]+>"_s);
+ static const QRegularExpression regex_html_tag_span_trans(u""_s);
+ static const QRegularExpression regex_html_tag_div_ellipsis(u""_s);
+ static const QRegularExpression regex_html_tag_span_contribs(u"
]*>[^<]*"_s);
+ static const QRegularExpression regex_html_tag_div_bio(u"
]*>.*?
"_s);
+ static const QRegularExpression regex_html_tag_h2(u"
]*>[^<]*
"_s);
+ static const QList
regex_removes{ regex_html_tag_span_trans, regex_html_tag_div_ellipsis, regex_html_tag_span_contribs, regex_html_tag_div_bio, regex_html_tag_h2 };
+
+ const QString lyrics = HtmlLyricsProvider::ParseLyricsFromHTML(QString::fromUtf8(data), start_tag, end_tag, lyrics_start, true, regex_removes);
if (!lyrics.isEmpty()) {
LyricsSearchResult result(lyrics);
result.artist = lyric.artist;
@@ -404,3 +416,17 @@ void GeniusLyricsProvider::EndSearch(const int id, const LyricsSearchRequest &re
Q_EMIT SearchFinished(id, results);
}
+
+bool GeniusLyricsProvider::StartsOrEndsMatch(QString s, QString t) {
+
+ constexpr Qt::CaseSensitivity cs = Qt::CaseInsensitive;
+
+ static const QRegularExpression puncts_regex(u"[!,.:;]"_s);
+ static const QRegularExpression quotes_regex(u"[’‘´`]"_s);
+
+ s.remove(puncts_regex).replace(quotes_regex, u"'"_s);
+ t.remove(puncts_regex).replace(quotes_regex, u"'"_s);
+
+ return (s.compare(t, cs) == 0 && !s.isEmpty()) || (!s.isEmpty() && !t.isEmpty() && (s.startsWith(t, cs) || t.startsWith(s, cs) || s.endsWith(t, cs) || t.endsWith(s, cs)));
+
+}
diff --git a/src/lyrics/geniuslyricsprovider.h b/src/lyrics/geniuslyricsprovider.h
index 949067f03..47771c519 100644
--- a/src/lyrics/geniuslyricsprovider.h
+++ b/src/lyrics/geniuslyricsprovider.h
@@ -79,6 +79,9 @@ class GeniusLyricsProvider : public JsonLyricsProvider {
void HandleSearchReply(QNetworkReply *reply, const int id);
void HandleLyricReply(QNetworkReply *reply, const int search_id, const QUrl &url);
+ private:
+ static bool StartsOrEndsMatch(QString s, QString t);
+
private:
OAuthenticator *oauth_;
mutable QMutex mutex_access_token_;
diff --git a/src/lyrics/htmllyricsprovider.cpp b/src/lyrics/htmllyricsprovider.cpp
index 4402415e2..1b2cbe165 100644
--- a/src/lyrics/htmllyricsprovider.cpp
+++ b/src/lyrics/htmllyricsprovider.cpp
@@ -109,7 +109,7 @@ void HtmlLyricsProvider::HandleLyricsReply(QNetworkReply *reply, const int id, c
}
-QString HtmlLyricsProvider::ParseLyricsFromHTML(const QString &content, const QRegularExpression &start_tag, const QRegularExpression &end_tag, const QRegularExpression &lyrics_start, const bool multiple) {
+QString HtmlLyricsProvider::ParseLyricsFromHTML(const QString &content, const QRegularExpression &start_tag, const QRegularExpression &end_tag, const QRegularExpression &lyrics_start, const bool multiple, const QList ®ex_removes) {
Q_ASSERT(QThread::currentThread() != qApp->thread());
@@ -153,30 +153,31 @@ QString HtmlLyricsProvider::ParseLyricsFromHTML(const QString &content, const QR
if (!lyrics.isEmpty()) {
lyrics.append(u'\n');
}
- static const QRegularExpression regex_html_tag_a(u"]*>[^<]*"_s);
- static const QRegularExpression regex_html_tag_script(u""_s);
- static const QRegularExpression regex_html_tag_div(u"]*>×
"_s);
- static const QRegularExpression regex_html_tag_br(u"]*>"_s);
- static const QRegularExpression regex_html_tag_p_close(u""_s);
- static const QRegularExpression regex_html_tags(u"<[^>]*>"_s);
- lyrics.append(content.mid(start_lyrics_idx, end_lyrics_idx - start_lyrics_idx)
- .remove(u'\r')
- .remove(u'\n')
- .remove(regex_html_tag_a)
- .remove(regex_html_tag_script)
- .remove(regex_html_tag_div)
- .replace(regex_html_tag_br, u"\n"_s)
- .replace(regex_html_tag_p_close, u"\n\n"_s)
- .remove(regex_html_tags)
- .trimmed());
- }
- else {
- start_idx = -1;
+ lyrics.append(content.mid(start_lyrics_idx, end_lyrics_idx - start_lyrics_idx).remove(u'\r').remove(u'\n'));
}
}
while (start_idx > 0 && multiple);
+ for (auto it = regex_removes.cbegin(); it != regex_removes.cend(); it++) {
+ lyrics.remove(*it);
+ }
+ static const QRegularExpression regex_html_tag_a(u"]*>[^<]*"_s);
+ static const QRegularExpression regex_html_tag_script(u""_s);
+ static const QRegularExpression regex_html_tag_div(u"]*>×
"_s);
+ static const QRegularExpression regex_html_tag_br(u"]*>"_s);
+ static const QRegularExpression regex_html_tag_p_close(u""_s);
+ static const QRegularExpression regex_html_tags(u"<[^>]*>"_s);
+ static const QRegularExpression regex_newlines_squash(u"\\n{3,}"_s);
+ lyrics.remove(regex_html_tag_a)
+ .remove(regex_html_tag_script)
+ .remove(regex_html_tag_div)
+ .replace(regex_html_tag_br, u"\n"_s)
+ .replace(regex_html_tag_p_close, u"\n\n"_s)
+ .remove(regex_html_tags)
+ .replace(regex_newlines_squash, u"\n\n"_s);
+ lyrics = lyrics.trimmed();
+
if (lyrics.length() > 6000 || lyrics.contains("there are no lyrics to"_L1, Qt::CaseInsensitive)) {
return QString();
}
diff --git a/src/lyrics/htmllyricsprovider.h b/src/lyrics/htmllyricsprovider.h
index 4022850dd..b8e31148f 100644
--- a/src/lyrics/htmllyricsprovider.h
+++ b/src/lyrics/htmllyricsprovider.h
@@ -41,7 +41,7 @@ class HtmlLyricsProvider : public LyricsProvider {
virtual bool StartSearchAsync(const int id, const LyricsSearchRequest &request) override;
- static QString ParseLyricsFromHTML(const QString &content, const QRegularExpression &start_tag, const QRegularExpression &end_tag, const QRegularExpression &lyrics_start, const bool multiple);
+ static QString ParseLyricsFromHTML(const QString &content, const QRegularExpression &start_tag, const QRegularExpression &end_tag, const QRegularExpression &lyrics_start, const bool multiple, const QList ®ex_removes = {});
protected:
virtual QUrl Url(const LyricsSearchRequest &request) = 0;