Merge origin/master into fix-629-webhttrack-argv-utf8

Resolve tests/Makefile.am: keep 66/67/69 (merged) alongside this branch's 68.

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

src/htscharset.c

@@ -373,7 +373,8 @@ char *hts_convertStringCPFromUTF8(const char *s, size_t size, UINT cp) {
   return NULL;
 }
 
-char *hts_convertStringToUTF8(const char *s, size_t size, const char *charset) {
+HTSEXT_API char *hts_convertStringToUTF8(const char *s, size_t size,
+                                         const char *charset) {
   const UINT cp = hts_getCodepage(charset);
 
   return hts_convertStringCPToUTF8(s, size, cp);
@@ -554,7 +555,8 @@ static char *hts_convertStringCharset(const char *s, size_t size,
   return NULL;
 }
 
-char *hts_convertStringToUTF8(const char *s, size_t size, const char *charset) {
+HTSEXT_API char *hts_convertStringToUTF8(const char *s, size_t size,
+                                         const char *charset) {
   /* Empty string ? */
   if (size == 0) {
     return strdup("");

src/htscharset.h

@@ -51,8 +51,8 @@ typedef unsigned int hts_UCS4;
  * Convert the string "s" from charset "charset" to UTF-8.
  * Return NULL upon error.
  **/
-extern char *hts_convertStringToUTF8(const char *s, size_t size,
-                                     const char *charset);
+HTSEXT_API char *hts_convertStringToUTF8(const char *s, size_t size,
+                                         const char *charset);
 
 /**
  * Convert the string "s" from UTF-8 to charset "charset".

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/htsselftest.c

@@ -2146,6 +2146,52 @@ static int st_cache_corrupt(httrackp *opt, int argc, char **argv) {
   return err;
 }
 
+/* Drives unzRepair over a damaged local file header whose CRC field's high
+   16-bit word has bit 15 set. Before the READ_32 fix that shifted an int and
+   overflowed, so UBSan aborts here; after it, repair recovers the one entry. */
+static int st_zip_repair_shift(httrackp *opt, int argc, char **argv) {
+  static const unsigned char zip[] = {
+      0x50, 0x4b, 0x03, 0x04, /* local file header signature */
+      0x14, 0x00,             /* version needed */
+      0x00, 0x00,             /* general purpose flag */
+      0x00, 0x00,             /* method */
+      0x00, 0x00,             /* time */
+      0x00, 0x00,             /* date */
+      0x00, 0x00, 0xe8, 0x8a, /* crc: high word 0x8ae8, bit 15 set */
+      0x00, 0x00, 0x00, 0x00, /* compressed size */
+      0x00, 0x00, 0x00, 0x00, /* uncompressed size */
+      0x01, 0x00,             /* filename length */
+      0x00, 0x00,             /* extra field length */
+      0x61                    /* filename "a" */
+  };
+  char in[HTS_URLMAXSIZE], out[HTS_URLMAXSIZE], tmp[HTS_URLMAXSIZE];
+  uLong nrec = 0, bytes = 0;
+  FILE *fp;
+  int err;
+
+  (void) opt;
+  if (argc < 1) {
+    fprintf(stderr, "zip-repair-shift: needs a directory\n");
+    return 1;
+  }
+  snprintf(in, sizeof(in), "%s/damaged.zip", argv[0]);
+  snprintf(out, sizeof(out), "%s/repair.zip", argv[0]);
+  snprintf(tmp, sizeof(tmp), "%s/repair.tmp", argv[0]);
+  fp = fopen(in, "wb");
+  if (fp == NULL || fwrite(zip, 1, sizeof(zip), fp) != sizeof(zip)) {
+    if (fp != NULL)
+      fclose(fp);
+    fprintf(stderr, "zip-repair-shift: cannot write %s\n", in);
+    return 1;
+  }
+  fclose(fp);
+  err = unzRepair(in, out, tmp, &nrec, &bytes);
+  printf("zip-repair-shift: %s (recovered %lu entr%s)\n",
+         (err == Z_OK && nrec == 1) ? "OK" : "FAIL", (unsigned long) nrec,
+         nrec == 1 ? "y" : "ies");
+  return (err == Z_OK && nrec == 1) ? 0 : 1;
+}
+
 static int st_cache_legacy(httrackp *opt, int argc, char **argv) {
   int err;
 
@@ -3253,6 +3299,9 @@ static const struct selftest_entry {
      st_cache_legacy},
     {"cache-corrupt", "<dir>", "cache read-side corruption self-test",
      st_cache_corrupt},
+    {"zip-repair-shift", "<dir>",
+     "cache zip-repair header read must not overflow a signed shift",
+     st_zip_repair_shift},
     {"dns", "", "DNS resolver/cache self-test", st_dns},
     {"dnstimeout", "", "a slow DNS resolve is bounded and holds no lock",
      st_dnstimeout},

src/htsserver.c

@@ -40,6 +40,7 @@ Please visit our Website: http://www.httrack.com
 
 #include "htsnet.h"
 #include "htslib.h"
+#include "htscharset.h"
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
@@ -749,7 +750,14 @@ int smallserver(T_SOC soc, char *url, char *method, char *data, char *path) {
                            RUN THE SERVER
                          */
                         if (strcmp((char *) adrcd, "start") == 0) {
-                          webhttrack_main((char *) adr + p);
+                          /* POST body is in the form's charset, not the
+                             UTF-8 argv the engine now assumes (#629). */
+                          char *const cmdl = (char *) adr + p;
+                          char *cmdlUtf8 = hts_convertStringToUTF8(
+                              cmdl, strlen(cmdl), LANGSEL("LANGUAGE_CHARSET"));
+
+                          webhttrack_main(cmdlUtf8 != NULL ? cmdlUtf8 : cmdl);
+                          freet(cmdlUtf8);
                         } else {
                           commandRunning = 0;
                           commandEnd = 1;

src/htsweb.c

@@ -65,6 +65,7 @@ Please visit our Website: http://www.httrack.com
 #include "htsserver.h"
 #include "htsurlport.h"
 #include "htsweb.h"
+#include "htscharset.h"
 
 #if USE_BEGINTHREAD==0
 #error fatal: no threads support
@@ -156,6 +157,7 @@ int main(int argc, char *argv[]) {
   printf("Initializing the server..\n");
 
 #ifdef _WIN32
+  hts_argv_utf8(&argc, &argv);
   {
     WORD wVersionRequested;     // requested version WinSock API
     WSADATA wsadata;            // Windows Sockets API data

src/minizip/mztools.c

@@ -14,7 +14,7 @@
 
 #define READ_8(adr)  ((unsigned char)*(adr))
 #define READ_16(adr) ( READ_8(adr) | (READ_8(adr+1) << 8) )
-#define READ_32(adr) ( READ_16(adr) | (READ_16((adr)+2) << 16) )
+#define READ_32(adr) ((uLong) READ_16(adr) | ((uLong) READ_16((adr) + 2) << 16))
 
 #define WRITE_8(buff, n) do { \
   *((unsigned char*)(buff)) = (unsigned char) ((n) & 0xff); \

tests/01_zlib-repair-shift.test

@@ -0,0 +1,21 @@
+#!/bin/bash
+#
+# Keep this POSIX-portable: the harness runs it via $(BASH), which is a plain
+# POSIX /bin/sh on some platforms (e.g. macOS), so avoid bashisms and GNU-only
+# tool flags despite the #!/bin/bash above.
+
+# unzRepair header read must not overflow a signed shift (-#test=zip-repair-shift
+# <dir>). A damaged local file header whose CRC high word has bit 15 set made
+# READ_32 shift an int past INT_MAX; UBSan aborts before the fix casts to uLong.
+
+set -eu
+
+dir=$(mktemp -d)
+trap 'rm -rf "$dir"' EXIT
+
+out=$(httrack -#test=zip-repair-shift "$dir")
+
+printf '%s\n' "$out" | grep -qx "zip-repair-shift: OK (recovered 1 entry)" || {
+    echo "expected 'zip-repair-shift: OK (recovered 1 entry)', got: $out" >&2
+    exit 1
+}

tests/68_webhttrack-outdir-charset.test

@@ -0,0 +1,90 @@
+#!/bin/bash
+#
+# webhttrack's POST body arrives in the form's declared charset, not UTF-8, yet
+# the engine now assumes UTF-8 argv. A non-ASCII -O output dir submitted through
+# the web UI must land under the UTF-8 directory, not an ISO-8859-1 twin (#629).
+# Drives the real htsserver over HTTP; the default (English) form is served
+# ISO-8859-1, so 'café' travels as the single byte 0xE9. The crawl target is a
+# dead port: only the -O directory name is under test, and htsserver creates it
+# from the decoded path before any fetch.
+
+set -euo pipefail
+
+testdir=$(cd "$(dirname "$0")" && pwd)
+distdir=${top_srcdir:-$(cd "${testdir}/.." && pwd)}
+distdir=$(cd "${distdir}" && pwd)
+# shellcheck source=tests/testlib.sh
+. "${testdir}/testlib.sh"
+
+fail() {
+    echo "FAIL: $*" >&2
+    exit 1
+}
+
+command -v htsserver >/dev/null || fail "no htsserver in PATH"
+python=$(find_python) || {
+    echo "python3 not found; skipping" >&2
+    exit 77
+}
+
+work=$(mktemp -d "${TMPDIR:-/tmp}/webhttrack_charset.XXXXXX") || fail "no tmpdir"
+srvlog=$(mktemp)
+srv=
+cleanup() {
+    # htsserver keeps SIGTERM ignored across its exec, so only -9 reaps it.
+    test -z "${srv}" || kill -9 "${srv}" 2>/dev/null || true
+    wait "${srv}" 2>/dev/null || true # absorb bash's async "Killed" notice
+    rm -rf "${work}" "${srvlog}"
+}
+trap cleanup EXIT HUP INT QUIT PIPE TERM
+
+# webhttrack server on a pre-picked port; an isolated HOME keeps a stray
+# ~/.httrack.ini out of it.
+sport=$("${python}" -c 'import socket
+s = socket.socket()
+s.bind(("127.0.0.1", 0))
+print(s.getsockname()[1])
+s.close()')
+(
+    trap '' TERM TTOU
+    export HOME="${work}"
+    exec htsserver "${distdir}/" --port "${sport}" >"${srvlog}" 2>&1
+) &
+srv=$!
+for _ in $(seq 1 40); do
+    url=$(sed -n 's/^URL=//p' "${srvlog}") && test -n "${url}" && break
+    kill -0 "${srv}" 2>/dev/null || break
+    sleep 0.25
+done
+test -n "${url:-}" || fail "htsserver did not start: $(cat "${srvlog}")"
+
+# Post a "start" whose -O dir is 'café' in the form's ISO-8859-1 charset.
+"${python}" - "${url}" "${work}" <<'PY'
+import sys, urllib.parse, urllib.request
+
+url, work = sys.argv[1], sys.argv[2]
+outdir = work + "/caf\xe9"  # 'café' as the single byte the browser would send
+# Port 1 refuses at once: only the decoded -O dir matters, not the fetch.
+cmd = "httrack --quiet --robots=0 http://127.0.0.1:1/x.html -O " + outdir
+fields = [("path", work), ("projname", "proj"), ("command_do", "start"),
+          ("winprofile", "x"), ("command", cmd)]
+body = "&".join("%s=%s" % (k, urllib.parse.quote(v, safe="", encoding="latin-1"))
+                for k, v in fields)
+req = urllib.request.Request(url + "step4.html", data=body.encode("latin-1"),
+                             method="POST")
+urllib.request.urlopen(req, timeout=20).read()
+PY
+
+# The crawl runs inside htsserver; wait for the -O dir, then check it chose the
+# UTF-8 name and not the ISO-8859-1 twin.
+utf8dir="${work}/café"
+latin1dir="${work}/caf"$'\xe9'
+for _ in $(seq 1 80); do
+    test -e "${utf8dir}" && break
+    kill -0 "${srv}" 2>/dev/null || break
+    sleep 0.25
+done
+test -e "${utf8dir}" || fail "-O dir never landed as UTF-8 café/: $(find "${work}" -maxdepth 2 2>/dev/null)"
+test ! -e "${latin1dir}" || fail "-O dir created as the ISO-8859-1 twin instead"
+
+echo "PASS"

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

@@ -84,6 +84,7 @@ TESTS = \
 	01_zlib-cache-legacy.test \
 	01_zlib-cache-golden.test \
 	01_zlib-cache-writefail.test \
+	01_zlib-repair-shift.test \
 	01_zlib-savename-cached.test \
 	02_manpage-regen.test \
 	02_update-cache.test \
@@ -148,6 +149,8 @@ TESTS = \
 	64_local-intl-outdir.test \
 	65_port-siblings.test \
 	66_engine-port80-strip.test \
-	67_engine-delayed-truncate.test
+	67_engine-delayed-truncate.test \
+	68_webhttrack-outdir-charset.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