Validate the RFC 1929 sub-negotiation version in the SOCKS5 test server

The server read the sub-negotiation version byte and dropped it on the floor,
so a client sending SOCKS5's 0x05 where RFC 1929 wants 0x01 kept the test
green while every real proxy rejected the handshake. Reject the mismatch the
way a real proxy does, and assert the version reached the proxy.

Verified by mutation: with htsproxy.c patched to send 0x05, the test on master
passes and the fixed test fails.

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

.github/workflows/windows-build.yml

@@ -158,13 +158,21 @@ jobs:
 
           pass=0 fail=0 skip=0 failed="" skipped=""
           for t in 00_runnable.test 01_engine-*.test 01_zlib-*.test \
-                   *_local-*.test 13_crawl_proxy_https.test 58_watchdog.test; do
+                   *_local-*.test 13_crawl_proxy_https.test 58_watchdog.test \
+                   60_crawl-log-salvage.test; do
             rc=0
             run_with_timeout 600 bash "$t" >"$t.log" 2>&1 || rc=$?
             case "$rc" in
               0) pass=$((pass + 1)); echo "PASS $t" ;;
               77) skip=$((skip + 1)) skipped="$skipped $t"; echo "SKIP $t" ;;
-              124) fail=$((fail + 1)) failed="$failed $t"; echo "FAIL $t (timed out, tree killed)" ;;
+              124)
+                fail=$((fail + 1)) failed="$failed $t"
+                echo "FAIL $t (timed out, tree killed)"
+                # Re-running a wedge traced would just hang again: salvage the
+                # killed crawl's own logs into the artifact instead.
+                dump_crawl_logs >>"$t.log"
+                tail -n 25 "$t.log" | sed 's/^/      /'
+                ;;
               *)
                 fail=$((fail + 1)) failed="$failed $t"
                 echo "FAIL $t (exit $rc)"

lang.indexes

@@ -23,7 +23,7 @@ pt:7
 ro:25
 ru:8
 sk:20
-si:24
+sl:24
 sv:17
 tr:10
 uk:22

lang/Finnish.txt

@@ -930,7 +930,6 @@ Server terminated
 Palvelin lopetettu
 A fatal error has occurred during this mirror
 Tällä peilillä tapahtui vakava virhe
-
 Proxy type:
 Välityspalvelimen tyyppi:
 Proxy protocol. HTTP: standard proxy. HTTP (CONNECT tunnel): sends every request through a CONNECT tunnel, for CONNECT-only proxies like Tor's HTTPTunnelPort. SOCKS5: default port 1080.

lang/Italiano.txt

@@ -374,7 +374,7 @@ Create error logging and report files
 Crea file di log per segnalare errori e informazioni
 Generate DOS 8-3 filenames ONLY
 Genera solo nomi di file in formato 8.3
-Generate ISO 9660 filenames ONLY for CDROM medias
+Generate ISO9660 filenames ONLY for CDROM medias
 Genera nomi di file in formato ISO9660 per i CDROM
 Do not create HTML error pages
 Non generare pagine d'errore HTML

lang/Portugues-Brasil.txt

@@ -918,7 +918,7 @@ normal\nextended\ndebug
 normal\nextendido\ncorrigir
 Download web site(s)\nDownload web site(s) + questions\nGet individual files\nDownload all sites in pages (multiple mirror)\nTest links in pages (bookmark test)\n* Continue interrupted download\n* Update existing download
 Copiar site(s) da Web\nCopiar site(s) interativos da Web (perguntas)\nReceber arquivos específicos\nCopiar todas as páginas do site (alternação múltipla)\nTestar links nas páginas (testar indicador)\n* Retomar download interrompido\n* Atualizar download existente
-Relative URL / Absolute URL (default)\nAbsolute URL / Absolute URL\nAbsolute URI / Absolute URL\nOriginal URL / Original URL
+Relative URI / Absolute URL (default)\nAbsolute URL / Absolute URL\nAbsolute URI / Absolute URL\nOriginal URL / Original URL
 URI Relativa / URL Absoluta (padrão)\nURL Absoluta / URL Absoluta\nURI Absoluta / URL Absoluta\nURL Original / URL Original
 Open Source offline browser
 Abrir origem offline no navegador

lang/Slovenian.txt

@@ -3,7 +3,7 @@ Slovenian
 LANGUAGE_FILE
 Slovenian
 LANGUAGE_ISO
-si
+sl
 LANGUAGE_AUTHOR
 Jadran Rudec,iur.\r\njrudec@email.si \r\n
 LANGUAGE_CHARSET

src/htsalias.c

@@ -567,24 +567,29 @@ int optinclude_file(const char *name, int *argc, char **argv, char *x_argvblk,
   return 0;
 }
 
-/* Get home directory, '.' if failed */
+/* Get home directory, '.' if unset or empty */
 /* example: /home/smith */
 const char *hts_gethome(void) {
   const char *home = getenv("HOME");
 
-  if (home)
-    return home;
-  else
-    return ".";
+  /* An empty $HOME would expand ~/foo into the absolute /foo */
+  return strnotempty(home) ? home : ".";
 }
 
-/* Convert ~/foo into /home/smith/foo */
+/* Convert ~/foo into /home/smith/foo (~user/ left alone: no getpwnam here) */
 void expand_home(String * str) {
-  if (StringSub(*str, 1) == '~') {
+  if (StringNotEmpty(*str) && StringSub(*str, 0) == '~' &&
+      (StringLength(*str) == 1 || StringSub(*str, 1) == '/')) {
     char BIGSTK tempo[HTS_URLMAXSIZE * 2];
+    const char *const home = hts_gethome();
+    const size_t homelen = strlen(home);
+    const size_t taillen = StringLength(*str) - 1;
 
-    strcpybuff(tempo, hts_gethome());
-    strcatbuff(tempo, StringBuff(*str) + 1);
-    StringCopy(*str, tempo);
+    /* Leave untouched rather than abort() in strcatbuff on a huge $HOME */
+    if (taillen < sizeof(tempo) && homelen < sizeof(tempo) - taillen) {
+      strcpybuff(tempo, home);
+      strcatbuff(tempo, StringBuff(*str) + 1);
+      StringCopy(*str, tempo);
+    }
   }
 }

src/htscoremain.c

@@ -113,6 +113,7 @@ HTSEXT_API int hts_main(int argc, char **argv) {
 }
 
 static int hts_main_internal(int argc, char **argv, httrackp * opt);
+static hts_boolean cmdl_shortopt_has(const char *s, char c);
 
 // Main, récupère les paramètres et appelle le robot
 HTSEXT_API int hts_main2(int argc, char **argv, httrackp * opt) {
@@ -304,12 +305,12 @@ static int hts_main_internal(int argc, char **argv, httrackp * opt) {
                      hts_get_version_info(opt));
               return 0;
             } else {
-              if (strncmp(tmp_argv[0], "--", 2)) {      /* pas */
-                if ((strchr(tmp_argv[0], 'q') != NULL))
-                  opt->quiet = 1;       // ne pas poser de questions! (nohup par exemple)
-                if ((strchr(tmp_argv[0], 'i') != NULL)) {       // doit.log!
-                  argv_url = -1;        /* forcer */
-                  opt->quiet = 1;
+              if (strncmp(tmp_argv[0], "--", 2)) { /* not a long option */
+                if (cmdl_shortopt_has(tmp_argv[0], 'q'))
+                  opt->quiet = HTS_TRUE; // never ask questions (nohup)
+                if (cmdl_shortopt_has(tmp_argv[0], 'i')) { // doit.log!
+                  argv_url = -1;
+                  opt->quiet = HTS_TRUE;
                 }
               } else if (strcmp(tmp_argv[0] + 2, "quiet") == 0) {
                 opt->quiet = 1; // ne pas poser de questions! (nohup par exemple)
@@ -2807,6 +2808,32 @@ int check_path(String * s, char *defaultname) {
   return return_value;
 }
 
+/* Does the short-option cluster s carry c from the main option set (-i, -iC2,
+   -%Mi)? Walked as the parser does below: %, &, @ and # each take the letter
+   after them into another set, so the i of -%i is not the main-set -i. */
+static hts_boolean cmdl_shortopt_has(const char *s, char c) {
+  const char *com;
+
+  if (s[0] != '-' || s[1] == '-')
+    return HTS_FALSE;
+  for (com = s + 1; *com != '\0'; com++) {
+    switch (*com) {
+    case '%':
+    case '&':
+    case '@':
+    case '#':
+      if (*(com + 1) != '\0')
+        com++; /* skip the other set's letter */
+      break;
+    default:
+      if (*com == c)
+        return HTS_TRUE;
+      break;
+    }
+  }
+  return HTS_FALSE;
+}
+
 // détermine si l'argument est une option
 int cmdl_opt(char *s) {
   if (s[0] == '-') {            // c'est peut être une option

src/htsselftest.c

@@ -45,6 +45,7 @@ Please visit our Website: http://www.httrack.com
 #include "htscore.h"
 #include "htsdefines.h"
 #include "htslib.h"
+#include "htsalias.h"
 #include "htsparse.h"
 #include "htscache.h"
 #include "htscache_selftest.h"
@@ -825,6 +826,21 @@ static int st_simplify(httrackp *opt, int argc, char **argv) {
   return 0;
 }
 
+static int st_expandhome(httrackp *opt, int argc, char **argv) {
+  String path = STRING_EMPTY;
+
+  (void) opt;
+  if (argc < 1) {
+    fprintf(stderr, "expandhome: needs a path\n");
+    return 1;
+  }
+  StringCopy(path, argv[0]);
+  expand_home(&path);
+  printf("expanded=%s\n", StringBuff(path));
+  StringFree(path);
+  return 0;
+}
+
 static int st_mime(httrackp *opt, int argc, char **argv) {
   char mime[256];
 
@@ -3079,6 +3095,7 @@ static const struct selftest_entry {
     {"filterbounds", "", "matcher length/work caps reject hostile patterns",
      st_filterbounds},
     {"simplify", "<path>", "collapse ./ and ../ in a path", st_simplify},
+    {"expandhome", "<path>", "expand a leading ~/ into $HOME", st_expandhome},
     {"stripquery", "", "--strip-query pattern/key stripping self-test",
      st_stripquery},
     {"urlhack", "", "-%u url-hack sub-flag (www/slash/query) self-test",

src/webhttrack

@@ -33,6 +33,19 @@ function log {
     return 0
 }
 
+# Map a POSIX locale ("zh_TW.UTF-8@euro") to its lang.indexes number, English (1) if unknown
+function lang_index {
+    local locale=$1 indexes=$2 tag n t
+    tag=$(echo "${locale}" | cut -f1 -d'.' | cut -f1 -d'@' | tr '[:upper:]' '[:lower:]' | tr -cd 'a-z_')
+    # a few languages carry a region-specific entry (zh_tw, pt_br); else use the bare language
+    for t in "${tag}" "${tag%%_*}"; do
+        n=$(grep -E "^${t}:" "${indexes}" | cut -f2 -d':')
+        test -n "${n}" && break
+    done
+    test -n "${n}" || n=1
+    echo "${n}"
+}
+
 function launch_browser {
     log "Launching $1"
     browser=$1
@@ -59,13 +72,11 @@ test -f "${DISTPATH}/lang.indexes" || ! log "Could not find ${DISTPATH}/lang.ind
 test -d "${DISTPATH}/lang" || ! log "Could not find ${DISTPATH}/lang" || exit 1
 test -d "${DISTPATH}/html" || ! log "Could not find ${DISTPATH}/html" || exit 1
 
-# Locale
-HTSLANG="${LC_MESSAGES}"
-! test -n "${HTSLANG}" && HTSLANG="${LC_ALL}"
+# Locale: POSIX precedence, LC_ALL overrides LC_MESSAGES overrides LANG
+HTSLANG="${LC_ALL}"
+! test -n "${HTSLANG}" && HTSLANG="${LC_MESSAGES}"
 ! test -n "${HTSLANG}" && HTSLANG="${LANG}"
-HTSLANG="$(echo "$LANG" | cut -f1 -d'.' | cut -f1 -d'_')"
-LANGN=$(grep -E "^${HTSLANG}:" "${DISTPATH}/lang.indexes" | cut -f2 -d':')
-! test -n "${LANGN}" && LANGN=1
+LANGN=$(lang_index "${HTSLANG}" "${DISTPATH}/lang.indexes")
 
 # Find the browser
 # note: not all systems have sensible-browser or www-browser alternative

tests/01_engine-cmdline.test

@@ -54,6 +54,13 @@ refused() {
         ! echo "FAIL: $1 (exit $RC)" || exit 1
 }
 
+# assert continue mode was entered: it drops the URL list, so with no cache to
+# resume the run ends on the usage screen rather than on any other error
+continued() {
+    { test "$RC" -ne 0 && grep -q 'usage:' "$1/.log"; } ||
+        ! echo "FAIL: $2 (exit $RC)" || exit 1
+}
+
 # a value past the old 126/256 caps but within the cap is accepted, on both the
 # short and long form of each option
 long=$(nchars 900)
@@ -102,4 +109,50 @@ for bad in nan nan:5 5:nan inf 10:5 99999; do
     refused "#185: invalid --pause '$bad' not refused cleanly"
 done
 
+# An option is not -i (continue) merely because its name contains an 'i' (#615).
+# These used to wipe the URL given before them and exit on the usage screen.
+run "$tmp/ord-bti" --build-top-index
+accepted "$tmp/ord-bti" "#615: --build-top-index after the URL wiped the URL list"
+run "$tmp/ord-bti-s" "-%i"
+accepted "$tmp/ord-bti-s" "#615: -%i after the URL wiped the URL list"
+run "$tmp/ord-proto" --protocol 2
+accepted "$tmp/ord-proto" "#615: --protocol after the URL wiped the URL list"
+run "$tmp/ord-proto-s" "-@i2"
+accepted "$tmp/ord-proto-s" "#615: -@i2 after the URL wiped the URL list"
+
+# %, &, @ and # take the letter after them into another option set, wherever
+# they sit in the cluster, so the i of -q%i is not the main-set -i either.
+run "$tmp/ord-qpi" "-q%i"
+accepted "$tmp/ord-qpi" "#615: -q%i after the URL wiped the URL list"
+run "$tmp/ord-qai" "-q@i"
+accepted "$tmp/ord-qai" "#615: -q@i after the URL wiped the URL list"
+
+# -%q only forces quiet mode, which nothing here can see: a run whose stdout is
+# not a tty is quiet from the start (htscoremain.c:174). Just check it crawls.
+run "$tmp/ord-iqs" "-%q"
+accepted "$tmp/ord-iqs" "#615: -%q after the URL broke the crawl"
+
+# The real -i still forces continue mode and drops the URL list, in every form:
+# a fix that merely stopped looking for 'i' would pass the checks above.
+run "$tmp/cont-s" -i
+continued "$tmp/cont-s" "#615: -i after the URL no longer forces continue mode"
+run "$tmp/cont-c" -iC2
+continued "$tmp/cont-c" "#615: -iC2 after the URL no longer forces continue mode"
+run "$tmp/cont-l" --continue
+continued "$tmp/cont-l" "#615: --continue after the URL no longer forces continue mode"
+run "$tmp/cont-u" --update
+continued "$tmp/cont-u" "#615: --update after the URL no longer forces continue mode"
+
+# ...including where another set's letter precedes it in the cluster.
+run "$tmp/cont-pq" "-%qi"
+continued "$tmp/cont-pq" "#615: -%qi after the URL no longer forces continue mode"
+run "$tmp/cont-pm" "-%Mi"
+continued "$tmp/cont-pm" "#615: -%Mi after the URL no longer forces continue mode"
+
+# Placed before the URL these always worked; they must keep working.
+run_only "$tmp/pre-bti" "-%i" "file://$tmp/index.html"
+accepted "$tmp/pre-bti" "#615: -%i before the URL broke the crawl"
+run_only "$tmp/pre-proto" "-@i2" "file://$tmp/index.html"
+accepted "$tmp/pre-proto" "#615: -@i2 before the URL broke the crawl"
+
 exit 0

tests/01_engine-expandhome.test

@@ -0,0 +1,53 @@
+#!/bin/bash
+#
+
+# SC2088: the literal, unexpanded '~' is the input under test.
+# shellcheck disable=SC2088
+
+set -euo pipefail
+
+# ~ expansion for the -O base path (expand_home). $HOME is pinned so the cases
+# cannot depend on the caller's, but MSYS rewrites it into a Windows path
+# before the native httrack.exe reads it: ask the engine what it saw rather
+# than hardcoding the value.
+ask() {
+    HOME=HTSHOME httrack -O /dev/null -#test=expandhome "$1"
+}
+exp() {
+    test "$(ask "$1")" == "expanded=$2" || exit 1
+}
+
+home=$(ask '~')
+home=${home#expanded=}
+
+# or every case below is vacuous: '~' means expansion never fired, '.' means it
+# fell back to the default without ever reading $HOME
+test "$home" != '~' || exit 1
+test "$home" != '.' || exit 1
+
+exp '~' "$home"
+exp '~/foo' "$home/foo"
+exp '~/foo/bar/#' "$home/foo/bar/#"
+exp '~/' "$home/"
+exp '~/../x' "$home/../x"
+
+# ~user/ needs getpwnam: left alone rather than mangled into $HOME + "user/"
+exp '~smith/foo' '~smith/foo'
+exp '~root' '~root'
+
+# only a leading '~' expands; anything else is a literal path component
+exp 'a~/x' 'a~/x'
+exp './~/x' './~/x'
+exp 'foo~bar' 'foo~bar'
+
+# an unset or empty $HOME must not turn ~/foo into the absolute /foo
+test "$(env -u HOME httrack -O /dev/null -#test=expandhome '~/foo')" == "expanded=./foo" || exit 1
+test "$(HOME='' httrack -O /dev/null -#test=expandhome '~/foo')" == "expanded=./foo" || exit 1
+
+# A $HOME past the 2 * HTS_URLMAXSIZE buffer is left alone: strcatbuff aborts
+# rather than truncates. Only meaningful where the engine sees the value we set:
+# MSYS rewrites HOME into a path of its own choosing, long or not.
+if test "$home" = HTSHOME; then
+    long=$(printf '%04096d' 0 | tr '0' 'A')
+    test "$(HOME="$long" httrack -O /dev/null -#test=expandhome '~/foo')" == "expanded=~/foo" || exit 1
+fi

tests/52_local-socks5.test

@@ -180,6 +180,12 @@ grep -q "^AUTH user secret\$" "$auth/socks.log" || {
     cat "$auth/socks.log" >&2
     exit 1
 }
+# the sub-negotiation carries RFC 1929's 0x01, not the 0x05 of the SOCKS5 greeting
+grep -q "^AUTHVER 1\$" "$auth/socks.log" || {
+    echo "FAIL: wrong RFC 1929 sub-negotiation version" >&2
+    cat "$auth/socks.log" >&2
+    exit 1
+}
 grep -qi "secret\|proxy-authorization" "$auth/origin-headers.log" && {
     echo "FAIL: socks credentials leaked into the origin request" >&2
     cat "$auth/origin-headers.log" >&2

tests/60_crawl-log-salvage.test

@@ -0,0 +1,46 @@
+#!/bin/bash
+#
+# Unit-tests dump_crawl_logs (testlib.sh): the killed test's hts-log.txt must
+# reach the uploaded logs, or a Windows wedge past --max-time stays
+# undiagnosable (#605).
+
+set -euo pipefail
+
+testdir=$(cd "$(dirname "$0")" && pwd)
+# shellcheck source=tests/testlib.sh
+. "${testdir}/testlib.sh"
+
+TMPDIR=$(mktemp -d "${TMPDIR:-/tmp}/httrack_salv.XXXXXX")
+export TMPDIR
+trap 'rm -rf "$TMPDIR"' EXIT
+fail() {
+    echo "FAIL: $*" >&2
+    exit 1
+}
+
+verdict="More than 120 seconds passed.. giving up"
+
+# Vacuity control: with nothing left behind the assertions below cannot pass on
+# ambient output.
+test -z "$(dump_crawl_logs)" || fail "dumped something with no crawl tmpdir"
+
+# A killed local-crawl.sh leaves exactly this shape.
+d="${TMPDIR}/httrack_local.probe"
+mkdir -p "$d/crawl"
+echo "$verdict" >"$d/crawl/hts-log.txt"
+echo "httrack stdout marker" >"$d/log"
+
+# Append to an unterminated line, as a killed test's own log always ends.
+art="${TMPDIR}/artifact.log"
+printf '[running httrack ...] ..\t' >"$art"
+dump_crawl_logs >>"$art"
+
+grep -qF "$verdict" "$art" || fail "hts-log.txt verdict not salvaged"
+grep -qF "httrack stdout marker" "$art" || fail "httrack stdout not salvaged"
+grep -q '^--- .*hts-log.txt' "$art" || fail "dump glued its header onto the unterminated line"
+
+# Cleared, so the next timing-out test cannot re-report this crawl as its own.
+test ! -d "$d" || fail "salvaged tmpdir left behind"
+test -z "$(dump_crawl_logs)" || fail "salvaged crawl reported twice"
+
+echo "crawl log salvage OK"

tests/61_webhttrack-locale.test

@@ -0,0 +1,91 @@
+#!/bin/bash
+#
+# webhttrack must map the caller's locale to the right lang.indexes number.
+# Guards: LC_ALL/LC_MESSAGES were ignored, and zh_tw/pt_br were unreachable.
+
+set -euo pipefail
+
+testdir=$(cd "$(dirname "$0")" && pwd)
+distdir=$(cd "${testdir}/.." && pwd)
+script="${distdir}/src/webhttrack"
+
+fail() {
+    echo "FAIL: $*" >&2
+    exit 1
+}
+
+test -r "${script}" || fail "no ${script}"
+test -r "${distdir}/lang.indexes" || fail "no ${distdir}/lang.indexes"
+
+# Lift the locale code out of the launcher, so this tests the shipped logic itself.
+funcs=$(sed -n '/^function lang_index/,/^}/p' "${script}")
+# End on the next section, not a blank line: a blank line grown inside the block
+# would truncate the lift and silently stop testing everything below it.
+grep -q '^# Find the browser' "${script}" || fail "locale block terminator moved in ${script}"
+block=$(sed -n '/^# Locale/,/^# Find the browser/p' "${script}" | sed '$d')
+echo "${block}" | grep -q 'LANGN=.*lang_index' || fail "could not lift the whole locale block from ${script}"
+
+# Run the lifted block against whatever locale vars the caller exported.
+run_block() {
+    # shellcheck disable=SC2034 # read by the eval'd locale block
+    DISTPATH="${distdir}"
+    # the launcher runs under neither -e ("! test -n x && y" returns 1) nor -u
+    set +eu
+    eval "${funcs}"
+    eval "${block}"
+    # shellcheck disable=SC2153 # LANGN is set by the eval'd block, not a LANG typo
+    echo "${LANGN}"
+}
+
+# Resolve LANG/LC_ALL/LC_MESSAGES (empty = unset) to the language number.
+lang_of() {
+    (
+        unset LANG LC_ALL LC_MESSAGES
+        # only the value matters here; hide bash's setlocale gripe when the locale is not installed
+        # shellcheck disable=SC2030 # confining the locale to this subshell is the point
+        {
+            if test -n "$1"; then export LANG="$1"; fi
+            if test -n "$2"; then export LC_ALL="$2"; fi
+            if test -n "$3"; then export LC_MESSAGES="$3"; fi
+        } 2>/dev/null
+        run_block
+    )
+}
+
+check() {
+    local want=$1 got desc
+    shift
+    desc="LANG='$1' LC_ALL='$2' LC_MESSAGES='$3'"
+    got=$(lang_of "$1" "$2" "$3")
+    test "${got}" = "${want}" || fail "${desc}: want ${want}, got ${got}"
+    echo "ok: ${desc} -> ${got}"
+}
+
+#      want  LANG            LC_ALL          LC_MESSAGES
+check 2 "fr_FR.UTF-8" "" ""
+check 14 "zh_TW.UTF-8" "" ""                      # region-qualified: Chinese-BIG5, not Simplified (13)
+check 12 "pt_BR.UTF-8" "" ""                      # region-qualified: Portugues-Brasil, not Portugues (7)
+check 13 "zh_CN.UTF-8" "" ""                      # unlisted region falls back to the bare language
+check 24 "sl_SI.UTF-8" "" ""                      # Slovenian is ISO 639-1 "sl"
+check 4 "" "de_DE.UTF-8" ""                       # LC_ALL alone
+check 4 "" "" "de_DE.UTF-8"                       # LC_MESSAGES alone
+check 4 "fr_FR.UTF-8" "de_DE.UTF-8" ""            # LC_ALL wins over LANG
+check 4 "fr_FR.UTF-8" "" "de_DE.UTF-8"            # LC_MESSAGES wins over LANG
+check 2 "de_DE.UTF-8" "fr_FR.UTF-8" "de_DE.UTF-8" # LC_ALL wins over LC_MESSAGES
+check 2 "fr_FR@euro" "" ""                        # @modifier without a charset
+check 1 "C" "" ""                                 # C, unknown and unset all mean English
+check 1 "xx_YY.UTF-8" "" ""
+check 1 "" "" ""
+
+# Exported but empty is unset (POSIX), so it must not mask LANG. lang_of cannot
+# express this: it exports nothing for an empty argument.
+got=$(
+    unset LANG LC_ALL LC_MESSAGES
+    # shellcheck disable=SC2031 # confining the locale to this subshell is the point
+    { export LANG="fr_FR.UTF-8" LC_ALL="" LC_MESSAGES=""; } 2>/dev/null
+    run_block
+)
+test "${got}" = 2 || fail "empty LC_ALL/LC_MESSAGES must not mask LANG: want 2, got ${got}"
+echo "ok: LANG='fr_FR.UTF-8' with LC_ALL='' exported -> ${got}"
+
+echo "PASS"

tests/62_lang-integrity.test

@@ -0,0 +1,90 @@
+#!/bin/bash
+#
+# lang/*.txt are line pairs: an English msgid, then its translation. A stray
+# blank or a drifted msgid silently unhooks every translation that follows,
+# rather than failing, so assert the pairing and the join against English.txt.
+# Pairs are physical lines here; a msgid ending in \ continues onto the next one
+# for the engine (linput_cpp) but not for us. None do today.
+
+set -euo pipefail
+
+langdir="${top_srcdir:-..}/lang"
+def="${top_srcdir:-..}/lang.def"
+eng="$langdir/English.txt"
+for f in "$eng" "$def"; do
+    [ -f "$f" ] || {
+        echo "cannot find $f"
+        exit 1
+    }
+done
+
+tmp=$(mktemp -d)
+trap 'rm -rf "$tmp"' EXIT
+keys="$tmp/english.keys"
+
+# Byte-wise: each file is in its own declared legacy charset, lang.def is CRLF.
+LC_ALL=C awk 'NR%2==1 { sub(/\r$/, ""); print }' "$eng" >"$keys"
+
+# An empty parse would make every check below pass vacuously.
+nkeys=$(wc -l <"$keys")
+if [ "$nkeys" -lt 400 ]; then
+    echo "only $nkeys msgids parsed from $eng; the parse is broken"
+    exit 1
+fi
+
+nfiles=0
+fail=0
+for f in "$langdir"/*.txt; do
+    nfiles=$((nfiles + 1))
+    LC_ALL=C awk -v keys="$keys" -v name="${f##*/}" '
+        BEGIN { while ((getline k < keys) > 0) known[k] = 1 }
+        { sub(/\r$/, "") }
+        NR % 2 == 1 {
+            if ($0 == "")
+                printf "%s:%d: blank line where an English msgid belongs\n", name, NR
+            else if (!($0 in known))
+                printf "%s:%d: msgid absent from English.txt: %s\n", name, NR, $0
+            else
+                next
+            bad++
+        }
+        END {
+            if (NR % 2) {
+                printf "%s: odd line count (%d): msgid/translation pairing is broken\n", name, NR
+                bad++
+            }
+            exit(bad > 0)
+        }
+    ' "$f" || fail=1
+done
+
+if [ "$nfiles" -lt 25 ]; then
+    echo "only $nfiles language files found in $langdir"
+    exit 1
+fi
+
+# lang.def maps each LANG_* macro to the English string the GUI compiles in,
+# which is the lookup key into every lang/*.txt: no msgid, no translations.
+LC_ALL=C awk -v keys="$keys" '
+    BEGIN { while ((getline k < keys) > 0) known[k] = 1 }
+    { sub(/\r$/, "") }
+    NR % 2 == 1 { macro = $0; next }
+    macro ~ /^(LANG_|LISTDEF_)/ {
+        n++
+        if (!($0 in known)) {
+            printf "lang.def:%d: %s maps to \"%s\", which is not a msgid in English.txt\n", NR, macro, $0
+            bad++
+        }
+    }
+    END {
+        if (n < 400) {
+            printf "lang.def: only %d LANG_ entries parsed; the parse is broken\n", n
+            bad++
+        }
+        exit(bad > 0)
+    }
+' "$def" || fail=1
+
+[ "$fail" -eq 0 ] || exit 1
+
+echo "$nfiles language files consistent with $nkeys English msgids"

tests/Makefile.am

@@ -59,6 +59,7 @@ TESTS = \
 	01_engine-pause.test \
 	01_engine-rcfile.test \
 	01_engine-reconcile.test \
+	01_engine-expandhome.test \
 	01_engine-fsize.test \
 	01_engine-redirect.test \
 	01_engine-relative.test \
@@ -138,6 +139,9 @@ TESTS = \
 	56_local-proxy-noleak.test \
 	57_local-proxy-connect.test \
 	58_watchdog.test \
-	59_local-tls-stall.test
+	59_local-tls-stall.test \
+	60_crawl-log-salvage.test \
+	61_webhttrack-locale.test \
+	62_lang-integrity.test
 
 CLEANFILES = check-network_sh.cache

tests/socks5-server.py

@@ -24,6 +24,7 @@ from proxytestlib import bind_ephemeral, pipe  # noqa: E402
 # The one name the proxy answers for; a .invalid TLD never resolves (RFC 6761),
 # so a locally-resolving client could not reach us -- success proves remote DNS.
 REMOTE_HOST = b"socks-origin.invalid"
+AUTH_VERSION = 0x01  # RFC 1929 sub-negotiation version
 # index links the subpages so an -r3 crawl reuses one keep-alive socket
 LINKS = "".join('<a href="/p%d.html">%d</a>' % (i, i) for i in range(1, 6))
 ORIGIN_BODY = ("<html><body>ORIGIN-PAGE-563 " + LINKS + "</body></html>").encode()
@@ -104,6 +105,13 @@ def negotiate_auth(conn, logdir):
     if 0x02 in methods:  # prefer user/pass so the auth test exercises RFC 1929
         conn.sendall(b"\x05\x02")
         (subver,) = recvn(conn, 1)
+        # RFC 1929's version byte is 0x01, not SOCKS5's 0x05: a real proxy
+        # rejects a mismatch instead of tunnelling anyway
+        if subver != AUTH_VERSION:
+            log(logdir, "AUTHVER-BAD %d" % subver)
+            conn.sendall(bytes([AUTH_VERSION, 0x01]))  # sub-negotiation failure
+            return False
+        log(logdir, "AUTHVER %d" % subver)
         (ulen,) = recvn(conn, 1)
         uname = recvn(conn, ulen)
         (plen,) = recvn(conn, 1)

tests/testlib.sh

@@ -44,6 +44,25 @@ stop_server() {
     return 0
 }
 
+# Dump and clear the crawl logs a hard-killed test leaves in TMPDIR (its cleanup
+# trap never ran): hts-log.txt alone records "More than N seconds passed.. giving
+# up", so a wedge past --max-time is undiagnosable without it (#605).
+dump_crawl_logs() {
+    local d f
+    for d in "${TMPDIR:-/tmp}"/httrack_local.*; do
+        test -d "$d" || continue
+        for f in "$d/crawl/hts-log.txt" "$d/log" "$d/log.2"; do
+            test -f "$f" || continue
+            # Leading newline: the killed test's last line has no terminator.
+            printf '\n--- %s (last 200 lines)\n' "$f"
+            tail -n 200 "$f"
+        done
+        # so a later test's dump cannot re-report this one; never fatal, the
+        # caller is already handling a failure and Windows may still hold a file
+        rm -rf "$d" || true
+    done
+}
+
 # Kill a backgrounded job and its whole descendant tree. POSIX: the caller must
 # have put the job in its own process group (run_with_timeout does) so we signal
 # the group; a bare kill would orphan the grandchildren. Windows: the tree is