X-Git-Url: http://mj.ucw.cz/gitweb/?a=blobdiff_plain;f=sherlock%2Fxml%2Fparse.c;h=27141b142287c973d7385a9a30c37d9df9bed16f;hb=89f86d97def2a365b1799f70e3f254264b9c7615;hp=2003d091091b51eef42bbd5207b05405577baca8;hpb=ccd6cdc3fe59f222ed3c4515fa9c4065479f6d9d;p=libucw.git

diff --git a/sherlock/xml/parse.c b/sherlock/xml/parse.c
index 2003d091..27141b14 100644
--- a/sherlock/xml/parse.c
+++ b/sherlock/xml/parse.c
@@ -13,11 +13,11 @@
 #include "sherlock/xml/xml.h"
 #include "sherlock/xml/dtd.h"
 #include "sherlock/xml/internals.h"
-#include "lib/fastbuf.h"
-#include "lib/ff-unicode.h"
-#include "lib/unicode.h"
-#include "lib/chartype.h"
-#include "lib/hashfunc.h"
+#include "ucw/fastbuf.h"
+#include "ucw/ff-unicode.h"
+#include "ucw/unicode.h"
+#include "ucw/chartype.h"
+#include "ucw/hashfunc.h"
 
 #include <setjmp.h>
 
@@ -622,7 +622,7 @@ xml_attrs_init_key(struct xml_attrs_table *t UNUSED, struct xml_attr *a, struct
 #define HASH_WANT_FIND
 #define HASH_GIVE_ALLOC
 XML_HASH_GIVE_ALLOC
-#include "lib/hashtable.h"
+#include "ucw/hashtable.h"
 
 static void
 xml_parse_attr(struct xml_context *ctx)
@@ -680,6 +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)
 {
@@ -705,12 +717,20 @@ xml_push_element(struct xml_context *ctx)
     xml_error(ctx, "Undefined element <%s>", e->name);
   else
     {
-      if (e->dtd->type == XML_DTD_ELEM_MIXED)
+      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;
-
-      // FIXME: validate regular expressions
+      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)
     {
@@ -729,6 +749,19 @@ xml_push_element(struct xml_context *ctx)
       xml_unget_char(ctx);
       xml_parse_attr(ctx);
     }
+  if (e->dtd)
+    SLIST_FOR_EACH(struct xml_dtd_attr *, a, e->dtd->attrs)
+      if (a->default_mode == XML_ATTR_REQUIRED)
+        {
+	  if (!xml_attrs_find(ctx->tab_attrs, e, a->name))
+	    xml_error(ctx, "Missing required attribute %s in element <%s>", a->name, e->name);
+	}
+      else if (a->default_mode != XML_ATTR_IMPLIED && ctx->flags & XML_ALLOC_DEFAULT_ATTRS)
+        {
+	  struct xml_attr *attr = xml_attrs_lookup(ctx->tab_attrs, e, a->name);
+	  if (!attr->val)
+	    attr->val = a->default_value;
+	}
   if ((ctx->flags & XML_REPORT_TAGS) && ctx->h_stag)
     ctx->h_stag(ctx);
 }
@@ -825,7 +858,11 @@ xml_parse_doctype_decl(struct xml_context *ctx)
       ctx->flags |= XML_HAS_EXTERNAL_SUBSET;
     }
   if (xml_peek_char(ctx) == '[')
-    ctx->flags |= XML_HAS_INTERNAL_SUBSET;
+    {
+      ctx->flags |= XML_HAS_INTERNAL_SUBSET;
+      xml_skip_char(ctx);
+      xml_inc(ctx);
+    }
   if (ctx->h_doctype_decl)
     ctx->h_doctype_decl(ctx);
 }
@@ -837,12 +874,19 @@ xml_parse_doctype_decl(struct xml_context *ctx)
 /* DTD: Internal subset */
 
 static void
-xml_parse_internal_subset(struct xml_context *ctx)
+xml_parse_subset(struct xml_context *ctx, uns external)
 {
-  // FIXME: comments/pi have no parent
+  // FIXME:
+  // -- comments/pi have no parent
+  // -- conditional sections in external subset
+  // -- check corectness of parameter entities
+
   /* '[' intSubset ']'
    * intSubset :== (markupdecl | DeclSep)
-   * Already parsed: ']' */
+   * Already parsed: '['
+   *
+   * extSubsetDecl ::= ( markupdecl | conditionalSect | DeclSep)*
+   */
   while (1)
     {
       xml_parse_white(ctx, 0);
@@ -890,16 +934,21 @@ xml_parse_internal_subset(struct xml_context *ctx)
 	  goto invalid_markup;
       else if (c == '%')
 	xml_parse_pe_ref(ctx);
-      else if (c == ']')
-	break;
+      else if (c == ']' && !external)
+        {
+	  break;
+	}
+      else if (c == '>' && external)
+        {
+	  break;
+	}
       else
 	goto invalid_markup;
     }
   xml_dec(ctx);
-  xml_dec(ctx);
   return;
-invalid_markup:
-  xml_fatal(ctx, "Invalid markup in the internal subset");
+invalid_markup: ;
+  xml_fatal(ctx, "Invalid markup in the %s subset", external ? "external" : "internal");
 }
 
 /*** The State Machine ***/
@@ -974,24 +1023,37 @@ error:
 		xml_unget_char(ctx);
 		xml_parse_doctype_decl(ctx);
 		PULL(DOCTYPE_DECL);
-		if (xml_peek_char(ctx) == '[')
-		  {
-		    xml_skip_char(ctx);
-		    xml_inc(ctx);
-		    if (ctx->flags & XML_PARSE_DTD)
-		      {
-			xml_dtd_init(ctx);
-			if (ctx->h_dtd_start)
-			  ctx->h_dtd_start(ctx);
-			// FIXME: pull iface?
-			xml_parse_internal_subset(ctx);
-			// FIXME: external subset
-			if (ctx->h_dtd_end)
-			  ctx->h_dtd_end(ctx);
-		      }
-		    else
-		      xml_skip_internal_subset(ctx);
-		  }
+		if (ctx->flags & XML_HAS_DTD)
+		  if (ctx->flags & XML_PARSE_DTD)
+		    {
+		      xml_dtd_init(ctx);
+		      if (ctx->h_dtd_start)
+		        ctx->h_dtd_start(ctx);
+		      if (ctx->flags & XML_HAS_INTERNAL_SUBSET)
+		        {
+		          xml_parse_subset(ctx, 0);
+			  xml_dec(ctx);
+			}
+		      if (ctx->flags & XML_HAS_EXTERNAL_SUBSET)
+		        {
+		          struct xml_dtd_entity ent = {
+			    .system_id = ctx->system_id,
+			    .public_id = ctx->public_id,
+			  };
+			  xml_parse_white(ctx, 0);
+			  xml_parse_char(ctx, '>');
+			  xml_unget_char(ctx);
+			  ASSERT(ctx->h_resolve_entity);
+			  ctx->h_resolve_entity(ctx, &ent);
+			  ctx->flags |= XML_SRC_EXPECTED_DECL;
+			  xml_parse_subset(ctx, 1);
+			  xml_unget_char(ctx);;
+			}
+		      if (ctx->h_dtd_end)
+		        ctx->h_dtd_end(ctx);
+		    }
+		  else if (ctx->flags & XML_HAS_INTERNAL_SUBSET)
+		    xml_skip_internal_subset(ctx);
 		xml_parse_white(ctx, 0);
 		xml_parse_char(ctx, '>');
 		xml_dec(ctx);
@@ -1182,3 +1244,44 @@ xml_parse(struct xml_context *ctx)
   while (xml_next(ctx));
   return ctx->err_code;
 }
+
+char *
+xml_merge_chars(struct xml_context *ctx UNUSED, struct xml_node *node, struct mempool *pool)
+{
+  ASSERT(node->type == XML_NODE_ELEM);
+  char *p = mp_start_noalign(pool, 1);
+  XML_NODE_FOR_EACH(son, node)
+    if (son->type == XML_NODE_CHARS)
+      {
+	p = mp_spread(pool, p, son->len + 1);
+	memcpy(p, son->text, son->len);
+	p += son->len;
+      }
+  *p++ = 0;
+  return mp_end(pool, p);
+}
+
+static char *
+xml_append_dom_chars(char *p, struct mempool *pool, struct xml_node *node)
+{
+  XML_NODE_FOR_EACH(son, node)
+    if (son->type == XML_NODE_CHARS)
+      {
+	p = mp_spread(pool, p, son->len + 1);
+	memcpy(p, son->text, son->len);
+	p += son->len;
+      }
+    else if (son->type == XML_NODE_ELEM)
+      p = xml_append_dom_chars(p, pool, son);
+  return p;
+}
+
+char *
+xml_merge_dom_chars(struct xml_context *ctx UNUSED, struct xml_node *node, struct mempool *pool)
+{
+  ASSERT(node->type == XML_NODE_ELEM);
+  char *p = mp_start_noalign(pool, 1);
+  p = xml_append_dom_chars(p, pool, node);
+  *p++ = 0;
+  return mp_end(pool, p);
+}