rpn2cql: less quotes in generated terms
[yaz-moved-to-github.git] / src / rpn2cql.c
index 09e137b..25ee138 100644 (file)
@@ -1,13 +1,14 @@
 /* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2009 Index Data
+ * Copyright (C) 1995-2012 Index Data
  * See the file LICENSE for details.
  */
-
 /**
  * \file
  * \brief Implements RPN to CQL conversion
- *
  */
+#if HAVE_CONFIG_H
+#include <config.h>
+#endif
 
 #include <assert.h>
 #include <stdlib.h>
 #include <yaz/diagbib1.h>
 #include <yaz/z-core.h>
 #include <yaz/wrbuf.h>
+#include <yaz/logrpn.h> /* For yaz_prox_unit_name() */
+
+static void wrbuf_vputs(const char *buf, void *client_data)
+{
+    wrbuf_write((WRBUF) client_data, buf, strlen(buf));
+}
 
 static const char *lookup_index_from_string_attr(Z_AttributeList *attributes)
 {
@@ -49,7 +56,6 @@ static const char *lookup_index_from_string_attr(Z_AttributeList *attributes)
 static const char *lookup_relation_index_from_attr(Z_AttributeList *attributes)
 {
     int j;
-    int server_choice = 1;
     for (j = 0; j < attributes->num_attributes; j++)
     {
         Z_AttributeElement *ae = attributes->attributes[j];
@@ -60,50 +66,43 @@ static const char *lookup_relation_index_from_attr(Z_AttributeList *attributes)
                 /* Only support for numeric relation */
                 Odr_int *relation = ae->value.numeric;
                 /* map this numeric to representation in CQL */
-                switch (*relation) {
+                switch (*relation)
+                {
                     /* Unsure on whether this is the relation attribute constants? */
-                    case Z_ProximityOperator_Prox_lessThan: 
-                        return "<";
-                    case Z_ProximityOperator_Prox_lessThanOrEqual: 
-                        return "<="; 
-                    case Z_ProximityOperator_Prox_equal: 
-                        return "="; 
-                    case Z_ProximityOperator_Prox_greaterThanOrEqual: 
-                        return ">="; 
-                    case Z_ProximityOperator_Prox_greaterThan: 
-                        return ">"; 
-                    case Z_ProximityOperator_Prox_notEqual: 
-                        return "<>"; 
-                                                
-                    case 100: 
-                        /* phonetic is not supported in CQL */
-                        return 0; 
-
-                    case 101: 
-                        /* stem is not supported in CQL */
-                        return 0; 
-
-                    case 102: 
-                        /* relevance is not supported in CQL ?*/
-                        return 0; 
-                otherwise: 
-                        /* Invalid relation */
-                        return 0;
+                case Z_ProximityOperator_Prox_lessThan:
+                    return "<";
+                case Z_ProximityOperator_Prox_lessThanOrEqual:
+                    return "<=";
+                case Z_ProximityOperator_Prox_equal:
+                    return "=";
+                case Z_ProximityOperator_Prox_greaterThanOrEqual:
+                    return ">=";
+                case Z_ProximityOperator_Prox_greaterThan:
+                    return ">";
+                case Z_ProximityOperator_Prox_notEqual:
+                    return "<>";
+                case 100:
+                    /* phonetic is not supported in CQL */
+                    return 0;
+                case 101:
+                    /* stem is not supported in CQL */
+                    return 0;
+                case 102:
+                    /* relevance is supported in CQL, but not implemented yet */
+                    return 0;
+                default:
+                    /* Invalid relation */
+                    return 0;
                 }
-
             }
             else {
-                /* Can we have a complex relation value? 
-                   Should we implement something? 
-                 */
+                /*  Can we have a complex relation value?
+                    Should we implement something?
+                */
             }
-                
-            server_choice = 0; /* Which ServerChoice if relation? */
         }
     }
-    if (server_choice)
-        return "cql.serverChoice";
-    return 0;
+    return "=";
 }
 
 static int rpn2cql_attr(cql_transform_t ct,
@@ -113,13 +112,12 @@ static int rpn2cql_attr(cql_transform_t ct,
     const char *index = cql_lookup_reverse(ct, "index.", attributes);
     const char *structure = cql_lookup_reverse(ct, "structure.", attributes);
 
-    /* if transform (properties) do not match, we'll just use a USE
-       string attribute (bug #2978) */
+    /* if transform (properties) do not match, we'll just use a USE string attribute (bug #2978) */
     if (!index)
         index = lookup_index_from_string_attr(attributes);
 
     /* Attempt to fix bug #2978: Look for a relation attribute */
-    if (!relation) 
+    if (!relation)
         relation = lookup_relation_index_from_attr(attributes);
 
     if (!index)
@@ -161,6 +159,22 @@ static int rpn2cql_attr(cql_transform_t ct,
     return 0;
 }
 
+static Odr_int lookup_truncation(Z_AttributeList *attributes)
+{
+    int j;
+    for (j = 0; j < attributes->num_attributes; j++)
+    {
+        Z_AttributeElement *ae = attributes->attributes[j];
+        if (*ae->attributeType == 5) /* truncation attribute */
+        {
+            if (ae->which == Z_AttributeValue_numeric)
+                return *(ae->value.numeric);
+        }
+    }
+    /* No truncation specified */
+    return 0;
+};
+
 static int rpn2cql_simple(cql_transform_t ct,
                           void (*pr)(const char *buf, void *client_data),
                           void *client_data,
@@ -178,11 +192,13 @@ static int rpn2cql_simple(cql_transform_t ct,
         Z_Term *term = apt->term;
         const char *sterm = 0;
         size_t lterm = 0;
+        Odr_int trunc = lookup_truncation(apt->attributes);
+        size_t i;
 
         wrbuf_rewind(w);
         ret = rpn2cql_attr(ct, apt->attributes, w);
 
-        switch(term->which)
+        switch (term->which)
         {
         case Z_Term_general:
             lterm = term->u.general->len;
@@ -196,29 +212,71 @@ static int rpn2cql_simple(cql_transform_t ct,
             lterm = strlen(sterm);
             break;
         default:
-            ret = -1;
             cql_transform_set_error(ct, YAZ_BIB1_TERM_TYPE_UNSUPP, 0);
+            return -1;
         }
 
-        if (term)
+        if (trunc <= 3 || trunc == 100 || trunc == 102 || trunc == 104)
         {
-            size_t i;
-            int must_quote = 0;
+            int quote_it = 0;
             for (i = 0 ; i < lterm; i++)
-                if (sterm[i] == ' ')
-                    must_quote = 1;
-            if (must_quote)
+                if (strchr(" ()=></", sterm[i]))
+                {
+                    quote_it = 1;
+                    break;
+                }
+            if (lterm == 0)
+                quote_it = 1;
+            if (quote_it)
                 wrbuf_puts(w, "\"");
-            wrbuf_write(w, sterm, lterm);
-            if (must_quote)
+            if (trunc == 2 || trunc == 3)
+                wrbuf_puts(w, "*");
+            for (i = 0; i < lterm; i++)
+            {
+                if (sterm[i] == '\\' && i < lterm - 1)
+                {
+                    i++;
+                    if (strchr("*?\"\\", sterm[i]))
+                        wrbuf_putc(w, '\\');
+                    wrbuf_putc(w, sterm[i]);
+                }
+                else if (trunc == 102 && sterm[i] == '.' && sterm[i+1] == '*')
+                {
+                    wrbuf_putc(w, '*');
+                    i++;
+                }
+                else if (trunc == 102 && sterm[i] == '.')
+                    wrbuf_putc(w, '?');
+                else if (trunc == 104 && sterm[i] == '?')
+                    wrbuf_putc(w, '*');
+                else if (trunc == 104 && sterm[i] == '#')
+                    wrbuf_putc(w, '?');
+                else if (strchr("*?\"", sterm[i]))
+                {
+                    wrbuf_putc(w, '\\');
+                    wrbuf_putc(w, sterm[i]);
+                }
+                else
+                    wrbuf_putc(w, sterm[i]);
+            }
+            if (trunc == 1 || trunc == 3)
+                wrbuf_puts(w, "*");
+            if (quote_it)
                 wrbuf_puts(w, "\"");
         }
+        else
+        {
+            cql_transform_set_error(
+                ct, YAZ_BIB1_UNSUPP_TRUNCATION_ATTRIBUTE, 0);
+            ret = -1;
+        }
         if (ret == 0)
             pr(wrbuf_cstr(w), client_data);
     }
     return ret;
 }
 
+
 static int rpn2cql_structure(cql_transform_t ct,
                              void (*pr)(const char *buf, void *client_data),
                              void *client_data,
@@ -230,6 +288,7 @@ static int rpn2cql_structure(cql_transform_t ct,
     else
     {
         Z_Operator *op = q->u.complex->roperator;
+        Z_ProximityOperator *prox;
         int r;
 
         if (nested)
@@ -249,9 +308,40 @@ static int rpn2cql_structure(cql_transform_t ct,
         case  Z_Operator_and_not:
             pr(" not ", client_data);
             break;
-        case  Z_Operator_prox:
-            cql_transform_set_error(ct, YAZ_BIB1_UNSUPP_SEARCH, 0);
-            return -1;
+        case  Z_Operator_prox: {
+            pr(" prox", client_data);
+            prox = op->u.prox;
+            /* No way to express Odr_bool *exclusion -- ignore it */
+            if (prox->distance) {
+                char buf[21]; /* Enough for any 64-bit int */
+                char *op2name[6] = { "<", "<=", "=", ">=", ">","<>" };
+                pr("/distance", client_data);
+                if (!prox->relationType ||
+                    *prox->relationType < Z_ProximityOperator_Prox_lessThan ||
+                    *prox->relationType > Z_ProximityOperator_Prox_notEqual) {
+                    cql_transform_set_error(ct, YAZ_BIB1_UNSUPP_SEARCH,
+                        "unrecognised proximity relationType");
+                    return -1;
+                }
+                pr(op2name[*prox->relationType-1], client_data);
+                sprintf(buf, "%ld", (long) *prox->distance);
+                pr(buf, client_data);
+            }
+            if (prox->ordered) {
+                if (*prox->ordered) {
+                    pr("/ordered", client_data);
+                } else {
+                    pr("/unordered", client_data);
+                }
+            }
+            if (prox->which != Z_ProximityOperator_known ||
+                *prox->u.known != Z_ProxUnit_word) {
+                    pr("/unit=", client_data);
+                    pr(yaz_prox_unit_name(prox), client_data);
+            }
+            pr(" ", client_data);
+            break;
+        }
         }
         r = rpn2cql_structure(ct, pr, client_data, q->u.complex->s2, 1, w);
         if (nested)