WOW... Any idea on how to implement it in an efficient way, i.e.
without putting the entire documnt in a buffer?

Well, since we depend on GNU regexes, we can take advantage of
re_search_2 which searches on split data, and perform the search on
small chunks.

*** Bug 369269 has been marked as a duplicate of this bug. ***

With GRegex, g_match_info_is_partial_match() can be useful.

The text of the buffer can be retrieved chunks by chunks with gtk_text_buffer_get_text(). If there is a full match, we return the corresponding GtkTextIter's. If there is a partial match, we take a bigger chunk.

For the search and replace feature, the performances should be OK.

Move GtkTextView -> GtkSourceView.

It will be easier to implement in gsv, if GTK_TEXT_SEARCH_VISIBLE_ONLY and GTK_TEXT_SEARCH_TEXT_ONLY are always enabled. In gtk+, we must support all flags combinations.

Implemented in GtkSourceView.