X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=dict%2Flookgrep.c;h=bde37714457bb4346517134ce43492c00a81a2c1;hb=1c3797bc503c1e7a109c8887d89d3ddda93bba71;hp=dc922695e93aca4448eb4162840473497d3a35ed;hpb=462c4c58fa0f21af1ffa7d44d5f2053aec671e08;p=idzebra-moved-to-github.git

diff --git a/dict/lookgrep.c b/dict/lookgrep.c
index dc92269..bde3771 100644
--- a/dict/lookgrep.c
+++ b/dict/lookgrep.c
@@ -1,10 +1,65 @@
 /*
- * Copyright (C) 1994, Index Data I/S 
+ * Copyright (C) 1994-1999, Index Data
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: lookgrep.c,v $
- * Revision 1.8  1995-10-19 14:57:21  adam
+ * Revision 1.24  1999-05-26 07:49:12  adam
+ * C++ compilation.
+ *
+ * Revision 1.23  1999/05/15 14:36:37  adam
+ * Updated dictionary. Implemented "compression" of dictionary.
+ *
+ * Revision 1.22  1999/02/02 14:50:23  adam
+ * Updated WIN32 code specific sections. Changed header.
+ *
+ * Revision 1.21  1998/06/24 12:16:12  adam
+ * Support for relations on text operands. Open range support in
+ * DFA module (i.e. [-j], [g-]).
+ *
+ * Revision 1.20  1997/10/27 14:33:03  adam
+ * Moved towards generic character mapping depending on "structure"
+ * field in abstract syntax file. Fixed a few memory leaks. Fixed
+ * bug with negative integers when doing searches with relational
+ * operators.
+ *
+ * Revision 1.19  1997/09/18 08:59:18  adam
+ * Extra generic handle for the character mapping routines.
+ *
+ * Revision 1.18  1997/09/05 15:29:58  adam
+ * Changed prototype for chr_map_input - added const.
+ * Added support for C++, headers uses extern "C" for public definitions.
+ *
+ * Revision 1.17  1996/06/04 10:20:06  adam
+ * Added support for character mapping.
+ *
+ * Revision 1.16  1996/05/24  14:46:04  adam
+ * Added dict_grep_cmap function to define user-mapping in grep lookups.
+ *
+ * Revision 1.15  1996/03/20  09:35:18  adam
+ * Function dict_lookup_grep got extra parameter, init_pos, which marks
+ * from which position in pattern approximate pattern matching should occur.
+ *
+ * Revision 1.14  1996/02/02  13:43:51  adam
+ * The public functions simply use char instead of Dict_char to represent
+ * search strings. Dict_char is used internally only.
+ *
+ * Revision 1.13  1996/01/08  09:09:30  adam
+ * Function dfa_parse got 'const' string argument.
+ *
+ * Revision 1.12  1995/12/11  09:04:48  adam
+ * Bug fix: the lookup/scan/lookgrep didn't handle empty dictionary.
+ *
+ * Revision 1.11  1995/12/06  14:43:02  adam
+ * New function: dict_delete.
+ *
+ * Revision 1.10  1995/11/16  17:00:44  adam
+ * Changed stupid log.
+ *
+ * Revision 1.9  1995/10/27  13:58:09  adam
+ * Makes 'Database unavailable' diagnostic.
+ *
+ * Revision 1.8  1995/10/19  14:57:21  adam
  * New feature: grep lookup saves length of longest prefix match.
  *
  * Revision 1.7  1995/10/17  18:01:22  adam
@@ -72,13 +127,13 @@ static INLINE MatchWord get_bit (MatchContext *mc, MatchWord *m, int ch,
 
 static MatchContext *mk_MatchContext (struct DFA *dfa, int range)
 {
-    MatchContext *mc = xmalloc (sizeof(*mc));
+    MatchContext *mc = (MatchContext *) xmalloc (sizeof(*mc));
     int s;
 
     mc->n = (dfa->no_states+WORD_BITS) / WORD_BITS;
     mc->range = range;
     mc->fact = (range+1)*mc->n;
-    mc->match_mask = xcalloc (mc->n, sizeof(*mc->match_mask));
+    mc->match_mask = (MatchWord *) xcalloc (mc->n, sizeof(*mc->match_mask));
 
     for (s = 0; s<dfa->no_states; s++)
         if (dfa->states[s]->rule_no)
@@ -213,7 +268,8 @@ static void or (MatchContext *mc, MatchWord *Rdst,
 }
 
 static INLINE int move (MatchContext *mc, MatchWord *Rj1, MatchWord *Rj,
-                 Dict_char ch, struct DFA *dfa, MatchWord *Rtmp)
+                        Dict_char ch, struct DFA *dfa, MatchWord *Rtmp,
+                        int range)
 {
     int d;
     MatchWord *Rtmp_2 = Rtmp + mc->n;
@@ -242,9 +298,9 @@ static INLINE int move (MatchContext *mc, MatchWord *Rj1, MatchWord *Rj,
 
 static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc,
                       MatchWord *Rj, int pos, void *client,
-                      int (*userfunc)(Dict_char *, const char *, void *),
+                      int (*userfunc)(char *, const char *, void *),
                       Dict_char *prefix, struct DFA *dfa,
-                      int *max_pos)
+                      int *max_pos, int init_pos)
 {
     int lo, hi, d;
     void *p;
@@ -254,7 +310,7 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc,
     dict_bf_readp (dict->dbf, ptr, &p);
     lo = 0;
     hi = DICT_nodir(p)-1;
-    indxp = (short*) ((char*) p+DICT_pagesize(dict)-sizeof(short));    
+    indxp = (short*) ((char*) p+DICT_bsize(p)-sizeof(short));
 
     while (lo <= hi)
     {
@@ -272,6 +328,7 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc,
                 MatchWord *Rj0 =    Rj + j    *mc->fact;
                 MatchWord *Rj1 =    Rj + (j+1)*mc->fact;
                 MatchWord *Rj_tmp = Rj + (j+2)*mc->fact;
+                int range;
 
                 memcpy (&ch, info+j*sizeof(Dict_char), sizeof(Dict_char));
                 prefix[pos+j] = ch;
@@ -280,20 +337,24 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc,
                 if (ch == DICT_EOS)
                 {
                     if (was_match)
-                        if ((*userfunc)(prefix, info+(j+1)*sizeof(Dict_char),
-                                        client))
+                        if ((*userfunc)((char*) prefix,
+                                        info+(j+1)*sizeof(Dict_char), client))
                             return 1;
                     break;
                 }
-                move (mc, Rj1, Rj0, ch, dfa, Rj_tmp);
+                if (pos+j >= init_pos)
+                    range = mc->range;
+                else
+                    range = 0;
+                move (mc, Rj1, Rj0, ch, dfa, Rj_tmp, range);
                 for (d = mc->n; --d >= 0; )
-                    if (Rj1[mc->range*mc->n + d])
+                    if (Rj1[range*mc->n + d])
                         break;
                 if (d < 0)
                     break;
                 was_match = 0;
                 for (d = mc->n; --d >= 0; )
-                    if (Rj1[mc->range*mc->n + d] & mc->match_mask[d])
+                    if (Rj1[range*mc->n + d] & mc->match_mask[d])
                     {
                         was_match = 1;
                         break;
@@ -305,6 +366,7 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc,
             MatchWord *Rj1 =    Rj+  mc->fact;
             MatchWord *Rj_tmp = Rj+2*mc->fact;
             Dict_char ch;
+            int range;
 
             /* Dict_ptr             subptr */
             /* Dict_char            sub char */
@@ -316,9 +378,13 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc,
             
             if (pos > *max_pos)
                 *max_pos = pos;
-            move (mc, Rj1, Rj, ch, dfa, Rj_tmp);
+            if (pos >= init_pos)
+                range = mc->range;
+            else
+                range = 0;
+            move (mc, Rj1, Rj, ch, dfa, Rj_tmp, range);
             for (d = mc->n; --d >= 0; )
-                if (Rj1[mc->range*mc->n + d])
+                if (Rj1[range*mc->n + d])
                     break;
             if (d >= 0)
             {
@@ -326,10 +392,11 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc,
                 if (info[sizeof(Dict_ptr)+sizeof(Dict_char)])
                 {
                     for (d = mc->n; --d >= 0; )
-                        if (Rj1[mc->range*mc->n + d] & mc->match_mask[d])
+                        if (Rj1[range*mc->n + d] & mc->match_mask[d])
                         {
                             prefix[pos+1] = DICT_EOS;
-                            if ((*userfunc)(prefix, info+sizeof(Dict_ptr)+
+                            if ((*userfunc)((char*) prefix,
+                                            info+sizeof(Dict_ptr)+
                                             sizeof(Dict_char), client))
                                 return 1;
                             break;
@@ -339,11 +406,11 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc,
                 if (subptr)
                 {
                     if (dict_grep (dict, subptr, mc, Rj1, pos+1,
-                                   client, userfunc, prefix, dfa, max_pos))
+                                   client, userfunc, prefix, dfa, max_pos,
+                                   init_pos))
                         return 1;
                     dict_bf_readp (dict->dbf, ptr, &p);
-                    indxp = (short*) ((char*) p+DICT_pagesize(dict)
-                                      -sizeof(short));
+                    indxp = (short*) ((char*) p+DICT_bsize(p)-sizeof(short));
                 }
             }
         }
@@ -352,16 +419,33 @@ static int dict_grep (Dict dict, Dict_ptr ptr, MatchContext *mc,
     return 0;
 }
 
-int dict_lookup_grep (Dict dict, Dict_char *pattern, int range, void *client,
-                      int (*userfunc)(Dict_char *name, const char *info,
+int dict_lookup_grep (Dict dict, const char *pattern, int range, void *client,
+                      int *max_pos, int init_pos,
+                      int (*userfunc)(char *name, const char *info,
                                       void *client))
 {
     MatchWord *Rj;
     Dict_char prefix[MAX_LENGTH+1];
-    char *this_pattern = pattern;
+    const char *this_pattern = pattern;
     MatchContext *mc;
     struct DFA *dfa = dfa_init();
-    int i, d, max_pos;
+    int i, d;
+
+#if 0
+    debug_dfa_trav = 1;
+    debug_dfa_tran = 1;
+    debug_dfa_followpos = 1;
+    dfa_verbose = 1;
+#endif
+
+    logf (LOG_DEBUG, "dict_lookup_grep range=%d", range);
+    for (i = 0; pattern[i]; i++)
+    {
+	logf (LOG_DEBUG, " %3d  %c", pattern[i],
+	      (pattern[i] > ' ' && pattern[i] < 127) ? pattern[i] : '?');
+    }
+   
+    dfa_set_cmap (dfa, dict->grep_cmap_data, dict->grep_cmap);
 
     i = dfa_parse (dfa, &this_pattern);
     if (i || *this_pattern)
@@ -373,9 +457,8 @@ int dict_lookup_grep (Dict dict, Dict_char *pattern, int range, void *client,
 
     mc = mk_MatchContext (dfa, range);
 
-    Rj = xcalloc ((MAX_LENGTH+1) * mc->n, sizeof(*Rj));
+    Rj = (MatchWord *) xcalloc ((MAX_LENGTH+1) * mc->n, sizeof(*Rj));
 
-    max_pos = 0;
     set_bit (mc, Rj, 0, 0);
     for (d = 1; d<=mc->range; d++)
     {
@@ -392,11 +475,23 @@ int dict_lookup_grep (Dict dict, Dict_char *pattern, int range, void *client,
             }
         }
     }
-    i = dict_grep (dict, 1, mc, Rj, 0, client, userfunc, prefix, dfa,
-                   &max_pos);
-    logf (LOG_DEBUG, "max_pos = %d", max_pos);
+    *max_pos = 0;
+    if (dict->head.root)
+        i = dict_grep (dict, dict->head.root, mc, Rj, 0, client,
+		       userfunc, prefix,
+                       dfa, max_pos, init_pos);
+    else
+        i = 0;
+    logf (LOG_DEBUG, "max_pos = %d", *max_pos);
     dfa_delete (&dfa);
     xfree (Rj);
     rm_MatchContext (&mc);
     return i;
 }
+
+void dict_grep_cmap (Dict dict, void *vp,
+		     const char **(*cmap)(void *vp, const char **from, int len))
+{
+    dict->grep_cmap = cmap;
+    dict->grep_cmap_data = vp;
+}