X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fcclfind.c;h=72f1aa6fd2627104c4cc79d9aea9c7b560af4c10;hb=be821514c869d68186361b5aab6bbfd1aa60e087;hp=08a4f79440fcd2c0ba2d35d302d77021bc654400;hpb=6e8a8f144dfc8225d07a56d6a7f18056cb18ee22;p=yaz-moved-to-github.git

diff --git a/src/cclfind.c b/src/cclfind.c
index 08a4f79..72f1aa6 100644
--- a/src/cclfind.c
+++ b/src/cclfind.c
@@ -41,10 +41,22 @@
  * USE OR PERFORMANCE OF THIS SOFTWARE.
  *
  */
+
+/** 
+ * \file cclfind.c
+ * \brief Implements parsing of a CCL FIND query.
+ *
+ * This source file implements parsing of a CCL Query (ISO8777).
+ * The parser uses predictive parsing, but it does several tokens
+ * of lookahead in the handling of relational operations.. So
+ * it's not really pure.
+ */
+
+
 /* CCL find (to rpn conversion)
  * Europagate, 1995
  *
- * $Id: cclfind.c,v 1.2 2004-08-11 09:02:48 adam Exp $
+ * $Id: cclfind.c,v 1.8 2005-06-25 15:46:03 adam Exp $
  *
  * Old Europagate log:
  *
@@ -109,7 +121,7 @@
 /* move one token forward */
 #define ADVANCE cclp->look_token = cclp->look_token->next
 
-/* 
+/**
  * qual_val_type: test for existance of attribute type/value pair.
  * qa:     Attribute array
  * type:   Type of attribute to search for
@@ -128,7 +140,7 @@ static int qual_val_type (struct ccl_rpn_attr **qa, int type, int value,
         while (q)
         {
             if (q->type == type && q->kind == CCL_RPN_ATTR_NUMERIC &&
-		q->value.numeric == value)
+                q->value.numeric == value)
             {
                 if (attset)
                     *attset = q->set;
@@ -139,7 +151,7 @@ static int qual_val_type (struct ccl_rpn_attr **qa, int type, int value,
     return 0;
 }
 
-/*
+/**
  * strxcat: concatenate strings.
  * n:      Null-terminated Destination string 
  * src:    Source string to be appended (not null-terminated)
@@ -154,7 +166,7 @@ static void strxcat (char *n, const char *src, int len)
     *n = '\0';
 }
 
-/*
+/**
  * copy_token_name: Return copy of CCL token name
  * tp:      Pointer to token info.
  * return:  malloc(3) allocated copy of token name.
@@ -168,7 +180,7 @@ static char *copy_token_name (struct ccl_token *tp)
     return str;
 }
 
-/*
+/**
  * mk_node: Create RPN node.
  * kind:   Type of node.
  * return: pointer to allocated node.
@@ -182,7 +194,7 @@ static struct ccl_rpn_node *mk_node (int kind)
     return p;
 }
 
-/*
+/**
  * ccl_rpn_delete: Delete RPN tree.
  * rpn:   Pointer to tree.
  */
@@ -204,8 +216,8 @@ void ccl_rpn_delete (struct ccl_rpn_node *rpn)
         for (attr = rpn->u.t.attr_list; attr; attr = attr1)
         {
             attr1 = attr->next;
-	    if (attr->kind == CCL_RPN_ATTR_STRING)
-		xfree(attr->value.str);
+            if (attr->kind == CCL_RPN_ATTR_STRING)
+                xfree(attr->value.str);
             if (attr->set)
                 xfree (attr->set);
             xfree (attr);
@@ -217,6 +229,7 @@ void ccl_rpn_delete (struct ccl_rpn_node *rpn)
     case CCL_RPN_PROX:
         ccl_rpn_delete (rpn->u.p[0]);
         ccl_rpn_delete (rpn->u.p[1]);
+        ccl_rpn_delete (rpn->u.p[2]);
         break;
     }
     xfree (rpn);
@@ -237,7 +250,7 @@ static struct ccl_rpn_node *search_terms (CCL_parser cclp,
                                           struct ccl_rpn_attr **qa);
 
 static struct ccl_rpn_attr *add_attr_node (struct ccl_rpn_node *p,
-					   const char *set, int type)
+                                           const char *set, int type)
 {
     struct ccl_rpn_attr *n;
     
@@ -259,7 +272,7 @@ static struct ccl_rpn_attr *add_attr_node (struct ccl_rpn_node *p,
     return n;
 }
 
-/*
+/**
  * add_attr_numeric: Add attribute (type/value) to RPN term node.
  * p:     RPN node of type term.
  * type:  Type of attribute
@@ -267,7 +280,7 @@ static struct ccl_rpn_attr *add_attr_node (struct ccl_rpn_node *p,
  * set: Attribute set name
  */
 static void add_attr_numeric (struct ccl_rpn_node *p, const char *set,
-			      int type, int value)
+                              int type, int value)
 {
     struct ccl_rpn_attr *n;
 
@@ -277,7 +290,7 @@ static void add_attr_numeric (struct ccl_rpn_node *p, const char *set,
 }
 
 static void add_attr_string (struct ccl_rpn_node *p, const char *set,
-			     int type, char *value)
+                             int type, char *value)
 {
     struct ccl_rpn_attr *n;
 
@@ -287,7 +300,7 @@ static void add_attr_string (struct ccl_rpn_node *p, const char *set,
 }
 
 
-/*
+/**
  * search_term: Parse CCL search term. 
  * cclp:   CCL Parser
  * qa:     Qualifier attributes already applied.
@@ -307,9 +320,9 @@ static struct ccl_rpn_node *search_term_x (CCL_parser cclp,
     const char *truncation_aliases;
 
     truncation_aliases =
-	ccl_qual_search_special(cclp->bibset, "truncation");
+        ccl_qual_search_special(cclp->bibset, "truncation");
     if (!truncation_aliases)
-	truncation_aliases = "?";
+        truncation_aliases = "?";
 
     if (qual_val_type (qa, CCL_BIB1_STR, CCL_BIB1_STR_AND_LIST, 0))
         and_list = 1;
@@ -332,13 +345,13 @@ static struct ccl_rpn_node *search_term_x (CCL_parser cclp,
         size_t max = 200;
         if (and_list || or_list || !multi)
             max = 1;
-	
-	/* ignore commas when dealing with and-lists .. */
+        
+        /* ignore commas when dealing with and-lists .. */
         if (and_list && lookahead && lookahead->kind == CCL_TOK_COMMA)
         {
-	    lookahead = lookahead->next;
+            lookahead = lookahead->next;
             ADVANCE;
-	    continue;
+            continue;
         }
         /* go through each TERM token. If no truncation attribute is yet
            met, then look for left/right truncation markers (?) and
@@ -347,8 +360,8 @@ 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, lookahead->name[i]))
+                    no_spaces++;
+                else if (strchr(truncation_aliases, lookahead->name[i]))
                 {
                     if (no == 0 && i == 0 && lookahead->len >= 1)
                         left_trunc = 1;
@@ -358,7 +371,7 @@ static struct ccl_rpn_node *search_term_x (CCL_parser cclp,
                     else
                         mid_trunc = 1;
                 }
-            len += 1+lookahead->len;
+            len += 1+lookahead->len+lookahead->ws_prefix_len;
             lookahead = lookahead->next;
         }
 
@@ -396,48 +409,48 @@ static struct ccl_rpn_node *search_term_x (CCL_parser cclp,
             struct ccl_rpn_attr *attr;
             
             for (attr = qa[i]; attr; attr = attr->next)
-		switch(attr->kind)
-		{
-		case CCL_RPN_ATTR_STRING:
-		    add_attr_string(p, attr->set, attr->type,
-				    attr->value.str);
-		    break;
-		case CCL_RPN_ATTR_NUMERIC:
-		    if (attr->value.numeric > 0)
-		    {   /* deal only with REAL attributes (positive) */
-			switch (attr->type)
-			{
-			case CCL_BIB1_REL:
-			    if (relation_value != -1)
-				continue;
-			    relation_value = attr->value.numeric;
-			    break;
-			case CCL_BIB1_POS:
-			    if (position_value != -1)
-				continue;
-			    position_value = attr->value.numeric;
-			    break;
-			case CCL_BIB1_STR:
-			    if (structure_value != -1)
-				continue;
-			    structure_value = attr->value.numeric;
-			    break;
-			case CCL_BIB1_TRU:
-			    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)
-				continue;
-			    completeness_value = attr->value.numeric;
-			    break;
-			}
-			add_attr_numeric(p, attr->set, attr->type,
-					 attr->value.numeric);
-		    }
-		}
+                switch(attr->kind)
+                {
+                case CCL_RPN_ATTR_STRING:
+                    add_attr_string(p, attr->set, attr->type,
+                                    attr->value.str);
+                    break;
+                case CCL_RPN_ATTR_NUMERIC:
+                    if (attr->value.numeric > 0)
+                    {   /* deal only with REAL attributes (positive) */
+                        switch (attr->type)
+                        {
+                        case CCL_BIB1_REL:
+                            if (relation_value != -1)
+                                continue;
+                            relation_value = attr->value.numeric;
+                            break;
+                        case CCL_BIB1_POS:
+                            if (position_value != -1)
+                                continue;
+                            position_value = attr->value.numeric;
+                            break;
+                        case CCL_BIB1_STR:
+                            if (structure_value != -1)
+                                continue;
+                            structure_value = attr->value.numeric;
+                            break;
+                        case CCL_BIB1_TRU:
+                            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)
+                                continue;
+                            completeness_value = attr->value.numeric;
+                            break;
+                        }
+                        add_attr_numeric(p, attr->set, attr->type,
+                                         attr->value.numeric);
+                    }
+                }
         }
         /* len now holds the number of characters in the RPN term */
         /* no holds the number of CCL tokens (1 or more) */
@@ -451,7 +464,7 @@ static struct ccl_rpn_node *search_term_x (CCL_parser cclp,
             else
                 add_attr_numeric (p, attset, CCL_BIB1_STR, 1);
         }
-        
+
         /* make the RPN token */
         p->u.t.term = (char *)xmalloc (len);
         ccl_assert (p->u.t.term);
@@ -466,17 +479,14 @@ static struct ccl_rpn_node *search_term_x (CCL_parser cclp,
                 src_len--;
                 src_str++;
             }
-            else if (i == no-1 && right_trunc)
+            if (i == no-1 && right_trunc)
                 src_len--;
-            if (src_len)
+            if (i && cclp->look_token->ws_prefix_len)
             {
-                int len = strlen(p->u.t.term);
-                if (len &&
-                    !strchr("-+", *src_str) &&
-                    !strchr("-+", p->u.t.term[len-1]))
-                {
-                    strcat (p->u.t.term, " ");
-                }
+                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, src_str, src_len);
             ADVANCE;
@@ -535,43 +545,13 @@ static struct ccl_rpn_node *search_term (CCL_parser cclp,
     return search_term_x(cclp, qa, list, 0);
 }
 
-static struct ccl_rpn_node *qualifiers2 (CCL_parser cclp,
-                                         struct ccl_rpn_attr **ap)
+static
+struct ccl_rpn_node *qualifiers_order (CCL_parser cclp,
+                                       struct ccl_rpn_attr **ap, char *attset)
 {
-    char *attset;
-    int rel;
+    int rel = 0;
+    struct ccl_rpn_node *p;
 
-    if (!qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER, &attset))
-    {                
-        /* unordered relation */
-        struct ccl_rpn_node *p;
-        if (KIND != CCL_TOK_EQ)
-        {
-            cclp->error_code = CCL_ERR_EQ_EXPECTED;
-            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;
-    }
-    /* ordered relation ... */
-    rel = 0;
     if (cclp->look_token->len == 1)
     {
         if (cclp->look_token->name[0] == '<')
@@ -591,78 +571,185 @@ static struct ccl_rpn_node *qualifiers2 (CCL_parser cclp,
             rel = 6;
     }
     if (!rel)
+    {
         cclp->error_code = CCL_ERR_BAD_RELATION;
-    else
+        return NULL;
+    }
+    ADVANCE;  /* skip relation */
+    if (rel == 3 &&
+        qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_PORDER, 0))
     {
-        struct ccl_rpn_node *p;
-        
-        ADVANCE;                      /* skip relation */
-        if (KIND == CCL_TOK_TERM &&
-            cclp->look_token->next && cclp->look_token->next->len == 1 &&
-            cclp->look_token->next->name[0] == '-')
+        /* allow - inside term and treat it as range _always_ */
+        /* relation is =. Extract "embedded" - to separate terms */
+        if (KIND == CCL_TOK_TERM)
         {
-            struct ccl_rpn_node *p1;
-            if (!(p1 = search_term (cclp, ap)))
-                return NULL;
-            ADVANCE;                   /* skip '-' */
-            if (KIND == CCL_TOK_TERM)  /* = term - term  ? */
+            size_t i;
+            for (i = 0; i<cclp->look_token->len; i++)
             {
-                struct ccl_rpn_node *p2;
-                
-                if (!(p2 = search_term (cclp, ap)))
-                {
-                    ccl_rpn_delete (p1);
-                    return NULL;
-                }
-                p = mk_node (CCL_RPN_AND);
-                p->u.p[0] = p1;
-                add_attr_numeric (p1, attset, CCL_BIB1_REL, 4);
-                p->u.p[1] = p2;
-                add_attr_numeric (p2, attset, CCL_BIB1_REL, 2);
-                return p;
+                if (cclp->look_token->name[i] == '-')
+                    break;
             }
-            else                       /* = term -    */
-            {
-                add_attr_numeric (p1, attset, CCL_BIB1_REL, 4);
-                return p1;
+            
+            if (cclp->look_token->len > 1 && i == 0)
+            {   /*  -xx*/
+                struct ccl_token *ntoken = ccl_token_add (cclp->look_token);
+
+                ntoken->kind = CCL_TOK_TERM;
+                ntoken->name = cclp->look_token->name + 1;
+                ntoken->len = cclp->look_token->len - 1;
+
+                cclp->look_token->len = 1;
+                cclp->look_token->name = "-";
+            }
+            else if (cclp->look_token->len > 1 && i == cclp->look_token->len-1)
+            {   /* xx- */
+                struct ccl_token *ntoken = ccl_token_add (cclp->look_token);
+
+                ntoken->kind = CCL_TOK_TERM;
+                ntoken->name = "-";
+                ntoken->len = 1;
+
+                (cclp->look_token->len)--;
+            }
+            else if (cclp->look_token->len > 2 && i < cclp->look_token->len)
+            {   /* xx-yy */
+                struct ccl_token *ntoken1 = ccl_token_add (cclp->look_token);
+                struct ccl_token *ntoken2 = ccl_token_add (ntoken1);
+
+                ntoken1->kind = CCL_TOK_TERM;  /* generate - */
+                ntoken1->name = "-";
+                ntoken1->len = 1;
+
+                ntoken2->kind = CCL_TOK_TERM;  /* generate yy */
+                ntoken2->name = cclp->look_token->name + (i+1);
+                ntoken2->len = cclp->look_token->len - (i+1);
+
+                cclp->look_token->len = i;     /* adjust xx */
+            }
+            else if (i == cclp->look_token->len &&
+                     cclp->look_token->next &&
+                     cclp->look_token->next->kind == CCL_TOK_TERM &&
+                     cclp->look_token->next->len > 1 &&
+                     cclp->look_token->next->name[0] == '-')
+                     
+            {   /* xx -yy */
+                /* we _know_ that xx does not have - in it */
+                struct ccl_token *ntoken = ccl_token_add (cclp->look_token);
+
+                ntoken->kind = CCL_TOK_TERM;    /* generate - */
+                ntoken->name = "-";
+                ntoken->len = 1;
+
+                (ntoken->next->name)++;        /* adjust yy */
+                (ntoken->next->len)--; 
             }
         }
-        else if (cclp->look_token->len == 1 &&
-                 cclp->look_token->name[0] == '-')   /* = - term  ? */
-        {
-            ADVANCE;
-            if (!(p = search_term (cclp, ap)))
-                return NULL;
-            add_attr_numeric (p, attset, CCL_BIB1_REL, 2);
-            return p;
-        }
-        else if (KIND == CCL_TOK_LP)
+    }
+        
+    if (rel == 3 &&
+        KIND == CCL_TOK_TERM &&
+        cclp->look_token->next && cclp->look_token->next->len == 1 &&
+        cclp->look_token->next->name[0] == '-')
+    {
+        struct ccl_rpn_node *p1;
+        if (!(p1 = search_term (cclp, ap)))
+            return NULL;
+        ADVANCE;                   /* skip '-' */
+        if (KIND == CCL_TOK_TERM)  /* = term - term  ? */
         {
-            ADVANCE;
-            if (!(p = find_spec (cclp, ap)))
-                return NULL;
-            if (KIND != CCL_TOK_RP)
+            struct ccl_rpn_node *p2;
+            
+            if (!(p2 = search_term (cclp, ap)))
             {
-                cclp->error_code = CCL_ERR_RP_EXPECTED;
-                ccl_rpn_delete (p);
+                ccl_rpn_delete (p1);
                 return NULL;
             }
-            ADVANCE;
+            p = mk_node (CCL_RPN_AND);
+            p->u.p[0] = p1;
+            add_attr_numeric (p1, attset, CCL_BIB1_REL, 4);
+            p->u.p[1] = p2;
+            add_attr_numeric (p2, attset, CCL_BIB1_REL, 2);
             return p;
         }
-        else
+        else                       /* = term -    */
         {
-            if (!(p = search_terms (cclp, ap)))
-                return NULL;
-            add_attr_numeric (p, attset, CCL_BIB1_REL, rel);
-            return p;
+            add_attr_numeric (p1, attset, CCL_BIB1_REL, 4);
+            return p1;
         }
-        cclp->error_code = CCL_ERR_TERM_EXPECTED;
     }
+    else if (rel == 3 &&
+             cclp->look_token->len == 1 &&
+             cclp->look_token->name[0] == '-')   /* = - term  ? */
+    {
+        ADVANCE;
+        if (!(p = search_term (cclp, ap)))
+            return NULL;
+        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)))
+            return NULL;
+        add_attr_numeric (p, attset, CCL_BIB1_REL, rel);
+        return p;
+    }
+    cclp->error_code = CCL_ERR_TERM_EXPECTED;
     return NULL;
 }
 
-/*
+static
+struct ccl_rpn_node *qualifiers2 (CCL_parser cclp, struct ccl_rpn_attr **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))
+        return qualifiers_order(cclp, ap, attset);
+
+    /* unordered relation */
+    if (KIND != CCL_TOK_EQ)
+    {
+        cclp->error_code = CCL_ERR_EQ_EXPECTED;
+        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;
+}
+
+/**
  * qualifiers1: Parse CCL qualifiers and search terms. 
  * cclp:   CCL Parser
  * la:     Token pointer to RELATION token.
@@ -814,7 +901,7 @@ static struct ccl_rpn_node *qualifiers1 (CCL_parser cclp, struct ccl_token *la,
 }
 
 
-/*
+/**
  * search_terms: Parse CCL search terms - including proximity.
  * cclp:   CCL Parser
  * qa:     Qualifier attributes already applied.
@@ -876,7 +963,7 @@ static struct ccl_rpn_node *search_terms (CCL_parser cclp,
     return p1;
 }
 
-/*
+/**
  * search_elements: Parse CCL search elements
  * cclp:   CCL Parser
  * qa:     Qualifier attributes already applied.
@@ -970,7 +1057,7 @@ static struct ccl_rpn_node *search_elements (CCL_parser cclp,
     }
 }
 
-/*
+/**
  * find_spec: Parse CCL find specification
  * cclp:   CCL Parser
  * qa:     Qualifier attributes already applied.
@@ -1038,8 +1125,6 @@ struct ccl_rpn_node *ccl_parser_find (CCL_parser cclp, struct ccl_token *list)
 {
     struct ccl_rpn_node *p;
 
-    
-
     cclp->look_token = list;
     p = find_spec (cclp, NULL);
     if (p && KIND != CCL_TOK_EOL)
@@ -1059,7 +1144,7 @@ struct ccl_rpn_node *ccl_parser_find (CCL_parser cclp, struct ccl_token *list)
     return p;
 }
 
-/*
+/**
  * ccl_find: Parse CCL find - token representation
  * bibset:  Bibset to be used for the parsing
  * list:    List of tokens
@@ -1085,7 +1170,7 @@ struct ccl_rpn_node *ccl_find (CCL_bibset bibset, struct ccl_token *list,
     return p;
 }
 
-/*
+/**
  * ccl_find_str: Parse CCL find - string representation
  * bibset:  Bibset to be used for the parsing
  * str:     String to be parsed
@@ -1112,3 +1197,11 @@ struct ccl_rpn_node *ccl_find_str (CCL_bibset bibset, const char *str,
     ccl_token_del (list);
     return p;
 }
+/*
+ * Local variables:
+ * c-basic-offset: 4
+ * indent-tabs-mode: nil
+ * End:
+ * vim: shiftwidth=4 tabstop=8 expandtab
+ */
+