Parse system entities

[yaz-moved-to-github.git] / retrieval / d1_read.c

diff --git a/retrieval/d1_read.c b/retrieval/d1_read.c
index e79ad72..7f3c4a3 100644 (file)
--- a/retrieval/d1_read.c
+++ b/retrieval/d1_read.c
@@ -3,7 +3,7 @@
  * See the file LICENSE for details.
  * Sebastian Hammer, Adam Dickmeiss
  *
- * $Id: d1_read.c,v 1.45 2002-07-05 16:04:28 adam Exp $
+ * $Id: d1_read.c,v 1.48 2002-08-19 21:09:10 adam Exp $
  */
 
 #include <assert.h>
@@ -24,9 +24,12 @@ data1_node *data1_get_root_tag (data1_handle dh, data1_node *n)
 {
     if (!n)
         return 0;
-    n = n->child;
-    while (n && n->which != DATA1N_tag)
-        n = n->next;
+    if (data1_is_xmlmode(dh))
+    {
+        n = n->child;
+        while (n && n->which != DATA1N_tag)
+            n = n->next;
+    }
     return n;
 }
         
@@ -36,9 +39,19 @@ data1_node *data1_get_root_tag (data1_handle dh, data1_node *n)
  */
 data1_node *get_parent_tag (data1_handle dh, data1_node *n)
 {
-    for (; n && n->which != DATA1N_root; n = n->parent)
-       if (n->which == DATA1N_tag)
-           return n;
+    if (data1_is_xmlmode(dh))
+    {
+        for (; n && n->which != DATA1N_root; n = n->parent)
+            if (n->which == DATA1N_tag && n->parent &&
+                n->parent->which != DATA1N_root)
+                return n;
+    }
+    else
+    {
+        for (; n && n->which != DATA1N_root; n = n->parent)
+            if (n->which == DATA1N_tag)
+                return n;
+    }
     return 0;
 }
 
@@ -182,16 +195,24 @@ data1_node *data1_mk_tag_n (data1_handle dh, NMEM nmem,
 {
     data1_node *partag = get_parent_tag(dh, at);
     data1_node *res = data1_mk_node2 (dh, nmem, DATA1N_tag, at);
-    data1_element *e = NULL;
     data1_xattr **p;
+    data1_element *e = 0;
     
     res->u.tag.tag = data1_insert_string_n (dh, res, nmem, tag, len);
     
-    if (partag)
-       e = partag->u.tag.element;
-    res->u.tag.element =
-       data1_getelementbytagname (dh, at->root->u.root.absyn,
-                                  e, res->u.tag.tag);
+    if (!partag)  /* top tag? */
+        e  = data1_getelementbytagname (dh, at->root->u.root.absyn,
+                                        0 /* index as local */,
+                                        res->u.tag.tag);
+    else
+    {
+        /* only set element for known tags */
+        e = partag->u.tag.element;
+        if (e)
+            e = data1_getelementbytagname (dh, at->root->u.root.absyn,
+                                           e, res->u.tag.tag);
+    }
+    res->u.tag.element = e;
     p = &res->u.tag.attributes;
     while (attr && *attr)
     {
@@ -335,9 +356,14 @@ static data1_node *data1_add_insert_taggeddata(data1_handle dh,
     data1_node *datn = 0;
     data1_node *tagn = 0;
 
-    if (partag)
+    if (!partag)
+        e = data1_getelementbytagname (dh, root->u.root.absyn, 0, tagname);
+    else 
+    {
        e = partag->u.tag.element;
-    e = data1_getelementbytagname (dh, root->u.root.absyn, e, tagname);
+        if (e)
+            e = data1_getelementbytagname (dh, root->u.root.absyn, e, tagname);
+    }
     if (local_allowed || e)
     {
         tagn = data1_mk_node2 (dh, m, DATA1N_tag, at);
@@ -622,8 +648,16 @@ data1_node *data1_read_nodex (data1_handle dh, NMEM m,
                        break;
                    }
                }
-               if (level <= 1)
-                   return d1_stack[0];
+                if (data1_is_xmlmode(dh))
+                {
+                    if (level <= 1)
+                        return d1_stack[0];
+                }
+                else
+                {
+                    if (level <= 0)
+                        return d1_stack[0];
+                }
                continue;
            }   
            else if (!strcmp(tag, "var"))
@@ -676,10 +710,20 @@ data1_node *data1_read_nodex (data1_handle dh, NMEM m,
                 if (level == 0)
                 {
                     parent = data1_mk_root (dh, m, tag);
-                    d1_stack[level++] = parent;
+                    res = d1_stack[level] = parent;
+
+                    if (data1_is_xmlmode(dh))
+                    {
+                        level++;
+                        res = data1_mk_tag (dh, m, tag, 0 /* attr */, parent);
+                        res->u.tag.attributes = xattr;
+                    }
+                }
+                else
+                {
+                    res = data1_mk_tag (dh, m, tag, 0 /* attr */, parent);
+                    res->u.tag.attributes = xattr;
                 }
-                res = data1_mk_tag (dh, m, tag, 0 /* attr */, parent);
-                res->u.tag.attributes = xattr;
             }
            d1_stack[level] = res;
            d1_stack[level+1] = 0;
@@ -834,7 +878,7 @@ static int conv_item (NMEM m, iconv_t t,
 }
 
 static void data1_iconv_s (data1_handle dh, NMEM m, data1_node *n,
-                           iconv_t t, WRBUF wrbuf)
+                           iconv_t t, WRBUF wrbuf, const char *tocode)
 {
     for (; n; n = n->next)
     {
@@ -863,7 +907,8 @@ static void data1_iconv_s (data1_handle dh, NMEM m, data1_node *n,
                 data1_xattr *p;
                 for (p = n->u.tag.attributes; p; p = p->next)
                 {
-                    if (conv_item(m, t, wrbuf, p->value, strlen(p->value))
+                    if (p->value &&
+                        conv_item(m, t, wrbuf, p->value, strlen(p->value))
                         == 0)
                     {
                         wrbuf_puts (wrbuf, "");
@@ -872,24 +917,55 @@ static void data1_iconv_s (data1_handle dh, NMEM m, data1_node *n,
                 }
             }
             break;
+        case DATA1N_preprocess:
+            if (strcmp(n->u.preprocess.target, "xml") == 0)
+            {
+                data1_xattr *p = n->u.preprocess.attributes;
+                for (; p; p = p->next)
+                    if (strcmp (p->name, "encoding") == 0)
+                        p->value = nmem_strdup (m, tocode);
+            }
+            break;
         }
-        data1_iconv_s (dh, m, n->child, t, wrbuf);
+        data1_iconv_s (dh, m, n->child, t, wrbuf, tocode);
     }
 }
 #endif
 
+const char *data1_get_encoding (data1_handle dh, data1_node *n)
+{
+    /* see if we have an xml header that specifies encoding */
+    if (n && n->child && n->child->which == DATA1N_preprocess &&
+        strcmp (n->child->u.preprocess.target, "xml") == 0)
+    {
+        data1_xattr *xp = n->child->u.preprocess.attributes;
+        for (; xp; xp = xp->next)
+            if (!strcmp (xp->name, "encoding") == 0)
+                return xp->value;
+    }
+    /* no encoding in header, so see if "encoding" was specified for abs */
+    if (n && n->which == DATA1N_root &&
+        n->u.root.absyn && n->u.root.absyn->encoding)
+        return n->u.root.absyn->encoding;
+    /* none of above, return a hard coded default */
+    return "ISO-8859-1";
+}
+
 int data1_iconv (data1_handle dh, NMEM m, data1_node *n,
                   const char *tocode, 
                   const char *fromcode)
 {
 #if HAVE_ICONV_H
-    WRBUF wrbuf = wrbuf_alloc();
-    iconv_t t = iconv_open (tocode, fromcode);
-    if (t == (iconv_t) (-1))
-        return -1;
-    data1_iconv_s (dh, m, n, t, wrbuf);
-    iconv_close (t);
-    wrbuf_free (wrbuf, 1);
+    if (strcmp (tocode, fromcode))
+    {
+        WRBUF wrbuf = wrbuf_alloc();
+        iconv_t t = iconv_open (tocode, fromcode);
+        if (t == (iconv_t) (-1))
+            return -1;
+        data1_iconv_s (dh, m, n, t, wrbuf, tocode);
+        iconv_close (t);
+        wrbuf_free (wrbuf, 1);
+    }
     return 0;
 #else
     return -2;