X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fcqltransform.c;h=35e421999bc7b5b92f327da319e8b8c45a71d7cd;hb=efecfe73da3c8c90fe749d95b4d68105ba320e72;hp=2c64c32829010162e95245b518d89d987d5b279d;hpb=8d691989077a0addcbd840d769dce6700f3d9622;p=yaz-moved-to-github.git

diff --git a/src/cqltransform.c b/src/cqltransform.c
index 2c64c32..35e4219 100644
--- a/src/cqltransform.c
+++ b/src/cqltransform.c
@@ -1,4 +1,4 @@
-/* $Id: cqltransform.c,v 1.26 2007-01-03 08:42:15 adam Exp $
+/* $Id: cqltransform.c,v 1.32 2008-01-06 19:34:34 adam Exp $
    Copyright (C) 1995-2007, Index Data ApS
    Index Data Aps
 
@@ -185,6 +185,18 @@ int cql_pr_attr_uri(cql_transform_t ct, const char *category,
     {
         if (!res)
             res = cql_lookup_property(ct, category, prefix, eval);
+        /* we have some aliases for some relations unfortunately.. */
+        if (!res && !prefix && !strcmp(category, "relation"))
+        {
+            if (!strcmp(val, "=="))
+                res = cql_lookup_property(ct, category, prefix, "exact");
+            if (!strcmp(val, "="))
+                res = cql_lookup_property(ct, category, prefix, "eq");
+            if (!strcmp(val, "<="))
+                res = cql_lookup_property(ct, category, prefix, "le");
+            if (!strcmp(val, ">="))
+                res = cql_lookup_property(ct, category, prefix, "ge");
+        }
         if (!res)
             res = cql_lookup_property(ct, category, prefix, "*");
     }
@@ -195,6 +207,7 @@ int cql_pr_attr_uri(cql_transform_t ct, const char *category,
         const char *cp0 = res, *cp1;
         while ((cp1 = strchr(cp0, '=')))
         {
+            int i;
             while (*cp1 && *cp1 != ' ')
                 cp1++;
             if (cp1 - cp0 >= sizeof(buf))
@@ -202,7 +215,19 @@ int cql_pr_attr_uri(cql_transform_t ct, const char *category,
             memcpy (buf, cp0, cp1 - cp0);
             buf[cp1-cp0] = 0;
             (*pr)("@attr ", client_data);
-            (*pr)(buf, client_data);
+
+            for (i = 0; buf[i]; i++)
+            {
+                if (buf[i] == '*')
+                    (*pr)(eval, client_data);
+                else
+                {
+                    char tmp[2];
+                    tmp[0] = buf[i];
+                    tmp[1] = '\0';
+                    (*pr)(tmp, client_data);
+                }
+            }
             (*pr)(" ", client_data);
             cp0 = cp1;
             while (*cp0 == ' ')
@@ -324,19 +349,30 @@ static int cql_pr_prox(cql_transform_t ct, struct cql_node *mods,
  * characters starting at `term', or a null pointer of there are
  * none -- like memchr().
  */
-static const char *wcchar(const char *term, int length)
+static const char *wcchar(int start, const char *term, int length)
 {
-    const char *best = 0;
-    const char *current;
-    char *whichp;
-
-    for (whichp = "*?"; *whichp != '\0'; whichp++) {
-        current = (const char *) memchr(term, *whichp, length);
-        if (current != 0 && (best == 0 || current < best))
-            best = current;
+    while (length > 0)
+    {
+        if (start || term[-1] != '\\')
+            if (strchr("*?", *term))
+                return term;
+        term++;
+        length--;
+        start = 0;
     }
+    return 0;
+}
+
 
-    return best;
+/* ### checks for CQL relation-name rather than Type-1 attribute */
+static int has_modifier(struct cql_node *cn, const char *name) {
+    struct cql_node *mod;
+    for (mod = cn->u.st.modifiers; mod != 0; mod = mod->u.st.modifiers) {
+        if (!strcmp(mod->u.st.index, name))
+            return 1;
+    }
+
+    return 0;
 }
 
 
@@ -348,10 +384,12 @@ void emit_term(cql_transform_t ct,
 {
     int i;
     const char *ns = cn->u.st.index_uri;
+    int process_term = !has_modifier(cn, "regexp");
+    char *z3958_mem = 0;
 
     assert(cn->which == CQL_NODE_ST);
 
-    if (length > 0)
+    if (process_term && length > 0)
     {
         if (length > 1 && term[0] == '^' && term[length-1] == '^')
         {
@@ -380,35 +418,40 @@ void emit_term(cql_transform_t ct,
         }
     }
 
-    if (length > 0)
+    if (process_term && length > 0)
     {
+        const char *first_wc = wcchar(1, term, length);
+        const char *second_wc = first_wc ?
+            wcchar(0, first_wc+1, length-(first_wc-term)-1) : 0;
+
         /* Check for well-known globbing patterns that represent
          * simple truncation attributes as expected by, for example,
          * Bath-compliant server.  If we find such a pattern but
          * there's no mapping for it, that's fine: we just use a
          * general pattern-matching attribute.
          */
-        if (length > 1 && term[0] == '*' && term[length-1] == '*' &&
-            wcchar(term+1, length-2) == 0 &&
-            cql_pr_attr(ct, "truncation", "both", 0,
-                        pr, client_data, 0)) {
+        if (first_wc == term && second_wc == term + length-1 
+            && *first_wc == '*' && *second_wc == '*' 
+            && cql_pr_attr(ct, "truncation", "both", 0, pr, client_data, 0)) 
+        {
             term++;
             length -= 2;
         }
-        else if (term[0] == '*' &&
-                 wcchar(term+1, length-1) == 0 &&
-                 cql_pr_attr(ct, "truncation", "left", 0,
-                             pr, client_data, 0)) {
+        else if (first_wc == term && second_wc == 0 && *first_wc == '*'
+                 && cql_pr_attr(ct, "truncation", "left", 0,
+                                pr, client_data, 0))
+        {
             term++;
             length--;
         }
-        else if (term[length-1] == '*' &&
-                 wcchar(term, length-1) == 0 &&
-                 cql_pr_attr(ct, "truncation", "right", 0,
-                             pr, client_data, 0)) {
+        else if (first_wc == term + length-1 && second_wc == 0
+                 && *first_wc == '*'
+                 && cql_pr_attr(ct, "truncation", "right", 0, 
+                                pr, client_data, 0))
+        {
             length--;
         }
-        else if (wcchar(term, length))
+        else if (first_wc)
         {
             /* We have one or more wildcard characters, but not in a
              * way that can be dealt with using only the standard
@@ -420,17 +463,22 @@ void emit_term(cql_transform_t ct,
              * supported".
              */
             int i;
-            char *mem;
             cql_pr_attr(ct, "truncation", "z3958", 0,
                         pr, client_data, 28);
-            mem = (char *) xmalloc(length+1);
-            for (i = 0; i < length; i++) {
-                if (term[i] == '*')      mem[i] = '?';
-                else if (term[i] == '?') mem[i] = '#';
-                else                     mem[i] = term[i];
+            z3958_mem = (char *) xmalloc(length+1);
+            for (i = 0; i < length; i++)
+            {
+                if (i > 0 && term[i-1] == '\\')
+                    z3958_mem[i] = term[i];
+                else if (term[i] == '*')
+                    z3958_mem[i] = '?';
+                else if (term[i] == '?')
+                    z3958_mem[i] = '#';
+                else
+                    z3958_mem[i] = term[i];
             }
-            mem[length] = '\0';
-            term = mem;
+            z3958_mem[length] = '\0';
+            term = z3958_mem;
         }
         else {
             /* No masking characters.  Use "truncation.none" if given. */
@@ -457,22 +505,44 @@ void emit_term(cql_transform_t ct,
     for (i = 0; i<length; i++)
     {
         /* pr(int) each character */
-        char buf[3];
-        const char *cp;
+        /* we do not need to deal with \-sequences because the
+           CQL and PQF terms have same \-format, bug #1988 */
+        char buf[2];
 
-        buf[1] = term[i];
-        buf[2] = 0;
-        /* do we have to escape this char? */
-        if (buf[1] == '"')
-        {
-            buf[0] = '\\';
-            cp = buf;
-        }
-        else
-            cp = buf+1;
-        (*pr)(cp, client_data);
+        buf[0] = term[i];
+        buf[1] = '\0';
+        (*pr)(buf, client_data);
     }
     (*pr)("\" ", client_data);
+    xfree(z3958_mem);
+}
+
+void emit_terms(cql_transform_t ct,
+                struct cql_node *cn,
+                void (*pr)(const char *buf, void *client_data),
+                void *client_data,
+                const char *op)
+{
+    struct cql_node *ne = cn->u.st.extra_terms;
+    if (ne)
+    {
+        (*pr)("@", client_data);
+        (*pr)(op, client_data);
+        (*pr)(" ", client_data);
+    }
+    emit_term(ct, cn, cn->u.st.term, strlen(cn->u.st.term),
+              pr, client_data);
+    for (; ne; ne = ne->u.st.extra_terms)
+    {
+        if (ne->u.st.extra_terms)
+        {
+            (*pr)("@", client_data);
+            (*pr)(op, client_data);
+            (*pr)(" ", client_data);
+        }            
+        emit_term(ct, cn, ne->u.st.term, strlen(ne->u.st.term),
+                  pr, client_data);
+    }
 }
 
 void emit_wordlist(cql_transform_t ct,
@@ -542,18 +612,7 @@ void cql_transform_r(cql_transform_t ct,
             }
         }
         cql_pr_attr(ct, "always", 0, 0, pr, client_data, 0);
-        if (cn->u.st.relation && !cql_strcmp(cn->u.st.relation, "="))
-            cql_pr_attr(ct, "relation", "eq", "scr",
-                        pr, client_data, 19);
-        else if (cn->u.st.relation && !cql_strcmp(cn->u.st.relation, "<="))
-            cql_pr_attr(ct, "relation", "le", "scr",
-                        pr, client_data, 19);
-        else if (cn->u.st.relation && !cql_strcmp(cn->u.st.relation, ">="))
-            cql_pr_attr(ct, "relation", "ge", "scr",
-                        pr, client_data, 19);
-        else
-            cql_pr_attr(ct, "relation", cn->u.st.relation, "eq",
-                        pr, client_data, 19);
+        cql_pr_attr(ct, "relation", cn->u.st.relation, 0, pr, client_data, 19);
         cql_pr_attr(ct, "structure", cn->u.st.relation, 0,
                     pr, client_data, 24);
         if (cn->u.st.relation && !cql_strcmp(cn->u.st.relation, "all"))
@@ -566,8 +625,7 @@ void cql_transform_r(cql_transform_t ct,
         }
         else
         {
-            emit_term(ct, cn, cn->u.st.term, strlen(cn->u.st.term),
-                      pr, client_data);
+            emit_terms(ct, cn, pr, client_data, "and");
         }
         break;
     case CQL_NODE_BOOL: