From dcf37eca8ae5cfd55c5e07df52ed0c28a4e79c53 Mon Sep 17 00:00:00 2001
From: "royqh1979@gmail.com" <Roy>
Date: Thu, 12 Aug 2021 11:29:26 +0800
Subject: [PATCH] work save: cpp parser - preprocessor : remove comments before
 preprocess

---
 RedPandaIDE/parser/cpppreprocessor.cpp | 115 ++++++++++++++++++++++++-
 RedPandaIDE/parser/cpppreprocessor.h   |  14 ++-
 2 files changed, 124 insertions(+), 5 deletions(-)

diff --git a/RedPandaIDE/parser/cpppreprocessor.cpp b/RedPandaIDE/parser/cpppreprocessor.cpp
index 8dbfb36e..66129d20 100644
--- a/RedPandaIDE/parser/cpppreprocessor.cpp
+++ b/RedPandaIDE/parser/cpppreprocessor.cpp
@@ -356,12 +356,12 @@ void CppPreprocessor::openInclude(const QString &fileName, QTextStream stream)
     // Process it
     mIndex = parsedFile->index;
     mFileName = parsedFile->fileName;
+    parsedFile->buffer = removeComments(parsedFile->buffer);
     mBuffer = parsedFile->buffer;
 
-    // Trim all lines
-    for (int i=0;i<mBuffer.count();i++) {
-        mBuffer[i] = mBuffer[i].trimmed();
-    }
+//    for (int i=0;i<mBuffer.count();i++) {
+//        mBuffer[i] = mBuffer[i].trimmed();
+//    }
 
     // Update result file
     QString includeLine = "#include " + fileName + ":1";
@@ -460,6 +460,113 @@ void CppPreprocessor::addDefinesInFile(const QString &fileName)
     }
 }
 
+QStringList CppPreprocessor::removeComments(const QStringList &text)
+{
+    QStringList result;
+    ContentType currentType = ContentType::Other;
+    QString delimiter;
+
+    for (QString line:text) {
+        QChar lastCh;
+        QString s;
+        int pos = 0;
+        bool stopProcess=false;
+        while (pos<line.length()) {
+            QChar ch =line[pos];
+            if (currentType == ContentType::AnsiCComment) {
+                if (ch=='*' && (pos+1<line.length()) && line[pos+1]=='/') {
+                    pos+=2;
+                    currentType = ContentType::Other;
+                } else {
+                    pos+=1;
+                }
+                continue;
+            }
+            switch (ch.unicode()) {
+            case '"':
+                switch (currentType) {
+                case ContentType::String:
+                    currentType=ContentType::Other;
+                    break;
+                case ContentType::RawString:
+                    if (line.mid(0,pos).endsWith(')'+delimiter))
+                        currentType = ContentType::Other;
+                case ContentType::Other:
+                    currentType=ContentType::String;
+                    break;
+                case ContentType::RawStringPrefix:
+                    delimiter+=ch;
+                    break;
+                }
+                s+=ch;
+                break;
+            case '\'':
+                switch (currentType) {
+                case ContentType::Character:
+                    currentType=ContentType::Other;
+                case ContentType::Other:
+                    currentType=ContentType::Character;
+                case ContentType::RawStringPrefix:
+                    delimiter+=ch;
+                    break;
+                }
+                s+=ch;
+                break;
+            case 'R':
+                if (currentType == ContentType::Other && pos+1<line.length() && line[pos+1]=='"') {
+                    s+=ch;
+                    pos++;
+                    ch = line[pos];
+                    currentType=ContentType::RawStringPrefix;
+                    delimiter = "";
+                }
+                if (currentType == ContentType::RawStringPrefix ) {
+                    delimiter += ch;
+                }
+                s+=ch;
+                break;
+            case '(':
+                switch(currentType) {
+                case ContentType::RawStringPrefix:
+                    currentType = ContentType::RawString;
+                    break;
+                }
+                s+=ch;
+                break;
+            case '/':
+                switch(currentType) {
+                case ContentType::Other:
+                    if (pos+1<line.length() && line[pos+1]=='/') {
+                        // line comment , skip all remainings of the current line
+                        stopProcess = true;
+                    } else if (pos+1<line.length() && line[pos+1]=='*') {
+                        /* ansi c comment */
+                        pos++;
+                        currentType = ContentType::AnsiCComment;
+                    }
+                    break;
+                }
+            case '\\':
+                switch (currentType) {
+                case ContentType::String:
+                case ContentType::Character:
+                    pos++;
+                    s+=ch;
+                    if (pos<line.length()) {
+                        ch = line[pos];
+                        s+=ch;
+                    }
+                }
+            }
+            if (stopProcess)
+                break;
+            pos++;
+        }
+        result.append(s.trimmed());
+    }
+    return result;
+}
+
 void CppPreprocessor::preprocessBuffer()
 {
     while (mIncludes.count() > 0) {
diff --git a/RedPandaIDE/parser/cpppreprocessor.h b/RedPandaIDE/parser/cpppreprocessor.h
index 1258f5d4..d2dc455e 100644
--- a/RedPandaIDE/parser/cpppreprocessor.h
+++ b/RedPandaIDE/parser/cpppreprocessor.h
@@ -19,6 +19,18 @@ using PParsedFile = std::shared_ptr<ParsedFile>;
 class CppPreprocessor : public QObject
 {
     Q_OBJECT
+
+    enum class ContentType {
+        AnsiCComment,
+        CppComment,
+        String,
+        Character,
+        EscapeSequence,
+        RawStringPrefix,
+        RawString,
+        Other
+    };
+
 public:
 
     explicit CppPreprocessor(QObject *parent = nullptr);
@@ -70,7 +82,7 @@ private:
     PFileIncludes getFileIncludesEntry(const QString& FileName);
     void addDefinesInFile(const QString& fileName);
 
-    void preprocessBuffer();
+    QStringList removeComments(const QStringList& text);
     /*
      * '_','a'..'z','A'..'Z','0'..'9'
      */