X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;ds=sidebyside;f=sherlock%2Fxml%2Fparse.c;h=6f5b192f0ab29e021f4e8e219e6bf6c5d9544cbb;hb=306801fc6da02fcbed53db6ff98cd0277b763d9a;hp=ef957bc6911632bb4553e63b8ca4f3cb4955255e;hpb=7ff16aadb50619ba67d2fe3fa0b7a958b86c3162;p=libucw.git

diff --git a/sherlock/xml/parse.c b/sherlock/xml/parse.c
index ef957bc6..6f5b192f 100644
--- a/sherlock/xml/parse.c
+++ b/sherlock/xml/parse.c
@@ -421,9 +421,16 @@ xml_flush_chars(struct xml_context *ctx)
   uns len = xml_end_chars(ctx, &text), rlen;
   if (len)
     {
+      if (ctx->flags & XML_NO_CHARS)
+        {
+          if ((ctx->flags & XML_REPORT_CHARS) && ctx->h_ignorable)
+            ctx->h_ignorable(ctx, text, len);
+	  mp_restore(ctx->pool, &ctx->chars_state);
+	  return 0;
+	}
       if ((ctx->flags & XML_REPORT_CHARS) && ctx->h_block && (rlen = xml_report_chars(ctx, &rtext)))
 	ctx->h_block(ctx, rtext, rlen);
-      if (!(ctx->flags & XML_ALLOC_CHARS) && (!(ctx->flags & XML_REPORT_CHARS) || !ctx->h_chars))
+      if (!(ctx->flags & XML_ALLOC_CHARS) && !(ctx->flags & XML_REPORT_CHARS))
         {
 	  mp_restore(ctx->pool, &ctx->chars_state);
 	  return 0;
@@ -450,25 +457,51 @@ xml_append_chars(struct xml_context *ctx)
 {
   TRACE(ctx, "append_chars");
   struct fastbuf *out = &ctx->chars;
-  while (xml_get_char(ctx) != '<')
-    if (xml_last_char(ctx) == '&')
-      {
-	xml_inc(ctx);
-        xml_parse_ref(ctx);
-      }
-    else
-      bput_utf8_32(out, xml_last_char(ctx));
+  if (ctx->flags & XML_NO_CHARS)
+    while (xml_get_char(ctx) != '<')
+      if (xml_last_cat(ctx) & XML_CHAR_WHITE)
+	bput_utf8_32(out, xml_last_char(ctx));
+      else
+        {
+	  xml_error(ctx, "This element must not contain character data");
+	  while (xml_get_char(ctx) != '<');
+	  break;
+	}
+  else
+    while (xml_get_char(ctx) != '<')
+      if (xml_last_char(ctx) == '&')
+        {
+	  xml_inc(ctx);
+          xml_parse_ref(ctx);
+        }
+      else
+        bput_utf8_32(out, xml_last_char(ctx));
   xml_unget_char(ctx);
 }
 
 /*** CDATA sections ***/
 
+static void
+xml_skip_cdata(struct xml_context *ctx)
+{
+  TRACE(ctx, "skip_cdata");
+  xml_parse_seq(ctx, "CDATA[");
+  while (xml_get_char(ctx) != ']' || xml_get_char(ctx) != ']' || xml_get_char(ctx) != '>');
+  xml_dec(ctx);
+}
+
 static void
 xml_append_cdata(struct xml_context *ctx)
 {
   /* CDSect :== '<![CDATA[' (Char* - (Char* ']]>' Char*)) ']]>'
    * Already parsed: '<![' */
   TRACE(ctx, "append_cdata");
+  if (ctx->flags & XML_NO_CHARS)
+    {
+      xml_error(ctx, "This element must not contain CDATA");
+      xml_skip_cdata(ctx);
+      return;
+    }
   xml_parse_seq(ctx, "CDATA[");
   struct fastbuf *out = &ctx->chars;
   uns rlen;
@@ -493,15 +526,6 @@ xml_append_cdata(struct xml_context *ctx)
   xml_dec(ctx);
 }
 
-static void UNUSED
-xml_skip_cdata(struct xml_context *ctx)
-{
-  TRACE(ctx, "skip_cdata");
-  xml_parse_seq(ctx, "CDATA[");
-  while (xml_get_char(ctx) != ']' || xml_get_char(ctx) != ']' || xml_get_char(ctx) != '>');
-  xml_dec(ctx);
-}
-
 /*** Attribute values ***/
 
 char *
@@ -630,6 +654,18 @@ xml_attr_find(struct xml_context *ctx, struct xml_node *node, char *name)
   return xml_attrs_find(ctx->tab_attrs, node, name);
 }
 
+char *
+xml_attr_value(struct xml_context *ctx, struct xml_node *node, char *name)
+{
+  struct xml_attr *attr = xml_attrs_find(ctx->tab_attrs, node, name);
+  if (attr)
+    return attr->val;
+  if (!node->dtd)
+    return NULL;
+  struct xml_dtd_attr *dtd = xml_dtd_find_attr(ctx, node->dtd, name);
+  return dtd ? dtd->default_value : NULL;
+}
+
 void
 xml_attrs_table_init(struct xml_context *ctx)
 {
@@ -644,6 +680,18 @@ xml_attrs_table_cleanup(struct xml_context *ctx)
 
 /*** Elements ***/
 
+static uns
+xml_validate_element(struct xml_dtd_elem_node *root, struct xml_dtd_elem *elem)
+{
+  if (root->elem)
+    return elem == root->elem;
+  else
+    SLIST_FOR_EACH(struct xml_dtd_elem_node *, son, root->sons)
+      if (xml_validate_element(son, elem))
+	return 1;
+  return 0;
+}
+
 static void
 xml_push_element(struct xml_context *ctx)
 {
@@ -669,7 +717,20 @@ xml_push_element(struct xml_context *ctx)
     xml_error(ctx, "Undefined element <%s>", e->name);
   else
     {
-      // FIXME: validate regular expressions
+      struct xml_dtd_elem *dtd = e->dtd, *parent_dtd = e->parent ? e->parent->dtd : NULL;
+      if (dtd->type == XML_DTD_ELEM_MIXED)
+        ctx->flags &= ~XML_NO_CHARS;
+      else
+	ctx->flags |= XML_NO_CHARS;
+      if (parent_dtd)
+        if (parent_dtd->type == XML_DTD_ELEM_EMPTY)
+	  xml_error(ctx, "Empty element must not contain children");
+        else if (parent_dtd->type != XML_DTD_ELEM_ANY)
+	  {
+	    // FIXME: validate regular expressions
+	    if (!xml_validate_element(parent_dtd->node, dtd))
+	      xml_error(ctx, "Unexpected element <%s>", e->name);
+	  }
     }
   while (1)
     {
@@ -1108,6 +1169,28 @@ epilog:
   ASSERT(0);
 }
 
+uns
+xml_next_state(struct xml_context *ctx, uns pull)
+{
+  uns saved = ctx->pull;
+  ctx->pull = pull;
+  uns res = xml_next(ctx);
+  ctx->pull = saved;
+  return res;
+}
+
+uns
+xml_skip_element(struct xml_context *ctx)
+{
+  ASSERT(ctx->state == XML_STATE_STAG);
+  struct xml_node *node = ctx->node;
+  uns saved = ctx->pull, res;
+  ctx->pull = XML_PULL_ETAG;
+  while ((res = xml_next(ctx)) && ctx->node != node);
+  ctx->pull = saved;
+  return res;
+}
+
 uns
 xml_parse(struct xml_context *ctx)
 {