X-Git-Url: http://git.indexdata.com/?p=yaz-moved-to-github.git;a=blobdiff_plain;f=src%2Frpn2cql.c;h=25ee13894976f56aea84e3ecf7425122e8d8d7ca;hp=3e581a4fdd334a757e388288b863b7240e59093d;hb=266cfeafc17930bfa6fb7eed92f80b8c751f43b4;hpb=b3d5a30f4d074053085693a8fc5381a4ebc41544

diff --git a/src/rpn2cql.c b/src/rpn2cql.c
index 3e581a4..25ee138 100644
--- a/src/rpn2cql.c
+++ b/src/rpn2cql.c
@@ -1,22 +1,14 @@
 /* This file is part of the YAZ toolkit.
- * Copyright (C) 1995-2008 Index Data
+ * Copyright (C) 1995-2012 Index Data
  * See the file LICENSE for details.
  */
-
 /**
  * \file
  * \brief Implements RPN to CQL conversion
- *
- * Evaluation order of rules:
- *
- * always
- * relation
- * structure
- * position
- * truncation
- * index
- * relationModifier
  */
+#if HAVE_CONFIG_H
+#include <config.h>
+#endif
 
 #include <assert.h>
 #include <stdlib.h>
@@ -26,11 +18,167 @@
 #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)
+{
+    int j;
+    int server_choice = 1;
+    for (j = 0; j < attributes->num_attributes; j++)
+    {
+        Z_AttributeElement *ae = attributes->attributes[j];
+        if (*ae->attributeType == 1) /* use attribute */
+        {
+            if (ae->which == Z_AttributeValue_complex)
+            {
+                Z_ComplexAttribute *ca = ae->value.complex;
+                int i;
+                for (i = 0; i < ca->num_list; i++)
+                {
+                    Z_StringOrNumeric *son = ca->list[i];
+                    if (son->which == Z_StringOrNumeric_string)
+                        return son->u.string;
+                }
+            }
+            server_choice = 0; /* not serverChoice because we have use attr */
+        }
+    }
+    if (server_choice)
+        return "cql.serverChoice";
+    return 0;
+}
+
+static const char *lookup_relation_index_from_attr(Z_AttributeList *attributes)
+{
+    int j;
+    for (j = 0; j < attributes->num_attributes; j++)
+    {
+        Z_AttributeElement *ae = attributes->attributes[j];
+        if (*ae->attributeType == 2) /* relation attribute */
+        {
+            if (ae->which == Z_AttributeValue_numeric)
+            {
+                /* Only support for numeric relation */
+                Odr_int *relation = ae->value.numeric;
+                /* map this numeric to representation in CQL */
+                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 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?
+                */
+            }
+        }
+    }
+    return "=";
+}
+
+static int rpn2cql_attr(cql_transform_t ct,
+                        Z_AttributeList *attributes, WRBUF w)
+{
+    const char *relation = cql_lookup_reverse(ct, "relation.", attributes);
+    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 (!index)
+        index = lookup_index_from_string_attr(attributes);
+
+    /* Attempt to fix bug #2978: Look for a relation attribute */
+    if (!relation)
+        relation = lookup_relation_index_from_attr(attributes);
+
+    if (!index)
+    {
+        cql_transform_set_error(ct,
+                                YAZ_BIB1_UNSUPP_USE_ATTRIBUTE, 0);
+        return -1;
+    }
+    /* for serverChoice we omit index+relation+structure */
+    if (strcmp(index, "cql.serverChoice"))
+    {
+        wrbuf_puts(w, index);
+        if (relation)
+        {
+            if (!strcmp(relation, "exact"))
+                relation = "==";
+            else if (!strcmp(relation, "eq"))
+                relation = "=";
+            else if (!strcmp(relation, "le"))
+                relation = "<=";
+            else if (!strcmp(relation, "ge"))
+                relation = ">=";
+            /* Missing mapping of not equal, phonetic, stem and relevance */
+            wrbuf_puts(w, relation);
+        }
+        else
+            wrbuf_puts(w, "=");
+
+        if (structure)
+        {
+            if (strcmp(structure, "*"))
+            {
+                wrbuf_puts(w, "/");
+                wrbuf_puts(w, structure);
+                wrbuf_puts(w, " ");
+            }
+        }
+    }
+    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,
-                          Z_Operand *q)
+                          Z_Operand *q, WRBUF w)
 {
     int ret = 0;
     if (q->which != Z_Operand_APT)
@@ -42,47 +190,111 @@ static int rpn2cql_simple(cql_transform_t ct,
     {
         Z_AttributesPlusTerm *apt = q->u.attributesPlusTerm;
         Z_Term *term = apt->term;
-        Z_AttributeList *attributes = apt->attributes;
-        WRBUF w = wrbuf_alloc();
+        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:
-            wrbuf_write(w, (const char *) term->u.general->buf, term->u.general->len);
+            lterm = term->u.general->len;
+            sterm = (const char *) term->u.general->buf;
             break;
         case Z_Term_numeric:
-            wrbuf_printf(w, "%d", *term->u.numeric);
+            wrbuf_printf(w, ODR_INT_PRINTF, *term->u.numeric);
             break;
         case Z_Term_characterString:
-            wrbuf_puts(w, term->u.characterString);
+            sterm = term->u.characterString;
+            lterm = strlen(sterm);
             break;
         default:
-            ret = -1;
             cql_transform_set_error(ct, YAZ_BIB1_TERM_TYPE_UNSUPP, 0);
+            return -1;
+        }
+
+        if (trunc <= 3 || trunc == 100 || trunc == 102 || trunc == 104)
+        {
+            int quote_it = 0;
+            for (i = 0 ; i < lterm; i++)
+                if (strchr(" ()=></", sterm[i]))
+                {
+                    quote_it = 1;
+                    break;
+                }
+            if (lterm == 0)
+                quote_it = 1;
+            if (quote_it)
+                wrbuf_puts(w, "\"");
+            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);
-        wrbuf_destroy(w);
     }
     return ret;
 }
 
+
 static int rpn2cql_structure(cql_transform_t ct,
                              void (*pr)(const char *buf, void *client_data),
                              void *client_data,
-                             Z_RPNStructure *q, int nested)
+                             Z_RPNStructure *q, int nested,
+                             WRBUF w)
 {
     if (q->which == Z_RPNStructure_simple)
-        return rpn2cql_simple(ct, pr, client_data, q->u.simple);
+        return rpn2cql_simple(ct, pr, client_data, q->u.simple, w);
     else
     {
         Z_Operator *op = q->u.complex->roperator;
+        Z_ProximityOperator *prox;
         int r;
 
         if (nested)
             pr("(", client_data);
 
-        r = rpn2cql_structure(ct, pr, client_data, q->u.complex->s1, 1);
+        r = rpn2cql_structure(ct, pr, client_data, q->u.complex->s1, 1, w);
         if (r)
             return r;
         switch(op->which)
@@ -96,30 +308,73 @@ 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);
+        r = rpn2cql_structure(ct, pr, client_data, q->u.complex->s2, 1, w);
         if (nested)
             pr(")", client_data);
         return r;
     }
 }
 
-int cql_transform_rpn2cql(cql_transform_t ct,
-                          void (*pr)(const char *buf, void *client_data),
-                          void *client_data,
-                          Z_RPNQuery *q)
+int cql_transform_rpn2cql_stream(cql_transform_t ct,
+                                 void (*pr)(const char *buf, void *client_data),
+                                 void *client_data,
+                                 Z_RPNQuery *q)
 {
+    int r;
+    WRBUF w = wrbuf_alloc();
     cql_transform_set_error(ct, 0, 0);
-    return rpn2cql_structure(ct, pr, client_data, q->RPNStructure, 0);
+    r = rpn2cql_structure(ct, pr, client_data, q->RPNStructure, 0, w);
+    wrbuf_destroy(w);
+    return r;
 }
 
 
+int cql_transform_rpn2cql_wrbuf(cql_transform_t ct,
+                                WRBUF w,
+                                Z_RPNQuery *q)
+{
+    return cql_transform_rpn2cql_stream(ct, wrbuf_vputs, w, q);
+}
+
 /*
  * Local variables:
  * c-basic-offset: 4
+ * c-file-style: "Stroustrup"
  * indent-tabs-mode: nil
  * End:
  * vim: shiftwidth=4 tabstop=8 expandtab