From e4baade57ba02d625ecd6452b0f4383b24c25f2d Mon Sep 17 00:00:00 2001
From: Adam Dickmeiss <adam@indexdata.dk>
Date: Wed, 29 Sep 2004 20:37:50 +0000
Subject: [PATCH] Added CCL facility r=r "range" which is similar to r=o
 "ordered" but does not require white-space before and after
 the dash, e.g. x=-1990 is equivalent to x <= 1990 iff r=r,
 but equivalent to x= -1980 iff r=o. Added CCL tests for
 ranges. Fixed memory leak in CCL parser that occurred when
 proximity was used.

---
 NEWS               |    5 +
 client/default.bib |    8 +-
 include/yaz/ccl.h  |    6 +-
 src/cclfind.c      |  280 +++++++++++++++++++++++++++++++++-------------------
 src/cclqfile.c     |    4 +-
 src/ccltoken.c     |   18 +++-
 test/tstccl.c      |   51 ++++++++--
 7 files changed, 253 insertions(+), 119 deletions(-)

diff --git a/NEWS b/NEWS
index e7740e2..9b9c19d 100644
--- a/NEWS
+++ b/NEWS
@@ -1,5 +1,10 @@
 Possible compatibility problems with earlier versions marked with '*'.
 
+Added CCL facility r=r "range" which is similar to r=o "ordered" but
+does not require white-space before and after the dash in a range, e.g.
+x=-1990 is equivalent to x <= 1990 iff r=r, but equivalent to x= -1980
+iff r=o.
+
 Fixed a few AC_TRY_LINK configure tests that did not operate properly
 due to new GCC removing "redundant" code.
 
diff --git a/client/default.bib b/client/default.bib
index 4d2046f..6fcac76 100644
--- a/client/default.bib
+++ b/client/default.bib
@@ -1,5 +1,5 @@
 # CCL field mappings
-# $Id: default.bib,v 1.8 2001-10-28 23:10:03 adam Exp $
+# $Id: default.bib,v 1.9 2004-09-29 20:37:50 adam Exp $
 # 
 # The rule below is used when no fields are specified
 term t=l,r  s=al
@@ -14,9 +14,9 @@ isbn u=7
 issn u=8
 cc   u=20
 su   u=21   s=pw
-date u=30   r=o
-dp   u=31   r=o
-da   u=32   r=o
+date u=30   r=r
+dp   u=31   r=r
+da   u=32   r=r
 la   u=54   s=pw
 ab   u=62   s=pw
 note u=63   s=pw
diff --git a/include/yaz/ccl.h b/include/yaz/ccl.h
index a52829b..7724ace 100644
--- a/include/yaz/ccl.h
+++ b/include/yaz/ccl.h
@@ -45,7 +45,7 @@
 /*
  * CCL - header file
  *
- * $Id: ccl.h,v 1.17 2004-09-22 12:17:24 adam Exp $
+ * $Id: ccl.h,v 1.18 2004-09-29 20:37:50 adam Exp $
  *
  * Old Europagate Log:
  *
@@ -182,6 +182,7 @@ typedef struct ccl_qualifiers *CCL_bibset;
 #define CCL_BIB1_STR_AND_LIST (-2)
 #define CCL_BIB1_STR_OR_LIST (-3)
 #define CCL_BIB1_REL_ORDER (-1)
+#define CCL_BIB1_REL_PORDER (-2)
 
 #define CCL_BIB1_TRU_CAN_LEFT (-1)
 #define CCL_BIB1_TRU_CAN_RIGHT (-2)
@@ -251,6 +252,9 @@ YAZ_EXPORT struct ccl_token *ccl_token_simple (const char *command);
 /* Delete token list */
 YAZ_EXPORT void ccl_token_del (struct ccl_token *list);
 
+/* Add single token after node at */
+YAZ_EXPORT struct ccl_token *ccl_token_add (struct ccl_token *at);
+
 /* Parse CCL Find command - NULL-terminated string */
 YAZ_EXPORT struct ccl_rpn_node *ccl_find_str (CCL_bibset bibset,
                                    const char *str, int *error, int *pos);
diff --git a/src/cclfind.c b/src/cclfind.c
index 82dd726..0883235 100644
--- a/src/cclfind.c
+++ b/src/cclfind.c
@@ -44,7 +44,7 @@
 /* CCL find (to rpn conversion)
  * Europagate, 1995
  *
- * $Id: cclfind.c,v 1.3 2004-08-11 20:13:36 adam Exp $
+ * $Id: cclfind.c,v 1.4 2004-09-29 20:37:50 adam Exp $
  *
  * Old Europagate log:
  *
@@ -217,6 +217,7 @@ void ccl_rpn_delete (struct ccl_rpn_node *rpn)
     case CCL_RPN_PROX:
         ccl_rpn_delete (rpn->u.p[0]);
         ccl_rpn_delete (rpn->u.p[1]);
+        ccl_rpn_delete (rpn->u.p[2]);
         break;
     }
     xfree (rpn);
@@ -535,43 +536,13 @@ static struct ccl_rpn_node *search_term (CCL_parser cclp,
     return search_term_x(cclp, qa, list, 0);
 }
 
-static struct ccl_rpn_node *qualifiers2 (CCL_parser cclp,
-                                         struct ccl_rpn_attr **ap)
+static
+struct ccl_rpn_node *qualifiers_order (CCL_parser cclp,
+				       struct ccl_rpn_attr **ap, char *attset)
 {
-    char *attset;
-    int rel;
+    int rel = 0;
+    struct ccl_rpn_node *p;
 
-    if (!qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER, &attset))
-    {                
-        /* unordered relation */
-        struct ccl_rpn_node *p;
-        if (KIND != CCL_TOK_EQ)
-        {
-            cclp->error_code = CCL_ERR_EQ_EXPECTED;
-            return NULL;
-        }
-        ADVANCE;
-        if (KIND == CCL_TOK_LP)
-        {
-            ADVANCE;
-            if (!(p = find_spec (cclp, ap)))
-            {
-                return NULL;
-            }
-            if (KIND != CCL_TOK_RP)
-            {
-                cclp->error_code = CCL_ERR_RP_EXPECTED;
-                ccl_rpn_delete (p);
-                return NULL;
-            }
-            ADVANCE;
-        }
-        else
-            p = search_terms (cclp, ap);
-        return p;
-    }
-    /* ordered relation ... */
-    rel = 0;
     if (cclp->look_token->len == 1)
     {
         if (cclp->look_token->name[0] == '<')
@@ -591,77 +562,184 @@ static struct ccl_rpn_node *qualifiers2 (CCL_parser cclp,
             rel = 6;
     }
     if (!rel)
+    {
         cclp->error_code = CCL_ERR_BAD_RELATION;
-    else
+	return NULL;
+    }
+    ADVANCE;  /* skip relation */
+    if (rel == 3 &&
+	qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_PORDER, 0))
     {
-        struct ccl_rpn_node *p;
+	/* allow - inside term and treat it as range _always_ */
+	/* relation is =. Extract "embedded" - to separate terms */
+	if (KIND == CCL_TOK_TERM)
+	{
+	    int i;
+	    for (i = 0; i<cclp->look_token->len; i++)
+	    {
+		if (cclp->look_token->name[i] == '-')
+		    break;
+	    }
+	    
+	    if (cclp->look_token->len > 1 && i == 0)
+	    {   /*  -xx*/
+		struct ccl_token *ntoken = ccl_token_add (cclp->look_token);
+
+		ntoken->kind = CCL_TOK_TERM;
+		ntoken->name = cclp->look_token->name + 1;
+		ntoken->len = cclp->look_token->len - 1;
+
+		cclp->look_token->len = 1;
+		cclp->look_token->name = "-";
+	    }
+	    else if (cclp->look_token->len > 1 && i == cclp->look_token->len-1)
+	    {	/* xx- */
+		struct ccl_token *ntoken = ccl_token_add (cclp->look_token);
+
+		ntoken->kind = CCL_TOK_TERM;
+		ntoken->name = "-";
+		ntoken->len = 1;
+
+		(cclp->look_token->len)--;
+	    }
+	    else if (cclp->look_token->len > 2 && i < cclp->look_token->len)
+	    {	/* xx-yy */
+		struct ccl_token *ntoken1 = ccl_token_add (cclp->look_token);
+		struct ccl_token *ntoken2 = ccl_token_add (ntoken1);
+
+		ntoken1->kind = CCL_TOK_TERM;  /* generate - */
+		ntoken1->name = "-";
+		ntoken1->len = 1;
+
+		ntoken2->kind = CCL_TOK_TERM;  /* generate yy */
+		ntoken2->name = cclp->look_token->name + (i+1);
+		ntoken2->len = cclp->look_token->len - (i+1);
+
+		cclp->look_token->len = i;     /* adjust xx */
+	    }
+	    else if (i == cclp->look_token->len &&
+		     cclp->look_token->next &&
+		     cclp->look_token->next->kind == CCL_TOK_TERM &&
+		     cclp->look_token->next->len > 1 &&
+		     cclp->look_token->next->name[0] == '-')
+		     
+	    {   /* xx -yy */
+		/* we _know_ that xx does not have - in it */
+		struct ccl_token *ntoken = ccl_token_add (cclp->look_token);
+
+		ntoken->kind = CCL_TOK_TERM;    /* generate - */
+		ntoken->name = "-";
+		ntoken->len = 1;
+
+		(ntoken->next->name)++;        /* adjust yy */
+		(ntoken->next->len)--; 
+	    }
+	}
+    }
         
-        ADVANCE;                      /* skip relation */
-        if (KIND == CCL_TOK_TERM &&
-            cclp->look_token->next && cclp->look_token->next->len == 1 &&
-            cclp->look_token->next->name[0] == '-')
-        {
-            struct ccl_rpn_node *p1;
-            if (!(p1 = search_term (cclp, ap)))
-                return NULL;
-            ADVANCE;                   /* skip '-' */
-            if (KIND == CCL_TOK_TERM)  /* = term - term  ? */
-            {
-                struct ccl_rpn_node *p2;
-                
-                if (!(p2 = search_term (cclp, ap)))
-                {
-                    ccl_rpn_delete (p1);
-                    return NULL;
-                }
-                p = mk_node (CCL_RPN_AND);
-                p->u.p[0] = p1;
-                add_attr_numeric (p1, attset, CCL_BIB1_REL, 4);
-                p->u.p[1] = p2;
-                add_attr_numeric (p2, attset, CCL_BIB1_REL, 2);
-                return p;
-            }
-            else                       /* = term -    */
-            {
-                add_attr_numeric (p1, attset, CCL_BIB1_REL, 4);
-                return p1;
-            }
-        }
-        else if (cclp->look_token->len == 1 &&
-                 cclp->look_token->name[0] == '-')   /* = - term  ? */
-        {
-            ADVANCE;
-            if (!(p = search_term (cclp, ap)))
-                return NULL;
-            add_attr_numeric (p, attset, CCL_BIB1_REL, 2);
-            return p;
-        }
-        else if (KIND == CCL_TOK_LP)
-        {
-            ADVANCE;
-            if (!(p = find_spec (cclp, ap)))
-                return NULL;
-            if (KIND != CCL_TOK_RP)
-            {
-                cclp->error_code = CCL_ERR_RP_EXPECTED;
-                ccl_rpn_delete (p);
-                return NULL;
-            }
-            ADVANCE;
-            return p;
-        }
-        else
-        {
-            if (!(p = search_terms (cclp, ap)))
-                return NULL;
-            add_attr_numeric (p, attset, CCL_BIB1_REL, rel);
-            return p;
-        }
-        cclp->error_code = CCL_ERR_TERM_EXPECTED;
+    if (rel == 3 &&
+	KIND == CCL_TOK_TERM &&
+	cclp->look_token->next && cclp->look_token->next->len == 1 &&
+	cclp->look_token->next->name[0] == '-')
+    {
+	struct ccl_rpn_node *p1;
+	if (!(p1 = search_term (cclp, ap)))
+	    return NULL;
+	ADVANCE;                   /* skip '-' */
+	if (KIND == CCL_TOK_TERM)  /* = term - term  ? */
+	{
+	    struct ccl_rpn_node *p2;
+	    
+	    if (!(p2 = search_term (cclp, ap)))
+	    {
+		ccl_rpn_delete (p1);
+		return NULL;
+	    }
+	    p = mk_node (CCL_RPN_AND);
+	    p->u.p[0] = p1;
+	    add_attr_numeric (p1, attset, CCL_BIB1_REL, 4);
+	    p->u.p[1] = p2;
+	    add_attr_numeric (p2, attset, CCL_BIB1_REL, 2);
+	    return p;
+	}
+	else                       /* = term -    */
+	{
+	    add_attr_numeric (p1, attset, CCL_BIB1_REL, 4);
+	    return p1;
+	}
+    }
+    else if (rel == 3 &&
+	     cclp->look_token->len == 1 &&
+	     cclp->look_token->name[0] == '-')   /* = - term  ? */
+    {
+	ADVANCE;
+	if (!(p = search_term (cclp, ap)))
+	    return NULL;
+	add_attr_numeric (p, attset, CCL_BIB1_REL, 2);
+	return p;
+    }
+    else if (KIND == CCL_TOK_LP)
+    {
+	ADVANCE;
+	if (!(p = find_spec (cclp, ap)))
+	    return NULL;
+	if (KIND != CCL_TOK_RP)
+	{
+	    cclp->error_code = CCL_ERR_RP_EXPECTED;
+	    ccl_rpn_delete (p);
+	    return NULL;
+	}
+	ADVANCE;
+	return p;
+    }
+    else
+    {
+	if (!(p = search_terms (cclp, ap)))
+	    return NULL;
+	add_attr_numeric (p, attset, CCL_BIB1_REL, rel);
+	return p;
     }
+    cclp->error_code = CCL_ERR_TERM_EXPECTED;
     return NULL;
 }
 
+static
+struct ccl_rpn_node *qualifiers2 (CCL_parser cclp, struct ccl_rpn_attr **ap)
+{
+    char *attset;
+    struct ccl_rpn_node *p;
+    
+    if (qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_ORDER, &attset)
+	|| qual_val_type(ap, CCL_BIB1_REL, CCL_BIB1_REL_PORDER, &attset))
+	return qualifiers_order(cclp, ap, attset);
+
+    /* unordered relation */
+    if (KIND != CCL_TOK_EQ)
+    {
+	cclp->error_code = CCL_ERR_EQ_EXPECTED;
+	return NULL;
+    }
+    ADVANCE;
+    if (KIND == CCL_TOK_LP)
+    {
+	ADVANCE;
+	if (!(p = find_spec (cclp, ap)))
+	{
+	    return NULL;
+	}
+	if (KIND != CCL_TOK_RP)
+	{
+	    cclp->error_code = CCL_ERR_RP_EXPECTED;
+	    ccl_rpn_delete (p);
+	    return NULL;
+	}
+	ADVANCE;
+    }
+    else
+	p = search_terms (cclp, ap);
+    return p;
+}
+
 /*
  * qualifiers1: Parse CCL qualifiers and search terms. 
  * cclp:   CCL Parser
@@ -1038,8 +1116,6 @@ struct ccl_rpn_node *ccl_parser_find (CCL_parser cclp, struct ccl_token *list)
 {
     struct ccl_rpn_node *p;
 
-    
-
     cclp->look_token = list;
     p = find_spec (cclp, NULL);
     if (p && KIND != CCL_TOK_EOL)
diff --git a/src/cclqfile.c b/src/cclqfile.c
index d46b730..0a618cb 100644
--- a/src/cclqfile.c
+++ b/src/cclqfile.c
@@ -44,7 +44,7 @@
 /* CCL qualifiers
  * Europagate, 1995
  *
- * $Id: cclqfile.c,v 1.3 2004-09-22 11:21:51 adam Exp $
+ * $Id: cclqfile.c,v 1.4 2004-09-29 20:37:50 adam Exp $
  *
  * Old Europagate Log:
  *
@@ -132,6 +132,8 @@ void ccl_qual_field (CCL_bibset bibset, const char *cp, const char *qual_name)
                 type = CCL_BIB1_REL;
                 if (!ccl_stricmp (qual_value, "o"))
                     value = CCL_BIB1_REL_ORDER;
+                else if (!ccl_stricmp (qual_value, "r"))
+                    value = CCL_BIB1_REL_PORDER;
                 break;                
             case 'p':
             case 'P':
diff --git a/src/ccltoken.c b/src/ccltoken.c
index ae8b4c0..06ff2a7 100644
--- a/src/ccltoken.c
+++ b/src/ccltoken.c
@@ -44,7 +44,7 @@
 /* CCL - lexical analysis
  * Europagate, 1995
  *
- * $Id: ccltoken.c,v 1.2 2004-08-11 11:44:30 adam Exp $
+ * $Id: ccltoken.c,v 1.3 2004-09-29 20:37:50 adam Exp $
  *
  * Old Europagate Log:
  *
@@ -317,6 +317,22 @@ struct ccl_token *ccl_parser_tokenize (CCL_parser cclp, const char *command)
     return first;
 }
 
+struct ccl_token *ccl_token_add (struct ccl_token *at)
+{
+    struct ccl_token *n = (struct ccl_token *)xmalloc (sizeof(*n));
+    ccl_assert(n);
+    n->next = at->next;
+    n->prev = at;
+    at->next = n;
+    if (n->next)
+	n->next->prev = n;
+
+    n->kind = CCL_TOK_TERM;
+    n->name = 0;
+    n->len = 0;
+    return n;
+}
+    
 struct ccl_token *ccl_tokenize (const char *command)
 {
     CCL_parser cclp = ccl_parser_create ();
diff --git a/test/tstccl.c b/test/tstccl.c
index 9f8bb1b..37cf0c9 100644
--- a/test/tstccl.c
+++ b/test/tstccl.c
@@ -1,12 +1,13 @@
 /*
- * Copyright (c) 2002-2003, Index Data
+ * Copyright (c) 2002-2004, Index Data
  * See the file LICENSE for details.
  *
- * $Id: tstccl.c,v 1.3 2004-09-22 11:21:51 adam Exp $
+ * $Id: tstccl.c,v 1.4 2004-09-29 20:37:51 adam Exp $
  */
 
 /* CCL test */
 
+#include <string.h>
 #include <yaz/ccl.h>
 
 struct ccl_tst {
@@ -23,10 +24,22 @@ static struct ccl_tst query_str[] = {
     { "x1 and", 0},
     { "tix=x5", 0},
     { "spid%æserne", "@prox 0 1 0 2 k 2 @attr 4=2 @attr 1=1016 spid @attr 4=2 @attr 1=1016 æserne "},
+    { "date=1980", "@attr 2=3 1980 "},
+    { "date=234-1990", "@and @attr 2=4 234 @attr 2=2 1990 "},
+    { "date=234- 1990", "@and @attr 2=4 234 @attr 2=2 1990 "},
+    { "date=234 -1990", "@and @attr 2=4 234 @attr 2=2 1990 "},
+    { "date=234 - 1990", "@and @attr 2=4 234 @attr 2=2 1990 "},
+    { "date=-1980", "@attr 2=2 1980 "},
+    { "date=- 1980", "@attr 2=2 1980 "},
+    { "x=-1980", "@attr 2=3 -1980 "},
+    { "x=- 1980", "@attr 2=2 1980 "},
+    { "x= -1980", "@attr 2=3 -1980 "},
+    { "x=234-1990", "@attr 2=3 234-1990 "},
+    { "x=234 - 1990", "@and @attr 2=4 234 @attr 2=2 1990 "},
     {0, 0}
 };
 
-void tst1(int pass)
+void tst1(int pass, int *number_of_errors)
 {
     CCL_parser parser = ccl_parser_create ();
     CCL_bibset bibset = ccl_qual_mk();
@@ -39,6 +52,8 @@ void tst1(int pass)
         ccl_qual_fitem(bibset, "u=4    s=pw t=l,r", "ti");
         ccl_qual_fitem(bibset, "1=1016 s=al,pw",    "term");
         ccl_qual_fitem(bibset, "1=/my/title",         "dc.title");
+        ccl_qual_fitem(bibset, "r=r",         "date");
+        ccl_qual_fitem(bibset, "r=o",         "x");
 	break;
     case 1:
 	strcpy(tstline, "ti u=4    s=pw t=l,r");
@@ -49,12 +64,21 @@ void tst1(int pass)
 
         strcpy(tstline, "dc.title 1=/my/title");
         ccl_qual_line(bibset, tstline);
+
+        strcpy(tstline, "date r=r # ordered relation");
+        ccl_qual_line(bibset, tstline);
+
+        strcpy(tstline, "x r=o # ordered relation");
+        ccl_qual_line(bibset, tstline);
 	break;
     case 2:
         ccl_qual_buf(bibset, "ti u=4    s=pw t=l,r\n"
 		     "term 1=1016 s=al,pw\r\n"
 		     "\n"
-		     "dc.title 1=/my/title\n");
+		     "dc.title 1=/my/title\n"
+		     "date r=r\n" 
+		     "x r=o\n"
+	    );
 	break;
     default:
 	exit(23);
@@ -78,14 +102,14 @@ void tst1(int pass)
 		printf ("Failed %s\n", query_str[i].query);
 		printf (" got:%s:\n", wrbuf_buf(wrbuf));
 		printf (" expected failure\n");
-		exit(3);
+		(*number_of_errors)++;
 	    }
 	    else if (strcmp(wrbuf_buf(wrbuf), query_str[i].result))
 	    {
 		printf ("Failed %s\n", query_str[i].query);
 		printf (" got:%s:\n", wrbuf_buf(wrbuf));
 		printf (" expected:%s:\n", query_str[i].result);
-		exit(2);
+		(*number_of_errors)++;
 	    }
 	    ccl_rpn_delete(rpn);
 	    wrbuf_free(wrbuf, 1);
@@ -95,7 +119,7 @@ void tst1(int pass)
 	    printf ("Failed %s\n", query_str[i].query);
 	    printf (" got failure\n");
 	    printf (" expected:%s:\n", query_str[i].result);
-	    exit(4);
+	    (*number_of_errors)++;
 	}
     }	
     ccl_parser_destroy (parser);
@@ -104,8 +128,15 @@ void tst1(int pass)
 
 int main(int argc, char **argv)
 {
-    tst1(0);
-    tst1(1);
-    tst1(2);
+    int number_of_errors = 0;
+    tst1(0, &number_of_errors);
+    if (number_of_errors)
+	exit(1);
+    tst1(1, &number_of_errors);
+    if (number_of_errors)
+	exit(1);
+    tst1(2, &number_of_errors);
+    if (number_of_errors)
+	exit(1);
     exit(0);
 }
-- 
1.7.10.4