tests: cover anchored-link (#frag) rewriting (#279) (#470)

An anchored hyperlink target.html#sec must fetch the target with the
fragment dropped yet keep the fragment in the rewritten local link so
the anchor still resolves. This already works; #279 is a stale
report from the Google Code era with no current repro.

Pin the behavior with a local-crawl test: the strict server 400s on a
'#' in the request-target (so a leaked fragment fails the fetch), and a
new --file-matches audit asserts the mirrored link keeps #sec/#sec2 for
both the unquoted and quoted forms.

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

src/htsparse.c

@@ -108,88 +108,97 @@ Please visit our Website: http://www.httrack.com
 
 #define HT_ADD_FOP
 
+/* Mutable extended-context fields: one source of truth so the DEFINE/SET/SAVE
+   load and store lists can't drift apart. */
+/* clang-format off */
+#define ENGINE_MUTABLE_FIELDS(X) \
+  X(int, error, stre->error_) \
+  X(int, store_errpage, stre->store_errpage_) \
+  X(int, makeindex_done, stre->makeindex_done_) \
+  X(FILE *, makeindex_fp, stre->makeindex_fp_) \
+  X(int, makeindex_links, stre->makeindex_links_) \
+  X(LLint, stat_fragment, stre->stat_fragment_)
+
+#define ENGINE_FIELD_DECLARE(type, name, src) type name = *(src);
+#define ENGINE_FIELD_LOAD(type, name, src) name = *(src);
+#define ENGINE_FIELD_STORE(type, name, src) *(src) = name;
+
 #define ENGINE_DEFINE_CONTEXT() \
   ENGINE_DEFINE_CONTEXT_BASE(); \
-  /* */ \
   htsblk* const r HTS_UNUSED = stre->r_; \
   hash_struct* const hash HTS_UNUSED = stre->hash_; \
   char* const codebase HTS_UNUSED = stre->codebase; \
   char* const base HTS_UNUSED = stre->base; \
-  /* */ \
   const char * const template_header HTS_UNUSED = stre->template_header_; \
   const char * const template_body HTS_UNUSED = stre->template_body_; \
   const char * const template_footer HTS_UNUSED = stre->template_footer_; \
-  /* */ \
   HTS_UNUSED char* const makeindex_firstlink = stre->makeindex_firstlink_; \
-  /* */ \
-  /* */ \
-  int error = * stre->error_; \
-  int store_errpage = * stre->store_errpage_; \
-  /* */ \
-  int makeindex_done = *stre->makeindex_done_; \
-  FILE* makeindex_fp = *stre->makeindex_fp_; \
-  int makeindex_links = *stre->makeindex_links_; \
-  /* */ \
-  LLint stat_fragment = *stre->stat_fragment_; \
+  ENGINE_MUTABLE_FIELDS(ENGINE_FIELD_DECLARE) \
+  /* load-once (kept out of SET/SAVE): re-reading would reset the throttle */ \
   HTS_UNUSED TStamp makestat_time = stre->makestat_time; \
   HTS_UNUSED FILE* makestat_fp = stre->makestat_fp
 
-/* clang-format off: an edit realigns all backslashes, churning the macro. */
-/* clang-format off */
-/* Load-once: re-reading resets makestat_time (mutated locally, never SAVEd). */
 #define ENGINE_SET_CONTEXT() \
   ENGINE_SET_CONTEXT_BASE(); \
-  /* */ \
-  error = * stre->error_; \
-  store_errpage = * stre->store_errpage_; \
-  /* */ \
-  makeindex_done = *stre->makeindex_done_; \
-  makeindex_fp = *stre->makeindex_fp_; \
-  makeindex_links = *stre->makeindex_links_; \
-  /* */ \
-  stat_fragment = *stre->stat_fragment_
-/* clang-format on */
+  ENGINE_MUTABLE_FIELDS(ENGINE_FIELD_LOAD)
 
 #define ENGINE_LOAD_CONTEXT() \
   ENGINE_DEFINE_CONTEXT()
 
 #define ENGINE_SAVE_CONTEXT() \
   ENGINE_SAVE_CONTEXT_BASE(); \
-  /* */ \
-  * stre->error_ = error; \
-  * stre->store_errpage_ = store_errpage; \
-  /* */ \
-  *stre->makeindex_done_ = makeindex_done; \
-  *stre->makeindex_fp_ = makeindex_fp; \
-  *stre->makeindex_links_ = makeindex_links; \
-  /* */ \
-  *stre->stat_fragment_ = stat_fragment
+  ENGINE_MUTABLE_FIELDS(ENGINE_FIELD_STORE)
+/* clang-format on */
 
 #define _ROBOTS      ((robots_wizard*)opt->robotsptr)
 
-/* Apply current *adr character for the script automate */
-#define AUTOMATE_LOOKUP_CURRENT_ADR() do { \
-  if (inscript) { \
-  int new_state_pos; \
-  new_state_pos=inscript_state[inscript_state_pos][(unsigned char)*html]; \
-  if (new_state_pos < 0) { \
-  new_state_pos=inscript_state[inscript_state_pos][INSCRIPT_DEFAULT]; \
-  } \
-  assertf(new_state_pos >= 0); \
-  assertf(new_state_pos*sizeof(inscript_state[0]) < sizeof(inscript_state)); \
-  inscript_state_pos=new_state_pos; \
-  } \
-} while(0)
+/* JS-detection automaton states; INSCRIPT_DEFAULT is the synthetic "any other
+   char" column of the transition table. */
+typedef enum {
+  INSCRIPT_START = 0,
+  INSCRIPT_ANTISLASH,
+  INSCRIPT_INQUOTE,
+  INSCRIPT_INQUOTE2,
+  INSCRIPT_SLASH,
+  INSCRIPT_SLASHSLASH,
+  INSCRIPT_COMMENT,
+  INSCRIPT_COMMENT2,
+  INSCRIPT_ANTISLASH_IN_QUOTE,
+  INSCRIPT_ANTISLASH_IN_QUOTE2,
+  INSCRIPT_DEFAULT = 256
+} INSCRIPT;
 
-/* Increment current pointer to 'steps' characters, modifying automate if necessary */
-#define INCREMENT_CURRENT_ADR(steps) do { \
-  int steps__ = (int) ( steps ); \
-  while(steps__ > 0) { \
-  html++; \
-  AUTOMATE_LOOKUP_CURRENT_ADR(); \
-  steps__ --; \
-  } \
-} while(0)
+#define INSCRIPT_NSTATES 10 /* rows in the transition table */
+
+/* Live view of the parser's automaton locals, set up once so the helpers below
+   can drive it without capturing them by lexical scope. */
+typedef struct {
+  const int *inscript;             /* nonzero while inside a script body */
+  const signed char (*table)[257]; /* [INSCRIPT_NSTATES][257] transitions */
+  INSCRIPT *pos;                   /* current state */
+  const char **html;               /* parse cursor */
+} script_automate;
+
+/* Feed the current *html byte to the automaton. No-op outside a script body. */
+static void hts_automate_lookup(const script_automate *aut) {
+  if (*aut->inscript) {
+    int next = aut->table[*aut->pos][(unsigned char) **aut->html];
+    if (next < 0) {
+      next = aut->table[*aut->pos][INSCRIPT_DEFAULT];
+    }
+    assertf(next >= 0 && next < INSCRIPT_NSTATES);
+    *aut->pos = (INSCRIPT) next;
+  }
+}
+
+/* Advance the cursor by 'steps' bytes, feeding each to the automaton. */
+static void hts_automate_increment(const script_automate *aut, int steps) {
+  while (steps > 0) {
+    (*aut->html)++;
+    hts_automate_lookup(aut);
+    steps--;
+  }
+}
 
 /* Percent-encode the angle brackets of a string so it is safe to embed inside
    an HTML comment (the default footer) or any other HTML context. A URL holding
@@ -334,20 +343,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
       int incomment = 0;        // dans un <!--
       int inscript = 0;         // dans un scipt pour applets javascript)
       int inscript_locked = 0;  // in locked script (ie. js file)
-      signed char inscript_state[10][257];
-      typedef enum {
-        INSCRIPT_START = 0,
-        INSCRIPT_ANTISLASH,
-        INSCRIPT_INQUOTE,
-        INSCRIPT_INQUOTE2,
-        INSCRIPT_SLASH,
-        INSCRIPT_SLASHSLASH,
-        INSCRIPT_COMMENT,
-        INSCRIPT_COMMENT2,
-        INSCRIPT_ANTISLASH_IN_QUOTE,
-        INSCRIPT_ANTISLASH_IN_QUOTE2,
-        INSCRIPT_DEFAULT = 256
-      } INSCRIPT;
+      signed char inscript_state[INSCRIPT_NSTATES][257];
       INSCRIPT inscript_state_pos = INSCRIPT_START;
       const char *inscript_name = NULL;       // script tag name
       int inscript_tag = 0;     // on est dans un <body onLoad="... terminé par >
@@ -408,6 +404,8 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
       inscript_state[INSCRIPT_COMMENT2]['*'] = INSCRIPT_COMMENT2;
       inscript_state[INSCRIPT_ANTISLASH_IN_QUOTE][INSCRIPT_DEFAULT] = INSCRIPT_INQUOTE; /* #8: escape in '' */
       inscript_state[INSCRIPT_ANTISLASH_IN_QUOTE2][INSCRIPT_DEFAULT] = INSCRIPT_INQUOTE2;       /* #9: escape in "" */
+      const script_automate saut = {&inscript, inscript_state,
+                                    &inscript_state_pos, &html};
 
       /* Primary list or URLs */
       if (ptr == 0) {
@@ -846,7 +844,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
             }
 
             /* automate */
-            AUTOMATE_LOOKUP_CURRENT_ADR();
+            hts_automate_lookup(&saut);
 
             // Note:
             // Certaines pages ne respectent pas le html
@@ -1762,7 +1760,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
             }
             // sauter espaces
             // adr+=p;
-            INCREMENT_CURRENT_ADR(p);
+            hts_automate_increment(&saut, p);
             while((is_space(*html)
                    || (inscriptgen && html[0] == '\\' && is_space(html[1])
                    )
@@ -1777,7 +1775,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
                 }
               // puis quitter
               // html++;    // sauter les espaces, "" et cie
-              INCREMENT_CURRENT_ADR(1);
+              hts_automate_increment(&saut, 1);
             }
 
             /* Stop at \n (LF) if primary links or link lists */
@@ -1792,7 +1790,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
               if (*html == '\\') {
                 if ((*(html + 1) == '\'') || (*(html + 1) == '"')) {      // \" ou \'
                   // html+=2;    // sauter
-                  INCREMENT_CURRENT_ADR(2);
+                  hts_automate_increment(&saut, 2);
                 }
               }
             }
@@ -1840,7 +1838,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
             if (srcset_p) {
               while(html < r->adr + r->size
                     && (is_realspace(*html) || *html == ','))
-                INCREMENT_CURRENT_ADR(1);
+                hts_automate_increment(&saut, 1);
             }
             eadr = html;
 
@@ -3300,7 +3298,7 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
 
             assertf(eadr - html >= 0);   // Should not go back
             if (eadr > html) {
-              INCREMENT_CURRENT_ADR(eadr - 1 - html);
+              hts_automate_increment(&saut, (int) (eadr - 1 - html));
             }
             // adr=eadr-1;  // ** sauter
 
@@ -3319,7 +3317,8 @@ int htsparse(htsmoduleStruct * str, htsmoduleStructExtended * stre) {
                   q++;          // skip whitespace and empty candidates
                 if (q < endp && *q != '\0' && *q != ',' && *q != quote
                     && *q != '<' && *q != '>' && (unsigned char) *q >= 32) {
-                  INCREMENT_CURRENT_ADR(q - html);   // keep the automate in sync
+                  hts_automate_increment(
+                      &saut, (int) (q - html)); // keep the automate in sync
                   ok = 1;
                   goto srcset_next;
                 }

tests/30_local-fragment-link.test

@@ -0,0 +1,13 @@
+#!/bin/bash
+# Issue #279: an anchored link (target.html#sec, quoted or bare) fetches the
+# target with the fragment dropped (strict server 400s on a '#' in the request)
+# but keeps it in the rewritten local link so the anchor still works.
+set -e
+
+: "${top_srcdir:=..}"
+
+bash "$top_srcdir/tests/local-crawl.sh" --errors 0 \
+    --found 'fraglink/target.html' \
+    --file-matches 'fraglink/index.html' 'href=target\.html#sec' \
+    --file-matches 'fraglink/index.html' 'href="target\.html#sec2"' \
+    httrack 'BASEURL/fraglink/index.html'

tests/Makefile.am

@@ -6,6 +6,7 @@ EXTRA_DIST = $(TESTS) crawl-test.sh run-all-tests.sh check-network.sh \
 	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 \
+	server-root/fraglink/index.html server-root/fraglink/target.html \
 	fixtures/cache-golden/hts-cache/new.zip
 
 TESTS_ENVIRONMENT =
@@ -83,6 +84,7 @@ TESTS = \
 	26_local-strip-query.test \
 	27_local-cookies-file.test \
 	28_local-pause.test \
-	29_local-redirect-fragment.test
+	29_local-redirect-fragment.test \
+	30_local-fragment-link.test
 
 CLEANFILES = check-network_sh.cache

tests/local-crawl.sh

@@ -15,8 +15,11 @@
 #   bash local-crawl.sh [--tls] [--root DIR] [--cookie NAME=VALUE ...] \
 #       --errors N --files N --found PATH ... --directory PATH ... \
 #       --log-found REGEX ... --log-not-found REGEX ... \
+#       --file-matches PATH REGEX ... --file-not-matches PATH REGEX ... \
 #       httrack BASEURL/some/path [httrack-args...]
 # --log-found/--log-not-found grep (ERE) the crawl's hts-log.txt.
+# --file-matches/--file-not-matches grep (ERE) a mirrored file (PATH under the
+# host root), to assert rewritten link/content survived the crawl.
 # --cookie writes a Netscape cookies.txt (scoped to the discovered host:port,
 # which the ephemeral port forces into the cookie domain) and passes it to
 # httrack via --cookies-file, to exercise preloaded cookies.
@@ -121,6 +124,10 @@ while test "$pos" -lt "$nargs"; do
         audit+=("${args[$pos]}" "${args[$((pos + 1))]}")
         pos=$((pos + 1))
         ;;
+    --file-matches | --file-not-matches)
+        audit+=("${args[$pos]}" "${args[$((pos + 1))]}" "${args[$((pos + 2))]}")
+        pos=$((pos + 2))
+        ;;
     httrack)
         pos=$((pos + 1))
         break
@@ -294,6 +301,24 @@ while test "$i" -lt "${#audit[@]}"; do
             exit 1
         else result "OK"; fi
         ;;
+    --file-matches)
+        path="${audit[$((i + 1))]}"
+        i=$((i + 2))
+        info "checking ${path} matches ${audit[$i]}"
+        if grep -aqE "${audit[$i]}" "${hostroot}/${path}"; then result "OK"; else
+            result "no match"
+            exit 1
+        fi
+        ;;
+    --file-not-matches)
+        path="${audit[$((i + 1))]}"
+        i=$((i + 2))
+        info "checking ${path} lacks ${audit[$i]}"
+        if grep -aqE "${audit[$i]}" "${hostroot}/${path}"; then
+            result "matched"
+            exit 1
+        else result "OK"; fi
+        ;;
     esac
     i=$((i + 1))
 done

tests/server-root/fraglink/index.html

@@ -0,0 +1,4 @@
+<html><body>
+<a href=target.html#sec>unquoted fragment link</a>
+<a href="target.html#sec2">quoted fragment link</a>
+</body></html>

tests/server-root/fraglink/target.html

@@ -0,0 +1 @@
+<html><body><a name="sec"></a><a name="sec2"></a>target</body></html>