Funkwhale/strawberry-audio-player-windows-mac-linux
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

LyricsProvider: Parse multiple sections of the same tag

Browse Source
This commit is contained in:
Jonas Kvinge 2022-10-16 01:27:54 +02:00
parent 43c14ae71b
commit bf904a6afa
3 changed files with 49 additions and 31 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/lyrics/geniuslyricsprovider.cpp
View File

@ -480,9 +480,9 @@ void GeniusLyricsProvider::HandleLyricReply(QNetworkReply *reply, const int sear
}
QString content = QString::fromUtf8(data);
QString lyrics = ParseLyricsFromHTML(content, QRegularExpression("<div[^>]*>"), QRegularExpression("<\\/div>"), QRegularExpression("<div data-lyrics-container=[^>]+>"));
QString lyrics = ParseLyricsFromHTML(content, QRegularExpression("<div[^>]*>"), QRegularExpression("<\\/div>"), QRegularExpression("<div data-lyrics-container=[^>]+>"), true);
if (lyrics.isEmpty()) {
lyrics = ParseLyricsFromHTML(content, QRegularExpression("<div[^>]*>"), QRegularExpression("<\\/div>"), QRegularExpression("<div class=\"lyrics\">"));
lyrics = ParseLyricsFromHTML(content, QRegularExpression("<div[^>]*>"), QRegularExpression("<\\/div>"), QRegularExpression("<div class=\"lyrics\">"), true);
}
if (!lyrics.isEmpty()) {

74
src/lyrics/lyricsprovider.cpp
View File

@ -30,40 +30,58 @@
LyricsProvider::LyricsProvider(const QString &name, const bool enabled, const bool authentication_required, NetworkAccessManager *network, QObject *parent)
: QObject(parent), network_(network), name_(name), enabled_(enabled), order_(0), authentication_required_(authentication_required) {}
QString LyricsProvider::ParseLyricsFromHTML(const QString &content, const QRegularExpression &start_tag, const QRegularExpression &end_tag, const QRegularExpression &lyrics_start) {
QString LyricsProvider::ParseLyricsFromHTML(const QString &content, const QRegularExpression &start_tag, const QRegularExpression &end_tag, const QRegularExpression &lyrics_start, const bool multiple) {
QRegularExpressionMatch rematch = lyrics_start.match(content);
if (!rematch.hasMatch()) return QString();
QString lyrics;
qint64 start_idx = 0;
const qint64 start_lyrics_tag_idx = rematch.capturedEnd();
// Find the index of the end tag.
qint64 start_tag_idx = 0;
qint64 end_tag_idx = 0;
qint64 idx = start_lyrics_tag_idx;
int tags = 1;
do {
QRegularExpressionMatch rematch_start_tag = QRegularExpression(start_tag).match(content, idx);
start_tag_idx = rematch_start_tag.hasMatch() ? rematch_start_tag.capturedStart() : -1;
QRegularExpressionMatch rematch_end_tag = QRegularExpression(end_tag).match(content, idx);
end_tag_idx = rematch_end_tag.hasMatch() ? rematch_end_tag.capturedStart() : -1;
if (rematch_start_tag.hasMatch() && start_tag_idx <= end_tag_idx) {
++tags;
idx = start_tag_idx + rematch_start_tag.capturedLength();
QRegularExpressionMatch rematch = lyrics_start.match(content, start_idx);
if (!rematch.hasMatch()) break;
const qint64 start_lyrics_tag_idx = rematch.capturedEnd();
// Find the index of the end tag.
qint64 start_tag_idx = 0;
qint64 end_tag_idx = 0;
qint64 end_tag_length = 0;
qint64 idx = start_lyrics_tag_idx;
int tags = 1;
do {
QRegularExpressionMatch rematch_start_tag = QRegularExpression(start_tag).match(content, idx);
start_tag_idx = rematch_start_tag.hasMatch() ? rematch_start_tag.capturedStart() : -1;
QRegularExpressionMatch rematch_end_tag = QRegularExpression(end_tag).match(content, idx);
if (rematch_end_tag.hasMatch()) {
end_tag_idx = rematch_end_tag.capturedStart();
end_tag_length = rematch_end_tag.capturedLength();
}
else {
end_tag_idx = -1;
end_tag_length = 0;
}
if (rematch_start_tag.hasMatch() && start_tag_idx <= end_tag_idx) {
++tags;
idx = start_tag_idx + rematch_start_tag.capturedLength();
}
else if (rematch_end_tag.hasMatch()) {
--tags;
idx = end_tag_idx + rematch_end_tag.capturedLength();
}
}
else if (rematch_end_tag.hasMatch()) {
--tags;
idx = end_tag_idx + rematch_end_tag.capturedLength();
while (tags > 0 || end_tag_idx >= start_tag_idx);
start_idx = end_tag_idx + end_tag_length;
if (end_tag_idx > 0 || start_lyrics_tag_idx < end_tag_idx) {
lyrics.append(content.mid(start_lyrics_tag_idx, end_tag_idx - start_lyrics_tag_idx)
.replace(QRegularExpression("<br[^>]+>"), "\n")
.remove(QRegularExpression("<[^>]*>"))
.trimmed());
}
}
while (tags > 0 || end_tag_idx >= start_tag_idx);
if (end_tag_idx <= 0 || start_lyrics_tag_idx >= end_tag_idx) return QString();
QString lyrics = content.mid(start_lyrics_tag_idx, end_tag_idx - start_lyrics_tag_idx)
.replace(QRegularExpression("<br[^>]+>"), "\n")
.remove(QRegularExpression("<[^>]*>"))
.trimmed();
while (start_idx > 0 && multiple);
if (lyrics.length() > 6000 || lyrics.contains("there are no lyrics to", Qt::CaseInsensitive)) {
return QString();

2
src/lyrics/lyricsprovider.h
View File

@ -57,7 +57,7 @@ class LyricsProvider : public QObject {
virtual void Error(const QString &error, const QVariant &debug = QVariant()) = 0;
protected:
QString ParseLyricsFromHTML(const QString &content, const QRegularExpression &start_tag, const QRegularExpression &end_tag, const QRegularExpression &lyrics_start);
QString ParseLyricsFromHTML(const QString &content, const QRegularExpression &start_tag, const QRegularExpression &end_tag, const QRegularExpression &lyrics_start, const bool multiple);
signals:
void AuthenticationComplete(bool, QStringList = QStringList());