Fixed hashtable corruption because of dirty code directly modifying the host address in memory, leading to have hashtable positions not anymore valid.

This issue was especially triggered when a redirect was processed ("Warning moved treated for .." messages)
  * closes: #43

4 files changed, 63 insertions, 3 deletions

diff --git a/src/htscore.c b/src/htscore.c
index 94ed251..6fe302a 100644
--- a/src/htscore.c
+++ b/src/htscore.c
@@ -2355,7 +2355,7 @@ void host_ban(httrackp * opt, lien_url ** liens, int ptr, int lien_tot,
           if (strfield2(jump_identification(liens[i]->adr), host)) {    // host
             hts_log_print(opt, LOG_DEBUG, "Cancel: %s%s", liens[i]->adr,
                           liens[i]->fil);
-            strcpybuff(liens[i]->adr, "!");     // cancel (invalide hash)
+            hash_invalidate_entry(opt->hash, i);  // invalidate hashtable entry
             // on efface pas le hash, because si on rencontre le lien, reverif sav..
           }
         } else {
diff --git a/src/htshash.c b/src/htshash.c
index 7bdf782..9389455 100644
--- a/src/htshash.c
+++ b/src/htshash.c
@@ -85,6 +85,14 @@ static int key_sav_equals(void *arg, const char *a, const char *b) {
   return strcasecmp(a, b) == 0;
 }
 
+static const char* key_sav_debug_print(void *arg, const char *a) {
+  return a;
+}
+
+static const char* value_sav_debug_print(void *arg, void *a) {
+  return (char*) a;
+}
+
 /* Pseudo-key (lien_url structure) hash function */
 static inthash_keys key_adrfil_hashes_generic(void *arg, const char *value_, 
                                               const int former) {
@@ -151,6 +159,32 @@ static int key_adrfil_equals_generic(void *arg, const char *a_, const char *b_,
   }
 }
 
+static const char* key_adrfil_debug_print_(void *arg, const char *a_, const int former) {
+  hash_struct *const hash = (hash_struct*) arg;
+  const int normalized = hash->normalized;
+  const lien_url*const a = (lien_url*) a_;
+  const char *const a_adr = !former ? a->adr : a->former_adr;
+  const char *const a_fil = !former ? a->fil : a->former_fil;
+  snprintf(hash->normfil, sizeof(hash->normfil), "%s%s", a_adr, a_fil);
+  return hash->normfil;
+}
+
+static const char* key_adrfil_debug_print(void *arg, const char *a_) {
+  return key_adrfil_debug_print_(arg, a_, 0);
+}
+
+static const char* key_former_adrfil_debug_print(void *arg, const char *a_) {
+  return key_adrfil_debug_print_(arg, a_, 1);
+}
+
+static const char* value_adrfil_debug_print(void *arg, void *value) {
+  hash_struct *const hash = (hash_struct*) arg;
+  inthash_value v;
+  v.ptr = value;
+  snprintf(hash->normfil2, sizeof(hash->normfil2), "%d", (int) v.intg);
+  return hash->normfil2;
+}
+
 /* "adr"/"fil" lien_url structure members hashing function */
 static inthash_keys key_adrfil_hashes(void *arg, const char *value_) {
   return key_adrfil_hashes_generic(arg, value_, 0);
@@ -207,6 +241,20 @@ void hash_init(httrackp *opt, hash_struct * hash, int normalized) {
                                 key_former_adrfil_hashes,
                                 key_former_adrfil_equals,
                                 hash);
+
+  /* pretty-printing */
+  inthash_set_print_handler(hash->sav,
+                            key_sav_debug_print,
+                            value_sav_debug_print,
+                            NULL);
+  inthash_set_print_handler(hash->adrfil,
+                            key_adrfil_debug_print,
+                            value_adrfil_debug_print,
+                            hash);
+  inthash_set_print_handler(hash->former_adrfil,
+                            key_former_adrfil_debug_print,
+                            value_adrfil_debug_print,
+                            hash);
 }
 
 void hash_free(hash_struct *hash) {
@@ -272,3 +320,14 @@ void hash_write(hash_struct * hash, int lpos) {
     inthash_write(hash->former_adrfil, (char*) hash->liens[lpos], lpos);
   }
 }
+
+void hash_invalidate_entry(hash_struct * hash, int lpos) {
+  if (inthash_remove(hash->adrfil, (char*) hash->liens[lpos])) {
+    /* devalidate entry now it is removed from hashtable */
+    strcpybuff(hash->liens[lpos]->adr, "!");
+    /* add back */
+    inthash_write(hash->adrfil, (char*) hash->liens[lpos], lpos);
+  } else {
+    assertf(! "error invalidating hash entry");
+  }
+}
diff --git a/src/htshash.h b/src/htshash.h
index 4810513..6a71cba 100644
--- a/src/htshash.h
+++ b/src/htshash.h
@@ -56,6 +56,7 @@ void hash_free(hash_struct *hash);
 int hash_read(const hash_struct * hash, const char *nom1, const char *nom2,
               hash_struct_type type);
 void hash_write(hash_struct * hash, int lpos);
+void hash_invalidate_entry(hash_struct * hash, int lpos);
 int *hash_calc_chaine(hash_struct * hash, hash_struct_type type, int pos);
 unsigned long int hash_cle(const char *nom1, const char *nom2);
 #endif
diff --git a/src/htsparse.c b/src/htsparse.c
index e12e94c..b3778bc 100644
--- a/src/htsparse.c
+++ b/src/htsparse.c
@@ -3606,7 +3606,7 @@ int hts_mirror_check_moved(htsmoduleStruct * str,
                           urladr, urlfil, mov_adr, mov_fil);
             // canceller lien actuel
             error = 1;
-            strcpybuff(liens[ptr]->adr, "!");   // caractère bidon (invalide hash)
+            hash_invalidate_entry(hashptr, ptr);  // invalidate hashtable entry
             // noter NOUVEAU lien
             //xxc xxc
             //  set_prio_to=0+1;  // protection if the moved URL is an html page!!
@@ -3742,7 +3742,7 @@ int hts_mirror_check_moved(htsmoduleStruct * str,
           //
           // canceller lien actuel
           error = 1;
-          strcpybuff(liens[ptr]->adr, "!");   // caractère bidon (invalide hash)
+          hash_invalidate_entry(hashptr, ptr);  // invalidate hashtable entry
           //
         } else {              // oups erreur, plus de mémoire!!
           printf("PANIC! : Not enough memory [%d]\n", __LINE__);