If we have no document charset on completion of parse, retrieve it from the binding.

Make the binding return Windows-1252 if it has no idea (as this is what the parser will have defaulted to).
Fix form_new to not require a document charset to be present -- it may not be known at this point.
Fixup form document charsets post-parse, so that form submission works correctly.

svn path=/trunk/netsurf/; revision=6575

3 files changed, 32 insertions, 8 deletions

diff --git a/render/form.c b/render/form.c
index 096d5d3dd..c9f1abe24 100644
--- a/render/form.c
+++ b/render/form.c
@@ -50,7 +50,7 @@ static char *form_encode_item(const char *item, const char *charset,
  * \param  target  Target frame of form, or NULL for default
  * \param  method  method and enctype
  * \param  charset acceptable encodings for form submission, or NULL
- * \param  doc_charset  encoding of containing document
+ * \param  doc_charset  encoding of containing document, or NULL
  * \return  a new structure, or NULL on memory exhaustion
  */
 struct form *form_new(void *node, const char *action, const char *target, 
@@ -59,8 +59,6 @@ struct form *form_new(void *node, const char *action, const char *target,
 {
 	struct form *form;
 
-	assert(doc_charset != NULL);
-
 	form = calloc(1, sizeof *form);
 	if (!form)
 		return NULL;
@@ -88,8 +86,9 @@ struct form *form_new(void *node, const char *action, const char *target,
 		return NULL;
 	}
 
-	form->document_charset = strdup(doc_charset);
-	if (form->document_charset == NULL) {
+	form->document_charset = doc_charset != NULL ? strdup(doc_charset)
+						     : NULL;
+	if (doc_charset && form->document_charset == NULL) {
 		free(form->accept_charsets);
 		free(form->target);
 		free(form->action);
diff --git a/render/html.c b/render/html.c
index 91f55b145..6887bdc80 100644
--- a/render/html.c
+++ b/render/html.c
@@ -333,6 +333,19 @@ bool html_convert(struct content *c, int width, int height)
 		return false;
 	}
 
+	if (c->data.html.encoding == NULL) {
+		const char *encoding = binding_get_encoding(
+				c->data.html.parser_binding, 
+				&c->data.html.encoding_source);
+
+		c->data.html.encoding = talloc_strdup(c, encoding);
+		if (c->data.html.encoding == NULL) {
+			msg_data.error = messages_get("NoMemory");
+			content_broadcast(c, CONTENT_MSG_ERROR, msg_data);
+			return false;
+		}
+	}
+
 	/* locate html and head elements */
 	html = xmlDocGetRootElement(c->data.html.document);
 	if (html == 0 || strcmp((const char *) html->name, "html") != 0) {
@@ -369,11 +382,11 @@ bool html_convert(struct content *c, int width, int height)
 #ifdef WITH_HUBBUB
 	/* Retrieve forms from parser */
 	c->data.html.forms = binding_get_forms(c->data.html.parser_binding);
-	/* Make all actions absolute */
 	for (f = c->data.html.forms; f != NULL; f = f->prev) {
 		char *action;
 		url_func_result res;
 
+		/* Make all actions absolute */
 		res = url_join(f->action, c->data.html.base_url, &action);
 		if (res != URL_FUNC_OK) {
 			msg_data.error = messages_get("NoMemory");
@@ -383,6 +396,17 @@ bool html_convert(struct content *c, int width, int height)
 
 		free(f->action);
 		f->action = action;
+
+		/* Ensure each form has a document encoding */
+		if (f->document_charset == NULL) {
+			f->document_charset = strdup(c->data.html.encoding);
+			if (f->document_charset == NULL) {
+				msg_data.error = messages_get("NoMemory");
+				content_broadcast(c, CONTENT_MSG_ERROR, 
+						msg_data);
+				return false;
+			}
+		}
 	}
 #endif
 
diff --git a/render/hubbub_binding.c b/render/hubbub_binding.c
index 79a3dba0f..b674d011c 100644
--- a/render/hubbub_binding.c
+++ b/render/hubbub_binding.c
@@ -139,7 +139,8 @@ binding_error binding_create_tree(void *arena, const char *charset, void **ctx)
 
 	c->parser = NULL;
 	c->encoding = charset;
-	c->encoding_source = ENCODING_SOURCE_HEADER;
+	c->encoding_source = charset != NULL ? ENCODING_SOURCE_HEADER
+					     : ENCODING_SOURCE_DETECTED;
 	c->document = NULL;
 	c->owns_doc = true;
 	c->forms = NULL;
@@ -232,7 +233,7 @@ const char *binding_get_encoding(void *ctx, binding_encoding_source *source)
 
 	*source = c->encoding_source;
 
-	return c->encoding;
+	return c->encoding != NULL ? c->encoding : "Windows-1252";
 }
 
 xmlDocPtr binding_get_document(void *ctx)