Merge origin/master into fix-630-pathlog-utf8

Resolve tests/Makefile.am: keep 66 (#632), 67 (#633) and this branch's 69.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Signed-off-by: Xavier Roche <roche@httrack.com>

src/htscoremain.c

@@ -2368,30 +2368,28 @@ static int hts_main_internal(int argc, char **argv, httrackp * opt) {
       opt->errlog = stderr;
     } else if (httrack_logmode >= 2) {
       // deux fichiers log
-      structcheck(StringBuff(opt->path_log));
-      if (fexist
-          (fconcat
-           (OPT_GET_BUFF(opt), OPT_GET_BUFF_SIZE(opt), StringBuff(opt->path_log), "hts-log.txt")))
-        remove(fconcat
-               (OPT_GET_BUFF(opt), OPT_GET_BUFF_SIZE(opt), StringBuff(opt->path_log), "hts-log.txt"));
-      if (fexist
-          (fconcat
-           (OPT_GET_BUFF(opt), OPT_GET_BUFF_SIZE(opt), StringBuff(opt->path_log), "hts-err.txt")))
-        remove(fconcat
-               (OPT_GET_BUFF(opt), OPT_GET_BUFF_SIZE(opt), StringBuff(opt->path_log), "hts-err.txt"));
+      // path_log holds UTF-8 bytes (argv is UTF-8): the ANSI file calls would
+      // read them as the codepage and drop the logs into a mangled twin (#630).
+      structcheck_utf8(StringBuff(opt->path_log));
+      if (fexist_utf8(fconcat(OPT_GET_BUFF(opt), OPT_GET_BUFF_SIZE(opt),
+                              StringBuff(opt->path_log), "hts-log.txt")))
+        UNLINK(fconcat(OPT_GET_BUFF(opt), OPT_GET_BUFF_SIZE(opt),
+                       StringBuff(opt->path_log), "hts-log.txt"));
+      if (fexist_utf8(fconcat(OPT_GET_BUFF(opt), OPT_GET_BUFF_SIZE(opt),
+                              StringBuff(opt->path_log), "hts-err.txt")))
+        UNLINK(fconcat(OPT_GET_BUFF(opt), OPT_GET_BUFF_SIZE(opt),
+                       StringBuff(opt->path_log), "hts-err.txt"));
 
       /* Check FS directory structure created */
-      structcheck(StringBuff(opt->path_log));
+      structcheck_utf8(StringBuff(opt->path_log));
 
-      opt->log =
-        fopen(fconcat
-              (OPT_GET_BUFF(opt), OPT_GET_BUFF_SIZE(opt), StringBuff(opt->path_log), "hts-log.txt"),
-              "w");
+      opt->log = FOPEN(fconcat(OPT_GET_BUFF(opt), OPT_GET_BUFF_SIZE(opt),
+                               StringBuff(opt->path_log), "hts-log.txt"),
+                       "w");
       if (httrack_logmode == 2)
-        opt->errlog =
-          fopen(fconcat
-                (OPT_GET_BUFF(opt), OPT_GET_BUFF_SIZE(opt), StringBuff(opt->path_log), "hts-err.txt"),
-                "w");
+        opt->errlog = FOPEN(fconcat(OPT_GET_BUFF(opt), OPT_GET_BUFF_SIZE(opt),
+                                    StringBuff(opt->path_log), "hts-err.txt"),
+                            "w");
       else
         opt->errlog = opt->log;
       if (opt->log == NULL) {

src/htsname.c

@@ -1532,8 +1532,30 @@ int url_savename(lien_adrfilsave *const afs,
       // last segment
       wsave[j++] = '/';
 #define MAX_UTF8_SEQ_CHARS 4
-      for(i = lastSeg; wsave[i] != '\0' && j < maxLen; i++) {
-        wsave[j++] = wsave[i];
+      {
+        // #623: the ".delayed" placeholder marker sits at the tail; cutting
+        // through it drops IS_DELAYED_EXT, so the file is never renamed to its
+        // final name. Reserve the trailing ".<id>.delayed" across the cut.
+        size_t markStart = wsaveLen;
+        if (IS_DELAYED_EXT(afs->save)) {
+          const size_t extDot = wsaveLen - strlen("." DELAYED_EXT);
+          size_t p = extDot; /* walk back over a dot-separated ".<hexid>" tag */
+
+          while (p > lastSeg && ((wsave[p - 1] >= '0' && wsave[p - 1] <= '9') ||
+                                 (wsave[p - 1] >= 'a' && wsave[p - 1] <= 'f')))
+            p--;
+          // keep the tag only if truly ".<hexid>.delayed"; else the bare marker
+          // (a wholly-hex base, e.g. a hashed #133 name, must not be absorbed)
+          markStart = (p > lastSeg && p < extDot && wsave[p - 1] == '.')
+                          ? p - 1
+                          : extDot;
+        }
+        // head, bounded so the marker still fits, then the marker itself
+        for (i = lastSeg; i < markStart && j + (wsaveLen - markStart) < maxLen;
+             i++)
+          wsave[j++] = wsave[i];
+        for (i = markStart; i < wsaveLen && j < maxLen; i++)
+          wsave[j++] = wsave[i];
       }
       // terminating \0
       wsave[j++] = '\0';

tests/67_engine-delayed-truncate.test

@@ -0,0 +1,72 @@
+#!/bin/bash
+#
+
+set -euo pipefail
+
+# #623: url_savename enforces the 236-char path ceiling by cutting the tail of
+# the last segment, where the mandatory ".delayed" placeholder marker lives. A
+# cut marker fails IS_DELAYED_EXT, so back_delayed_rename never renames the file
+# to its final name and the download is lost. The marker must survive the cut.
+
+# statuscode=302 status=-1 = a redirect answer still downloading: no type is
+# resolved, so the name gets a ".<id>.delayed" placeholder (see 01_engine-savename).
+CEIL=236
+
+httrack_bin=$(cd "$(dirname "$(command -v httrack)")" && pwd)/httrack
+scratch=$(mktemp -d)
+trap 'rm -rf "$scratch"' EXIT
+cd "$scratch"
+
+run() {
+    "$httrack_bin" -O /dev/null -#test=savename "$@" | sed -n 's/^savename: //p'
+}
+
+# A deep path ending in a long segment that overruns the ceiling.
+deep="/d1$(printf 'a%.0s' {1..40})/d2$(printf 'b%.0s' {1..40})"
+deep="$deep/d3$(printf 'c%.0s' {1..40})/d4$(printf 'd%.0s' {1..40})"
+long="$deep/$(printf 'z%.0s' {1..90})"
+
+out="$(run "$long" text/html statuscode=302 status=-1)"
+case "$out" in
+*.delayed) ;;
+*)
+    echo "FAIL: delayed marker cut by truncation: '$out'"
+    exit 1
+    ;;
+esac
+test "${#out}" -le "$CEIL" || {
+    echo "FAIL: truncated name ${#out} > $CEIL ceiling: '$out'"
+    exit 1
+}
+
+# #133-style hashed name: an all-hex last segment must keep the marker too (the
+# ".<id>." tag is not mistaken for part of the hash).
+hexseg=$(printf 'a1b2c3d4e5f60718%.0s' {1..8})
+hexpath="/d1$(printf 'x%.0s' {1..30})/d2$(printf 'y%.0s' {1..30})/$hexseg"
+out="$(run "$hexpath" text/html statuscode=302 status=-1)"
+case "$out" in
+*.delayed) ;;
+*)
+    echo "FAIL: hashed-name marker cut by truncation: '$out'"
+    exit 1
+    ;;
+esac
+test "${#out}" -le "$CEIL" || {
+    echo "FAIL: hashed name ${#out} > $CEIL ceiling: '$out'"
+    exit 1
+}
+
+# A non-delayed name of the same shape still truncates, with no marker to keep.
+out="$(run "$long.html" text/html)"
+test "${#out}" -le "$CEIL" || {
+    echo "FAIL: non-delayed name ${#out} > $CEIL ceiling: '$out'"
+    exit 1
+}
+case "$out" in
+*.delayed)
+    echo "FAIL: non-delayed name grew a .delayed marker: '$out'"
+    exit 1
+    ;;
+esac
+
+echo "delayed-truncate OK"

tests/69_local-intl-logdir.test

@@ -0,0 +1,15 @@
+#!/bin/bash
+#
+# A single non-ASCII -O sets both path_html and path_log to "café" (#630). The
+# logs (hts-log.txt/hts-err.txt) must land there, not in an ANSI-mangled twin:
+# on Windows path_log holds UTF-8 bytes the raw file calls read as the codepage.
+# The --errors/--files/--log-found audits all grep logroot=café/hts-log.txt, so
+# a log written to the twin fails them. POSIX has no twin, so this bites on the
+# Windows CI leg (like test 64). The cache twin is a separate, larger fix.
+
+: "${top_srcdir:=..}"
+
+bash "$top_srcdir/tests/local-crawl.sh" --outdir-intl 'café' --errors 0 --files 5 \
+    --found 'simple/basic.html' \
+    --log-found 'mirror complete in' \
+    httrack 'BASEURL/simple/basic.html'

tests/Makefile.am

@@ -147,6 +147,8 @@ TESTS = \
 	63_webhttrack-home.test \
 	64_local-intl-outdir.test \
 	65_port-siblings.test \
-	66_engine-port80-strip.test
+	66_engine-port80-strip.test \
+	67_engine-delayed-truncate.test \
+	69_local-intl-logdir.test
 
 CLEANFILES = check-network_sh.cache

tests/local-crawl.sh

@@ -48,6 +48,7 @@ key="${testdir}/server.key"
 tls=
 verbose=
 html_subdir=
+outdir_intl=
 rerun=
 rerun_args=
 rerun_dead=
@@ -140,6 +141,13 @@ while test "$pos" -lt "$nargs"; do
         pos=$((pos + 1))
         html_subdir="${args[$pos]}"
         ;;
+    --outdir-intl)
+        # Single non-ASCII -O "$out/NAME": path_html AND path_log are NAME, so
+        # the logs (and the harness reads of them) go through the non-ASCII path
+        # (#630). Distinct from --html-subdir, which keeps path_log ASCII.
+        pos=$((pos + 1))
+        outdir_intl="${args[$pos]}"
+        ;;
     --errors | --errors-content | --files)
         audit+=("${args[$pos]}" "${args[$((pos + 1))]}")
         pos=$((pos + 1))
@@ -216,13 +224,18 @@ test -n "$ver" || die "could not run httrack"
 out="${tmpdir}/crawl"
 mkdir "$out" || die "could not create $out"
 # path_html holds the mirror + index; path_log holds hts-cache/hts-log.txt.
-# Default: both are "$out". With --html-subdir, path_html becomes "$out/NAME"
-# (the mirror root the audits inspect) while path_log stays "$out".
+# Default: both are "$out". --html-subdir moves path_html to "$out/NAME" while
+# path_log (logroot) stays "$out"; --outdir-intl moves both to "$out/NAME".
 mirrorroot="$out"
+logroot="$out"
 odir="$out"
 if test -n "$html_subdir"; then
     mirrorroot="${out}/${html_subdir}"
     odir="${mirrorroot},${out}"
+elif test -n "$outdir_intl"; then
+    mirrorroot="${out}/${outdir_intl}"
+    logroot="$mirrorroot"
+    odir="$mirrorroot"
 fi
 # Localhost is fast; disable the rate/bandwidth safety limits but keep a
 # max-time backstop so a hang cannot wedge the suite.
@@ -241,7 +254,7 @@ test "$crawlres" -eq 0 || ! result "httrack exited $crawlres" || {
     exit 1
 }
 result "OK"
-grep -iE "^[0-9:]*[[:space:]]Error:" "${out}/hts-log.txt" >&2
+grep -iE "^[0-9:]*[[:space:]]Error:" "${logroot}/hts-log.txt" >&2
 
 # --- optional second pass: re-mirror into the same dir (cache/update path) ----
 if test -n "$rerun"; then
@@ -260,7 +273,7 @@ if test -n "$rerun"; then
     # 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" "${out}/hts-log.txt"; then
+    if grep -aqE "mirror complete in .*files updated" "${logroot}/hts-log.txt"; then
         result "OK"
     else
         result "update pass did not report cache activity"
@@ -292,7 +305,7 @@ if test -n "$rerun_dead"; then
     zip="${out}/hts-cache/new.zip"
     test -s "$zip" || die "no cache was written by the first pass"
     cp "$zip" "${tmpdir}/cache-before.zip"
-    cp "${out}/hts-log.txt" "${tmpdir}/log-before.txt"
+    cp "${logroot}/hts-log.txt" "${tmpdir}/log-before.txt"
     stop_server "$serverpid"
     serverpid=
     info "re-running httrack against the stopped server"
@@ -305,7 +318,7 @@ if test -n "$rerun_dead"; then
     # The dead pass must have gone through the no-data rollback, not bailed out
     # before the mirror loop (which would leave the cache trivially untouched).
     info "checking the dead pass hit the rollback"
-    if grep -aq "No data seems to have been transferred" "${out}/hts-log.txt"; then
+    if grep -aq "No data seems to have been transferred" "${logroot}/hts-log.txt"; then
         result "OK"
     else
         result "rollback notice not found in hts-log.txt"
@@ -320,7 +333,7 @@ if test -n "$rerun_dead"; then
         exit 1
     fi
     # Audits below describe the healthy crawl, not the dead pass.
-    cp "${tmpdir}/log-before.txt" "${out}/hts-log.txt"
+    cp "${tmpdir}/log-before.txt" "${logroot}/hts-log.txt"
 fi
 
 # --- discover the single host root (127.0.0.1_<port> or 127.0.0.1) -----------
@@ -350,19 +363,19 @@ while test "$i" -lt "${#audit[@]}"; do
     --errors)
         i=$((i + 1))
         assert_equals "checking errors" "${audit[$i]}" \
-            "$(grep -iEc "^[0-9:]*[[:space:]]Error:" "${out}/hts-log.txt")"
+            "$(grep -iEc "^[0-9:]*[[:space:]]Error:" "${logroot}/hts-log.txt")"
         ;;
     --errors-content)
         i=$((i + 1))
-        total=$(grep -icE "^[0-9:]*[[:space:]]Error:" "${out}/hts-log.txt")
+        total=$(grep -icE "^[0-9:]*[[:space:]]Error:" "${logroot}/hts-log.txt")
         # transient network failures (statuscode -2..-6) flake on busy loopback;
         # the code parens are followed by " at link" or " after N retries at link"
-        transient=$(grep -cE '\(-[2-6]\) (at link|after )' "${out}/hts-log.txt" || true)
+        transient=$(grep -cE '\(-[2-6]\) (at link|after )' "${logroot}/hts-log.txt" || true)
         assert_equals "checking content errors" "${audit[$i]}" "$((total - transient))"
         ;;
     --files)
         i=$((i + 1))
-        nFiles=$(grep -E "^HTTrack Website Copier/[^ ]* mirror complete in " "${out}/hts-log.txt" |
+        nFiles=$(grep -E "^HTTrack Website Copier/[^ ]* mirror complete in " "${logroot}/hts-log.txt" |
             sed -e 's/.*[[:space:]]\([^ ]*\)[[:space:]]files written.*/\1/g')
         assert_equals "checking files" "${audit[$i]}" "$nFiles"
         ;;
@@ -393,7 +406,7 @@ while test "$i" -lt "${#audit[@]}"; do
     --log-found)
         i=$((i + 1))
         info "checking log matches ${audit[$i]}"
-        if grep -aqE "${audit[$i]}" "${out}/hts-log.txt"; then result "OK"; else
+        if grep -aqE "${audit[$i]}" "${logroot}/hts-log.txt"; then result "OK"; else
             result "not in log"
             exit 1
         fi
@@ -401,7 +414,7 @@ while test "$i" -lt "${#audit[@]}"; do
     --log-not-found)
         i=$((i + 1))
         info "checking log lacks ${audit[$i]}"
-        if grep -aqE "${audit[$i]}" "${out}/hts-log.txt"; then
+        if grep -aqE "${audit[$i]}" "${logroot}/hts-log.txt"; then
             result "present in log"
             exit 1
         else result "OK"; fi