlpid"; then kill -9 "$crawlpid" 2>/dev/null || true; fi
+    if test -n "$serverpid"; then
+        kill "$serverpid" 2>/dev/null || true
+        wait "$serverpid" 2>/dev/null || true
+    fi
+    rm -rf "$tmpdir"
+}
+trap cleanup EXIT HUP INT QUIT PIPE TERM
+
+# OVERLAP_COUNTER gets a byte per flaky.bin request so pass 1 knows when to interrupt.
+serverlog="${tmpdir}/server.log"
+counter="${tmpdir}/hits"
+resumed="${tmpdir}/resumed" # gets a byte when the server serves a resume 206
+OVERLAP_COUNTER="$counter" OVERLAP_RESUMED="$resumed" \
+    python3 "$server" --root "${testdir}/server-root" \
+    >"$serverlog" 2>&1 &
+serverpid=$!
+port=
+for _ in $(seq 1 50); do
+    line=$(head -n1 "$serverlog" 2>/dev/null)
+    if test "${line%% *}" == "PORT"; then
+        port="${line#PORT }"
+        break
+    fi
+    kill -0 "$serverpid" 2>/dev/null || {
+        echo "server exited early: $(cat "$serverlog")"
+        exit 1
+    }
+    sleep 0.1
+done
+test -n "$port" || {
+    echo "could not discover server port"
+    exit 1
+}
+base="http://127.0.0.1:${port}"
+
+which httrack >/dev/null || {
+    echo "could not find httrack"
+    exit 1
+}
+out="${tmpdir}/crawl"
+common=(-O "$out" --quiet --disable-security-limits --robots=0 --timeout=30 --retries=0 -c1)
+refdir="${out}/hts-cache/ref"
+
+# pass 1: interrupt once flaky.bin's prefix is streaming (partial + temp-ref).
+printf '[pass 1: interrupt flaky.bin] ..\t'
+httrack "${common[@]}" "${base}/overlap/index.html" >"${tmpdir}/log1" 2>&1 &
+crawlpid=$!
+for _ in $(seq 1 300); do
+    test -s "$counter" && break
+    kill -0 "$crawlpid" 2>/dev/null || break
+    sleep 0.1
+done
+sleep 0.5
+kill -TERM "$crawlpid" 2>/dev/null || true
+wait "$crawlpid" 2>/dev/null || true
+crawlpid=
+test -n "$(find "$refdir" -name '*.ref' 2>/dev/null)" || {
+    echo "FAIL: no temp-ref survived pass 1; cannot drive the resume"
+    exit 1
+}
+echo "OK (temp-ref present)"
+
+# pass 2: --continue -> resume Range -> 206 that starts 8 bytes early.
+printf '[pass 2: resume flaky.bin] ..\t'
+httrack "${common[@]}" --continue "${base}/overlap/index.html" >"${tmpdir}/log2" 2>&1 || true
+echo "OK"
+
+# Guard against a silent full re-download: the byte-compare below only tests the
+# fix if pass 2 actually went through the resume Range -> 206 path.
+printf '[resume path was exercised] ..\t'
+if ! test -s "$resumed"; then
+    echo "FAIL: pass 2 never triggered a resume 206; the overlap fix was not exercised"
+    exit 1
+fi
+echo "OK"
+
+printf '[resumed file is not corrupted] ..\t'
+dir=$(find "$out" -maxdepth 1 -type d -name '127.0.0.1*' | head -1)
+flaky="${dir}/overlap/flaky.bin"
+full="${dir}/overlap/full.bin"
+if ! test -f "$flaky" || ! test -f "$full"; then
+    echo "FAIL: flaky.bin or full.bin missing after pass 2"
+    exit 1
+fi
+if ! cmp -s "$flaky" "$full"; then
+    echo "FAIL: resumed flaky.bin ($(wc -c <"$flaky")) != full.bin ($(wc -c <"$full")); overlap duplicated"
+    exit 1
+fi
+echo "OK ($(wc -c <"$flaky") bytes, byte-identical)"

tests/25_local-mime-exclude.test

@@ -0,0 +1,16 @@
+#!/bin/bash
+#
+# A -mime: exclusion must abort the transfer on the response Content-Type, not
+# fetch the whole 1 MB body then discard it (#58). The bytes-received guard is
+# the real one: the file is absent either way, but only the fix keeps the count
+# tiny (header only) instead of pulling the body. Match it positively (a small,
+# <=4-digit count) so a vanished/reworded summary line fails rather than passes.
+
+: "${top_srcdir:=..}"
+
+bash "$top_srcdir/tests/local-crawl.sh" --errors 0 \
+    --found 'mimex/real.html' \
+    --not-found 'mimex/blob.pdf' \
+    --log-found 'excluded by MIME type filter' \
+    --log-found '\[[0-9]{1,4} bytes received' \
+    httrack 'BASEURL/mimex/index.html' '-mime:application/pdf'

tests/26_local-strip-query.test

@@ -0,0 +1,23 @@
+#!/bin/bash
+#
+# End-to-end --strip-query (#112): two links to one resource differing only by
+# ?utm_source dedup to a single saved file (2 files written: index + resource);
+# the control crawl without the option keeps both variants (3 files). Locks the
+# CLI->opt->hash plumbing the engine self-test can't reach.
+
+set -e
+
+: "${top_srcdir:=..}"
+
+# stripped: the two ?utm_source variants collapse to one resource
+bash "$top_srcdir/tests/local-crawl.sh" --errors 0 --files 2 \
+    httrack 'BASEURL/stripquery/index.html' --strip-query 'utm_source'
+
+# control: no stripping -> both query-named variants are saved
+bash "$top_srcdir/tests/local-crawl.sh" --errors 0 --files 3 \
+    httrack 'BASEURL/stripquery/index.html'
+
+# strip still applies with url-hack off (-%u0): exercises the urlhack-off
+# savename branch, which must normalize the dedup key the same way the hash does
+bash "$top_srcdir/tests/local-crawl.sh" --errors 0 --files 2 \
+    httrack 'BASEURL/stripquery/index.html' -%u0 --strip-query 'utm_source'

tests/Makefile.am

@@ -5,6 +5,7 @@ EXTRA_DIST = $(TESTS) crawl-test.sh run-all-tests.sh check-network.sh \
 	proxy-https-server.py \
 	local-crawl.sh local-server.py server.crt server.key \
 	server-root/simple/basic.html server-root/simple/link.html \
+	server-root/stripquery/index.html server-root/stripquery/a.html \
 	fixtures/cache-golden/hts-cache/new.zip
 
 TESTS_ENVIRONMENT =
@@ -13,6 +14,7 @@ TESTS_ENVIRONMENT += PATH=$(top_builddir)/src$(PATH_SEPARATOR)$$PATH
 ### TESTS_ENVIRONMENT += $(SHLIBPATH_VAR)="$(top_builddir)/src/$(LT_CV_OBJDIR)$${$(SHLIBPATH_VAR):+$(PATH_SEPARATOR)}$$$(SHLIBPATH_VAR)"
 TESTS_ENVIRONMENT += ONLINE_UNIT_TESTS=$(ONLINE_UNIT_TESTS)
 TESTS_ENVIRONMENT += HTTPS_SUPPORT=$(HTTPS_SUPPORT)
+TESTS_ENVIRONMENT += V6_SUPPORT=$(V6_SUPPORT)
 TESTS_ENVIRONMENT += top_srcdir=$(top_srcdir)
 
 TEST_EXTENSIONS = .test
@@ -25,6 +27,7 @@ TESTS = \
 	00_runnable.test \
 	01_engine-cache.test \
 	01_engine-cache-golden.test \
+	01_engine-cache-writefail.test \
 	01_engine-charset.test \
 	01_engine-cmdline.test \
 	01_engine-cookies.test \
@@ -32,6 +35,7 @@ TESTS = \
 	01_engine-dns.test \
 	01_engine-doitlog.test \
 	01_engine-entities.test \
+	01_engine-filelist.test \
 	01_engine-filter.test \
 	01_engine-hashtable.test \
 	01_engine-idna.test \
@@ -39,8 +43,12 @@ TESTS = \
 	01_engine-parse.test \
 	01_engine-rcfile.test \
 	01_engine-relative.test \
+	01_engine-savename.test \
+	01_engine-selftest-dispatch.test \
 	01_engine-simplify.test \
+	01_engine-stripquery.test \
 	01_engine-strsafe.test \
+	01_engine-urlhack.test \
 	02_manpage-regen.test \
 	02_update-cache.test \
 	10_crawl-simple.test \
@@ -56,6 +64,14 @@ TESTS = \
 	15_local-types.test \
 	16_local-assume.test \
 	17_local-empty-ct.test \
-	18_local-update.test
+	18_local-update.test \
+	19_local-connect-fallback.test \
+	20_local-resume-loop.test \
+	21_local-intl-update.test \
+	22_local-broken-size.test \
+	23_local-errpage.test \
+	24_local-resume-overlap.test \
+	25_local-mime-exclude.test \
+	26_local-strip-query.test
 
 CLEANFILES = check-network_sh.cache

tests/local-crawl.sh

@@ -14,7 +14,9 @@
 # Usage:
 #   bash local-crawl.sh [--tls] [--root DIR] \
 #       --errors N --files N --found PATH ... --directory PATH ... \
+#       --log-found REGEX ... --log-not-found REGEX ... \
 #       httrack BASEURL/some/path [httrack-args...]
+# --log-found/--log-not-found grep (ERE) the crawl's hts-log.txt.
 
 set -u
 
@@ -107,7 +109,7 @@ while test "$pos" -lt "$nargs"; do
         audit+=("${args[$pos]}" "${args[$((pos + 1))]}")
         pos=$((pos + 1))
         ;;
-    --found | --not-found | --directory)
+    --found | --not-found | --directory | --log-found | --log-not-found)
         audit+=("${args[$pos]}" "${args[$((pos + 1))]}")
         pos=$((pos + 1))
         ;;
@@ -196,6 +198,15 @@ if test -n "$rerun"; then
         exit 1
     }
     result "OK (update)"
+    # The update summary reports "files updated"; a fresh crawl never does. Assert
+    # it so a regression that bypasses the cache (re-crawls fresh) can't pass.
+    info "checking update used the cache"
+    if grep -aqE "mirror complete in .*files updated" "${out}/hts-log.txt"; then
+        result "OK"
+    else
+        result "update pass did not report cache activity"
+        exit 1
+    fi
 fi
 
 # --- discover the single host root (127.0.0.1_<port> or 127.0.0.1) -----------
@@ -248,6 +259,22 @@ while test "$i" -lt "${#audit[@]}"; do
             exit 1
         fi
         ;;
+    --log-found)
+        i=$((i + 1))
+        info "checking log matches ${audit[$i]}"
+        if grep -aqE "${audit[$i]}" "${out}/hts-log.txt"; then result "OK"; else
+            result "not in log"
+            exit 1
+        fi
+        ;;
+    --log-not-found)
+        i=$((i + 1))
+        info "checking log lacks ${audit[$i]}"
+        if grep -aqE "${audit[$i]}" "${out}/hts-log.txt"; then
+            result "present in log"
+            exit 1
+        else result "OK"; fi
+        ;;
     esac
     i=$((i + 1))
 done

tests/local-server.py

@@ -15,6 +15,7 @@ stdlib only (http.server + ssl) -- no new build or runtime dependency.
 
 import argparse
 import os
+import time
 from http.server import SimpleHTTPRequestHandler, ThreadingHTTPServer
 from urllib.parse import quote, unquote, urlsplit
 
@@ -176,6 +177,170 @@ class Handler(SimpleHTTPRequestHandler):
         body, ctype = self.TYPE_MATRIX[path]
         self.send_raw(body, ctype)
 
+    # --- MIME-type exclusion abort (issue #58) -----------------------------
+    # A -mime:application/pdf filter must abort the transfer once the header
+    # arrives, not download the whole body and discard it.
+    def route_mimex_index(self):
+        self.send_html(
+            '\t<a href="blob.pdf">pdf</a>\n' '\t<a href="real.html">real</a>\n'
+        )
+
+    # 1 MB body: the fix aborts after the header, so httrack's "bytes received"
+    # stays tiny; without it the engine reads the body and the count jumps.
+    MIMEX_BLOB = b"%PDF-1.4\n" + b"\x00" * (1024 * 1024)
+
+    def route_mimex_blob(self):
+        self.send_raw(self.MIMEX_BLOB, "application/pdf")
+
+    def route_mimex_real(self):
+        self.send_raw(b"<html><body>real</body></html>", "text/html")
+
+    # --- special chars in URLs across an update (issue #157) ---------------
+    # A dotless, accented basename served as text/html (MediaWiki style). The
+    # name the first crawl picks (.html) must survive the update pass.
+    INTL_NAME = "Instalação_CVS_no_Ubuntu"
+
+    def route_intl_index(self):
+        self.send_html('\t<a href="%s">accented</a>\n' % self.INTL_NAME)
+
+    def route_intl_page(self):
+        self.send_raw(b"<html><body>accented page</body></html>\n", "text/html")
+
+    # resume / 416 loop (#206): the first GET stalls after a prefix so the crawl
+    # can be interrupted (partial + temp-ref); every later request is 416.
+    RESUME_PREFIX = b"PARTIAL-" + b"x" * 4096  # flushed before the stall
+    RESUME_LEN = len(RESUME_PREFIX) + 4096  # declared length never delivered
+    _resume_started = False
+
+    def route_resume_index(self):
+        self.send_html('\t<a href="blob.txt">blob</a>')
+
+    def route_resume(self):
+        counter = os.environ.get("RESUME_COUNTER")
+        if counter:
+            with open(counter, "a") as fp:
+                fp.write("x")
+        # First GET: stall mid-body so the crawl can be interrupted with a partial.
+        if not Handler._resume_started:
+            Handler._resume_started = True
+            self.send_response(200)
+            self.send_header("Content-Type", "image/png")
+            self.send_header("Content-Length", str(self.RESUME_LEN))
+            self.send_header("Accept-Ranges", "bytes")
+            self.end_headers()
+            if self.command != "HEAD":
+                self.wfile.write(self.RESUME_PREFIX)
+                self.wfile.flush()
+                try:
+                    while True:
+                        time.sleep(3600)
+                except OSError:
+                    pass
+            return
+        self.send_response(416, "Requested Range Not Satisfiable")
+        self.send_header("Content-Type", "image/png")
+        self.send_header("Content-Range", "bytes */%d" % self.RESUME_LEN)
+        self.send_header("Content-Length", "0")
+        self.end_headers()
+
+    # 206 resume must honor the server's Content-Range, not the offset we asked
+    # for (#198): a server resuming a few bytes *before* the request must not
+    # leave httrack duplicating the overlap onto the partial. flaky.bin
+    # interrupts once then resumes OVERLAP_EARLY bytes early; full.bin serves
+    # the identical bytes in one shot, so the test can compare the two.
+    OVERLAP_BLOB = b"%PDF-1.4\n" + bytes((i * 37 + 11) % 256 for i in range(8000))
+    OVERLAP_EARLY = 8
+    OVERLAP_PREFIX_LEN = 4000  # flushed before the stall
+    _overlap_started = False
+
+    def route_overlap_index(self):
+        self.send_html('\t<a href="flaky.bin">flaky</a>\n\t<a href="full.bin">full</a>')
+
+    def route_overlap_full(self):
+        self.send_raw(self.OVERLAP_BLOB, "application/octet-stream")
+
+    def route_overlap(self):
+        counter = os.environ.get("OVERLAP_COUNTER")
+        if counter:
+            with open(counter, "a") as fp:
+                fp.write("x")
+        blob = self.OVERLAP_BLOB
+        rng = self.headers.get("Range")
+        # First GET: stream a prefix then stall, so the crawl can be interrupted
+        # mid-body (partial + temp-ref on disk).
+        if rng is None and not Handler._overlap_started:
+            Handler._overlap_started = True
+            self.send_response(200)
+            self.send_header("Content-Type", "application/octet-stream")
+            self.send_header("Content-Length", str(len(blob)))
+            self.send_header("Accept-Ranges", "bytes")
+            self.end_headers()
+            if self.command != "HEAD":
+                self.wfile.write(blob[: self.OVERLAP_PREFIX_LEN])
+                self.wfile.flush()
+                try:
+                    while True:
+                        time.sleep(3600)
+                except OSError:
+                    pass
+            return
+        if rng is None:  # no resume request: serve the whole file
+            return self.route_overlap_full()
+        # Resume: honor the Range, but back up OVERLAP_EARLY bytes.
+        start = (
+            int(rng[len("bytes=") :].split("-")[0]) if rng.startswith("bytes=") else 0
+        )
+        start = max(0, start - self.OVERLAP_EARLY)
+        # Signal that the resume Range -> 206 path actually fired, so the test
+        # can prove it was exercised (not a silent full re-download).
+        resumed = os.environ.get("OVERLAP_RESUMED")
+        if resumed:
+            with open(resumed, "a") as fp:
+                fp.write("x")
+        part = blob[start:]
+        self.send_response(206, "Partial Content")
+        self.send_header("Content-Type", "application/octet-stream")
+        self.send_header("Content-Length", str(len(part)))
+        self.send_header(
+            "Content-Range", "bytes %d-%d/%d" % (start, len(blob) - 1, len(blob))
+        )
+        self.end_headers()
+        if self.command != "HEAD":
+            self.wfile.write(part)
+
+    # error pages / 0-byte files (#17): -o0 ("no error pages") must keep 4xx/5xx
+    # bodies off disk; a genuine 0-byte 200 is a valid file and stays.
+    def route_errpage_index(self):
+        self.send_html(
+            '\t<a href="good.html">good</a>\n'
+            '\t<a href="missing.html">missing</a>\n'
+            '\t<a href="empty.html">empty</a>\n'
+        )
+
+    def route_errpage_good(self):
+        self.send_raw(b"<html><body>good page</body></html>\n", "text/html")
+
+    def route_errpage_missing(self):
+        self.send_html("\t404 error body", status=404, extra_status="Not Found")
+
+    def route_errpage_empty(self):
+        self.send_raw(b"", "text/html")
+
+    # broken Content-Length (#32/#41): declared size != bytes sent. httrack
+    # warns "bogus state (broken size)" and skips the cache unless -%B.
+    def route_size_index(self):
+        self.send_html('\t<a href="oversize.bin">over</a>\n')
+
+    def route_size_oversize(self):
+        body = b"A" * 100
+        self.send_response(200)
+        self.send_header("Content-Type", "application/octet-stream")
+        self.send_header("Content-Length", str(len(body) - 2))  # lie: too short
+        self.send_header("Connection", "close")
+        self.end_headers()
+        if self.command != "HEAD":
+            self.wfile.write(body)
+
     ROUTES = {
         "/cookies/entrance.php": route_entrance,
         "/cookies/second.php": route_second,
@@ -195,6 +360,22 @@ class Handler(SimpleHTTPRequestHandler):
         "/types/style.css": route_types,
         "/types/data.json": route_types,
         "/types/gen.php": route_types,
+        "/intl/index.html": route_intl_index,
+        "/intl/" + INTL_NAME: route_intl_page,
+        "/resume/index.html": route_resume_index,
+        "/resume/blob.txt": route_resume,
+        "/overlap/index.html": route_overlap_index,
+        "/overlap/flaky.bin": route_overlap,
+        "/overlap/full.bin": route_overlap_full,
+        "/size/index.html": route_size_index,
+        "/size/oversize.bin": route_size_oversize,
+        "/errpage/index.html": route_errpage_index,
+        "/errpage/good.html": route_errpage_good,
+        "/errpage/missing.html": route_errpage_missing,
+        "/errpage/empty.html": route_errpage_empty,
+        "/mimex/index.html": route_mimex_index,
+        "/mimex/blob.pdf": route_mimex_blob,
+        "/mimex/real.html": route_mimex_real,
     }
 
     # --- dispatch ----------------------------------------------------------
@@ -202,7 +383,8 @@ class Handler(SimpleHTTPRequestHandler):
     def dispatch(self):
         self._set_cookies = []
         path = urlsplit(self.path).path
-        handler = self.ROUTES.get(path)
+        # Match percent-encoded paths (accented #157 route) by their decoded form.
+        handler = self.ROUTES.get(path) or self.ROUTES.get(unquote(path))
         if handler is not None:
             handler(self)
             return True

tests/server-root/stripquery/a.html

@@ -0,0 +1 @@
+<html><body>resource A</body></html>

tests/server-root/stripquery/index.html

@@ -0,0 +1,5 @@
+<html><body>
+Two links to one resource, differing only by a tracking parameter.
+<a href="a.html?utm_source=x">x</a>
+<a href="a.html?utm_source=y">y</a>
+</body></html>