Better logging for hashtables

11 files changed, 92 insertions, 12 deletions

diff --git a/src/htsback.c b/src/htsback.c
index 2f6df7b..d5d7c10 100644
--- a/src/htsback.c
+++ b/src/htsback.c
@@ -67,13 +67,15 @@ static int slot_can_be_cached_on_disk(const lien_back * back);
 static int slot_can_be_cleaned(const lien_back * back);
 static int slot_can_be_finalized(httrackp * opt, const lien_back * back);
 
-struct_back *back_new(int back_max) {
+struct_back *back_new(httrackp *opt, int back_max) {
   int i;
   struct_back *sback = calloct(1, sizeof(struct_back));
 
   sback->count = back_max;
   sback->lnk = (lien_back *) calloct((back_max + 1), sizeof(lien_back));
   sback->ready = inthash_new(0);
+  hts_set_hash_handler(sback->ready, opt);
+  inthash_set_name(sback->ready, "back_new");
   sback->ready_size_bytes = 0;
   inthash_value_is_malloc(sback->ready, 1);
   // init
diff --git a/src/htsback.h b/src/htsback.h
index 2203dd5..1da5450 100644
--- a/src/htsback.h
+++ b/src/htsback.h
@@ -68,7 +68,7 @@ typedef struct htsblk htsblk;
 #ifdef HTS_INTERNAL_BYTECODE
 
 // create/destroy
-struct_back *back_new(int back_max);
+struct_back *back_new(httrackp *opt, int back_max);
 void back_free(struct_back ** sback);
 
 // backing
diff --git a/src/htscore.c b/src/htscore.c
index bd2580b..94ed251 100644
--- a/src/htscore.c
+++ b/src/htscore.c
@@ -401,6 +401,10 @@ int httpmirror(char *url1, httrackp * opt) {
     XH_extuninit;
     return 0;
   }
+  hts_set_hash_handler(cache_hashtable, opt);
+  hts_set_hash_handler(cache_tests, opt);
+  inthash_set_name(cache_hashtable, "cache_hashtable");
+  inthash_set_name(cache_tests, "cache_tests");
   inthash_value_is_malloc(cache_tests, 1);      /* malloc */
   cache.hashtable = (void *) cache_hashtable;   /* copy backcache hash */
   cache.cached_tests = (void *) cache_tests;    /* copy of cache_tests */
@@ -445,7 +449,7 @@ int httpmirror(char *url1, httrackp * opt) {
   lien_tot = 0;
 
   // initialiser hachage
-  hash_init(&hash, opt->urlhack);
+  hash_init(opt, &hash, opt->urlhack);
   // note: we need a cast because of the const
   hash.liens = (const lien_url **) liens;
 
@@ -683,7 +687,7 @@ int httpmirror(char *url1, httrackp * opt) {
     // On prévoit large: les fichiers HTML ne prennent que peu de place en mémoire, et les
     // fichiers non html sont sauvés en direct sur disque.
     // --> 1024 entrées + 32 entrées par socket en supplément
-    sback = back_new(opt->maxsoc * 32 + 1024);
+    sback = back_new(opt, opt->maxsoc * 32 + 1024);
     if (sback == NULL) {
       hts_log_print(opt, LOG_PANIC,
                     "Not enough memory, can not allocate %d bytes",
diff --git a/src/htshash.c b/src/htshash.c
index 15af190..7bdf782 100644
--- a/src/htshash.c
+++ b/src/htshash.c
@@ -34,11 +34,11 @@ Please visit our Website: http://www.httrack.com
 /* Internal engine bytecode */
 #define HTS_INTERNAL_BYTECODE
 
+#include "htsopt.h"
 #include "htshash.h"
 
 /* specific definitions */
 #include "htsbase.h"
-#include "htsopt.h"
 #include "htsglobal.h"
 #include "htsmd5.h"
 #include "htscore.h"
@@ -171,12 +171,20 @@ static int key_former_adrfil_equals(void *arg, const char *a, const char *b) {
   return key_adrfil_equals_generic(arg, a, b, 1);
 }
 
-void hash_init(hash_struct * hash, int normalized) {
+void hash_init(httrackp *opt, hash_struct * hash, int normalized) {
   hash->sav = inthash_new(0);
   hash->adrfil = inthash_new(0);
   hash->former_adrfil = inthash_new(0);
   hash->normalized = normalized;
 
+  hts_set_hash_handler(hash->sav, opt);
+  hts_set_hash_handler(hash->adrfil, opt);
+  hts_set_hash_handler(hash->former_adrfil, opt);
+
+  inthash_set_name(hash->sav, "hash->sav");
+  inthash_set_name(hash->adrfil, "hash->adrfil");
+  inthash_set_name(hash->former_adrfil, "hash->former_adrfil");
+
   /* Case-insensitive comparison ; keys are direct char* filenames */
   inthash_value_set_key_handler(hash->sav,
                                 key_duphandler,
diff --git a/src/htshash.h b/src/htshash.h
index f8150c6..4810513 100644
--- a/src/htshash.h
+++ b/src/htshash.h
@@ -51,7 +51,7 @@ typedef enum hash_struct_type {
 } hash_struct_type;
 
 // tables de hachage
-void hash_init(hash_struct *hash, int normalized);
+void hash_init(httrackp *opt, hash_struct *hash, int normalized);
 void hash_free(hash_struct *hash);
 int hash_read(const hash_struct * hash, const char *nom1, const char *nom2,
               hash_struct_type type);
diff --git a/src/htslib.c b/src/htslib.c
index 9248cf9..d1936cc 100644
--- a/src/htslib.c
+++ b/src/htslib.c
@@ -5116,8 +5116,33 @@ static void default_inthash_asserthandler(void *arg, const char* exp, const char
   abortf_(exp, file, line);
 }
 
+static int get_loglevel_from_inthash(inthash_loglevel level) {
+  switch(level) {
+  case inthash_log_critical:
+    return LOG_PANIC;
+    break;
+  case inthash_log_warning:
+    return LOG_WARNING;
+    break;
+  case inthash_log_info:
+    return LOG_INFO;
+    break;
+  case inthash_log_debug:
+    return LOG_DEBUG;
+    break;
+  case inthash_log_trace:
+    return LOG_TRACE;
+    break;
+  default:
+    return LOG_ERROR;
+    break;
+  }
+}
+
+/* log to default console */
 static void default_inthash_loghandler(void *arg, inthash_loglevel level, 
                                        const char* format, va_list args) {
+
   if (level <= inthash_log_warning) {
     fprintf(stderr, "** warning: ");
   }
@@ -5125,6 +5150,27 @@ static void default_inthash_loghandler(void *arg, inthash_loglevel level,
   fprintf(stderr, "\n");
 }
 
+/* log to project log */
+static void htsopt_inthash_loghandler(void *arg, inthash_loglevel level, 
+                                      const char* format, va_list args) {
+  httrackp *const opt = (httrackp*) arg;
+  if (opt != NULL && opt->log != NULL) {
+    hts_log_vprint(opt, get_loglevel_from_inthash(level), 
+      format, args);
+  } else {
+    default_inthash_loghandler(NULL, level, format, args);
+  }
+}
+
+/* attach hashtable logger to project log */
+void hts_set_hash_handler(inthash hashtable, httrackp *opt) {
+  /* Init hashtable default assertion handler. */
+  inthash_set_assert_handler(hashtable,
+    htsopt_inthash_loghandler, 
+    default_inthash_asserthandler,
+    opt);
+}
+
 static int hts_init_ok = 0;
 HTSEXT_API int hts_init(void) {
   const char *dbg_env;
@@ -5232,10 +5278,9 @@ HTSEXT_API int hts_log(httrackp * opt, const char *prefix, const char *msg) {
   return 1;                     /* Error */
 }
 
-HTSEXT_API void hts_log_print(httrackp * opt, int type, const char *format, ...) {
+HTSEXT_API void hts_log_vprint(httrackp * opt, int type, const char *format, va_list args) {
   assertf(format != NULL);
   if (opt != NULL && opt->log != NULL) {
-    va_list args;
     const int save_errno = errno;
     const char *s_type = "unknown";
     const int level = type & 0xff;
@@ -5267,9 +5312,7 @@ HTSEXT_API void hts_log_print(httrackp * opt, int type, const char *format, ...)
       break;
     }
     fspc(opt, opt->log, s_type);
-    va_start(args, format);
     (void) vfprintf(opt->log, format, args);
-    va_end(args);
     if ((type & LOG_ERRNO) != 0) {
       fprintf(opt->log, ": %s", strerror(save_errno));
     }
@@ -5281,6 +5324,16 @@ HTSEXT_API void hts_log_print(httrackp * opt, int type, const char *format, ...)
   }
 }
 
+HTSEXT_API void hts_log_print(httrackp * opt, int type, const char *format, ...) {
+  assertf(format != NULL);
+  if (opt != NULL && opt->log != NULL) {
+    va_list args;
+    va_start(args, format);
+    hts_log_vprint(opt, type, format, args);
+    va_end(args);
+  }
+}
+
 HTSEXT_API void set_wrappers(httrackp * opt) {  // LEGACY
 }
 
diff --git a/src/htslib.h b/src/htslib.h
index 8a6c206..2b75dc4 100644
--- a/src/htslib.h
+++ b/src/htslib.h
@@ -234,6 +234,9 @@ struct t_dnscache {
 // initialize an htsblk structure
 void hts_init_htsblk(htsblk * r);
 
+// attach specific project log to hachtable logger
+void hts_set_hash_handler(inthash hashtable, httrackp *opt);
+
 // version
 HTSEXT_API const char* hts_version(void);
 
diff --git a/src/htsserver.c b/src/htsserver.c
index 1388110..65329f3 100644
--- a/src/htsserver.c
+++ b/src/htsserver.c
@@ -1443,6 +1443,7 @@ int smallserver(T_SOC soc, char *url, char *method, char *data, char *path) {
 int htslang_init(void) {
   if (NewLangList == NULL) {
     NewLangList = inthash_new(0);
+    inthash_set_name(NewLangList, "NewLangList");
     if (NewLangList == NULL) {
       abortLog("Error in lang.h: not enough memory");
     } else {
@@ -1487,6 +1488,8 @@ static int htslang_load(char *limit_to, char *path) {
     LANG_DELETE();
     NewLangStr = inthash_new(0);
     NewLangStrKeys = inthash_new(0);
+    inthash_set_name(NewLangStr, "NewLangStr");
+    inthash_set_name(NewLangStrKeys, "NewLangStrKeys");
     if ((NewLangStr == NULL) || (NewLangStrKeys == NULL)) {
       abortLog("Error in lang.h: not enough memory");
     } else {
diff --git a/src/htstools.c b/src/htstools.c
index 78fb827..de81d57 100644
--- a/src/htstools.c
+++ b/src/htstools.c
@@ -980,6 +980,7 @@ HTSEXT_API char *hts_getcategories(char *path, int type) {
 
     if (type == 1) {
       hashCateg = inthash_new(0);
+      inthash_set_name(hashCateg, "hashCateg");
       StringCat(categ, "Test category 1");
       StringCat(categ, "\r\nTest category 2");
     }
diff --git a/src/httrack-library.h b/src/httrack-library.h
index 21ff79c..8c63fb4 100644
--- a/src/httrack-library.h
+++ b/src/httrack-library.h
@@ -116,6 +116,8 @@ HTSEXT_API int plug_wrapper(httrackp * opt, const char *moduleName,
 HTSEXT_API int hts_log(httrackp * opt, const char *prefix, const char *msg);
 HTSEXT_API void hts_log_print(httrackp * opt, int type, const char *format,
                               ...) HTS_PRINTF_FUN(3, 4);
+HTSEXT_API void hts_log_vprint(httrackp * opt, int type, const char *format,
+                               va_list args);
 
 /* Infos */
 HTSEXT_API const char *hts_get_version_info(httrackp * opt);
diff --git a/src/proxy/store.c b/src/proxy/store.c
index 9d756b7..daa9952 100644
--- a/src/proxy/store.c
+++ b/src/proxy/store.c
@@ -238,6 +238,7 @@ PT_Indexes PT_New(void) {
   PT_Indexes index = (PT_Indexes) calloc(sizeof(_PT_Indexes), 1);
 
   index->cil = inthash_new(0);
+  inthash_set_name(index->cil, "index->cil");
   index->index_size = 0;
   index->index = NULL;
   return index;
@@ -432,8 +433,10 @@ char **PT_Enumerate(PT_Indexes indexes, const char *url, int subtree) {
     inthash_chain *chain;
     inthash hdupes = NULL;
 
-    if (!subtree)
+    if (!subtree) {
       hdupes = inthash_new(0);
+      inthash_set_name(hdupes, "hdupes");
+    }
     StringClear(list);
     StringClear(listindexes);
     StringClear(subitem);
@@ -551,6 +554,7 @@ PT_Index PT_LoadCache(const char *filename) {
       index->slots.common.timestamp = (time_t) time(NULL);
       index->slots.common.startUrl[0] = '\0';
       index->slots.common.hash = inthash_new(0);
+      inthash_set_name(index->slots.common.hash, "index->slots.common.hash");
       if (!_IndexFuncts[type].PT_LoadCache(index, filename)) {
         proxytrack_print_log(DEBUG,
                              "reading httrack cache (format #%d) %s : error",