Add SpellChecking to NeoChat

Fix #98

src/CMakeLists.txt

@@ -43,8 +43,8 @@ ecm_add_app_icon(NEOCHAT_ICON ICONS ${CMAKE_SOURCE_DIR}/128-logo.png)
 target_sources(neochat PRIVATE ${NEOCHAT_ICON})
 
 if(NOT ANDROID)
-    target_sources(neochat PRIVATE trayicon.cpp colorschemer.cpp)
-    target_link_libraries(neochat PRIVATE KF5::ConfigWidgets)
+    target_sources(neochat PRIVATE trayicon.cpp colorschemer.cpp spellcheckhighlighter.cpp)
+    target_link_libraries(neochat PRIVATE KF5::ConfigWidgets KF5::SonnetCore)
     target_compile_definitions(neochat PRIVATE -DHAVE_COLORSCHEME)
 endif()
 

src/chatdocumenthandler.cpp

@@ -11,6 +11,9 @@
 #include <QTextDocument>
 
 #include "neochatroom.h"
+#ifndef Q_OS_ANDROID
+#include "spellcheckhighlighter.h"
+#endif
 
 ChatDocumentHandler::ChatDocumentHandler(QObject *parent)
     : QObject(parent)
@@ -34,6 +37,9 @@ void ChatDocumentHandler::setDocument(QQuickTextDocument *document)
         m_document->textDocument()->disconnect(this);
     }
     m_document = document;
+#ifndef Q_OS_ANDROID
+    new SpellcheckHighlighter(m_document->textDocument());
+#endif
     Q_EMIT documentChanged();
 }
 

src/spellcheckhighlighter.cpp

@@ -0,0 +1,109 @@
+// Copyright (c) 2020 Christian Mollekopf <mollekopf@kolabsystems.com>
+// SPDX-License-Identifier: LGPL-2.0-or-later
+
+#include "spellcheckhighlighter.h"
+
+#include <QDebug>
+#include <QTextBoundaryFinder>
+
+QVector<QStringRef> split(QTextBoundaryFinder::BoundaryType boundary, const QString &text, int reasonMask = 0)
+{
+    QVector<QStringRef> parts;
+    QTextBoundaryFinder boundaryFinder(boundary, text);
+
+    while (boundaryFinder.position() < text.length()) {
+        const int start = boundaryFinder.position();
+
+        //Advance until we find a break that matches the mask or are at the end
+        for (;;) {
+            if (boundaryFinder.toNextBoundary() == -1) {
+                boundaryFinder.toEnd();
+                break;
+            }
+            if (!reasonMask || boundaryFinder.boundaryReasons() & reasonMask) {
+                break;
+            }
+        }
+
+        const auto length = boundaryFinder.position() - start;
+
+        if (length < 1) {
+            continue;
+        }
+        parts << QStringRef{&text, start, length};
+    }
+    return parts;
+}
+
+
+SpellcheckHighlighter::SpellcheckHighlighter(QTextDocument *parent)
+    : QSyntaxHighlighter(parent),
+    mSpellchecker{new Sonnet::Speller()},
+    mLanguageGuesser{new Sonnet::GuessLanguage()}
+{
+    //Danger red from our color scheme
+    mErrorFormat.setForeground(QColor{"#ed1515"});
+    mErrorFormat.setUnderlineColor(QColor{"#ed1515"});
+    mErrorFormat.setUnderlineStyle(QTextCharFormat::SingleUnderline);
+    mQuoteFormat.setForeground(QColor{"#7f8c8d"});
+
+
+    if (!mSpellchecker->isValid()) {
+        qWarning() << "Spellchecker is invalid";
+    }
+    qDebug() << "Available dictionaries: " << mSpellchecker->availableDictionaries();
+}
+
+void SpellcheckHighlighter::autodetectLanguage(const QString &sentence)
+{
+    const auto lang = mLanguageGuesser->identify(sentence, mSpellchecker->availableLanguages());
+    if (lang.isEmpty()) {
+        return;
+    }
+    mSpellchecker->setLanguage(lang);
+}
+
+static bool isSpellcheckable(const QStringRef &token)
+{
+    if (token.isNull() || token.isEmpty()) {
+        return false;
+    }
+    if (!token.at(0).isLetter() || token.at(0).isUpper() || token.startsWith(QStringLiteral("http"))) {
+        return false;
+    }
+    //TODO ignore urls and uppercase?
+    return true;
+}
+
+void SpellcheckHighlighter::highlightBlock(const QString &text)
+{
+    //Avoid spellchecking quotes
+    if (text.isEmpty() || text.at(0) == QChar{'>'}) {
+        setFormat(0, text.length(), mQuoteFormat);
+        return;
+    }
+    for (const auto &sentenceRef : split(QTextBoundaryFinder::Sentence, text)) {
+        //Avoid spellchecking quotes
+        if (sentenceRef.isEmpty() || sentenceRef.at(0) == QChar{'>'}) {
+            continue;
+        }
+
+        const auto sentence = QString::fromRawData(sentenceRef.data(), sentenceRef.length());
+
+        autodetectLanguage(sentence);
+
+        const int offset = sentenceRef.position();
+        for (const auto &wordRef : split(QTextBoundaryFinder::Word, sentence)) {
+            //Avoid spellchecking words in progress
+            //FIXME this will also prevent spellchecking a single word on a line.
+            if (offset + wordRef.position() + wordRef.length() >= text.length()) {
+                continue;
+            }
+            if (isSpellcheckable(wordRef)) {
+                const auto word = QString::fromRawData(wordRef.data(), wordRef.length());
+                const auto format = mSpellchecker->isMisspelled(word) ? mErrorFormat : QTextCharFormat{};
+                setFormat(offset + wordRef.position(), wordRef.length(), format);
+            }
+        }
+    }
+}

src/spellcheckhighlighter.h

@@ -0,0 +1,26 @@
+// Copyright (c) 2020 Christian Mollekopf <mollekopf@kolabsystems.com>
+// SPDX-License-Identifier: LGPL-2.0-or-later
+
+#pragma once
+
+#include <QTextDocument>
+#include <QSyntaxHighlighter>
+#include <Sonnet/Speller>
+#include <Sonnet/GuessLanguage>
+
+class SpellcheckHighlighter: public QSyntaxHighlighter
+{
+public:
+    SpellcheckHighlighter(QTextDocument *parent);
+
+protected:
+    void highlightBlock(const QString &text) override;
+
+private:
+    void autodetectLanguage(const QString &sentence);
+    QTextCharFormat mErrorFormat;
+    QTextCharFormat mQuoteFormat;
+    QScopedPointer<Sonnet::Speller> mSpellchecker;
+    QScopedPointer<Sonnet::GuessLanguage> mLanguageGuesser;
+};
+