refactor: improved HTML-formatting mechanism for text messages

Added class provides text font formatting with HTML font tags
Supported stuff:
  - nested formatting;
  - several text pieces formatted with the same formatting style at the
    one message;
  - styling applies only if non-whitespace symbol follows opening
    formatting symbol and also non-whitespace symbol preceds closing
    symbol;
  - only multiline code font formatting supports new line inside of
    message text.
Also fix #3804

CMakeLists.txt

@@ -188,6 +188,8 @@ set(${PROJECT_NAME}_SOURCES
   src/chatlog/documentcache.h
   src/chatlog/pixmapcache.cpp
   src/chatlog/pixmapcache.h
+  src/chatlog/textformatter.cpp
+  src/chatlog/textformatter.h
   src/core/coreav.cpp
   src/core/coreav.h
   src/core/core.cpp

qtox.pro

@@ -436,6 +436,7 @@ HEADERS  += \
     src/chatlog/customtextdocument.h \
     src/chatlog/documentcache.h \
     src/chatlog/pixmapcache.h \
+    src/chatlog/textformatter.h \
     src/core/core.h \
     src/core/coreav.h \
     src/core/coredefines.h \
@@ -554,6 +555,7 @@ SOURCES += \
     src/chatlog/customtextdocument.cpp\
     src/chatlog/documentcache.cpp \
     src/chatlog/pixmapcache.cpp \
+    src/chatlog/textformatter.cpp \
     src/core/core.cpp \
     src/core/coreav.cpp \
     src/core/corefile.cpp \

src/chatlog/chatmessage.cpp

@@ -19,6 +19,7 @@
 
 #include "chatmessage.h"
 #include "chatlinecontentproxy.h"
+#include "textformatter.h"
 #include "content/text.h"
 #include "content/timestamp.h"
 #include "content/spinner.h"
@@ -56,8 +57,12 @@ ChatMessage::Ptr ChatMessage::createChatMessage(const QString &sender, const QSt
     text = detectQuotes(detectAnchors(text), type);
 
     //text styling
-    if (Settings::getInstance().getStylePreference() != Settings::StyleType::NONE)
+    auto styleType = Settings::getInstance().getStylePreference();
-        text = detectStyle(text);
+    if (styleType != Settings::StyleType::NONE) {
+        TextFormatter tf = TextFormatter(text);
+        text = tf.applyStyling(styleType == Settings::StyleType::WITHOUT_CHARS);
+    }
+
 
     switch(type)
     {
@@ -199,71 +204,6 @@ void ChatMessage::hideDate()
         c->hide();
 }
 
-QString ChatMessage::detectStyle(const QString &str)
-{
-    QString out = str;
-
-    // Create regex for text styling syntax
-    QRegExp exp("(\\*)([^\\*]{2,})(\\*)"            		// Bold         *text*
-                "|(\\*\\*)([^\\*\\*]{2,})(\\*\\*)"  		// Bold         **text**
-                "|(\\/)([^\\/]{2,})(\\/)"           		// Italics      /text/
-                "|(\\/\\/)([^\\/\\/]{2,})(\\/\\/)"  		// Italics      //text//
-                "|(\\_)([^\\_]{2,})(\\_)"           		// Underline    _text_
-                "|(\\_\\_)([^\\_\\_]{2,})(\\_\\_)"  		// Underline    __text__
-                "|(\\~)([^\\~]{2,})(\\~)"           		// Strike       ~text~
-                "|(\\~\\~)([^\\~\\~]{2,})(\\~\\~)"      	// Strike       ~~text~~
-                "|(\\`)([^\\`]{2,})(\\`)"                   // Codeblock    `text`
-                "|(\\`\\`\\`)([^\\`\\`\\`]{2,})(\\`\\`\\`)" // Codeblock    ```\ntext\n```
-                );
-
-    int offset = 0;
-    while ((offset = exp.indexIn(out, offset)) != -1)
-    {
-        QString snipCheck = out.mid(offset-1,exp.cap(0).length()+2);
-        QString snippet = exp.cap(0).trimmed();
-        QString htmledSnippet;
-
-        // Only parse if surrounded by spaces, newline(s) and/or beginning/end of line
-        if ((snipCheck.startsWith(' ') || snipCheck.startsWith('>') || offset == 0)
-            && ((snipCheck.endsWith(' ') || snipCheck.endsWith('<')) || offset + snippet.length() == out.length()))
-        {
-            int mul = 0; // Determines how many characters to strip from text
-            // Set mul depending on styleownPreference
-            if (Settings::getInstance().getStylePreference() == Settings::StyleType::WITHOUT_CHARS)
-                mul = 2;
-
-            // Match captured string to corresponding style format
-            if (exp.cap(1) == "*" && snippet.length() > 2) // Bold *text*
-                htmledSnippet = QString("<b>%1</b>").arg(snippet.mid(mul/2,snippet.length()-mul));
-            else if (exp.cap(4) == "**" && snippet.length() > 4) // Bold **text**
-                htmledSnippet = QString("<b>%1</b>").arg(snippet.mid(mul,snippet.length()-2*mul));
-            else if (exp.cap(7) == "/" && snippet.length() > 2) // Italics /text/
-                htmledSnippet = QString("<i>%1</i>").arg(snippet.mid(mul/2,snippet.length()-mul));
-            else if (exp.cap(10) == "//" && snippet.length() > 4) // Italics //text//
-                htmledSnippet = QString("<i>%1</i>").arg(snippet.mid(mul,snippet.length()-2*mul));
-            else if (exp.cap(13) == "_"&& snippet.length() > 2) // Underline _text_
-                htmledSnippet = QString("<u>%1</u>").arg(snippet.mid(mul/2,snippet.length()-mul));
-            else if (exp.cap(16) == "__" && snippet.length() > 4) // Underline __text__
-                htmledSnippet = QString("<u>%1</u>").arg(snippet.mid(mul,snippet.length()-2*mul));
-            else if (exp.cap(19) == "~" && snippet.length() > 2) // Strike ~text~
-                htmledSnippet = QString("<s>%1</s>").arg(snippet.mid(mul/2,snippet.length()-mul));
-            else if (exp.cap(22) == "~~" && snippet.length() > 4) // Strike ~~text~~
-                htmledSnippet = QString("<s>%1</s>").arg(snippet.mid(mul,snippet.length()-2*mul));
-            else if (exp.cap(25) == "`" && snippet.length() > 2) // Codeblock `text`
-                htmledSnippet = QString("<font color=#595959><code>%1</code></font>").arg(snippet.mid(mul/2,snippet.length()-mul));
-            else if (exp.cap(28) == "```" && snippet.length() > 6) // Codeblock ```text```
-                htmledSnippet = QString("<font color=#595959><code>%1</code></font>").arg(snippet.mid(4*mul,snippet.length()-8*mul));
-            else
-                htmledSnippet = snippet;
-            out.replace(offset, exp.cap().length(), htmledSnippet);
-            offset += htmledSnippet.length();
-        } else
-            offset += snippet.length();
-    }
-
-    return out;
-}
-
 QString ChatMessage::detectAnchors(const QString &str)
 {
     QString out = str;
@@ -324,7 +264,7 @@ QString ChatMessage::detectQuotes(const QString& str, MessageType type)
         }
 
         if (i < messageLines.size() - 1)
-            quotedText += "<br/>";
+            quotedText += '\n';
     }
 
     return quotedText;

src/chatlog/chatmessage.h

@@ -61,7 +61,6 @@ public:
     void hideDate();
 
 protected:
-    static QString detectStyle(const QString& str);
     static QString detectAnchors(const QString& str);
     static QString detectQuotes(const QString& str, MessageType type);
     static QString wrapDiv(const QString& str, const QString& div);

src/chatlog/textformatter.cpp

@@ -0,0 +1,124 @@
+#include "textformatter.h"
+
+#include <QMap>
+#include <QPair>
+#include <QRegularExpression>
+#include <QVector>
+
+enum TextStyle {
+    BOLD = 0,
+    ITALIC,
+    UNDERLINE,
+    STRIKE,
+    CODE
+};
+
+static const QString SINGLE_SIGN_FORMATTING_TEXT_FONT_PATTERN = QStringLiteral("(?:(^|[^\\%1]))(\\%1[^\\s\\%1])([^\\%1\\n]+)([^\\s\\%1]\\%1)(?:($|[^\\%1]))");
+
+//Pattern for escaping slashes from inserted HTML tags
+static const QString SINGLE_SLASH_FORMATTING_TEXT_FONT_PATTERN = QStringLiteral("(?:(^|[^/<]))(\\/[^\\s/])([^\\n/]+)([^<\\s/]\\/)(?:($|[^/]))");
+
+static const QString DOUBLE_SIGN_FORMATTING_TEXT_FONT_PATTERN = QStringLiteral("(?:(^|[^\\%1]))([\\%1]{2}[^\\s\\%1])([^\\n]+)([^\\s\\%1][\\%1]{2})(?:($|[^\\%1]))");
+
+static const QString MULTILINE_CODE_FORMATTING_TEXT_FONT_PATTERN = QStringLiteral("(?:(^|[^`]))([`]{3})((\\n|.)+)([`]{3})(?:($|[^`]))");
+
+// Items in vector associated with TextStyle values respectively. Do NOT change this order
+static const QVector<QString> fontStylePatterns {
+    QStringLiteral("<b>%1</b>"),
+    QStringLiteral("<i>%1</i>"),
+    QStringLiteral("<u>%1</u>"),
+    QStringLiteral("<s>%1</s>"),
+    QStringLiteral("<font color=#595959><code>%1</code></font>")
+};
+
+// Unfortunately, can't use simple QMap because ordered applying of styles is required
+static const QVector<QPair<QRegularExpression, QString>> textPatternStyle {
+    { QRegularExpression(SINGLE_SLASH_FORMATTING_TEXT_FONT_PATTERN), fontStylePatterns[ITALIC] },
+    { QRegularExpression(SINGLE_SIGN_FORMATTING_TEXT_FONT_PATTERN.arg('*')), fontStylePatterns[BOLD] },
+    { QRegularExpression(SINGLE_SIGN_FORMATTING_TEXT_FONT_PATTERN.arg('_')), fontStylePatterns[UNDERLINE] },
+    { QRegularExpression(SINGLE_SIGN_FORMATTING_TEXT_FONT_PATTERN.arg('~')), fontStylePatterns[STRIKE] },
+    { QRegularExpression(SINGLE_SIGN_FORMATTING_TEXT_FONT_PATTERN.arg('`')), fontStylePatterns[CODE] },
+    { QRegularExpression(DOUBLE_SIGN_FORMATTING_TEXT_FONT_PATTERN.arg('*')), fontStylePatterns[BOLD] },
+    { QRegularExpression(DOUBLE_SIGN_FORMATTING_TEXT_FONT_PATTERN.arg('/')), fontStylePatterns[ITALIC] },
+    { QRegularExpression(DOUBLE_SIGN_FORMATTING_TEXT_FONT_PATTERN.arg('_')), fontStylePatterns[UNDERLINE] },
+    { QRegularExpression(DOUBLE_SIGN_FORMATTING_TEXT_FONT_PATTERN.arg('~')), fontStylePatterns[STRIKE] },
+    { QRegularExpression(MULTILINE_CODE_FORMATTING_TEXT_FONT_PATTERN), fontStylePatterns[CODE] }
+};
+
+TextFormatter::TextFormatter(const QString &str)
+    : sourceString(str) {}
+
+/**
+ * @brief TextFormatter::patternEscapeSignsCount Counts equal symbols at the beginning of the string
+ * @param str Source string
+ * @return Amount of equal symbols at the beginning of the string
+ */
+int TextFormatter::patternEscapeSignsCount(const QString &str) {
+    QChar escapeSign = str.at(0);
+    int result = 0;
+    for (const QChar c : str) {
+        if (c == escapeSign)
+            ++result;
+        else
+            break;
+    }
+    return result;
+}
+
+/**
+ * @brief TextFormatter::getCapturedLength Get length of string captured by subexpression with appropriate checks
+ * @param match Global match of QRegularExpression
+ * @param exprNumber Number of subexpression
+ * @return Length of captured string. If nothing was captured, returns 0
+ */
+int TextFormatter::getCapturedLength(const QRegularExpressionMatch &match, const int exprNumber) {
+    QString captured = match.captured(exprNumber);
+    return captured.isNull() || captured.isEmpty() ? 0 : captured.length();
+}
+
+/**
+ * @brief TextFormatter::applyHtmlFontStyling Applies styles to the font of text that was passed to the constructor
+ * @param dontShowFormattingSymbols True, if it does not suppose to include formatting symbols into resulting string
+ * @return Source text with styled font
+ */
+QString TextFormatter::applyHtmlFontStyling(bool dontShowFormattingSymbols){
+    QString out = sourceString;
+    int choppingSignsCountMultiplier = dontShowFormattingSymbols ? 0 : 1;
+
+    for (QPair<QRegularExpression, QString> pair : textPatternStyle) {
+        QRegularExpression exp = pair.first;
+        QRegularExpressionMatchIterator matchesIterator = exp.globalMatch(out);
+        int insertedTagSymbolsCount = 0;
+
+        while (matchesIterator.hasNext()) {
+            QRegularExpressionMatch match = matchesIterator.next();
+
+            // Regular expressions may capture one redundant symbol from both sides because of extra check, so we don't need to handle them
+            int firstCheckResultLength = getCapturedLength(match, 1);
+            int matchStart = match.capturedStart() + firstCheckResultLength + insertedTagSymbolsCount;
+            int matchLength = match.capturedLength() - firstCheckResultLength - getCapturedLength(match, exp.captureCount());
+
+            int choppingSignsCount = patternEscapeSignsCount(out.mid(matchStart, matchLength));
+            int textStart = matchStart + choppingSignsCount;
+            int textLength = matchLength - choppingSignsCount * 2;
+
+            QString styledText = pair.second.arg(out.mid(textStart, textLength));
+
+            textStart = matchStart + choppingSignsCount * choppingSignsCountMultiplier;
+            textLength = matchLength - choppingSignsCount * choppingSignsCountMultiplier * 2;
+
+            out.replace(textStart, textLength, styledText);
+            insertedTagSymbolsCount += pair.second.length() - 2 - choppingSignsCount * (1 - choppingSignsCountMultiplier) * 2;
+        }
+    }
+    return out;
+}
+
+/**
+ * @brief TextFormatter::applyStyling Applies all styling for the text
+ * @param dontShowFormattingSymbols True, if it does not suppose to include formatting symbols into resulting string
+ * @return Styled string
+ */
+QString TextFormatter::applyStyling(bool dontShowFormattingSymbols) {
+    return applyHtmlFontStyling(dontShowFormattingSymbols);
+}

src/chatlog/textformatter.h

@@ -0,0 +1,24 @@
+#ifndef TEXTFORMATTER_H
+#define TEXTFORMATTER_H
+
+#include <QString>
+
+class TextFormatter
+{
+private:
+
+    QString sourceString;
+
+    int patternEscapeSignsCount(const QString& str);
+
+    int getCapturedLength(const QRegularExpressionMatch& match, const int exprNumber);
+
+    QString applyHtmlFontStyling(bool dontShowFormattingSymbols);
+
+public:
+    explicit TextFormatter(const QString& str);
+
+    QString applyStyling(bool dontShowFormattingSymbols);
+};
+
+#endif // TEXTFORMATTER_H