Refactor SRU/Z39.50 diag map utils to diag_map.c

[yaz-moved-to-github.git] / src / cclfind.c

diff --git a/src/cclfind.c b/src/cclfind.c
index f242169..3cbeacf 100644 (file)
--- a/src/cclfind.c
+++ b/src/cclfind.c
@@ -212,7 +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
@@ -260,6 +275,7 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
         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;
@@ -363,6 +379,11 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
         {
             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 * 2 + 2);
@@ -377,31 +398,44 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
 
             if (p->u.t.term[0] && cclp->look_token->ws_prefix_len)
             {
-                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';
+                strxcat(p->u.t.term, cclp->look_token->ws_prefix_buf,
+                        cclp->look_token->ws_prefix_len);
             }
             for (j = 0; j < src_len; j++)
             {
+                size_t op_size;
                 if (j > 0 && src_str[j-1] == '\\')
                 {
-                    if (regex_trunc && strchr("()[]?*.", src_str[j]))
+                    if (regex_trunc && strchr(REGEX_CHARS "\\", src_str[j]))
                     {
                         regex_trunc = 2;
-                        strcat(p->u.t.term, "\\\\");
+                        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 && src_str[j] == '?')
+                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)
@@ -420,6 +454,11 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
                         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;
@@ -429,10 +468,15 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
                 }
                 else if (src_str[j] != '\\')
                 {
-                    if (regex_trunc && strchr("()[]?*.", src_str[j]))
+                    if (regex_trunc && strchr(REGEX_CHARS, src_str[j]))
                     {
                         regex_trunc = 2;
-                        strcat(p->u.t.term, "\\\\");
+                        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);                    
                 }
@@ -497,6 +541,10 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
         {
             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,