Refactor Decompiled Code Structure (#1717)

src/common/Decompiler.cpp

@@ -5,6 +5,42 @@
 #include <QJsonObject>
 #include <QJsonArray>
 
+
+ut64 AnnotatedCode::OffsetForPosition(size_t pos) const
+{
+    size_t closestPos = SIZE_MAX;
+    ut64 closestOffset = UT64_MAX;
+    for (const auto &annotation : annotations) {
+        if (annotation.type != CodeAnnotation::Type::Offset || annotation.start > pos || annotation.end <= pos) {
+            continue;
+        }
+        if (closestPos != SIZE_MAX && closestPos >= annotation.start) {
+            continue;
+        }
+        closestPos = annotation.start;
+        closestOffset = annotation.offset.offset;
+    }
+    return closestOffset;
+}
+
+size_t AnnotatedCode::PositionForOffset(ut64 offset) const
+{
+    size_t closestPos = SIZE_MAX;
+    ut64 closestOffset = UT64_MAX;
+    for (const auto &annotation : annotations) {
+        if (annotation.type != CodeAnnotation::Type::Offset || annotation.offset.offset > offset) {
+            continue;
+        }
+        if (closestOffset != UT64_MAX && closestOffset >= annotation.offset.offset) {
+            continue;
+        }
+        closestPos = annotation.start;
+        closestOffset = annotation.offset.offset;
+    }
+    return closestPos;
+}
+
+
 Decompiler::Decompiler(const QString &id, const QString &name, QObject *parent)
     : QObject(parent),
     id(id),
@@ -22,9 +58,9 @@ bool R2DecDecompiler::isAvailable()
     return Core()->cmdList("e cmd.pdc=?").contains(QStringLiteral("pdd"));
 }
 
-DecompiledCode R2DecDecompiler::decompileAt(RVA addr)
+AnnotatedCode R2DecDecompiler::decompileAt(RVA addr)
 {
-    DecompiledCode code;
+    AnnotatedCode code = {};
     QString s;
 
     QJsonObject json = Core()->cmdj("pddj @ " + QString::number(addr)).object();
@@ -36,31 +72,32 @@ DecompiledCode R2DecDecompiler::decompileAt(RVA addr)
         if (!line.isString()) {
             continue;
         }
-        code.lines.append(DecompiledCode::Line(line.toString()));
+        code.code.append(line.toString() + "\n");
     }
 
     auto linesArray = json["lines"].toArray();
-    code.lines.reserve(code.lines.size() + linesArray.size());
     for (const auto &line : linesArray) {
         QJsonObject lineObject = line.toObject();
         if (lineObject.isEmpty()) {
             continue;
         }
-        DecompiledCode::Line codeLine;
+        CodeAnnotation annotation = {};
-        codeLine.str = lineObject["str"].toString();
+        annotation.type = CodeAnnotation::Type::Offset;
+        annotation.start = code.code.length();
+        code.code.append(lineObject["str"].toString() + "\n");
+        annotation.end = code.code.length();
         bool ok;
-        codeLine.addr = lineObject["offset"].toVariant().toULongLong(&ok);
+        annotation.offset.offset = lineObject["offset"].toVariant().toULongLong(&ok);
-        if (!ok) {
+        if (ok) {
-            codeLine.addr = RVA_INVALID;
+            code.annotations.push_back(annotation);
         }
-        code.lines.append(codeLine);
     }
 
     for (const auto &line : json["errors"].toArray()) {
         if (!line.isString()) {
             continue;
         }
-        code.lines.append(DecompiledCode::Line(line.toString()));
+        code.code.append(line.toString() + "\n");
     }
 
     return code;

src/common/Decompiler.h

@@ -6,34 +6,37 @@
 #include <QString>
 #include <QObject>
 
+struct CodeAnnotation
+{
+    size_t start;
+    size_t end;
+
+    enum class Type { Offset };
+    Type type;
+
+    union
+    {
+        struct
+        {
+            ut64 offset;
+        } offset;
+    };
+};
+
 /**
  * Describes the result of a Decompilation Process with optional metadata
  */
-struct DecompiledCode {
+struct AnnotatedCode
+{
     /**
-     * A single line of decompiled code
+     * The entire decompiled code
      */
-    struct Line
+    QString code;
-    {
-        QString str;
 
-        /**
+    QList<CodeAnnotation> annotations;
-         * Offset of the original instruction
-         */
-        RVA addr;
 
-        Line()
+    ut64 OffsetForPosition(size_t pos) const;
-        {
+    size_t PositionForOffset(ut64 offset) const;
-            this->addr = RVA_INVALID;
-        }
-
-        explicit Line(const QString &str, RVA addr = RVA_INVALID)
-        {
-            this->str = str;
-            this->addr = addr;
-        }
-    };
-    QList<Line> lines = {};
 };
 
 /**
@@ -54,7 +57,7 @@ public:
     QString getId() const   { return id; }
     QString getName() const { return name; }
 
-    virtual DecompiledCode decompileAt(RVA addr) =0;
+    virtual AnnotatedCode decompileAt(RVA addr) =0;
 };
 
 class R2DecDecompiler: public Decompiler
@@ -63,7 +66,7 @@ class R2DecDecompiler: public Decompiler
 
 public:
     explicit R2DecDecompiler(QObject *parent = nullptr);
-    DecompiledCode decompileAt(RVA addr) override;
+    AnnotatedCode decompileAt(RVA addr) override;
 
     static bool isAvailable();
 };

src/widgets/PseudocodeWidget.cpp

@@ -14,27 +14,6 @@
 #include <QObject>
 #include <QTextBlockUserData>
 
-/**
- * Represents a single line of decompiled code as part of the displayed text,
- * including the position inside the QTextDocument
- */
-struct DecompiledCodeTextLine
-{
-    /**
-     * position inside the QTextDocument
-     */
-    int position;
-
-    DecompiledCode::Line line;
-
-    DecompiledCodeTextLine(int position, const DecompiledCode::Line &line)
-    {
-        this->position = position;
-        this->line = line;
-    }
-};
-
-
 PseudocodeWidget::PseudocodeWidget(MainWindow *main, QAction *action) :
     MemoryDockWidget(MemoryWidgetType::Pseudocode, main, action),
     mCtxMenu(new DisassemblyContextMenu(this, main)),
@@ -104,24 +83,14 @@ void PseudocodeWidget::doRefresh(RVA addr)
         return;
     }
 
-    DecompiledCode decompiledCode = dec->decompileAt(addr);
+    code = dec->decompileAt(addr);
-
+    if (code.code.isEmpty()) {
-    textLines = {};
-    textLines.reserve(decompiledCode.lines.size());
-
-    if (decompiledCode.lines.isEmpty()) {
         ui->textEdit->setPlainText(tr("Cannot decompile at") + " " + RAddressString(
                                   addr) + " " + tr("(Not a function?)"));
         return;
     } else {
         connectCursorPositionChanged(true);
-        ui->textEdit->document()->clear();
+        ui->textEdit->setPlainText(code.code);
-        QTextCursor cursor(ui->textEdit->document());
-        for (const DecompiledCode::Line &line : decompiledCode.lines) {
-            textLines.append(DecompiledCodeTextLine(cursor.position(), line));
-            // Can't use cursor.block()->setUserData() here, because the Syntax Highlighter will mess it up.
-            cursor.insertText(line.str + "\n");
-        }
         connectCursorPositionChanged(false);
         seekChanged();
     }
@@ -148,7 +117,8 @@ void PseudocodeWidget::connectCursorPositionChanged(bool disconnect)
 
 void PseudocodeWidget::cursorPositionChanged()
 {
-    RVA offset = getOffsetAtLine(ui->textEdit->textCursor());
+    size_t pos = ui->textEdit->textCursor().position();
+    RVA offset = code.OffsetForPosition(pos);
     if (offset != RVA_INVALID && offset != Core()->getOffset()) {
         seekFromCursor = true;
         Core()->seek(offset);
@@ -169,61 +139,18 @@ void PseudocodeWidget::seekChanged()
 void PseudocodeWidget::updateCursorPosition()
 {
     RVA offset = Core()->getOffset();
+    size_t pos = code.PositionForOffset(offset);
+    if (pos == SIZE_MAX) {
+        return;
+    }
     connectCursorPositionChanged(true);
-
-    auto it = findLineByOffset(offset);
-    if (it != textLines.end()) {
-        // move back if the offset is identical (so we don't land on closing braces for example)
-        while (it != textLines.begin()) {
-            auto prev = it - 1;
-            if (prev->line.addr != it->line.addr) {
-                break;
-            }
-            it = prev;
-        }
     QTextCursor cursor = ui->textEdit->textCursor();
-        cursor.setPosition((*it).position);
+    cursor.setPosition(pos);
     ui->textEdit->setTextCursor(cursor);
     updateSelection();
-    }
-
     connectCursorPositionChanged(false);
 }
 
-QList<DecompiledCodeTextLine>::iterator PseudocodeWidget::findLine(int position)
-{
-    return std::upper_bound(textLines.begin(), textLines.end(), position,
-                            [](int pos, const DecompiledCodeTextLine &line) {
-                                return pos < line.position;
-                            });
-}
-
-QList<DecompiledCodeTextLine>::iterator PseudocodeWidget::findLineByOffset(RVA offset)
-{
-    auto it = textLines.begin();
-    auto candidate = it;
-    for (; it != textLines.end(); it++) {
-        RVA lineOffset = it->line.addr;
-        if (lineOffset != RVA_INVALID && lineOffset > offset) {
-            break;
-        }
-        if (candidate->line.addr == RVA_INVALID || (lineOffset != RVA_INVALID && lineOffset > candidate->line.addr)) {
-            candidate = it;
-        }
-    }
-    return candidate;
-}
-
-RVA PseudocodeWidget::getOffsetAtLine(const QTextCursor &tc)
-{
-    auto it = findLine(tc.position());
-    if (it == textLines.begin()) {
-        return RVA_INVALID;
-    }
-    it--;
-    return (*it).line.addr;
-}
-
 void PseudocodeWidget::setupFonts()
 {
     QFont font = Config()->getFont();
@@ -236,20 +163,7 @@ void PseudocodeWidget::updateSelection()
 
     // Highlight the current line
     auto cursor = ui->textEdit->textCursor();
-
-    RVA cursorOffset = getOffsetAtLine(cursor);
-    if (cursorOffset != RVA_INVALID) {
-        for (auto it = findLineByOffset(cursorOffset);
-                it != textLines.end() && it->line.addr != RVA_INVALID && it->line.addr <= cursorOffset;
-                it++) {
-            auto lineCursor = cursor;
-            lineCursor.setPosition(it->position);
-            extraSelections.append(createLineHighlightSelection(lineCursor));
-        }
-    } else {
-        // if the cursor position has no valid offset, just highlight the line
     extraSelections.append(createLineHighlightSelection(cursor));
-    }
 
     // Highlight all the words in the document same as the current one
     cursor.select(QTextCursor::WordUnderCursor);

src/widgets/PseudocodeWidget.h

@@ -5,6 +5,7 @@
 
 #include "core/Cutter.h"
 #include "MemoryDockWidget.h"
+#include "Decompiler.h"
 
 namespace Ui {
 class PseudocodeWidget;
@@ -41,10 +42,7 @@ private:
 
     QSyntaxHighlighter *syntaxHighlighter;
 
-    /**
+    AnnotatedCode code;
-     * Index of all lines that are currently displayed, ordered by the position in the text
-     */
-    QList<DecompiledCodeTextLine> textLines;
 
     bool seekFromCursor = false;
 
@@ -54,17 +52,6 @@ private:
     void connectCursorPositionChanged(bool disconnect);
     void updateCursorPosition();
 
-    /**
-     * @return Iterator to the first line that is after position or last if not found.
-     */
-    QList<DecompiledCodeTextLine>::iterator findLine(int position);
-
-    /**
-     * @return Iterator to the first line that is considered to contain offset
-     */
-    QList<DecompiledCodeTextLine>::iterator findLineByOffset(RVA offset);
-    RVA getOffsetAtLine(const QTextCursor &tc);
-
     QString getWindowTitle() const override;
 };