X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=src%2Fcqltransform.c;h=3f54faa53d3de4f96a32338cf6c17bb3e1f80693;hb=ecd6d03e10dfbe4b6bb5b1b599c23bdb7ff305ca;hp=8494046ccec76a1a16802648a8460a68c23af389;hpb=e9db3cf52b6df66badd97391a51e85f9c78a8c91;p=yaz-moved-to-github.git

diff --git a/src/cqltransform.c b/src/cqltransform.c
index 8494046..3f54faa 100644
--- a/src/cqltransform.c
+++ b/src/cqltransform.c
@@ -1,5 +1,5 @@
-/* $Id: cqltransform.c,v 1.22 2006-04-05 12:04:51 mike Exp $
-   Copyright (C) 1995-2005, Index Data ApS
+/* $Id: cqltransform.c,v 1.29 2007-10-31 21:58:07 adam Exp $
+   Copyright (C) 1995-2007, Index Data ApS
    Index Data Aps
 
 This file is part of the YAZ toolkit.
@@ -10,8 +10,19 @@ See the file LICENSE.
 /**
  * \file cqltransform.c
  * \brief Implements CQL transform (CQL to RPN conversion).
+ *
+ * Evaluation order of rules:
+ *
+ * always
+ * relation
+ * structure
+ * position
+ * truncation
+ * index
+ * relationModifier
  */
 
+#include <assert.h>
 #include <stdlib.h>
 #include <string.h>
 #include <yaz/cql.h>
@@ -42,36 +53,48 @@ cql_transform_t cql_transform_open_FILE(FILE *f)
     {
         const char *cp_value_start;
         const char *cp_value_end;
+        const char *cp_pattern_start;
         const char *cp_pattern_end;
         const char *cp = line;
-        while (*cp && !strchr(" \t=\r\n#", *cp))
+
+        while (*cp && strchr(" \t", *cp))
+            cp++;
+        cp_pattern_start = cp;
+        
+        while (*cp && !strchr(" \t\r\n=#", *cp))
             cp++;
         cp_pattern_end = cp;
-        if (cp == line)
+        if (cp == cp_pattern_start)
             continue;
-        while (*cp && strchr(" \t\r\n", *cp))
+        while (*cp && strchr(" \t", *cp))
             cp++;
         if (*cp != '=')
-            continue;
+        {
+            *pp = 0;
+            cql_transform_close(ct);
+            return 0;
+        }
         cp++;
         while (*cp && strchr(" \t\r\n", *cp))
             cp++;
         cp_value_start = cp;
-        if (!(cp_value_end = strchr(cp, '#')))
+        cp_value_end = strchr(cp, '#');
+        if (!cp_value_end)
             cp_value_end = strlen(line) + line;
 
         if (cp_value_end != cp_value_start &&
             strchr(" \t\r\n", cp_value_end[-1]))
             cp_value_end--;
         *pp = (struct cql_prop_entry *) xmalloc (sizeof(**pp));
-        (*pp)->pattern = (char *) xmalloc (cp_pattern_end - line + 1);
-        memcpy ((*pp)->pattern, line, cp_pattern_end - line);
-        (*pp)->pattern[cp_pattern_end-line] = 0;
+        (*pp)->pattern = (char *) xmalloc(cp_pattern_end-cp_pattern_start + 1);
+        memcpy ((*pp)->pattern, cp_pattern_start,
+                cp_pattern_end-cp_pattern_start);
+        (*pp)->pattern[cp_pattern_end-cp_pattern_start] = '\0';
 
-        (*pp)->value = (char *) xmalloc (cp_value_end - cp_value_start + 1);
+        (*pp)->value = (char *) xmalloc (cp_value_end-cp_value_start + 1);
         if (cp_value_start != cp_value_end)
             memcpy ((*pp)->value, cp_value_start, cp_value_end-cp_value_start);
-        (*pp)->value[cp_value_end - cp_value_start] = 0;
+        (*pp)->value[cp_value_end - cp_value_start] = '\0';
         pp = &(*pp)->next;
     }
     *pp = 0;
@@ -172,6 +195,7 @@ 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))
@@ -179,7 +203,19 @@ 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 == ' ')
@@ -301,29 +337,47 @@ 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;
+}
+
+
+/* ### 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 best;
+    return 0;
 }
 
 
 void emit_term(cql_transform_t ct,
+               struct cql_node *cn,
                const char *term, int length,
                void (*pr)(const char *buf, void *client_data),
                void *client_data)
 {
     int i;
-    if (length > 0)
+    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 (process_term && length > 0)
     {
         if (length > 1 && term[0] == '^' && term[length-1] == '^')
         {
@@ -352,35 +406,40 @@ 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
@@ -392,17 +451,22 @@ 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. */
@@ -410,16 +474,42 @@ void emit_term(cql_transform_t ct,
                         pr, client_data, 0);
         }
     }
+    if (ns) {
+        cql_pr_attr_uri(ct, "index", ns,
+                        cn->u.st.index, "serverChoice",
+                        pr, client_data, 16);
+    }
+    if (cn->u.st.modifiers)
+    {
+        struct cql_node *mod = cn->u.st.modifiers;
+        for (; mod; mod = mod->u.st.modifiers)
+        {
+            cql_pr_attr(ct, "relationModifier", mod->u.st.index, 0,
+                        pr, client_data, 20);
+        }
+    }
 
     (*pr)("\"", client_data);
     for (i = 0; i<length; i++)
     {
-        char buf[2];
-        buf[0] = term[i];
-        buf[1] = 0;
-        (*pr)(buf, client_data);
+        /* pr(int) each character */
+        char buf[3];
+        const char *cp;
+
+        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);
     }
     (*pr)("\" ", client_data);
+    xfree(z3958_mem);
 }
 
 void emit_wordlist(cql_transform_t ct,
@@ -442,7 +532,7 @@ void emit_wordlist(cql_transform_t ct,
             (*pr)("@", client_data);
             (*pr)(op, client_data);
             (*pr)(" ", client_data);
-            emit_term(ct, last_term, last_length, pr, client_data);
+            emit_term(ct, cn, last_term, last_length, pr, client_data);
         }
         last_term = cp0;
         if (cp1)
@@ -452,7 +542,7 @@ void emit_wordlist(cql_transform_t ct,
         cp0 = cp1;
     }
     if (last_term)
-        emit_term(ct, last_term, last_length, pr, client_data);
+        emit_term(ct, cn, last_term, last_length, pr, client_data);
 }
 
 void cql_transform_r(cql_transform_t ct,
@@ -501,22 +591,8 @@ void cql_transform_r(cql_transform_t ct,
         else
             cql_pr_attr(ct, "relation", cn->u.st.relation, "eq",
                         pr, client_data, 19);
-        if (cn->u.st.modifiers)
-        {
-            struct cql_node *mod = cn->u.st.modifiers;
-            for (; mod; mod = mod->u.st.modifiers)
-            {
-                cql_pr_attr(ct, "relationModifier", mod->u.st.index, 0,
-                            pr, client_data, 20);
-            }
-        }
         cql_pr_attr(ct, "structure", cn->u.st.relation, 0,
                     pr, client_data, 24);
-        if (ns) {
-            cql_pr_attr_uri(ct, "index", ns,
-                            cn->u.st.index, "serverChoice",
-                            pr, client_data, 16);
-        }
         if (cn->u.st.relation && !cql_strcmp(cn->u.st.relation, "all"))
         {
             emit_wordlist(ct, cn, pr, client_data, "and");
@@ -527,7 +603,7 @@ void cql_transform_r(cql_transform_t ct,
         }
         else
         {
-            emit_term(ct, cn->u.st.term, strlen(cn->u.st.term),
+            emit_term(ct, cn, cn->u.st.term, strlen(cn->u.st.term),
                       pr, client_data);
         }
         break;