From 2d4d55cfedccfa80d283592af349e93d0968f58e Mon Sep 17 00:00:00 2001
From: Antti Ajanki <antti.ajanki@iki.fi>
Date: Fri, 9 Aug 2013 16:28:16 +0300
Subject: Assume that incoming HTML is UTF-8 encoded until a proper fix is
 implemented

---
 src/libwebvi/linkextractor.c | 1 +
 src/libwebvi/request.c       | 2 ++
 2 files changed, 3 insertions(+)

diff --git a/src/libwebvi/linkextractor.c b/src/libwebvi/linkextractor.c
index b0e4270..1691a46 100644
--- a/src/libwebvi/linkextractor.c
+++ b/src/libwebvi/linkextractor.c
@@ -63,6 +63,7 @@ GPtrArray *link_extractor_get_links(LinkExtractor *self) {
   tidyOptSetInt(tdoc, TidyWrapLen, 4096);
   tidyBufInit(&errbuf);
   tidySetErrorBuffer(tdoc, &errbuf);
+  tidySetInCharEncoding(tdoc, "utf8");
 
   err = tidyParseBuffer(tdoc, &self->html_buffer);
   if (err >= 0) {
diff --git a/src/libwebvi/request.c b/src/libwebvi/request.c
index fc2b635..baca467 100644
--- a/src/libwebvi/request.c
+++ b/src/libwebvi/request.c
@@ -133,6 +133,8 @@ PipeComponent *build_and_start_menu_pipe(const WebviRequest *self) {
   pipe_component_set_next(p2, p3);
   pipe_downloader_start(p1);
 
+  // TODO: the downloaded HTML should be converted into UTF-8
+
   return (PipeComponent *)p1;
 }
 
-- 
cgit v1.2.3