On Sun, 2009-04-05 at 19:03 +0200, Mark wrote:
> For now, I suppose HTML content is the most relevant, unless XML content
> is parsed differently? I could try calling detect_encoding() in
> render/libxml_binding.c
No. XML content isn't (yet) parsed differently. Additionally, the parser
will have been destroyed, so you can't call binding_get_encoding.
detect_encoding() is private to the libxml2 parser binding (this parser
binding will be disappearing imminently, as hubbub is far superior) so
cannot be called from anywhere else.
In nsgtk_source_dialog_init, you have the following:
> //may need to g_utf_validate() then strcpy the url
> thiswindow->url = bw->current_content->url;
All strings in the core are UTF-8 encoded.
> char * closing = strstr(bw->current_content->source_data, "</html>");
> if (!closing) {
> closing = strstr(bw->current_content->source_data, "</HTML>");
> }
> if (closing)
> *(closing + 7) = 0;
>
> int len = strlen(bw->current_content->source_data);
> *(bw->current_content->source_data + len) = 0;
Ugh! If you want the length (in bytes) of the source data, use
content->source_size. The above needlessly corrupts the source data, and
will fail for all pages with source after the </html>. It also doesn't
handle </Html>, for example. All-in-all, not very nice.
> char ** data = (char **)malloc(sizeof(char *));
> utf8_from_enc(bw->current_content->source_data,
> bw->current_content->data.html.encoding,
> strlen(bw->current_content->source_data), data);
> thiswindow->data = *data;
The following is far cleaner:
utf8_from_enc(bw->current_content->source_data,
bw->current_content->data.html.encoding,
bw->current_content->source_size,
&thiswindow->data);
You must check the return value from utf8_from_enc, too.
> char *title = malloc(strlen(bw->current_content->url) + 10);
It's clearer to use "+ SLEN("Source of ") + 1" than "+ 10".
> strcpy(title, "Source of ");
> strcat(title, bw->current_content->url);
Consider using memcpy here -- it's faster. Alternatively, sprintf(title,
"Source of %s", bw->current_content->url); is clearer.
> gtk_window_set_title(wndSource, title);
Where is title freed?
I'll leave the rest for someone else, as it's GTK specific.
J.
