Salastil/ncmpcpp
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

lyricwiki fetcher: deal with utf8 characters properly

Browse Source
This commit is contained in:
Andrzej Rybczak
2010-08-09 13:46:24 +02:00
parent 26b7218f8e
commit 2eddb37497
2 changed files with 16 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
src/lyrics_fetcher.cpp
View File

@@ -120,7 +120,7 @@ LyricsFetcher::Result LyricwikiFetcher::fetch(const std::string &artist, const s
Replace(data, "<br />", "\n");
result.second = unescape(data);
result.second = unescapeHtmlUtf8(data);
result.first = true;
}
return result;
@@ -131,7 +131,7 @@ bool LyricwikiFetcher::notLyrics(const std::string &data)
return data.find("action=edit") != std::string::npos;
}
std::string LyricwikiFetcher::unescape(const std::string &data)
std::string LyricwikiFetcher::unescapeHtmlUtf8(const std::string &data)
{
std::string result;
for (size_t i = 0, j; i < data.length(); ++i)
@@ -139,7 +139,19 @@ std::string LyricwikiFetcher::unescape(const std::string &data)
if (data[i] == '&' && data[i+1] == '#' && (j = data.find(';', i)) != std::string::npos)
{
int n = atoi(&data.c_str()[i+2]);
result += char(n);
if (n >= 0x800)
{
result += (0xe0 | ((n >> 12) & 0x0f));
result += (0x80 | ((n >> 6) & 0x3f));
result += (0x80 | (n & 0x3f));
}
if (n >= 0x80)
{
result += (0xc0 | ((n >> 6) & 0x1f));
result += (0x80 | (n & 0x3f));
}
else
result += n;
i = j;
}
else

2
src/lyrics_fetcher.h
View File

@@ -70,7 +70,7 @@ struct LyricwikiFetcher : public LyricsFetcher
virtual bool notLyrics(const std::string &data);
private:
std::string unescape(const std::string &data);
std::string unescapeHtmlUtf8(const std::string &data);
};
struct LyricsflyFetcher : public LyricsFetcher