initial state based parser for gemini

martinrotter · Dec 18, 2024 · 65bb665 · 65bb665
1 parent bdc71f3
commit 65bb665
Show file tree

Hide file tree

Showing 2 changed files with 102 additions and 20 deletions.
diff --git a/src/librssguard/network-web/gemini/geminiparser.cpp b/src/librssguard/network-web/gemini/geminiparser.cpp
@@ -10,50 +10,119 @@ QString GeminiParser::geminiToHtml(const QByteArray& gemini_data) {
   QString gemini_hypertext =
     QString::fromUtf8(gemini_data).replace(QSL("\r\n"), QSL("\n")).replace(QSL("\r"), QSL("\n"));
   QStringList lines = gemini_hypertext.split(QL1C('\n'));
-  bool normal_mode = true;
+  mode = State::Normal;
 
   static QRegularExpression exp_link(R"(^=>\s+([^\s]+)(?:\s+(\w.+))?$)");
   static QRegularExpression exp_heading(R"(^(#{1,6})\s+(.+)$)");
   static QRegularExpression exp_list(R"(^\*\s(.+)$)");
   static QRegularExpression exp_quote(R"((?:^>$|^>\s?(.+)$))");
   static QRegularExpression exp_pre(R"(^```.*$)");
-  static QRegularExpression exp_text(R"()");
 
   QRegularExpressionMatch mtch;
   QString title;
 
   for (const QString& line : lines) {
     if ((mtch = exp_pre.match(line)).hasMatch()) {
-      normal_mode = !normal_mode;
+      // Begin or end PRE block.
+      switch (mode) {
+        case State::Pre:
+          // Ending of a PRE block.
+          html += endBlock(State::Normal);
+          break;
+
+        default:
+          // Beginning of a PRE block.
+          html += endBlock(State::Normal);
+          html += beginBlock(State::Pre);
+          break;
+      }
       continue;
     }
 
-    if (normal_mode) {
+    if (mode != State::Pre) {
       if ((mtch = exp_link.match(line)).hasMatch()) {
+        html += endBlock(State::Normal);
         html += parseLink(mtch);
       }
       else if ((mtch = exp_heading.match(line)).hasMatch()) {
+        html += endBlock(State::Normal);
         html += parseHeading(mtch, title.isEmpty() ? &title : nullptr);
       }
       else if ((mtch = exp_list.match(line)).hasMatch()) {
+        html += beginBlock(State::List);
         html += parseList(mtch);
       }
       else if ((mtch = exp_quote.match(line)).hasMatch()) {
+        html += beginBlock(State::Quote);
         html += parseQuote(mtch);
       }
       else {
+        html += endBlock(State::Normal);
         html += parseTextInNormalMode(line);
       }
     }
     else {
+      // Add new line in PRE mode.
       html += parseInPreMode(line);
     }
   }
 
+  html += endBlock(State::Normal);
+
+  // IOFactory::writeFile("aa", html.toUtf8());
+
   return QSL("<html>"
              "<head><title>%1</title></head>"
              "<body>%2</body>"
-             "</html>").arg(title, html);
+             "</html>")
+    .arg(title, html);
+}
+
+QString GeminiParser::beginBlock(State new_mode) {
+  if (new_mode != mode) {
+    mode = new_mode;
+
+    switch (new_mode) {
+      case State::List:
+        return "<ul>\n";
+
+      case State::Quote:
+        return "<div align=\"center\" style=\""
+               "background-color: #E1E5EE;"
+               "font-style: italic;"
+               "margin-left: 20px;"
+               "margin-right: 20px;\">\n";
+
+      case State::Pre:
+        return "<pre style=\"background-color: #E1E5EE;\">\n";
+    }
+  }
+
+  return QString();
+}
+
+QString GeminiParser::endBlock(State new_mode) {
+  QString to_return;
+
+  if (new_mode != mode) {
+    switch (mode) {
+      case State::List:
+        to_return = "</ul>\n";
+        break;
+
+      case State::Quote:
+        to_return = "</div>\n";
+        break;
+
+      case State::Pre:
+        to_return = "</pre>\n";
+        break;
+    }
+
+    mode = new_mode;
+  }
+
+  return to_return;
 }
 
 QString GeminiParser::parseLink(const QRegularExpressionMatch& mtch) const {
@@ -75,29 +144,22 @@ QString GeminiParser::parseHeading(const QRegularExpressionMatch& mtch, QString*
   return QSL("<h%1>%2</h%1>\n").arg(QString::number(level), header);
 }
 
-QString GeminiParser::parseQuote(const QRegularExpressionMatch &mtch) const {
+QString GeminiParser::parseQuote(const QRegularExpressionMatch& mtch) const {
   QString text = mtch.captured(1);
 
-  return QSL("<p align=\"center\" style=\""
-             "background-color: #E1E5EE;"
-             "font-style: italic;"
-             "margin-left: 20px;"
-             "margin-right: 20px;"
-             "\">%1</p>\n").arg(text.isEmpty() ? QString() : QSL("“%1”").arg(text));
+  return QSL("<div>%1</div>\n").arg(text.isEmpty() ? QString() : QSL("“%1”").arg(text));
 }
 
-QString GeminiParser::parseList(const QRegularExpressionMatch &mtch) const {
+QString GeminiParser::parseList(const QRegularExpressionMatch& mtch) const {
   QString text = mtch.captured(1);
 
-  return QSL("<p style=\""
-             "margin-left: 20px;"
-             "\">• %1</p>\n").arg(text);
+  return QSL("<li>%1</li>\n").arg(text);
 }
 
-QString GeminiParser::parseTextInNormalMode(const QString &line) const{
+QString GeminiParser::parseTextInNormalMode(const QString& line) const {
   return QSL("<p>%1</p>\n").arg(line);
 }
 
 QString GeminiParser::parseInPreMode(const QString& line) const {
-  return QSL("<pre>%1</pre>\n").arg(line);
+  return QSL("%1\n").arg(line.toHtmlEscaped());
 }
diff --git a/src/librssguard/network-web/gemini/geminiparser.h b/src/librssguard/network-web/gemini/geminiparser.h
@@ -3,20 +3,40 @@
 #ifndef GEMINIPARSER_H
 #define GEMINIPARSER_H
 
-#include <QString>
 #include <QRegularExpressionMatch>
+#include <QString>
 
 class GeminiParser {
   public:
     QString geminiToHtml(const QByteArray& gemini_data);
 
   private:
+    enum class State {
+      // Regular state.
+      Normal,
+
+      // Inside list.
+      List,
+
+      // Inside quote.
+      Quote,
+
+      // Inside PRE.
+      Pre
+    };
+
     QString parseLink(const QRegularExpressionMatch& mtch) const;
-    QString parseHeading(const QRegularExpressionMatch& mtch, QString *clean_header = nullptr) const;
+    QString parseHeading(const QRegularExpressionMatch& mtch, QString* clean_header = nullptr) const;
     QString parseQuote(const QRegularExpressionMatch& mtch) const;
     QString parseList(const QRegularExpressionMatch& mtch) const;
     QString parseTextInNormalMode(const QString& line) const;
     QString parseInPreMode(const QString& line) const;
+
+    QString beginBlock(State new_mode);
+    QString endBlock(State new_mode);
+
+  private:
+    State mode;
 };
 
 #endif // GEMINIPARSER_H