fix: support captures in ignores (#5126)

CHANGELOG.md

@@ -53,10 +53,10 @@
 - Bugfix: Fixed thread popup window missing messages for nested threads. (#4923)
 - Bugfix: Fixed an occasional crash for channel point redemptions with text input. (#4949)
 - Bugfix: Fixed triple click on message also selecting moderation buttons. (#4961)
-- Bugfix: Fixed a freeze from a bad regex in _Ignores_. (#4965)
+- Bugfix: Fixed a freeze from a bad regex in _Ignores_. (#4965, #5126)
 - Bugfix: Fixed badge highlight changes not immediately being reflected. (#5110)
-- Bugfix: Fixed some emotes not appearing when using _Ignores_. (#4965)
+- Bugfix: Fixed some emotes not appearing when using _Ignores_. (#4965, #5126)
-- Bugfix: Fixed lookahead/-behind not working in _Ignores_. (#4965)
+- Bugfix: Fixed lookahead/-behind not working in _Ignores_. (#4965, #5126)
 - Bugfix: Fixed Image Uploader accidentally deleting images with some hosts when link resolver was enabled. (#4971)
 - Bugfix: Fixed rare crash with Image Uploader when closing a split right after starting an upload. (#4971)
 - Bugfix: Fixed an issue on macOS where the image uploader would keep prompting the user even after they clicked "Yes, don't ask again". (#5011)

src/providers/twitch/TwitchMessageBuilder.cpp

@@ -269,6 +269,128 @@ namespace {
         builder->message().badgeInfos = badgeInfos;
     }
 
+    /**
+     * Computes (only) the replacement of @a match in @a source.
+     * The parts before and after the match in @a source are ignored.
+     *
+     * Occurrences of \b{\\1}, \b{\\2}, ..., in @a replacement are replaced
+     * with the string captured by the corresponding capturing group.
+     * This function should only be used if the regex contains capturing groups.
+     * 
+     * Since Qt doesn't provide a way of replacing a single match with some replacement
+     * while supporting both capturing groups and lookahead/-behind in the regex,
+     * this is included here. It's essentially the implementation of 
+     * QString::replace(const QRegularExpression &, const QString &).
+     * @see https://github.com/qt/qtbase/blob/97bb0ecfe628b5bb78e798563212adf02129c6f6/src/corelib/text/qstring.cpp#L4594-L4703
+     */
+    QString makeRegexReplacement(QStringView source,
+                                 const QRegularExpression &regex,
+                                 const QRegularExpressionMatch &match,
+                                 const QString &replacement)
+    {
+        using SizeType = QString::size_type;
+        struct QStringCapture {
+            SizeType pos;
+            SizeType len;
+            int captureNumber;
+        };
+
+        qsizetype numCaptures = regex.captureCount();
+
+        // 1. build the backreferences list, holding where the backreferences
+        //    are in the replacement string
+        QVarLengthArray<QStringCapture> backReferences;
+
+        SizeType replacementLength = replacement.size();
+        for (SizeType i = 0; i < replacementLength - 1; i++)
+        {
+            if (replacement[i] != u'\\')
+            {
+                continue;
+            }
+
+            int no = replacement[i + 1].digitValue();
+            if (no <= 0 || no > numCaptures)
+            {
+                continue;
+            }
+
+            QStringCapture backReference{.pos = i, .len = 2};
+
+            if (i < replacementLength - 2)
+            {
+                int secondDigit = replacement[i + 2].digitValue();
+                if (secondDigit != -1 &&
+                    ((no * 10) + secondDigit) <= numCaptures)
+                {
+                    no = (no * 10) + secondDigit;
+                    ++backReference.len;
+                }
+            }
+
+            backReference.captureNumber = no;
+            backReferences.append(backReference);
+        }
+
+        // 2. iterate on the matches.
+        //    For every match, copy the replacement string in chunks
+        //    with the proper replacements for the backreferences
+
+        // length of the new string, with all the replacements
+        SizeType newLength = 0;
+        QVarLengthArray<QStringView> chunks;
+        QStringView replacementView{replacement};
+
+        // Initially: empty, as we only care about the replacement
+        SizeType len = 0;
+        SizeType lastEnd = 0;
+        for (const QStringCapture &backReference :
+             std::as_const(backReferences))
+        {
+            // part of "replacement" before the backreference
+            len = backReference.pos - lastEnd;
+            if (len > 0)
+            {
+                chunks << replacementView.mid(lastEnd, len);
+                newLength += len;
+            }
+
+            // backreference itself
+            len = match.capturedLength(backReference.captureNumber);
+            if (len > 0)
+            {
+                chunks << source.mid(
+                    match.capturedStart(backReference.captureNumber), len);
+                newLength += len;
+            }
+
+            lastEnd = backReference.pos + backReference.len;
+        }
+
+        // add the last part of the replacement string
+        len = replacementView.size() - lastEnd;
+        if (len > 0)
+        {
+            chunks << replacementView.mid(lastEnd, len);
+            newLength += len;
+        }
+
+        // 3. assemble the chunks together
+        QString dst;
+        dst.reserve(newLength);
+        for (const QStringView &chunk : std::as_const(chunks))
+        {
+#if QT_VERSION < QT_VERSION_CHECK(5, 15, 2)
+            static_assert(sizeof(QChar) == sizeof(decltype(*chunk.utf16())));
+            dst.append(reinterpret_cast<const QChar *>(chunk.utf16()),
+                       chunk.length());
+#else
+            dst += chunk;
+#endif
+        }
+        return dst;
+    }
+
 }  // namespace
 
 TwitchMessageBuilder::TwitchMessageBuilder(
@@ -419,7 +541,9 @@ MessagePtr TwitchMessageBuilder::build()
         this->tags, this->originalMessage_, this->messageOffset_);
 
     // This runs through all ignored phrases and runs its replacements on this->originalMessage_
-    this->runIgnoreReplaces(twitchEmotes);
+    TwitchMessageBuilder::processIgnorePhrases(
+        *getSettings()->ignoredMessages.readOnly(), this->originalMessage_,
+        twitchEmotes);
 
     std::sort(twitchEmotes.begin(), twitchEmotes.end(),
               [](const auto &a, const auto &b) {
@@ -960,12 +1084,12 @@ void TwitchMessageBuilder::appendUsername()
     }
 }
 
-void TwitchMessageBuilder::runIgnoreReplaces(
+void TwitchMessageBuilder::processIgnorePhrases(
+    const std::vector<IgnorePhrase> &phrases, QString &originalMessage,
     std::vector<TwitchEmoteOccurrence> &twitchEmotes)
 {
     using SizeType = QString::size_type;
 
-    auto phrases = getSettings()->ignoredMessages.readOnly();
     auto removeEmotesInRange = [&twitchEmotes](SizeType pos, SizeType len) {
         // all emotes outside the range come before `it`
         // all emotes in the range start at `it`
@@ -1034,20 +1158,20 @@ void TwitchMessageBuilder::runIgnoreReplaces(
     auto replaceMessageAt = [&](const IgnorePhrase &phrase, SizeType from,
                                 SizeType length, const QString &replacement) {
         auto removedEmotes = removeEmotesInRange(from, length);
-        this->originalMessage_.replace(from, length, replacement);
+        originalMessage.replace(from, length, replacement);
         auto wordStart = from;
         while (wordStart > 0)
         {
-            if (this->originalMessage_[wordStart - 1] == ' ')
+            if (originalMessage[wordStart - 1] == ' ')
             {
                 break;
             }
             --wordStart;
         }
         auto wordEnd = from + replacement.length();
-        while (wordEnd < this->originalMessage_.length())
+        while (wordEnd < originalMessage.length())
         {
-            if (this->originalMessage_[wordEnd] == ' ')
+            if (originalMessage[wordEnd] == ' ')
             {
                 break;
             }
@@ -1058,11 +1182,11 @@ void TwitchMessageBuilder::runIgnoreReplaces(
                           static_cast<int>(replacement.length() - length));
 
 #if QT_VERSION >= QT_VERSION_CHECK(5, 15, 0)
-        auto midExtendedRef = QStringView{this->originalMessage_}.mid(
+        auto midExtendedRef =
-            wordStart, wordEnd - wordStart);
+            QStringView{originalMessage}.mid(wordStart, wordEnd - wordStart);
 #else
         auto midExtendedRef =
-            this->originalMessage_.midRef(wordStart, wordEnd - wordStart);
+            originalMessage.midRef(wordStart, wordEnd - wordStart);
 #endif
 
         for (auto &emote : removedEmotes)
@@ -1088,7 +1212,7 @@ void TwitchMessageBuilder::runIgnoreReplaces(
         addReplEmotes(phrase, midExtendedRef, wordStart);
     };
 
-    for (const auto &phrase : *phrases)
+    for (const auto &phrase : phrases)
     {
         if (phrase.isBlock())
         {
@@ -1110,16 +1234,22 @@ void TwitchMessageBuilder::runIgnoreReplaces(
             QRegularExpressionMatch match;
             size_t iterations = 0;
             SizeType from = 0;
-            while ((from = this->originalMessage_.indexOf(regex, from,
+            while ((from = originalMessage.indexOf(regex, from, &match)) != -1)
-                                                          &match)) != -1)
             {
+                auto replacement = phrase.getReplace();
+                if (regex.captureCount() > 0)
+                {
+                    replacement = makeRegexReplacement(originalMessage, regex,
+                                                       match, replacement);
+                }
+
                 replaceMessageAt(phrase, from, match.capturedLength(),
-                                 phrase.getReplace());
+                                 replacement);
                 from += phrase.getReplace().length();
                 iterations++;
                 if (iterations >= 128)
                 {
-                    this->originalMessage_ =
+                    originalMessage =
                         u"Too many replacements - check your ignores!"_s;
                     return;
                 }
@@ -1129,8 +1259,8 @@ void TwitchMessageBuilder::runIgnoreReplaces(
         }
 
         SizeType from = 0;
-        while ((from = this->originalMessage_.indexOf(
+        while ((from = originalMessage.indexOf(pattern, from,
-                    pattern, from, phrase.caseSensitivity())) != -1)
+                                               phrase.caseSensitivity())) != -1)
         {
             replaceMessageAt(phrase, from, pattern.length(),
                              phrase.getReplace());

src/providers/twitch/TwitchMessageBuilder.hpp

@@ -20,6 +20,7 @@ using EmotePtr = std::shared_ptr<const Emote>;
 class Channel;
 class TwitchChannel;
 class MessageThread;
+class IgnorePhrase;
 struct HelixVip;
 using HelixModerator = HelixVip;
 struct ChannelPointReward;
@@ -108,6 +109,10 @@ public:
         const QVariantMap &tags, const QString &originalMessage,
         int messageOffset);
 
+    static void processIgnorePhrases(
+        const std::vector<IgnorePhrase> &phrases, QString &originalMessage,
+        std::vector<TwitchEmoteOccurrence> &twitchEmotes);
+
 private:
     void parseUsernameColor() override;
     void parseUsername() override;
@@ -118,8 +123,6 @@ private:
     void parseThread();
     void appendUsername();
 
-    void runIgnoreReplaces(std::vector<TwitchEmoteOccurrence> &twitchEmotes);
-
     Outcome tryAppendEmote(const EmoteName &name) override;
 
     void addWords(const QStringList &words,

tests/src/TwitchMessageBuilder.cpp

@@ -3,6 +3,7 @@
 #include "common/Channel.hpp"
 #include "controllers/accounts/AccountController.hpp"
 #include "controllers/highlights/HighlightController.hpp"
+#include "controllers/ignores/IgnorePhrase.hpp"
 #include "messages/MessageBuilder.hpp"
 #include "mocks/Channel.hpp"
 #include "mocks/ChatterinoBadges.hpp"
@@ -478,3 +479,142 @@ TEST_F(TestTwitchMessageBuilder, ParseMessage)
         delete privmsg;
     }
 }
+
+TEST_F(TestTwitchMessageBuilder, IgnoresReplace)
+{
+    struct TestCase {
+        std::vector<IgnorePhrase> phrases;
+        QString input;
+        std::vector<TwitchEmoteOccurrence> twitchEmotes;
+        QString expectedMessage;
+        std::vector<TwitchEmoteOccurrence> expectedTwitchEmotes;
+    };
+
+    auto *twitchEmotes = this->mockApplication->getEmotes()->getTwitchEmotes();
+
+    auto emoteAt = [&](int at, const QString &name) {
+        return TwitchEmoteOccurrence{
+            .start = at,
+            .end = static_cast<int>(at + name.size() - 1),
+            .ptr =
+                twitchEmotes->getOrCreateEmote(EmoteId{name}, EmoteName{name}),
+            .name = EmoteName{name},
+        };
+    };
+
+    auto regularReplace = [](auto pattern, auto replace,
+                             bool caseSensitive = true) {
+        return IgnorePhrase(pattern, false, false, replace, caseSensitive);
+    };
+    auto regexReplace = [](auto pattern, auto regex,
+                           bool caseSensitive = true) {
+        return IgnorePhrase(pattern, true, false, regex, caseSensitive);
+    };
+
+    std::vector<TestCase> testCases{
+        {
+            {regularReplace("foo1", "baz1")},
+            "foo1 Kappa",
+            {emoteAt(4, "Kappa")},
+            "baz1 Kappa",
+            {emoteAt(4, "Kappa")},
+        },
+        {
+            {regularReplace("foo1", "baz1", false)},
+            "FoO1 Kappa",
+            {emoteAt(4, "Kappa")},
+            "baz1 Kappa",
+            {emoteAt(4, "Kappa")},
+        },
+        {
+            {regexReplace("f(o+)1", "baz1[\\1]")},
+            "foo1 Kappa",
+            {emoteAt(4, "Kappa")},
+            "baz1[oo] Kappa",
+            {emoteAt(8, "Kappa")},
+        },
+
+        {
+            {regexReplace("f(o+)1", R"(baz1[\0][\1][\2])")},
+            "foo1 Kappa",
+            {emoteAt(4, "Kappa")},
+            "baz1[\\0][oo][\\2] Kappa",
+            {emoteAt(16, "Kappa")},
+        },
+        {
+            {regexReplace("f(o+)(\\d+)", "baz1[\\1+\\2]")},
+            "foo123 Kappa",
+            {emoteAt(6, "Kappa")},
+            "baz1[oo+123] Kappa",
+            {emoteAt(12, "Kappa")},
+        },
+        {
+            {regexReplace("(?<=foo)(\\d+)", "[\\1]")},
+            "foo123 Kappa",
+            {emoteAt(6, "Kappa")},
+            "foo[123] Kappa",
+            {emoteAt(8, "Kappa")},
+        },
+        {
+            {regexReplace("a(?=a| )", "b")},
+            "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"
+            "aaaa"
+            "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa "
+            "Kappa",
+            {emoteAt(127, "Kappa")},
+            "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb"
+            "bbbb"
+            "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb "
+            "Kappa",
+            {emoteAt(127, "Kappa")},
+        },
+        {
+            {regexReplace("abc", "def", false)},
+            "AbC Kappa",
+            {emoteAt(3, "Kappa")},
+            "def Kappa",
+            {emoteAt(3, "Kappa")},
+        },
+        {
+            {
+                regexReplace("abc", "def", false),
+                regularReplace("def", "ghi"),
+            },
+            "AbC Kappa",
+            {emoteAt(3, "Kappa")},
+            "ghi Kappa",
+            {emoteAt(3, "Kappa")},
+        },
+        {
+            {
+                regexReplace("a(?=a| )", "b"),
+                regexReplace("b(?=b| )", "c"),
+            },
+            "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"
+            "aaaa"
+            "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa "
+            "Kappa",
+            {emoteAt(127, "Kappa")},
+            "cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc"
+            "cccccccccccccccccccccccccccccccccccccccccccccccccccccccccccc "
+            "Kappa",
+            {emoteAt(127, "Kappa")},
+        },
+    };
+
+    for (const auto &test : testCases)
+    {
+        auto message = test.input;
+        auto emotes = test.twitchEmotes;
+        TwitchMessageBuilder::processIgnorePhrases(test.phrases, message,
+                                                   emotes);
+
+        EXPECT_EQ(message, test.expectedMessage)
+            << "Message not equal for input '" << test.input.toStdString()
+            << "' - expected: '" << test.expectedMessage.toStdString()
+            << "' got: '" << message.toStdString() << "'";
+        EXPECT_EQ(emotes, test.expectedTwitchEmotes)
+            << "Twitch emotes not equal for input '" << test.input.toStdString()
+            << "' and output '" << message.toStdString() << "'";
+    }
+}