New source grs1disp.c
[yaz-moved-to-github.git] / retrieval / d1_map.c
index 98d6b8b..d3c79e5 100644 (file)
@@ -1,70 +1,59 @@
 /*
- * Copyright (c) 1995, Index Data.
+ * Copyright (c) 1995-2002, Index Data.
  * See the file LICENSE for details.
  * Sebastian Hammer, Adam Dickmeiss
  *
- * $Log: d1_map.c,v $
- * Revision 1.4  1995-12-11 15:22:37  quinn
- * Added last_child field to the node.
- * Rewrote schema-mapping.
- *
- * Revision 1.3  1995/11/01  16:34:56  quinn
- * Making data1 look for tables in data1_tabpath
- *
- * Revision 1.2  1995/11/01  13:54:46  quinn
- * Minor adjustments
- *
- * Revision 1.1  1995/11/01  11:56:08  quinn
- * Added Retrieval (data management) functions en masse.
- *
- *
+ * $Id: d1_map.c,v 1.25 2002-09-24 08:05:41 adam Exp $
  */
 
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 
-#include <oid.h>
-#include <xmalloc.h>
-#include <log.h>
-#include <readconf.h>
-
-#include <tpath.h>
-#include <data1.h>
-#include "d1_map.h"
+#include <yaz/oid.h>
+#include <yaz/log.h>
+#include <yaz/readconf.h>
+#include <yaz/tpath.h>
+#include <yaz/data1.h>
+#include <yaz/d1_map.h>
 
-data1_maptab *data1_read_maptab(char *file)
+data1_maptab *data1_read_maptab (data1_handle dh, const char *file)
 {
-    data1_maptab *res = xmalloc(sizeof(*res));
+    NMEM mem = data1_nmem_get (dh);
+    data1_maptab *res = (data1_maptab *)nmem_malloc(mem, sizeof(*res));
     FILE *f;
+    int lineno = 0;
     int argc;
     char *argv[50], line[512];
     data1_mapunit **mapp;
+    int local_numeric = 0;
 
-    if (!(f = yaz_path_fopen(data1_tabpath, file, "r")))
+    if (!(f = data1_path_fopen(dh, file, "r")))
     {
-       logf(LOG_WARN|LOG_ERRNO, "%s", file);
+       yaz_log(LOG_WARN|LOG_ERRNO, "%s", file);
        return 0;
     }
 
     res->name = 0;
-    res->target_absyn_ref = ODR_NONE;
+    res->target_absyn_ref = VAL_NONE;
     res->map = 0;
     mapp = &res->map;
     res->next = 0;
 
-    while ((argc = readconf_line(f, line, 512, argv, 50)))
+    while ((argc = readconf_line(f, &lineno, line, 512, argv, 50)))
        if (!strcmp(argv[0], "targetref"))
        {
            if (argc != 2)
            {
-               logf(LOG_WARN, "%s: one argument required for targetref",
-                   file);
+               yaz_log(LOG_WARN, "%s:%d: Bad # args for targetref",
+                       file, lineno);
                continue;
            }
-           if ((res->target_absyn_ref = oid_getvalbyname(argv[1])) == ODR_NONE)
+           if ((res->target_absyn_ref = oid_getvalbyname(argv[1]))
+               == VAL_NONE)
            {
-               logf(LOG_WARN, "%s: Unknown reference '%s'", file, argv[1]);
+               yaz_log(LOG_WARN, "%s:%d: Unknown reference '%s'",
+                       file, lineno, argv[1]);
                continue;
            }
        }
@@ -72,22 +61,24 @@ data1_maptab *data1_read_maptab(char *file)
        {
            if (argc != 2)
            {
-               logf(LOG_WARN, "%s: one argument required for targetref",
-                   file);
+               yaz_log(LOG_WARN, "%s:%d: Bad # args for targetname",
+                       file, lineno);
                continue;
            }
-           res->target_absyn_name = xmalloc(strlen(argv[1])+1);
+           res->target_absyn_name =
+               (char *)nmem_malloc(mem, strlen(argv[1])+1);
            strcpy(res->target_absyn_name, argv[1]);
        }
+       else if (!yaz_matchstr(argv[0], "localnumeric"))
+           local_numeric = 1;
        else if (!strcmp(argv[0], "name"))
        {
            if (argc != 2)
            {
-               logf(LOG_WARN, "%s: one argument required for name",
-                   file);
+               yaz_log(LOG_WARN, "%s:%d: Bad # args for name", file, lineno);
                continue;
            }
-           res->name = xmalloc(strlen(argv[1])+1);
+           res->name = (char *)nmem_malloc(mem, strlen(argv[1])+1);
            strcpy(res->name, argv[1]);
        }
        else if (!strcmp(argv[0], "map"))
@@ -97,17 +88,18 @@ data1_maptab *data1_read_maptab(char *file)
 
            if (argc < 3)
            {
-               logf(LOG_WARN, "%s: At least 2 arguments required for map",
-                   file);
+               yaz_log(LOG_WARN, "%s:%d: Bad # of args for map",
+                       file, lineno);
                continue;
            }
-           *mapp = xmalloc(sizeof(**mapp));
+           *mapp = (data1_mapunit *)nmem_malloc(mem, sizeof(**mapp));
            (*mapp)->next = 0;
            if (argc > 3 && !data1_matchstr(argv[3], "nodata"))
                (*mapp)->no_data = 1;
            else
                (*mapp)->no_data = 0;
-           (*mapp)->source_element_name = xmalloc(strlen(argv[1])+1);
+           (*mapp)->source_element_name =
+               (char *)nmem_malloc(mem, strlen(argv[1])+1);
            strcpy((*mapp)->source_element_name, argv[1]);
            mtp = &(*mapp)->target_path;
            if (*path == '/')
@@ -120,51 +112,47 @@ data1_maptab *data1_read_maptab(char *file)
 
                if (ep)
                    ep++;
-               if ((np = sscanf(path, "(%d,%[^)]):%[^/]", &type, valstr,
+               if ((np = sscanf(path, "(%d,%511[^)]):%511[^/]", &type, valstr,
                    parm)) < 2)
                {
-                   logf(LOG_WARN, "%s: Syntax error in map directive: %s",
-                       file, argv[2]);
+                   yaz_log(LOG_WARN, "%s:%d: Syntax error in map "
+                           "directive: %s", file, lineno, argv[2]);
                    fclose(f);
                    return 0;
                }
-               *mtp = xmalloc(sizeof(**mtp));
+               *mtp = (data1_maptag *)nmem_malloc(mem, sizeof(**mtp));
                (*mtp)->next = 0;
                (*mtp)->type = type;
                if (np > 2 && !data1_matchstr(parm, "new"))
                    (*mtp)->new_field = 1;
                else
                    (*mtp)->new_field = 0;
-#if 0
-               if ((numval = atoi(valstr)))
-               {
+               if ((type != 3 || local_numeric) && d1_isdigit(*valstr))
+                {
                    (*mtp)->which = D1_MAPTAG_numeric;
-                   (*mtp)->value.numeric = numval;
+                   (*mtp)->value.numeric = atoi(valstr);
                }
                else
                {
-#endif
                    (*mtp)->which = D1_MAPTAG_string;
-                   (*mtp)->value.string = xmalloc(strlen(valstr)+1);
+                   (*mtp)->value.string =
+                       (char *)nmem_malloc(mem, strlen(valstr)+1);
                    strcpy((*mtp)->value.string, valstr);
-#if 0
                }
-#endif 
                mtp = &(*mtp)->next;
            }
            mapp = &(*mapp)->next;
        }
        else 
-           logf(LOG_WARN, "%s: Unknown directive '%s'", argv[0]);
+           yaz_log(LOG_WARN, "%s:%d: Unknown directive '%s'",
+                   file, lineno, argv[0]);
 
     fclose(f);
     return res;
 }
 
-#if 0
-
 /*
- * Locate node with givel elementname.
+ * Locate node with given elementname.
  * NOTE: This is stupid - we don't find repeats this way.
  */
 static data1_node *find_node(data1_node *p, char *elementname)
@@ -189,7 +177,12 @@ static int tagmatch(data1_node *n, data1_maptag *t)
        return 0;
     if (n->u.tag.element)
     {
-       if (n->u.tag.element->tag->tagset->type != t->type)
+       if (n->u.tag.element->tag->tagset)
+       {
+           if (n->u.tag.element->tag->tagset->type != t->type)
+               return 0;
+       }
+       else if (t->type != 3)
            return 0;
        if (n->u.tag.element->tag->which == DATA1T_numeric)
        {
@@ -223,7 +216,29 @@ static int tagmatch(data1_node *n, data1_maptag *t)
     return 1;
 }
 
-static int map_children(data1_node *n, data1_maptab *map, data1_node *res)
+static data1_node *dup_child (data1_handle dh, data1_node *n,
+                              data1_node **last, NMEM mem,
+                              data1_node *parent)
+{
+    data1_node *first = 0;
+    data1_node **m = &first;
+
+    for (; n; n = n->next)
+    {
+        *last = *m = (data1_node *) nmem_malloc (mem, sizeof(**m));
+        memcpy (*m, n, sizeof(**m));
+        
+        (*m)->parent = parent;
+        (*m)->root = parent->root;
+        (*m)->child = dup_child(dh, n->child, &(*m)->last_child, mem, *m);
+        m = &(*m)->next;
+    }
+    *m = 0;
+    return first;
+}
+
+static int map_children(data1_handle dh, data1_node *n, data1_maptab *map,
+                       data1_node *res, NMEM mem)
 {
     data1_node *c;
     data1_mapunit *m;
@@ -239,198 +254,69 @@ static int map_children(data1_node *n, data1_maptab *map, data1_node *res)
                    c->u.tag.element->name))
                {
                    data1_node *pn = res;
+                   data1_node *cur = pn->last_child;
                    data1_maptag *mt;
-                   data1_node *l;
 
                    /*
                     * process the target path specification.
                     */
                    for (mt = m->target_path; mt; mt = mt->next)
                    {
-                       int match = 0;
-                       data1_node *cur;
-                       data1_node *last;
-
-                       for (l = pn->child, last = 0; l; last = l, l = l->next)
-                           if (!match)
-                               match = tagmatch(l, mt);
-                           else
-                               if (!tagmatch(l, mt))
-                                   break;
-                       if (!match || !mt->next || mt->new_field)
+                       if (!cur || mt->new_field || !tagmatch(cur, mt))
                        {
-                           cur = data1_mk_node();
-                           cur->which = DATA1N_tag;
-                           cur->u.tag.element = 0;
+                           cur = data1_mk_node2 (dh, mem, DATA1N_tag, pn);
                            cur->u.tag.tag = mt->value.string;
-                           cur->u.tag.node_selected = 0;
-                           cur->parent = pn;
-                           cur->root = pn->root;
-                           if (!last)
-                           {
-                               cur->next = pn->child;
-                               pn->child = cur;
-                           }
-                           else
-                           {
-                               cur->next = last->next;
-                               last->next = cur;
-                           }
-                           pn->num_children++;
                        }
-                       else
-                           cur = last ? last : pn->child;
                        
-                       if (mt ->next)
+                       if (mt->next)
                            pn = cur;
                        else if (!m->no_data)
                        {
-                           cur->child = c->child;
-                           cur->num_children = c->num_children;
-                           c->child = 0;
-                           c->num_children = 0;
+                            cur->child =
+                                dup_child (dh, c->child,
+                                           &cur->last_child, mem, cur);
                        }
                    }
                }
            }
-           if (map_children(c, map, res) < 0)
+           if (map_children(dh, c, map, res, mem) < 0)
                return -1;
        }
     return 0;
 }
 
-
-#else
-
-/*
- * See if the node n is equivalent to the tag t.
- */
-static int tagmatch(data1_node *n, data1_maptag *t)
-{
-    if (n->which != DATA1N_tag)
-       return 0;
-    if (n->u.tag.element)
-    {
-       if (n->u.tag.element->tag->tagset->type != t->type)
-           return 0;
-       if (n->u.tag.element->tag->which == DATA1T_numeric)
-       {
-           if (t->which != D1_MAPTAG_numeric)
-               return 0;
-           if (n->u.tag.element->tag->value.numeric != t->value.numeric)
-               return 0;
-       }
-       else
-       {
-           if (t->which != D1_MAPTAG_string)
-               return 0;
-           if (data1_matchstr(n->u.tag.element->tag->value.string,
-               t->value.string))
-               return 0;
-       }
-    }
-    else /* local tag */
-    {
-       char str[10];
-
-       if (t->type != 3)
-           return 0;
-       if (t->which == D1_MAPTAG_numeric)
-           sprintf(str, "%d", t->value.numeric);
-       else
-           strcpy(str, t->value.string);
-       if (data1_matchstr(n->u.tag.tag, str))
-           return 0;
-    }
-    return 1;
-}
-
-static int map_elements(data1_node *res, data1_node *n, data1_mapunit *m)
-{
-    data1_node *c;
-
-    for (c = n->child; c; c = c->next)
-    {
-       if (c->which == DATA1N_tag)
-       {
-           if (c->u.tag.element && !data1_matchstr(c->u.tag.element->name,
-               m->source_element_name))
-           {
-               /* Process target path specification */
-               data1_maptag *mt;
-               data1_node *pn = res, *cur = pn->last_child;
-
-               for (mt = m->target_path; mt; mt = mt->next)
-               {
-                   if (!cur || !tagmatch(cur, mt))
-                   {
-                       cur = data1_mk_node();
-                       cur->which = DATA1N_tag;
-                       cur->u.tag.element = 0;
-                       cur->u.tag.tag = mt->value.string;
-                       cur->u.tag.node_selected = 0;
-                       cur->parent = pn;
-                       cur->root = pn->root;
-                       if (!pn->child)
-                           pn->child = cur;
-                       if (pn->last_child)
-                           pn->last_child->next = cur;
-                       pn->last_child = cur;
-                       pn->num_children++;
-                   }
-                   if (mt->next)
-                       pn = cur;
-                   else if (!m->no_data)
-                   {
-                       cur->child = c->child;
-                       cur->num_children = c->num_children;
-                       c->child = 0;
-                       c->num_children = 0;
-                   }
-               }
-           }
-           else if (map_elements(res, c, m) < 0)
-               return -1;
-       }
-    }
-    return 0;
-}
-
-static int map_record(data1_node *res, data1_node *n, data1_maptab *map)
-{
-    data1_mapunit *m;
-
-    for (m = map->map; m; m = m->next)
-       if (map_elements(res, n, m) < 0)
-           return -1;
-    return 0;
-}
-
 /*
  * Create a (possibly lossy) copy of the given record based on the
  * table. The new copy will refer back to the data of the original record,
  * which should not be discarded during the lifetime of the copy.
  */
-data1_node *data1_map_record(data1_node *n, data1_maptab *map)
+data1_node *data1_map_record (data1_handle dh, data1_node *n,
+                             data1_maptab *map, NMEM m)
 {
-    data1_node *res = data1_mk_node();
+    data1_node *res1, *res = data1_mk_node2 (dh, m, DATA1N_root, 0);
 
     res->which = DATA1N_root;
     res->u.root.type = map->target_absyn_name;
-    if (!(res->u.root.absyn = data1_get_absyn(map->target_absyn_name)))
+    if (!(res->u.root.absyn = data1_get_absyn(dh, map->target_absyn_name)))
+    {
+       yaz_log(LOG_WARN, "%s: Failed to load target absyn '%s'",
+               map->name, map->target_absyn_name);
+    }
+    if (data1_is_xmlmode(dh))
     {
-       logf(LOG_WARN, "%s: Failed to load target absyn '%s'",
-           map->name, map->target_absyn_name);
+        n = n->child;
+        if (!n)
+            return 0;
+        res1 = data1_mk_tag (dh, m, map->target_absyn_name, 0, res);
     }
-    res->parent = 0;
-    res->root = res;
+    else
+        res1 = res;
 
-    if (map_record(res, n, map) < 0)
+    if (map_children(dh, n, map, res1, m) < 0)
     {
-       data1_free_tree(res);
+       data1_free_tree(dh, res);
        return 0;
     }
     return res;
 }
 
-#endif