Rename yaz_use_attribute_create to zget_AttributeList_use_string
[yaz-moved-to-github.git] / src / pquery.c
index f9f51f1..f913a0f 100644 (file)
@@ -1,15 +1,18 @@
 /* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2010 Index Data
+ * Copyright (C) Index Data
  * See the file LICENSE for details.
  */
 /**
  * \file pquery.c
  * \brief Implements PQF parsing
  */
+#if HAVE_CONFIG_H
+#include <config.h>
+#endif
+
 #include <stdio.h>
 #include <string.h>
 #include <stdlib.h>
-#include <ctype.h>
 
 #include <yaz/proto.h>
 #include <yaz/oid_db.h>
@@ -30,7 +33,7 @@ struct yaz_pqf_parser {
 };
 
 static Z_RPNStructure *rpn_structure(struct yaz_pqf_parser *li, ODR o,
-                                     int num_attr, int max_attr, 
+                                     int num_attr, int max_attr,
                                      Odr_int *attr_list, char **attr_clist,
                                      Odr_oid **attr_set);
 
@@ -49,7 +52,7 @@ 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;
@@ -72,8 +75,8 @@ static int query_token(struct yaz_pqf_parser *li)
         ++(*qptr);
     }
     li->lex_buf = *qptr;
-   
-    if (**qptr == li->escape_char && isdigit(((const unsigned char *) *qptr)[1]))
+
+    if (**qptr == li->escape_char && yaz_isdigit((*qptr)[1]))
     {
         ++(li->lex_len);
         ++(*qptr);
@@ -81,7 +84,7 @@ static int query_token(struct yaz_pqf_parser *li)
     }
     while (**qptr && **qptr != sep_char)
     {
-        if (**qptr == '\\')
+        if (**qptr == '\\' && (*qptr)[1])
         {
             ++(li->lex_len);
             ++(*qptr);
@@ -163,7 +166,7 @@ static int escape_string(char *out_buf, const char *in, int len)
                     char s[4];
                     int n = 0;
                     s[0] = *in;
-                    s[1] = *++in;                   
+                    s[1] = *++in;
                     s[2] = *++in;
                     s[3] = '\0';
                     len = len - 2;
@@ -209,7 +212,7 @@ static int p_query_parse_attr(struct yaz_pqf_parser *li, ODR o,
             return 0;
         }
     }
-    else 
+    else
     {
         if (num_attr > 0)
             attr_set[num_attr] = attr_set[num_attr-1];
@@ -240,98 +243,99 @@ static int p_query_parse_attr(struct yaz_pqf_parser *li, ODR o,
     return 1;
 }
 
-static Z_AttributesPlusTerm *rpn_term(struct yaz_pqf_parser *li, ODR o,
-                                      int num_attr, Odr_int *attr_list,
-                                      char **attr_clist, Odr_oid **attr_set)
+static Z_AttributeList *get_attributeList(ODR o,
+                                          int num_attr, Odr_int *attr_list,
+                                          char **attr_clist, Odr_oid **attr_set)
 {
-    Z_AttributesPlusTerm *zapt;
-    Odr_oct *term_octet;
-    Z_Term *term;
+    int i, k = 0;
+    Odr_int *attr_tmp;
     Z_AttributeElement **elements;
-
-    zapt = (Z_AttributesPlusTerm *)odr_malloc(o, sizeof(*zapt));
-    term_octet = (Odr_oct *)odr_malloc(o, sizeof(*term_octet));
-    term = (Z_Term *)odr_malloc(o, sizeof(*term));
-
+    Z_AttributeList *attributes= (Z_AttributeList *)
+        odr_malloc(o, sizeof(*attributes));
+    attributes->num_attributes = num_attr;
     if (!num_attr)
-        elements = (Z_AttributeElement**)odr_nullval();
-    else
     {
-        int i, k = 0;
-        Odr_int *attr_tmp;
-
-        elements = (Z_AttributeElement**)
-            odr_malloc (o, num_attr * sizeof(*elements));
+        attributes->attributes = (Z_AttributeElement**)odr_nullval();
+        return attributes;
+    }
+    elements = (Z_AttributeElement**)
+        odr_malloc(o, num_attr * sizeof(*elements));
 
-        attr_tmp = (Odr_int *)odr_malloc(o, num_attr * 2 * sizeof(*attr_tmp));
-        memcpy(attr_tmp, attr_list, num_attr * 2 * sizeof(*attr_tmp));
-        for (i = num_attr; --i >= 0; )
+    attr_tmp = (Odr_int *)odr_malloc(o, num_attr * 2 * sizeof(*attr_tmp));
+    memcpy(attr_tmp, attr_list, num_attr * 2 * sizeof(*attr_tmp));
+    for (i = num_attr; --i >= 0; )
+    {
+        int j;
+        for (j = i+1; j<num_attr; j++)
+            if (attr_tmp[2*j] == attr_tmp[2*i])
+                break;
+        if (j < num_attr)
+            continue;
+        elements[k] =
+            (Z_AttributeElement*)odr_malloc(o,sizeof(**elements));
+        elements[k]->attributeType = &attr_tmp[2*i];
+        elements[k]->attributeSet = attr_set[i];
+
+        if (attr_clist[i])
         {
-            int j;
-            for (j = i+1; j<num_attr; j++)
-                if (attr_tmp[2*j] == attr_tmp[2*i])
-                    break;
-            if (j < num_attr)
-                continue;
-            elements[k] =
-                (Z_AttributeElement*)odr_malloc(o,sizeof(**elements));
-            elements[k]->attributeType = &attr_tmp[2*i];
-            elements[k]->attributeSet = attr_set[i];
-
-            if (attr_clist[i])
-            {
-                elements[k]->which = Z_AttributeValue_complex;
-                elements[k]->value.complex = (Z_ComplexAttribute *)
-                    odr_malloc(o, sizeof(Z_ComplexAttribute));
-                elements[k]->value.complex->num_list = 1;
-                elements[k]->value.complex->list =
-                    (Z_StringOrNumeric **)
-                    odr_malloc(o, 1 * sizeof(Z_StringOrNumeric *));
-                elements[k]->value.complex->list[0] =
-                    (Z_StringOrNumeric *)
-                    odr_malloc(o, sizeof(Z_StringOrNumeric));
-                elements[k]->value.complex->list[0]->which =
-                    Z_StringOrNumeric_string;
-                elements[k]->value.complex->list[0]->u.string =
-                    attr_clist[i];
-                elements[k]->value.complex->semanticAction = 0;
-                elements[k]->value.complex->num_semanticAction = 0;
-            }
-            else
-            {
-                elements[k]->which = Z_AttributeValue_numeric;
-                elements[k]->value.numeric = &attr_tmp[2*i+1];
-            }
-            k++;
+            elements[k]->which = Z_AttributeValue_complex;
+            elements[k]->value.complex = (Z_ComplexAttribute *)
+                odr_malloc(o, sizeof(Z_ComplexAttribute));
+            elements[k]->value.complex->num_list = 1;
+            elements[k]->value.complex->list =
+                (Z_StringOrNumeric **)
+                odr_malloc(o, 1 * sizeof(Z_StringOrNumeric *));
+            elements[k]->value.complex->list[0] =
+                (Z_StringOrNumeric *)
+                odr_malloc(o, sizeof(Z_StringOrNumeric));
+            elements[k]->value.complex->list[0]->which =
+                Z_StringOrNumeric_string;
+            elements[k]->value.complex->list[0]->u.string =
+                attr_clist[i];
+            elements[k]->value.complex->semanticAction = 0;
+            elements[k]->value.complex->num_semanticAction = 0;
         }
-        num_attr = k;
+        else
+        {
+            elements[k]->which = Z_AttributeValue_numeric;
+            elements[k]->value.numeric = &attr_tmp[2*i+1];
+        }
+        k++;
     }
-    zapt->attributes = (Z_AttributeList *)
-        odr_malloc(o, sizeof(*zapt->attributes));
-    zapt->attributes->num_attributes = num_attr;
-    zapt->attributes->attributes = elements;
+    attributes->num_attributes = k;
+    attributes->attributes = elements;
+    return attributes;
+}
 
-    zapt->term = term;
+Z_AttributeList *zget_AttributeList_use_string(ODR o, const char *name)
+{
+    Odr_int attr_list[2];
+    char *attr_clist[1];
+    Odr_oid *attr_set[1];
+
+    attr_list[0] = 1;
+    attr_list[1] = 0; /* not used */
+    attr_clist[0] = odr_strdup(o, name);
+    attr_set[0] = 0;
+    return get_attributeList(o, 1, attr_list, attr_clist, attr_set);
+}
 
-    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);
-    term_octet->buf[term_octet->size] = 0;  /* null terminate */
-    
-    switch (li->term_type)
+Z_Term *z_Term_create(ODR o, int term_type, const char *buf, size_t len)
+{
+    Z_Term *term = (Z_Term *)odr_malloc(o, sizeof(*term));
+    switch (term_type)
     {
     case Z_Term_general:
         term->which = Z_Term_general;
-        term->u.general = term_octet;
+        term->u.general = odr_create_Odr_oct(o, buf, len);
         break;
     case Z_Term_characterString:
         term->which = Z_Term_characterString;
-        term->u.characterString = (char*) term_octet->buf; 
-        /* null terminated above */
+        term->u.characterString = odr_strdupn(o, buf, len);
         break;
     case Z_Term_numeric:
         term->which = Z_Term_numeric;
-        term->u.numeric = odr_intdup(o, odr_atoi((const char*) term_octet->buf));
+        term->u.numeric = odr_intdup(o, odr_atoi(odr_strdupn(o, buf, len)));
         break;
     case Z_Term_null:
         term->which = Z_Term_null;
@@ -346,9 +350,30 @@ static Z_AttributesPlusTerm *rpn_term(struct yaz_pqf_parser *li, ODR o,
         term->u.null = odr_nullval();
         break;
     }
+    return term;
+}
+
+static Z_AttributesPlusTerm *rpn_term_attributes(
+    struct yaz_pqf_parser *li, ODR o, Z_AttributeList *attributes)
+{
+    char *es_str = odr_malloc(o, li->lex_len+1);
+    int es_len = escape_string(es_str, li->lex_buf, li->lex_len);
+    Z_Term *term = z_Term_create(o, li->term_type, es_str, es_len);
+    Z_AttributesPlusTerm *zapt = (Z_AttributesPlusTerm *)
+        odr_malloc(o, sizeof(*zapt));
+
+    zapt->term = term;
+    zapt->attributes = attributes;
     return zapt;
 }
 
+static Z_AttributesPlusTerm *rpn_term(struct yaz_pqf_parser *li, ODR o,
+                                      int num_attr, Odr_int *attr_list,
+                                      char **attr_clist, Odr_oid **attr_set)
+{
+    return rpn_term_attributes(li, o, get_attributeList(o, num_attr, attr_list, attr_clist, attr_set));
+}
+
 static Z_Operand *rpn_simple(struct yaz_pqf_parser *li, ODR o,
                              int num_attr, Odr_int *attr_list,
                              char **attr_clist,
@@ -374,9 +399,7 @@ static Z_Operand *rpn_simple(struct yaz_pqf_parser *li, ODR o,
             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);
-        zo->u.resultSetId[li->lex_len] = '\0';
+        zo->u.resultSetId = odr_strdupn(o, li->lex_buf, li->lex_len);
         lex(li);
         break;
     default:
@@ -436,7 +459,7 @@ static Z_ProximityOperator *rpn_proximity(struct yaz_pqf_parser *li, ODR o)
         li->error = YAZ_PQF_ERROR_PROXIMITY;
         return NULL;
     }
-    
+
     if (!lex (li))
     {
         li->error = YAZ_PQF_ERROR_MISSING;
@@ -485,7 +508,7 @@ static Z_ProximityOperator *rpn_proximity(struct yaz_pqf_parser *li, ODR o)
 }
 
 static Z_Complex *rpn_complex(struct yaz_pqf_parser *li, ODR o,
-                              int num_attr, int max_attr, 
+                              int num_attr, int max_attr,
                               Odr_int *attr_list, char **attr_clist,
                               Odr_oid **attr_set)
 {
@@ -559,9 +582,9 @@ static void rpn_term_type(struct yaz_pqf_parser *li)
 #endif
     lex(li);
 }
-                           
+
 static Z_RPNStructure *rpn_structure(struct yaz_pqf_parser *li, ODR o,
-                                     int num_attr, int max_attr, 
+                                     int num_attr, int max_attr,
                                      Odr_int *attr_list,
                                      char **attr_clist,
                                      Odr_oid **attr_set)
@@ -667,6 +690,12 @@ static Z_RPNQuery *p_query_rpn_mk(ODR o, struct yaz_pqf_parser *li)
     return zq;
 }
 
+static void pqf_parser_begin(struct yaz_pqf_parser *li, const char *buf)
+{
+    li->query_buf = li->query_ptr = buf;
+    li->lex_buf = 0;
+}
+
 Z_RPNQuery *p_query_rpn(ODR o, const char *qbuf)
 {
     struct yaz_pqf_parser li;
@@ -676,13 +705,12 @@ Z_RPNQuery *p_query_rpn(ODR o, const char *qbuf)
     li.right_sep = "}\"";
     li.escape_char = '@';
     li.term_type = Z_Term_general;
-    li.query_buf = li.query_ptr = qbuf;
-    li.lex_buf = 0;
+
+    pqf_parser_begin(&li, qbuf);
     return p_query_rpn_mk(o, &li);
 }
 
-
-static Z_AttributesPlusTerm *p_query_scan_mk(struct yaz_pqf_parser *li,
+static Z_AttributeList *p_query_scan_attributes_mk(struct yaz_pqf_parser *li,
                                              ODR o,
                                              Odr_oid **attributeSetP)
 {
@@ -692,7 +720,6 @@ static Z_AttributesPlusTerm *p_query_scan_mk(struct yaz_pqf_parser *li,
     int num_attr = 0;
     int max_attr = 512;
     Odr_oid *top_set = 0;
-    Z_AttributesPlusTerm *apt;
 
     lex(li);
     if (li->query_look == 'r')
@@ -741,12 +768,22 @@ static Z_AttributesPlusTerm *p_query_scan_mk(struct yaz_pqf_parser *li,
         else
             break;
     }
+    return get_attributeList(o, num_attr, attr_list, attr_clist, attr_set);
+}
+
+static Z_AttributesPlusTerm *p_query_scan_mk(struct yaz_pqf_parser *li,
+                                             ODR o,
+                                             Odr_oid **attributeSetP)
+{
+    Z_AttributeList *attr_list = p_query_scan_attributes_mk(li, o, attributeSetP);
+    Z_AttributesPlusTerm *apt;
+
     if (!li->query_look)
     {
         li->error = YAZ_PQF_ERROR_MISSING;
         return 0;
     }
-    apt = rpn_term(li, o, num_attr, attr_list, attr_clist, attr_set);
+    apt = rpn_term_attributes(li, o, attr_list);
 
     lex(li);
 
@@ -780,8 +817,7 @@ 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;
+    pqf_parser_begin(p, qbuf);
     return p_query_rpn_mk(o, p);
 }
 
@@ -791,11 +827,83 @@ Z_AttributesPlusTerm *yaz_pqf_scan(YAZ_PQF_Parser p, ODR o,
 {
     if (!p)
         return 0;
-    p->query_buf = p->query_ptr = qbuf;
-    p->lex_buf = 0;
+    pqf_parser_begin(p, qbuf);
     return p_query_scan_mk(p, o, attributeSetP);
 }
 
+Z_AttributeList *yaz_pqf_scan_attribute_list(YAZ_PQF_Parser p, ODR o,
+                                             Odr_oid **attributeSetP,
+                                             const char *qbuf)
+{
+    if (!p)
+        return 0;
+    pqf_parser_begin(p, qbuf);
+    return p_query_scan_attributes_mk(p, o, attributeSetP);
+}
+
+static Z_FacetField* parse_facet(ODR odr, const char *facet)
+{
+    YAZ_PQF_Parser pqf_parser = yaz_pqf_create();
+    struct yaz_pqf_parser *li = pqf_parser;
+    Odr_oid *attributeSetId = 0;
+    Z_FacetField *facet_field = 0;
+    Z_AttributeList *attribute_list;
+
+    pqf_parser_begin(pqf_parser, facet);
+    attribute_list = p_query_scan_attributes_mk(li, odr, &attributeSetId);
+    if (attribute_list)
+    {
+        facet_field = (Z_FacetField *) odr_malloc(odr, sizeof(*facet_field));
+        facet_field->attributes = attribute_list;
+        facet_field->num_terms = 0;
+        facet_field->terms = odr_malloc(odr, 10 * sizeof(*facet_field->terms));
+        while (li->query_look == 't')
+        {
+            if (facet_field->num_terms < 10)
+            {
+                char *es_str = odr_malloc(odr, li->lex_len+1);
+                int es_len = escape_string(es_str, li->lex_buf, li->lex_len);
+                Z_Term *term = z_Term_create(odr, li->term_type, es_str, es_len);
+
+                facet_field->terms[facet_field->num_terms] =
+                    (Z_FacetTerm *) odr_malloc(odr, sizeof(Z_FacetTerm));
+                facet_field->terms[facet_field->num_terms]->term = term;
+                facet_field->terms[facet_field->num_terms]->count =
+                    odr_intdup(odr, 0);
+                facet_field->num_terms++;
+            }
+            lex(li);
+        }
+    }
+    yaz_pqf_destroy(pqf_parser);
+    return facet_field;
+}
+
+Z_FacetList *yaz_pqf_parse_facet_list(ODR o, const char *qbuf)
+{
+    char **darray;
+    int num;
+
+    nmem_strsplit(odr_getmem(o), ",", qbuf, &darray, &num);
+    if (num > 0)
+    {
+        int i;
+        Z_FacetList *fl = (Z_FacetList*) odr_malloc(o, sizeof(*fl));
+        fl->num = num;
+        fl->elements = (Z_FacetField **)
+            odr_malloc(o, num * sizeof(*fl->elements));
+        for (i = 0; i < num; i++)
+        {
+            fl->elements[i] = parse_facet(o, darray[i]);
+            if (!fl->elements[i])
+                return 0;
+        }
+        return fl;
+    }
+    else
+        return 0;
+}
+
 int yaz_pqf_error(YAZ_PQF_Parser p, const char **msg, size_t *off)
 {
     switch (p->error)