LyricsProvider: Add function for parsing lyrics from HTML

src/lyrics/lyricsprovider.cpp

@@ -21,9 +21,55 @@
 
 #include <QObject>
 #include <QString>
+#include <QRegularExpression>
 
+#include "core/utilities.h"
 #include "core/networkaccessmanager.h"
 #include "lyricsprovider.h"
 
 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) {
+
+  QRegularExpressionMatch rematch = lyrics_start.match(content);
+  if (!rematch.hasMatch()) return QString();
+
+  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();
+    }
+    else if (rematch_end_tag.hasMatch()) {
+      --tags;
+      idx = end_tag_idx + rematch_end_tag.capturedLength();
+    }
+  }
+  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();
+
+  if (lyrics.length() > 6000 || lyrics.contains("there are no lyrics to", Qt::CaseInsensitive)) {
+    return QString();
+  }
+
+  return Utilities::DecodeHtmlEntities(lyrics);
+  
+}
+

src/lyrics/lyricsprovider.h

@@ -28,6 +28,7 @@
 #include <QList>
 #include <QVariant>
 #include <QString>
+#include <QRegularExpression>
 
 #include "lyricsfetcher.h"
 
@@ -55,6 +56,9 @@ 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);
+
  signals:
   void AuthenticationComplete(bool, QStringList = QStringList());
   void AuthenticationSuccess();