X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=retrieval%2Fd1_absyn.c;h=c8479a43234a2af290cdebed3710445d4eceb545;hb=7674cd40af86e353235cb1b509f4b4687167cdf2;hp=f93d39b7320f783a67931e12972e5d6e295021b9;hpb=cea212fba995de3fa0d6d6723d828f6fd7ca31df;p=yaz-moved-to-github.git

diff --git a/retrieval/d1_absyn.c b/retrieval/d1_absyn.c
index f93d39b..c8479a4 100644
--- a/retrieval/d1_absyn.c
+++ b/retrieval/d1_absyn.c
@@ -1,10 +1,53 @@
 /*
- * Copyright (c) 1995, Index Data.
+ * Copyright (c) 1995-1997, Index Data.
  * See the file LICENSE for details.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: d1_absyn.c,v $
- * Revision 1.1  1995-11-01 11:56:06  quinn
+ * Revision 1.14  1997-10-31 12:20:09  adam
+ * Improved memory debugging for xmalloc/nmem.c. References to NMEM
+ * instead of ODR in n ESPEC-1 handling in source d1_espec.c.
+ * Bug fix: missing fclose in data1_read_espec1.
+ *
+ * Revision 1.13  1997/10/27 13:54:18  adam
+ * Changed structure field in data1 node to be simple string which
+ * is "unknown" to the retrieval system itself.
+ *
+ * Revision 1.12  1997/09/17 12:10:34  adam
+ * YAZ version 1.4.
+ *
+ * Revision 1.11  1997/09/05 09:50:55  adam
+ * Removed global data1_tabpath - uses data1_get_tabpath() instead.
+ *
+ * Revision 1.10  1997/05/14 06:54:01  adam
+ * C++ support.
+ *
+ * Revision 1.9  1997/02/19 14:46:15  adam
+ * The "all" specifier only affects elements that are indexed (and not
+ * all elements).
+ *
+ * Revision 1.8  1997/01/02 10:47:59  quinn
+ * Added optional, physical ANY
+ *
+ * Revision 1.7  1996/06/10 08:56:01  quinn
+ * Work on Summary.
+ *
+ * Revision 1.6  1996/05/31  13:52:21  quinn
+ * Fixed uninitialized variable for local tags in abstract syntax.
+ *
+ * Revision 1.5  1996/05/09  07:27:43  quinn
+ * Multiple local attributes values supported.
+ *
+ * Revision 1.4  1996/05/01  12:45:28  quinn
+ * Support use of local tag names in abs file.
+ *
+ * Revision 1.3  1995/11/01  16:34:55  quinn
+ * Making data1 look for tables in data1_tabpath
+ *
+ * Revision 1.2  1995/11/01  13:54:44  quinn
+ * Minor adjustments
+ *
+ * Revision 1.1  1995/11/01  11:56:06  quinn
  * Added Retrieval (data management) functions en masse.
  *
  *
@@ -16,46 +59,61 @@
 #include <stdlib.h>
 #include <string.h>
 
-#include <xmalloc.h>
 #include <oid.h>
 #include <log.h>
+#include <tpath.h>
 
-#include "data1.h"
+#include <data1.h>
 
 #define D1_MAX_NESTING  128
-#define DATA1_MAX_SYNTAXES 30 /* max no of syntaxes to handle in one session */
 
-static struct /* cache of abstract syntaxes */
+struct data1_absyn_cache_info 
 {
     char *name;
     data1_absyn *absyn;
-} syntaxes[DATA1_MAX_SYNTAXES] = {{0,0}};
+    data1_absyn_cache next;
+};
 
-data1_absyn *data1_get_absyn(char *name)
+data1_absyn *data1_absyn_search (data1_handle dh, const char *name)
 {
-    char fname[512];
-    int i;
-
-    for (i = 0; syntaxes[i].name; i++)
-	if (!strcmp(name, syntaxes[i].name))
-	    return syntaxes[i].absyn;
+    data1_absyn_cache p = *data1_absyn_cache_get (dh);
 
-    if (i >= DATA1_MAX_SYNTAXES - 1)
+    while (p)
     {
-	logf(LOG_WARN, "Too many abstract syntaxes loaded");
-	return 0;
+	if (!strcmp (name, p->name))
+	    return p->absyn;
+	p = p->next;
     }
+    return NULL;
+}
+
+data1_absyn *data1_absyn_add (data1_handle dh, const char *name)
+{
+    char fname[512];
+    NMEM mem = data1_nmem_get (dh);
+
+    data1_absyn_cache p = nmem_malloc (mem, sizeof(*p));
+    data1_absyn_cache *pp = data1_absyn_cache_get (dh);
+
     sprintf(fname, "%s.abs", name);
-    if (!(syntaxes[i].absyn = data1_read_absyn(fname)))
-	return 0;
-    if (!(syntaxes[i].name = xmalloc(strlen(name)+1)))
-	abort();
-    strcpy(syntaxes[i].name, name);
-    syntaxes[i+1].name = 0;
-    return syntaxes[i].absyn;
+    p->absyn = data1_read_absyn (dh, fname);
+    p->name = nmem_strdup (mem, name);
+    p->next = *pp;
+    *pp = p;
+    return p->absyn;
+}
+
+data1_absyn *data1_get_absyn (data1_handle dh, char *name)
+{
+    data1_absyn *absyn;
+
+    if (!(absyn = data1_absyn_search (dh, name)))
+	absyn = data1_absyn_add (dh, name);
+    return absyn;
 }
 
-data1_esetname *data1_getesetbyname(data1_absyn *a, char *name)
+data1_esetname *data1_getesetbyname(data1_handle dh, data1_absyn *a,
+				    char *name)
 {
     data1_esetname *r;
 
@@ -65,8 +123,9 @@ data1_esetname *data1_getesetbyname(data1_absyn *a, char *name)
     return 0;
 }
 
-data1_element *data1_getelementbytagname(data1_absyn *abs,
-    data1_element *parent, char *tagname)
+data1_element *data1_getelementbytagname (data1_handle dh, data1_absyn *abs,
+					  data1_element *parent,
+					  char *tagname)
 {
     data1_element *r;
 
@@ -85,7 +144,8 @@ data1_element *data1_getelementbytagname(data1_absyn *abs,
     return 0;
 }
 
-data1_element *data1_getelementbyname(data1_absyn *absyn, char *name)
+data1_element *data1_getelementbyname (data1_handle dh, data1_absyn *absyn,
+				       char *name)
 {
     data1_element *r;
 
@@ -95,25 +155,26 @@ data1_element *data1_getelementbyname(data1_absyn *absyn, char *name)
     return 0;
 }
 
-data1_absyn *data1_read_absyn(char *file)
+data1_absyn *data1_read_absyn (data1_handle dh, const char *file)
 {
     char line[512], *r, cmd[512], args[512];
     data1_absyn *res = 0;
     FILE *f;
-    data1_element **ppl[D1_MAX_NESTING];
+    data1_element **ppl[D1_MAX_NESTING], *cur[D1_MAX_NESTING];
     data1_esetname **esetpp;
     data1_maptab **maptabp;
     data1_marctab **marcp;
+    data1_termlist *all = 0;
     int level = 0;
 
-    if (!(f = fopen(file, "r")))
+    logf (LOG_DEBUG, "begin data1_read_absyn file=%s", file);
+    if (!(f = yaz_path_fopen(data1_get_tabpath (dh), file, "r")))
     {
 	logf(LOG_WARN|LOG_ERRNO, "%s", file);
 	return 0;
     }
 
-    if (!(res = xmalloc(sizeof(*res))))
-	abort();
+    res = nmem_malloc(data1_nmem_get(dh), sizeof(*res));
     res->name = 0;
     res->reference = VAL_NONE;
     res->tagset = 0;
@@ -126,6 +187,7 @@ data1_absyn *data1_read_absyn(char *file)
     marcp = &res->marc;
     res->elements = 0;
     ppl[0] = &res->elements;
+    cur[0] = 0;
     esetpp = &res->esetnames;
 
     for (;;)
@@ -138,20 +200,18 @@ data1_absyn *data1_read_absyn(char *file)
 		break;
 	}
 	if (!r)
-	{
-	    fclose(f);
-	    return res;
-	}
+            break;
 	if (sscanf(r, "%s %[^\n]", cmd, args) < 2)
 	    *args = '\0';
 	if (!strcmp(cmd, "elm"))
 	{
-	    data1_element *new;
+	    data1_element *new_element;
 	    int i;
-	    char path[512], name[512], att[512], *p;
+	    char path[512], name[512], termlists[512], *p;
 	    int type, value;
+	    data1_termlist **tp;
 
-	    if (sscanf(args, "%s %s %s", path, name, att) < 3)
+	    if (sscanf(args, "%511s %511s %511s", path, name, termlists) < 3)
 	    {
 		logf(LOG_WARN, "Bad # of args to elm in %s: '%s'", 
 		    file, args);
@@ -170,40 +230,66 @@ data1_absyn *data1_read_absyn(char *file)
 	    }
 	    if (i > level + 1)
 	    {
-		logf(LOG_WARN, "Bad level inc in %s in '%'", file, args);
+		logf(LOG_WARN, "Bad level inc in %s in '%s'", file, args);
 		fclose(f);
 		return 0;
 	    }
 	    level = i;
-	    if (!(new = *ppl[level] = xmalloc(sizeof(*new))))
-		abort;
-	    new ->next = new->children = 0;
-	    ppl[level] = &new->next;
-	    ppl[level+1] = &new->children;
-
-	    if (sscanf(p, "(%d,%d)", &type, &value) < 2)
+	    new_element = cur[level] = *ppl[level] =
+		nmem_malloc(data1_nmem_get(dh), sizeof(*new_element));
+	    new_element->next = new_element->children = 0;
+	    new_element->tag = 0;
+	    new_element->termlists = 0;
+	    new_element->parent = level ? cur[level - 1] : 0;
+	    tp = &new_element->termlists;
+	    ppl[level] = &new_element->next;
+	    ppl[level+1] = &new_element->children;
+	    
+	    /* well-defined tag */
+	    if (sscanf(p, "(%d,%d)", &type, &value) == 2)
 	    {
-		logf(LOG_WARN, "Malformed element '%s' in %s", p, file);
-		fclose(f);
-		return 0;
+		if (!res->tagset)
+		{
+		    logf(LOG_WARN, "No tagset loaded in %s", file);
+		    fclose(f);
+		    return 0;
+		}
+		if (!(new_element->tag = data1_gettagbynum (dh, res->tagset,
+							    type, value)))
+		{
+		    logf(LOG_WARN, "Couldn't find tag %s in tagset in %s",
+			p, file);
+		    fclose(f);
+		    return 0;
+		}
 	    }
-	    if (!res->tagset)
+	    /* private tag */
+	    else if (*p)
 	    {
-		logf(LOG_WARN, "No tagset loaded in %s", file);
-		fclose(f);
-		return 0;
+		data1_tag *nt =
+		    new_element->tag = nmem_malloc(data1_nmem_get (dh),
+						   sizeof(*new_element->tag));
+		nt->which = DATA1T_string;
+		nt->value.string = nmem_strdup(data1_nmem_get (dh), p);
+		nt->names = nmem_malloc(data1_nmem_get(dh), 
+					sizeof(*new_element->tag->names));
+		nt->names->name = nt->value.string;
+		nt->names->next = 0;
+		nt->kind = DATA1K_string;
+		nt->next = 0;
+		nt->tagset = 0;
 	    }
-	    if (!(new->tag = data1_gettagbynum(res->tagset, type, value)))
+	    else
 	    {
-		logf(LOG_WARN, "Couldn't find tag %s in tagset in %s",
-		    p, file);
+		logf(LOG_WARN, "Bad element is %s", file);
 		fclose(f);
 		return 0;
 	    }
-	    if (*att == '!')
-		strcpy(att, name);
-	    if (*att == '-')
-		new->att = 0;
+
+	    /* parse termList definitions */
+	    p = termlists;
+	    if (*p == '-')
+		new_element->termlists = 0;
 	    else
 	    {
 		if (!res->attset)
@@ -212,16 +298,100 @@ data1_absyn *data1_read_absyn(char *file)
 		    fclose(f);
 		    return 0;
 		}
-		if (!(new->att = data1_getattbyname(res->attset, att)))
+		do
+		{
+		    char attname[512], structure[512];
+		    int r;
+
+		    if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
+			structure)))
+		    {
+			logf(LOG_WARN, "Syntax error in termlistspec in %s",
+			    file);
+			fclose(f);
+			return 0;
+		    }
+		    if (*attname == '!')
+			strcpy(attname, name);
+		    *tp = nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
+		    (*tp)->next = 0;
+		    if (!((*tp)->att = data1_getattbyname(dh, res->attset,
+							  attname)))
+		    {
+			logf(LOG_WARN, "Couldn't find att '%s' in attset",
+			     attname);
+			fclose(f);
+			return 0;
+		    }
+		    if (r < 2) /* is the structure qualified? */
+			(*tp)->structure = "w";
+		    else 
+		    {
+			(*tp)->structure = nmem_malloc (data1_nmem_get (dh),
+							strlen(structure)+1);
+			strcpy ((*tp)->structure, structure);
+		    }
+		    tp = &(*tp)->next;
+		}
+		while ((p = strchr(p, ',')) && *(++p));
+	        *tp = all; /* append any ALL entries to the list */
+	    }
+
+	    new_element->name = nmem_strdup(data1_nmem_get (dh), name);
+	}
+	else if (!strcmp(cmd, "all"))
+	{
+	    char *p;
+	    data1_termlist **tp = &all;
+
+	    if (all)
+	    {
+		logf(LOG_WARN, "Too many ALL declarations in %s - ignored",
+		    file);
+		continue;
+	    }
+
+	    p = args;
+	    if (!res->attset)
+	    {
+		logf(LOG_WARN, "No attset loaded in %s", file);
+		fclose(f);
+		return 0;
+	    }
+	    do
+	    {
+		char attname[512], structure[512];
+		int r;
+
+		if (!(r = sscanf(p, "%511[^:,]:%511[^,]", attname,
+		    structure)))
+		{
+		    logf(LOG_WARN, "Syntax error in termlistspec in %s",
+			file);
+		    fclose(f);
+		    return 0;
+		}
+		*tp = nmem_malloc(data1_nmem_get(dh), sizeof(**tp));
+		if (!((*tp)->att = data1_getattbyname (dh, res->attset,
+						       attname)))
 		{
-		    logf(LOG_WARN, "Couldn't find att '%s' in attset", att);
+		    logf(LOG_WARN, "Couldn't find att '%s' in attset",
+			 attname);
 		    fclose(f);
 		    return 0;
 		}
+		if (r < 2) /* is the structure qualified? */
+		    (*tp)->structure = "w";
+		else 
+		{
+		    (*tp)->structure = nmem_malloc (data1_nmem_get (dh),
+						    strlen(structure)+1);
+		    strcpy ((*tp)->structure, structure);
+		}
+		(*tp)->next = 0;
+		tp = &(*tp)->next;
 	    }
-	    if (!(new->name = xmalloc(strlen(name)+1)))
-		abort();
-	    strcpy(new->name, name);
+	    while ((p = strchr(p, ',')) && *(++p));
 	}
 	else if (!strcmp(cmd, "name"))
 	{
@@ -233,9 +403,7 @@ data1_absyn *data1_read_absyn(char *file)
 		fclose(f);
 		return 0;
 	    }
-	    if (!(res->name = xmalloc(strlen(args)+1)))
-		abort();
-	    strcpy(res->name, name);
+	    res->name = nmem_strdup(data1_nmem_get(dh), args);
 	}
 	else if (!strcmp(cmd, "reference"))
 	{
@@ -264,7 +432,7 @@ data1_absyn *data1_read_absyn(char *file)
 		fclose(f);
 		return 0;
 	    }
-	    if (!(res->attset = data1_read_attset(name)))
+	    if (!(res->attset = data1_read_attset (dh, name)))
 	    {
 		logf(LOG_WARN, "Attset failed in %s", file);
 		fclose(f);
@@ -281,7 +449,7 @@ data1_absyn *data1_read_absyn(char *file)
 		fclose(f);
 		return 0;
 	    }
-	    if (!(res->tagset = data1_read_tagset(name)))
+	    if (!(res->tagset = data1_read_tagset (dh, name)))
 	    {
 		logf(LOG_WARN, "Tagset failed in %s", file);
 		fclose(f);
@@ -298,7 +466,7 @@ data1_absyn *data1_read_absyn(char *file)
 		fclose(f);
 		return 0;
 	    }
-	    if (!(res->varset = data1_read_varset(name)))
+	    if (!(res->varset = data1_read_varset (dh, name)))
 	    {
 		logf(LOG_WARN, "Varset failed in %s", file);
 		fclose(f);
@@ -315,18 +483,17 @@ data1_absyn *data1_read_absyn(char *file)
 		fclose(f);
 		return 0;
 	    }
-	    *esetpp = xmalloc(sizeof(**esetpp));
-	    (*esetpp)->name = xmalloc(strlen(name)+1);
-	    strcpy((*esetpp)->name, name);
+	    *esetpp = nmem_malloc(data1_nmem_get(dh), sizeof(**esetpp));
+	    (*esetpp)->name = nmem_strdup(data1_nmem_get(dh), name);
+	    (*esetpp)->next = 0;
 	    if (*fname == '@')
 		(*esetpp)->spec = 0;
-	    else if (!((*esetpp)->spec = data1_read_espec1(fname, 0)))
+	    else if (!((*esetpp)->spec = data1_read_espec1 (dh, fname)))
 	    {
 		logf(LOG_WARN, "%s: Espec-1 read failed", file);
 		fclose(f);
 		return 0;
 	    }
-	    (*esetpp)->next = 0;
 	    esetpp = &(*esetpp)->next;
 	}
 	else if (!strcmp(cmd, "maptab"))
@@ -339,7 +506,7 @@ data1_absyn *data1_read_absyn(char *file)
 		    file);
 		continue;
 	    }
-	    if (!(*maptabp = data1_read_maptab(name)))
+	    if (!(*maptabp = data1_read_maptab (dh, name)))
 	    {
 		logf(LOG_WARN, "%s: Failed to read maptab.");
 		continue;
@@ -356,7 +523,7 @@ data1_absyn *data1_read_absyn(char *file)
 		    file);
 		continue;
 	    }
-	    if (!(*marcp = data1_read_marctab(name)))
+	    if (!(*marcp = data1_read_marctab (dh, name)))
 	    {
 		logf(LOG_WARN, "%s: Failed to read marctab.");
 		continue;
@@ -370,4 +537,7 @@ data1_absyn *data1_read_absyn(char *file)
 	    return 0;
 	}
     }
+    fclose(f);
+    logf (LOG_DEBUG, "end data1_read_absyn file=%s", file);
+    return res;
 }