X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=zutil%2Fpquery.c;h=284055736593af5fac9fe62c8e2dbc2c740c4db7;hb=f512aa33473ae86f8e8efd8da749a82c398e87f0;hp=0df4c1b4a3d2703c50a54831fa5866ae9446450f;hpb=fd4adcc80d442f84c6a1894e890b1b6ccd02615e;p=yaz-moved-to-github.git

diff --git a/zutil/pquery.c b/zutil/pquery.c
index 0df4c1b..2840557 100644
--- a/zutil/pquery.c
+++ b/zutil/pquery.c
@@ -1,13 +1,14 @@
 /*
- * Copyright (c) 1995-2001, Index Data.
+ * Copyright (c) 1995-2003, Index Data.
  * See the file LICENSE for details.
  *
- * $Id: pquery.c,v 1.11 2001-11-13 23:00:43 adam Exp $
+ * $Id: pquery.c,v 1.20 2003-01-06 08:20:29 adam Exp $
  */
 
 #include <stdio.h>
 #include <string.h>
 #include <stdlib.h>
+#include <ctype.h>
 
 #include <yaz/proto.h>
 #include <yaz/oid.h>
@@ -15,8 +16,9 @@
 
 static oid_value p_query_dfset = VAL_NONE;
 
-struct lex_info {
+struct yaz_pqf_parser {
     const char *query_buf;
+    const char *query_ptr;
     const char *lex_buf;
     size_t lex_len;
     int query_look;
@@ -24,14 +26,17 @@ struct lex_info {
     char *right_sep;
     int escape_char;
     int term_type;
+    int external_type;
+    int error;
 };
 
-static Z_RPNStructure *rpn_structure (struct lex_info *li, ODR o, oid_proto, 
+static Z_RPNStructure *rpn_structure (struct yaz_pqf_parser *li, ODR o,
+                                      oid_proto, 
                                       int num_attr, int max_attr, 
                                       int *attr_list, char **attr_clist,
 				      oid_value *attr_set);
 
-static enum oid_value query_oid_getvalbyname (struct lex_info *li)
+static enum oid_value query_oid_getvalbyname (struct yaz_pqf_parser *li)
 {
     enum oid_value value;
     char buf[32];
@@ -44,20 +49,21 @@ static enum oid_value query_oid_getvalbyname (struct lex_info *li)
     return value;
 }
 
-static int compare_term (struct lex_info *li, const char *src, size_t off)
+static int compare_term (struct yaz_pqf_parser *li, const char *src,
+                         size_t off)
 {
     size_t len=strlen(src);
-
+    
     if (li->lex_len == len+off && !memcmp (li->lex_buf+off, src, len-off))
 	return 1;
     return 0;
 }
 
-static int query_token (struct lex_info *li)
+static int query_token (struct yaz_pqf_parser *li)
 {
     int sep_char = ' ';
     const char *sep_match;
-    const char **qptr = &li->query_buf;
+    const char **qptr = &li->query_ptr;
 
     while (**qptr == ' ')
         (*qptr)++;
@@ -70,7 +76,13 @@ static int query_token (struct lex_info *li)
         ++(*qptr);
     }
     li->lex_buf = *qptr;
-    
+   
+    if (**qptr == li->escape_char && isdigit ((*qptr)[1]))
+    {
+	++(li->lex_len);
+	++(*qptr);
+	return 'l';
+    }
     while (**qptr && **qptr != sep_char)
     {
 	if (**qptr == '\\')
@@ -83,7 +95,8 @@ static int query_token (struct lex_info *li)
     }
     if (**qptr)
 	++(*qptr);
-    if (li->lex_len >= 1 && li->lex_buf[0] == li->escape_char)
+    if (sep_char == ' ' &&
+        li->lex_len >= 1 && li->lex_buf[0] == li->escape_char)
     {
 	if (compare_term (li, "and", 1))
 	    return 'a';
@@ -105,7 +118,7 @@ static int query_token (struct lex_info *li)
     return 't';
 }
 
-static int lex (struct lex_info *li)
+static int lex (struct yaz_pqf_parser *li)
 {
     return li->query_look = query_token (li);
 }
@@ -173,7 +186,7 @@ static int escape_string(char *out_buf, const char *in, int len)
     return out - out_buf;
 }
 
-static int p_query_parse_attr(struct lex_info *li, ODR o,
+static int p_query_parse_attr(struct yaz_pqf_parser *li, ODR o,
 			      int num_attr, int *attr_list,
 			      char **attr_clist, oid_value *attr_set)
 {
@@ -183,11 +196,20 @@ static int p_query_parse_attr(struct lex_info *li, ODR o,
     {
 	attr_set[num_attr] = query_oid_getvalbyname (li);
 	if (attr_set[num_attr] == VAL_NONE)
+        {
+            li->error = YAZ_PQF_ERROR_ATTSET;
 	    return 0;
-	lex (li);
-	
+        }
+	if (!lex (li))
+        {
+            li->error = YAZ_PQF_ERROR_MISSING;
+            return 0;
+        }
 	if (!(cp = strchr (li->lex_buf, '=')))
+        {
+            li->error = YAZ_PQF_ERROR_BADATTR;
 	    return 0;
+        }
     }
     else 
     {
@@ -214,7 +236,7 @@ static int p_query_parse_attr(struct lex_info *li, ODR o,
     return 1;
 }
 
-static Z_AttributesPlusTerm *rpn_term (struct lex_info *li, ODR o,
+static Z_AttributesPlusTerm *rpn_term (struct yaz_pqf_parser *li, ODR o,
                                        oid_proto proto, 
                                        int num_attr, int *attr_list,
 				       char **attr_clist, oid_value *attr_set)
@@ -253,6 +275,7 @@ static Z_AttributesPlusTerm *rpn_term (struct lex_info *li, ODR o,
             elements[k]->attributeType = &attr_tmp[2*i];
 	    elements[k]->attributeSet =
 		yaz_oidval_to_z3950oid(o, CLASS_ATTSET, attr_set[i]);
+
 	    if (attr_clist[i])
 	    {
 		elements[k]->which = Z_AttributeValue_complex;
@@ -288,15 +311,44 @@ static Z_AttributesPlusTerm *rpn_term (struct lex_info *li, ODR o,
     zapt->attributes->attributes = elements;
 
     zapt->term = term;
-    term->which = Z_Term_general;
-    term->u.general = term_octet;
-    term_octet->buf = (unsigned char *)odr_malloc (o, li->lex_len);
+
+    term_octet->buf = (unsigned char *)odr_malloc (o, 1 + li->lex_len);
     term_octet->size = term_octet->len =
-	escape_string ((char *) (term_octet->buf), li->lex_buf, li->lex_len);
+        escape_string ((char *) (term_octet->buf), li->lex_buf, li->lex_len);
+    term_octet->buf[term_octet->size] = 0;  /* null terminate */
+    
+    switch (li->term_type)
+    {
+    case Z_Term_general:
+        term->which = Z_Term_general;
+        term->u.general = term_octet;
+        break;
+    case Z_Term_characterString:
+        term->which = Z_Term_characterString;
+        term->u.characterString = (char*) term_octet->buf; 
+                                    /* null terminated above */
+        break;
+    case Z_Term_numeric:
+        term->which = Z_Term_numeric;
+        term->u.numeric = odr_intdup (o, atoi((char*) (term_octet->buf)));
+        break;
+    case Z_Term_null:
+        term->which = Z_Term_null;
+        term->u.null = odr_nullval();
+        break;
+    case Z_Term_external:
+        term->which = Z_Term_external;
+	term->u.external = 0;
+	break;
+    default:
+        term->which = Z_Term_null;
+        term->u.null = odr_nullval();
+        break;
+    }
     return zapt;
 }
 
-static Z_Operand *rpn_simple (struct lex_info *li, ODR o, oid_proto proto,
+static Z_Operand *rpn_simple (struct yaz_pqf_parser *li, ODR o, oid_proto proto,
                               int num_attr, int *attr_list, char **attr_clist,
                               oid_value *attr_set)
 {
@@ -310,13 +362,16 @@ static Z_Operand *rpn_simple (struct lex_info *li, ODR o, oid_proto proto,
         if (!(zo->u.attributesPlusTerm =
               rpn_term (li, o, proto, num_attr, attr_list, attr_clist,
 			attr_set)))
-            return NULL;
+            return 0;
         lex (li);
         break;
     case 's':
         lex (li);
         if (!li->query_look)
-            return NULL;
+        {
+            li->error = YAZ_PQF_ERROR_MISSING;
+            return 0;
+        }
         zo->which = Z_Operand_resultSetId;
         zo->u.resultSetId = (char *)odr_malloc (o, li->lex_len+1);
         memcpy (zo->u.resultSetId, li->lex_buf, li->lex_len);
@@ -324,17 +379,23 @@ static Z_Operand *rpn_simple (struct lex_info *li, ODR o, oid_proto proto,
         lex (li);
         break;
     default:
-        return NULL;
+        /* we're only called if one of the above types are seens so
+           this shouldn't happen */
+        li->error = YAZ_PQF_ERROR_INTERNAL;
+        return 0;
     }
     return zo;
 }
 
-static Z_ProximityOperator *rpn_proximity (struct lex_info *li, ODR o)
+static Z_ProximityOperator *rpn_proximity (struct yaz_pqf_parser *li, ODR o)
 {
     Z_ProximityOperator *p = (Z_ProximityOperator *)odr_malloc (o, sizeof(*p));
 
     if (!lex (li))
+    {
+        li->error = YAZ_PQF_ERROR_MISSING;
         return NULL;
+    }
     if (*li->lex_buf == '1')
     {
         p->exclusion = (int *)odr_malloc (o, sizeof(*p->exclusion));
@@ -349,22 +410,34 @@ static Z_ProximityOperator *rpn_proximity (struct lex_info *li, ODR o)
         p->exclusion = NULL;
 
     if (!lex (li))
+    {
+        li->error = YAZ_PQF_ERROR_MISSING;
         return NULL;
+    }
     p->distance = (int *)odr_malloc (o, sizeof(*p->distance));
     *p->distance = atoi (li->lex_buf);
 
     if (!lex (li))
+    {
+        li->error = YAZ_PQF_ERROR_MISSING;
         return NULL;
+    }
     p->ordered = (int *)odr_malloc (o, sizeof(*p->ordered));
     *p->ordered = atoi (li->lex_buf);
     
     if (!lex (li))
+    {
+        li->error = YAZ_PQF_ERROR_MISSING;
         return NULL;
+    }
     p->relationType = (int *)odr_malloc (o, sizeof(*p->relationType));
     *p->relationType = atoi (li->lex_buf);
 
     if (!lex (li))
+    {
+        li->error = YAZ_PQF_ERROR_MISSING;
         return NULL;
+    }
     if (*li->lex_buf == 'k')
         p->which = 0;
     else if (*li->lex_buf == 'p')
@@ -373,14 +446,17 @@ static Z_ProximityOperator *rpn_proximity (struct lex_info *li, ODR o)
         p->which = atoi (li->lex_buf);
 
     if (!lex (li))
+    {
+        li->error = YAZ_PQF_ERROR_MISSING;
         return NULL;
+    }
     p->which = Z_ProximityOperator_known;
     p->u.known = (int *)odr_malloc (o, sizeof(*p->u.known));
     *p->u.known = atoi (li->lex_buf);
     return p;
 }
 
-static Z_Complex *rpn_complex (struct lex_info *li, ODR o, oid_proto proto,
+static Z_Complex *rpn_complex (struct yaz_pqf_parser *li, ODR o, oid_proto proto,
                                int num_attr, int max_attr, 
                                int *attr_list, char **attr_clist,
 			       oid_value *attr_set)
@@ -412,6 +488,9 @@ static Z_Complex *rpn_complex (struct lex_info *li, ODR o, oid_proto proto,
             return NULL;
         break;
     default:
+        /* we're only called if one of the above types are seens so
+           this shouldn't happen */
+        li->error = YAZ_PQF_ERROR_INTERNAL;
         return NULL;
     }
     lex (li);
@@ -426,7 +505,32 @@ static Z_Complex *rpn_complex (struct lex_info *li, ODR o, oid_proto proto,
     return zc;
 }
 
-static Z_RPNStructure *rpn_structure (struct lex_info *li, ODR o,
+static void rpn_term_type (struct yaz_pqf_parser *li, ODR o)
+{
+    if (!li->query_look)
+        return ;
+    if (compare_term (li, "general", 0))
+        li->term_type = Z_Term_general;
+    else if (compare_term (li, "numeric", 0))
+        li->term_type = Z_Term_numeric;
+    else if (compare_term (li, "string", 0))
+        li->term_type = Z_Term_characterString;
+    else if (compare_term (li, "oid", 0))
+        li->term_type = Z_Term_oid;
+    else if (compare_term (li, "datetime", 0))
+        li->term_type = Z_Term_dateTime;
+    else if (compare_term (li, "null", 0))
+        li->term_type = Z_Term_null;
+    else if (compare_term(li, "range", 0))
+    {
+	/* prepare for external: range search .. */
+        li->term_type = Z_Term_external;
+	li->external_type = VAL_MULTISRCH2;
+    }
+    lex (li);
+}
+                           
+static Z_RPNStructure *rpn_structure (struct yaz_pqf_parser *li, ODR o,
                                       oid_proto proto, 
                                       int num_attr, int max_attr, 
                                       int *attr_list,
@@ -459,10 +563,18 @@ static Z_RPNStructure *rpn_structure (struct lex_info *li, ODR o,
     case 'l':
         lex (li);
         if (!li->query_look)
-            return NULL;
+        {
+            li->error = YAZ_PQF_ERROR_MISSING;
+            return 0;
+        }
         if (num_attr >= max_attr)
-            return NULL;
-	p_query_parse_attr(li, o, num_attr, attr_list, attr_clist, attr_set);
+        {
+            li->error = YAZ_PQF_ERROR_TOOMANY;
+            return 0;
+        }
+	if (!p_query_parse_attr(li, o, num_attr, attr_list,
+                                attr_clist, attr_set))
+            return 0;
 	num_attr++;
         lex (li);
         return
@@ -470,31 +582,18 @@ static Z_RPNStructure *rpn_structure (struct lex_info *li, ODR o,
 			   attr_clist,  attr_set);
     case 'y':
 	lex (li);
-	if (!li->query_look)
-	    return NULL;
-	if (compare_term (li, "general", 0))
-	    li->term_type = Z_Term_general;
-	else if (compare_term (li, "numeric", 0))
-	    li->term_type = Z_Term_numeric;
-	else if (compare_term (li, "string", 0))
-	    li->term_type = Z_Term_characterString;
-	else if (compare_term (li, "oid", 0))
-	    li->term_type = Z_Term_oid;
-	else if (compare_term (li, "datetime", 0))
-	    li->term_type = Z_Term_dateTime;
-	else if (compare_term (li, "null", 0))
-	    li->term_type = Z_Term_null;
-	lex (li);
+        rpn_term_type (li, o);
         return
             rpn_structure (li, o, proto, num_attr, max_attr, attr_list,
 			   attr_clist, attr_set);
     case 0:                /* operator/operand expected! */
-        return NULL;
+        li->error = YAZ_PQF_ERROR_MISSING;
+        return 0;
     }
     return sz;
 }
 
-Z_RPNQuery *p_query_rpn_mk (ODR o, struct lex_info *li, oid_proto proto,
+Z_RPNQuery *p_query_rpn_mk (ODR o, struct yaz_pqf_parser *li, oid_proto proto,
                             const char *qbuf)
 {
     Z_RPNQuery *zq;
@@ -510,7 +609,10 @@ Z_RPNQuery *p_query_rpn_mk (ODR o, struct lex_info *li, oid_proto proto,
         lex (li);
         topSet = query_oid_getvalbyname (li);
         if (topSet == VAL_NONE)
+        {
+            li->error = YAZ_PQF_ERROR_ATTSET;
             return NULL;
+        }
 
         lex (li);
     }
@@ -522,29 +624,39 @@ Z_RPNQuery *p_query_rpn_mk (ODR o, struct lex_info *li, oid_proto proto,
     zq->attributeSetId = yaz_oidval_to_z3950oid(o, CLASS_ATTSET, topSet);
 
     if (!zq->attributeSetId)
+    {
+        li->error = YAZ_PQF_ERROR_ATTSET;
 	return 0;
+    }
 
     if (!(zq->RPNStructure = rpn_structure (li, o, proto, 0, 512,
                                             attr_array, attr_clist, attr_set)))
-        return NULL;
+        return 0;
+    if (li->query_look)
+    {
+        li->error = YAZ_PQF_ERROR_EXTRA;
+        return 0;
+    }
     return zq;
 }
 
 Z_RPNQuery *p_query_rpn (ODR o, oid_proto proto,
                          const char *qbuf)
 {
-    struct lex_info li;
-    
+    struct yaz_pqf_parser li;
+
+    li.error = 0;
     li.left_sep = "{\"";
     li.right_sep = "}\"";
     li.escape_char = '@';
     li.term_type = Z_Term_general;
-    li.query_buf = qbuf;
+    li.query_buf = li.query_ptr = qbuf;
+    li.lex_buf = 0;
     return p_query_rpn_mk (o, &li, proto, qbuf);
 }
 
 
-Z_AttributesPlusTerm *p_query_scan_mk (struct lex_info *li,
+Z_AttributesPlusTerm *p_query_scan_mk (struct yaz_pqf_parser *li,
                                        ODR o, oid_proto proto,
                                        Odr_oid **attributeSetP,
                                        const char *qbuf)
@@ -555,6 +667,7 @@ Z_AttributesPlusTerm *p_query_scan_mk (struct lex_info *li,
     int num_attr = 0;
     int max_attr = 512;
     oid_value topSet = VAL_NONE;
+    Z_AttributesPlusTerm *apt;
 
     lex (li);
     if (li->query_look == 'r')
@@ -571,33 +684,65 @@ Z_AttributesPlusTerm *p_query_scan_mk (struct lex_info *li,
 
     *attributeSetP = yaz_oidval_to_z3950oid (o, CLASS_ATTSET, topSet);
 
-    while (li->query_look == 'l')
+    while (1)
     {
-        lex (li);
-        if (!li->query_look)
-            return NULL;
-        if (num_attr >= max_attr)
-            return NULL;
-        p_query_parse_attr(li, o, num_attr, attr_list, attr_clist, attr_set);
-        num_attr++;
-        lex (li);
+        if (li->query_look == 'l')
+        {
+            lex (li);
+            if (!li->query_look)
+            {
+                li->error = YAZ_PQF_ERROR_MISSING;
+                return 0;
+            }
+            if (num_attr >= max_attr)
+            {
+                li->error = YAZ_PQF_ERROR_TOOMANY;
+                return 0;
+            }
+            if (!p_query_parse_attr(li, o, num_attr, attr_list,
+                                    attr_clist, attr_set))
+                return 0;
+            num_attr++;
+            lex (li);
+        }
+        else if (li->query_look == 'y')
+        {
+            lex (li);
+            rpn_term_type (li, o);
+        }
+        else
+            break;
     }
     if (!li->query_look)
-        return NULL;
-    return rpn_term (li, o, proto, num_attr, attr_list, attr_clist, attr_set);
+    {
+        li->error = YAZ_PQF_ERROR_MISSING;
+        return 0;
+    }
+    apt = rpn_term (li, o, proto, num_attr, attr_list, attr_clist, attr_set);
+
+    lex (li);
+
+    if (li->query_look != 0)
+    {
+        li->error = YAZ_PQF_ERROR_EXTRA;
+        return 0;
+    }
+    return apt;
 }
 
 Z_AttributesPlusTerm *p_query_scan (ODR o, oid_proto proto,
                                     Odr_oid **attributeSetP,
                                     const char *qbuf)
 {
-    struct lex_info li;
+    struct yaz_pqf_parser li;
 
+    li.error = 0;
     li.left_sep = "{\"";
     li.right_sep = "}\"";
     li.escape_char = '@';
     li.term_type = Z_Term_general;
-    li.query_buf = qbuf;
+    li.query_buf = li.query_ptr = qbuf;
+    li.lex_buf = 0;
 
     return p_query_scan_mk (&li, o, proto, attributeSetP, qbuf);
 }
@@ -608,3 +753,65 @@ int p_query_attset (const char *arg)
     return (p_query_dfset == VAL_NONE) ? -1 : 0;
 }
 
+YAZ_PQF_Parser yaz_pqf_create (void)
+{
+    YAZ_PQF_Parser p = (YAZ_PQF_Parser) xmalloc (sizeof(*p));
+
+    p->error = 0;
+    p->left_sep = "{\"";
+    p->right_sep = "}\"";
+    p->escape_char = '@';
+    p->term_type = Z_Term_general;
+
+    return p;
+}
+
+void yaz_pqf_destroy (YAZ_PQF_Parser p)
+{
+    xfree (p);
+}
+
+Z_RPNQuery *yaz_pqf_parse (YAZ_PQF_Parser p, ODR o, const char *qbuf)
+{
+    if (!p)
+        return 0;
+    p->query_buf = p->query_ptr = qbuf;
+    p->lex_buf = 0;
+    return p_query_rpn_mk (o, p, PROTO_Z3950, qbuf);
+}
+
+Z_AttributesPlusTerm *yaz_pqf_scan (YAZ_PQF_Parser p, ODR o,
+                                    Odr_oid **attributeSetP,
+                                    const char *qbuf)
+{
+    if (!p)
+        return 0;
+    p->query_buf = p->query_ptr = qbuf;
+    p->lex_buf = 0;
+    return p_query_scan_mk (p, o, PROTO_Z3950, attributeSetP, qbuf);
+}
+
+int yaz_pqf_error (YAZ_PQF_Parser p, const char **msg, size_t *off)
+{
+    switch (p->error)
+    {
+    case YAZ_PQF_ERROR_NONE:
+        *msg = "no error"; break;
+    case YAZ_PQF_ERROR_EXTRA:
+        *msg = "extra token"; break;
+    case YAZ_PQF_ERROR_MISSING:
+        *msg = "missing token"; break;
+    case YAZ_PQF_ERROR_ATTSET:
+        *msg = "unknown attribute set"; break;
+    case YAZ_PQF_ERROR_TOOMANY:
+        *msg = "too many attributes"; break;
+    case YAZ_PQF_ERROR_BADATTR:
+        *msg = "bad attribute specification"; break;
+    case YAZ_PQF_ERROR_INTERNAL:
+        *msg = "internal error"; break;
+    default:
+        *msg = "unknown error"; break;
+    }
+    *off = p->query_ptr - p->query_buf;
+    return p->error;
+}