X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fcclfind.c;h=072fdabc0aa86d23c5a5e4459ddbf53568c835a9;hp=6207b55700f7d443598cbfaa9cef8a4ed8017719;hb=0e7fdbc857d4905e67a7bdf8fe5b6c773ee47a7b;hpb=379504a233e3e2cc85bca1e7b6d864f1395aec7c

diff --git a/src/cclfind.c b/src/cclfind.c
index 6207b55..072fdab 100644
--- a/src/cclfind.c
+++ b/src/cclfind.c
@@ -1,5 +1,5 @@
 /* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2009 Index Data
+ * Copyright (C) 1995-2012 Index Data
  * See the file LICENSE for details.
  */
 /** 
@@ -11,9 +11,13 @@
  * of lookahead in the handling of relational operations.. So
  * it's not really pure.
  */
+#if HAVE_CONFIG_H
+#include <config.h>
+#endif
 
 #include <stdlib.h>
 #include <string.h>
+#include <assert.h>
 
 #include "cclp.h"
 
@@ -208,7 +212,22 @@ void ccl_add_attr_string(struct ccl_rpn_node *p, const char *set,
     n->value.str = xstrdup(value);
 }
 
+static size_t cmp_operator(const char **aliases, const char *input)
+{
+    for (; *aliases; aliases++)
+    {
+        const char *cp = *aliases;
+        size_t i;
+        for (i = 0; *cp && *cp == input[i]; i++, cp++)
+            ;
+        if (*cp == '\0')
+            return i;
+    }
+    return 0;
+}
 
+#define REGEX_CHARS "^[]{}()|.*+?!$"
+#define CCL_CHARS "#?\\"
 /**
  * search_term: Parse CCL search term. 
  * cclp:   CCL Parser
@@ -247,15 +266,16 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
         struct ccl_rpn_node *p;
         size_t no, i;
         int no_spaces = 0;
-        int left_trunc = 0;
-        int right_trunc = 0;
-        int mid_trunc = 0;
         int relation_value = -1;
         int position_value = -1;
         int structure_value = -1;
         int truncation_value = -1;
         int completeness_value = -1;
         int len = 0;
+        int left_trunc = 0;
+        int right_trunc = 0;
+        int regex_trunc = 0;
+        int z3958_trunc = 0;
         size_t max = 200;
         if (and_list || or_list || !multi)
             max = 1;
@@ -275,16 +295,6 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
             for (i = 0; i<lookahead->len; i++)
                 if (lookahead->name[i] == ' ')
                     no_spaces++;
-                else if (strchr(truncation_aliases[0], lookahead->name[i]))
-                {
-                    if (no == 0 && i == 0 && lookahead->len >= 1)
-                        left_trunc = 1;
-                    else if (!is_term_ok(lookahead->next->kind, term_list) &&
-                             i == lookahead->len-1 && i >= 1)
-                        right_trunc = 1;
-                    else
-                        mid_trunc = 1;
-                }
             len += 1+lookahead->len+lookahead->ws_prefix_len;
             lookahead = lookahead->next;
         }
@@ -339,7 +349,6 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
                             if (truncation_value != -1)
                                 continue;
                             truncation_value = attr->value.numeric;
-                            left_trunc = right_trunc = mid_trunc = 0;
                             break;
                         case CCL_BIB1_COM:
                             if (completeness_value != -1)
@@ -365,30 +374,113 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
                 ccl_add_attr_numeric(p, attset, CCL_BIB1_STR, 1);
         }
 
+        if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_REGEX,
+                          &attset))
+        {
+            regex_trunc = 1; /* regex trunc (102) allowed */
+        }
+        else if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_Z3958,
+                          &attset))
+        {
+            z3958_trunc = 1; /* Z39.58 trunc (CCL) trunc allowed */
+        }
+
         /* make the RPN token */
-        p->u.t.term = (char *)xmalloc(len);
+        p->u.t.term = (char *)xmalloc(len * 2 + 2);
         ccl_assert(p->u.t.term);
         p->u.t.term[0] = '\0';
         for (i = 0; i<no; i++)
         {
             const char *src_str = cclp->look_token->name;
             size_t src_len = cclp->look_token->len;
-            
-            if (i == 0 && left_trunc)
+            int j;
+            int quote_mode = 0;
+
+            if (p->u.t.term[0] && cclp->look_token->ws_prefix_len)
             {
-                src_len--;
-                src_str++;
+                strxcat(p->u.t.term, cclp->look_token->ws_prefix_buf,
+                        cclp->look_token->ws_prefix_len);
             }
-            if (i == no-1 && right_trunc)
-                src_len--;
-            if (p->u.t.term[0] && cclp->look_token->ws_prefix_len)
+            for (j = 0; j < src_len; j++)
             {
-                size_t len = strlen(p->u.t.term);
-                memcpy(p->u.t.term + len, cclp->look_token->ws_prefix_buf,
-                       cclp->look_token->ws_prefix_len);
-                p->u.t.term[len + cclp->look_token->ws_prefix_len] = '\0';
+                size_t op_size;
+                if (j > 0 && src_str[j-1] == '\\')
+                {
+                    if (regex_trunc && strchr(REGEX_CHARS "\\", src_str[j]))
+                    {
+                        regex_trunc = 2;
+                        strcat(p->u.t.term, "\\");
+                    }
+                    else if (z3958_trunc && strchr(CCL_CHARS "\\", src_str[j]))
+                    {
+                        z3958_trunc = 2;
+                        strcat(p->u.t.term, "\\");
+                    }
+                    strxcat(p->u.t.term, src_str + j, 1);
+                }
+                else if (src_str[j] == '"')
+                    quote_mode = !quote_mode;
+                else if (!quote_mode &&
+                         (op_size = cmp_operator(truncation_aliases,
+                                                 src_str + j))
+                    )
+                {
+                    j += (op_size - 1);  /* j++ in for loop */
+                    if (regex_trunc)
+                    {
+                        strcat(p->u.t.term, ".*");
+                        regex_trunc = 2; /* regex trunc is really needed */
+                    }
+                    else if (z3958_trunc)
+                    {
+                        strcat(p->u.t.term, "?");
+                        z3958_trunc = 2;
+                    }
+                    else if (i == 0 && j == 0)
+                        left_trunc = 1;
+                    else if (i == no - 1 && j == src_len - 1)
+                        right_trunc = 1;
+                    else
+                    {
+                        cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
+                        ccl_rpn_delete(p);
+                        return NULL;
+                    }
+                }
+                else if (!quote_mode && src_str[j] == '#')
+                {
+                    if (regex_trunc)
+                    {
+                        strcat(p->u.t.term, ".");
+                        regex_trunc = 2; /* regex trunc is really needed */
+                    }
+                    else if (z3958_trunc)
+                    {
+                        strcat(p->u.t.term, "#");
+                        z3958_trunc = 2;
+                    }
+                    else
+                    {
+                        cclp->error_code = CCL_ERR_TRUNC_NOT_BOTH;
+                        ccl_rpn_delete(p);
+                        return NULL;
+                    }
+                }
+                else if (src_str[j] != '\\')
+                {
+                    if (regex_trunc && strchr(REGEX_CHARS, src_str[j]))
+                    {
+                        regex_trunc = 2;
+                        strcat(p->u.t.term, "\\");
+                    }
+                    else if (z3958_trunc && strchr(CCL_CHARS, src_str[j]))
+                    {
+                        z3958_trunc = 2;
+                        strcat(p->u.t.term, "\\");
+                    }
+                    strxcat(p->u.t.term, src_str + j, 1);                    
+                }
             }
-            strxcat(p->u.t.term, src_str, src_len);
             ADVANCE;
         }
 
@@ -445,6 +537,14 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
             }
             ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 2);
         }
+        else if (regex_trunc == 2)
+        {
+            ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 102);
+        }
+        else if (z3958_trunc == 2)
+        {
+            ccl_add_attr_numeric(p, attset, CCL_BIB1_TRU, 104);
+        }
         else
         {
             if (qual_val_type(qa, CCL_BIB1_TRU, CCL_BIB1_TRU_CAN_NONE,
@@ -465,6 +565,37 @@ static struct ccl_rpn_node *search_term(CCL_parser cclp, ccl_qualifier_t *qa)
     return search_term_x(cclp, qa, list, 0);
 }
 
+
+static struct ccl_rpn_node *search_terms2(CCL_parser cclp,
+                                          ccl_qualifier_t *qa)
+{
+    if (KIND == CCL_TOK_LP)
+    {
+        struct ccl_rpn_node *p;
+        ADVANCE;
+        if (!(p = find_spec(cclp, qa)))
+            return NULL;
+        if (KIND != CCL_TOK_RP)
+        {
+            cclp->error_code = CCL_ERR_RP_EXPECTED;
+            ccl_rpn_delete(p);
+            return NULL;
+        }
+        ADVANCE;
+        return p;
+    }
+    else
+    {
+        static int list[] = {
+            CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ,
+            CCL_TOK_REL, CCL_TOK_SET, -1};
+        
+        return search_term_x(cclp, qa, list, 1);
+    }
+}
+
+
+
 static
 struct ccl_rpn_node *qualifiers_order(CCL_parser cclp,
                                       ccl_qualifier_t *ap, char *attset)
@@ -607,20 +738,6 @@ struct ccl_rpn_node *qualifiers_order(CCL_parser cclp,
         ccl_add_attr_numeric(p, attset, CCL_BIB1_REL, 2);
         return p;
     }
-    else if (KIND == CCL_TOK_LP)
-    {
-        ADVANCE;
-        if (!(p = find_spec(cclp, ap)))
-            return NULL;
-        if (KIND != CCL_TOK_RP)
-        {
-            cclp->error_code = CCL_ERR_RP_EXPECTED;
-            ccl_rpn_delete(p);
-            return NULL;
-        }
-        ADVANCE;
-        return p;
-    }
     else
     {
         if (!(p = search_terms(cclp, ap)))
@@ -636,7 +753,6 @@ static
 struct ccl_rpn_node *qualifier_relation(CCL_parser cclp, ccl_qualifier_t *ap)
 {
     char *attset;
-    struct ccl_rpn_node *p;
     
     if (qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER, &attset)
         || qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_PORDER, &attset))
@@ -649,24 +765,7 @@ struct ccl_rpn_node *qualifier_relation(CCL_parser cclp, ccl_qualifier_t *ap)
         return NULL;
     }
     ADVANCE;
-    if (KIND == CCL_TOK_LP)
-    {
-        ADVANCE;
-        if (!(p = find_spec(cclp, ap)))
-        {
-            return NULL;
-        }
-        if (KIND != CCL_TOK_RP)
-        {
-            cclp->error_code = CCL_ERR_RP_EXPECTED;
-            ccl_rpn_delete(p);
-            return NULL;
-        }
-        ADVANCE;
-    }
-    else
-        p = search_terms(cclp, ap);
-    return p;
+    return search_terms(cclp, ap);
 }
 
 /**
@@ -833,9 +932,10 @@ static struct ccl_rpn_node *qualifier_list(CCL_parser cclp,
 static struct ccl_rpn_node *search_terms(CCL_parser cclp, ccl_qualifier_t *qa)
 {
     static int list[] = {
-        CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ, CCL_TOK_REL, CCL_TOK_SET, -1};
+        CCL_TOK_TERM, CCL_TOK_COMMA,CCL_TOK_EQ,
+        CCL_TOK_REL, CCL_TOK_SET, -1};
     struct ccl_rpn_node *p1, *p2, *pn;
-    p1 = search_term_x(cclp, qa, list, 1);
+    p1 = search_terms2(cclp, qa);
     if (!p1)
         return NULL;
     while (1)
@@ -853,7 +953,7 @@ static struct ccl_rpn_node *search_terms(CCL_parser cclp, ccl_qualifier_t *qa)
             p_prox->u.t.attr_list = 0;
 
             ADVANCE;
-            p2 = search_term_x(cclp, qa, list, 1);
+            p2 = search_terms2(cclp, qa);
             if (!p2)
             {
                 ccl_rpn_delete(p1);
@@ -867,7 +967,7 @@ static struct ccl_rpn_node *search_terms(CCL_parser cclp, ccl_qualifier_t *qa)
         }
         else if (is_term_ok(KIND, list))
         {
-            p2 = search_term_x(cclp, qa, list, 1);
+            p2 = search_terms2(cclp, qa);
             if (!p2)
             {
                 ccl_rpn_delete(p1);
@@ -896,22 +996,7 @@ static struct ccl_rpn_node *search_elements(CCL_parser cclp,
 {
     struct ccl_rpn_node *p1;
     struct ccl_token *lookahead;
-    if (KIND == CCL_TOK_LP)
-    {
-        ADVANCE;
-        p1 = find_spec(cclp, qa);
-        if (!p1)
-            return NULL;
-        if (KIND != CCL_TOK_RP)
-        {
-            cclp->error_code = CCL_ERR_RP_EXPECTED;
-            ccl_rpn_delete(p1);
-            return NULL;
-        }
-        ADVANCE;
-        return p1;
-    }
-    else if (KIND == CCL_TOK_SET)
+    if (KIND == CCL_TOK_SET)
     {
         ADVANCE;
         if (KIND == CCL_TOK_EQ)
@@ -937,7 +1022,7 @@ static struct ccl_rpn_node *search_elements(CCL_parser cclp,
             break;
         lookahead = lookahead->next;
     }
-    if (qa)
+    if (qa || lookahead->kind == CCL_TOK_LP)
         return search_terms(cclp, qa);
     else
     {