X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Fcclfind.c;h=51f06c444c755dd0e506cafca1f2439093ea4929;hp=3cbeacf04054c7db31b2c5691c92ad14c2714f1e;hb=818a9b6c9a5dfae6176d350346da0a9c55a393d6;hpb=e7ee503f39a77a3b79d590feef34e0bd959ece0c

diff --git a/src/cclfind.c b/src/cclfind.c
index 3cbeacf..51f06c4 100644
--- a/src/cclfind.c
+++ b/src/cclfind.c
@@ -1,5 +1,5 @@
 /* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2011 Index Data
+ * Copyright (C) 1995-2012 Index Data
  * See the file LICENSE for details.
  */
 /** 
@@ -226,8 +226,99 @@ static size_t cmp_operator(const char **aliases, const char *input)
     return 0;
 }
 
+
 #define REGEX_CHARS "^[]{}()|.*+?!$"
 #define CCL_CHARS "#?\\"
+static int append_term(CCL_parser cclp, const char *src_str, size_t src_len,
+                       char *dst_term, int *regex_trunc, int *z3958_trunc,
+                       const char **truncation_aliases,
+                       int is_first, int is_last,
+                       int *left_trunc, int *right_trunc)
+{
+    size_t j;
+    int quote_mode = 0;
+
+    for (j = 0; j < src_len; j++)
+    {
+        size_t op_size;
+        if (j > 0 && src_str[j-1] == '\\')
+        {
+            if (*regex_trunc && strchr(REGEX_CHARS "\\", src_str[j]))
+            {
+                *regex_trunc = 2;
+                strcat(dst_term, "\\");
+            }
+            else if (*z3958_trunc && strchr(CCL_CHARS "\\", src_str[j]))
+            {
+                *z3958_trunc = 2;
+                strcat(dst_term, "\\");
+            }
+            strxcat(dst_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(dst_term, ".*");
+                *regex_trunc = 2; /* regex trunc is really needed */
+            }
+            else if (*z3958_trunc)
+            {
+                strcat(dst_term, "?");
+                *z3958_trunc = 2;
+            }
+            else if (is_first && j == 0)
+                *left_trunc = 1;
+            else if (is_last && j == src_len - 1)
+                *right_trunc = 1;
+            else
+            {
+                cclp->error_code = CCL_ERR_TRUNC_NOT_EMBED;
+                return -1;
+            }
+        }
+        else if (!quote_mode && src_str[j] == '#')
+        {
+            if (*regex_trunc)
+            {
+                strcat(dst_term, ".");
+                *regex_trunc = 2; /* regex trunc is really needed */
+            }
+            else if (*z3958_trunc)
+            {
+                strcat(dst_term, "#");
+                *z3958_trunc = 2;
+            }
+            else
+            {
+                cclp->error_code = CCL_ERR_TRUNC_NOT_SINGLE;
+                return -1;
+            }
+        }
+        else if (src_str[j] != '\\')
+        {
+            if (*regex_trunc && strchr(REGEX_CHARS, src_str[j]))
+            {
+                *regex_trunc = 2;
+                strcat(dst_term, "\\");
+            }
+            else if (*z3958_trunc && strchr(CCL_CHARS, src_str[j]))
+            {
+                *z3958_trunc = 2;
+                strcat(dst_term, "\\");
+            }
+            strxcat(dst_term, src_str + j, 1);                    
+        }
+    }
+    return 0;
+}
+
 /**
  * search_term: Parse CCL search term. 
  * cclp:   CCL Parser
@@ -243,6 +334,7 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
     struct ccl_rpn_node *p_top = 0;
     struct ccl_token *lookahead = cclp->look_token;
     int and_list = 0;
+    int auto_group = 0;
     int or_list = 0;
     char *attset;
     const char **truncation_aliases;
@@ -259,13 +351,15 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
 
     if (qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_AND_LIST, 0))
         and_list = 1;
+    if (qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_AUTO_GROUP, 0))
+        auto_group = 1;
     if (qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_OR_LIST, 0))
         or_list = 1;
     while (1)
     {
         struct ccl_rpn_node *p;
         size_t no, i;
-        int no_spaces = 0;
+        int is_phrase = 0;
         int relation_value = -1;
         int position_value = -1;
         int structure_value = -1;
@@ -287,14 +381,21 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
             ADVANCE;
             continue;
         }
-        /* go through each TERM token. If no truncation attribute is yet
-           met, then look for left/right truncation markers (?) and
-           set left_trunc/right_trunc/mid_trunc accordingly */
         for (no = 0; no < max && is_term_ok(lookahead->kind, term_list); no++)
         {
+            int this_is_phrase = 0;
             for (i = 0; i<lookahead->len; i++)
                 if (lookahead->name[i] == ' ')
-                    no_spaces++;
+                    this_is_phrase = 1;
+
+            if (auto_group)
+            {
+                if (no > 0 && (is_phrase || is_phrase != this_is_phrase))
+                    break;
+                is_phrase = this_is_phrase;
+            }
+            else if (this_is_phrase || no > 0)
+                is_phrase = 1;
             len += 1+lookahead->len+lookahead->ws_prefix_len;
             lookahead = lookahead->next;
         }
@@ -368,7 +469,7 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
             qual_val_type(qa, CCL_BIB1_STR, CCL_BIB1_STR_WP, &attset))
         {   /* no structure attribute met. Apply either structure attribute 
                WORD or PHRASE depending on number of CCL tokens */
-            if (no == 1 && no_spaces == 0)
+            if (!is_phrase)
                 ccl_add_attr_numeric(p, attset, CCL_BIB1_STR, 2);
             else
                 ccl_add_attr_numeric(p, attset, CCL_BIB1_STR, 1);
@@ -393,97 +494,21 @@ static struct ccl_rpn_node *search_term_x(CCL_parser cclp,
         {
             const char *src_str = cclp->look_token->name;
             size_t src_len = cclp->look_token->len;
-            int j;
-            int quote_mode = 0;
 
             if (p->u.t.term[0] && cclp->look_token->ws_prefix_len)
             {
                 strxcat(p->u.t.term, cclp->look_token->ws_prefix_buf,
                         cclp->look_token->ws_prefix_len);
             }
-            for (j = 0; j < src_len; j++)
+            if (append_term(cclp, src_str, src_len, p->u.t.term, &regex_trunc,
+                            &z3958_trunc, truncation_aliases, i == 0, i == no - 1,
+                            &left_trunc, &right_trunc))
             {
-                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);                    
-                }
+                ccl_rpn_delete(p);
+                return NULL;
             }
             ADVANCE;
         }
-
         /* make the top node point to us.. */
         if (p_top)
         {
@@ -565,6 +590,37 @@ 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)
@@ -707,20 +763,6 @@ 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)))
@@ -736,7 +778,6 @@ 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))
@@ -749,24 +790,7 @@ 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);
 }
 
 /**
@@ -933,9 +957,10 @@ 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)
@@ -953,7 +978,7 @@ 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);
@@ -967,7 +992,7 @@ 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);
@@ -996,22 +1021,7 @@ 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)
@@ -1037,7 +1047,7 @@ 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
     {