feat: support for ANSI color sequences(#338)

Author: nowhszh

src/ui/CMakeLists.txt

@@ -46,6 +46,9 @@ add_library(
   ${CMAKE_CURRENT_SOURCE_DIR}/include/decompressor.h
   ${CMAKE_CURRENT_SOURCE_DIR}/include/fontutils.h
   ${CMAKE_CURRENT_SOURCE_DIR}/include/colorlabelsmanager.h
+  ${CMAKE_CURRENT_SOURCE_DIR}/include/ANSI.h
+  ${CMAKE_CURRENT_SOURCE_DIR}/include/SGRParser.h
+  ${CMAKE_CURRENT_SOURCE_DIR}/include/ColorfulTextParser.h
   ${CMAKE_CURRENT_SOURCE_DIR}/include/highlighteredit.ui
   ${CMAKE_CURRENT_SOURCE_DIR}/include/highlightersetedit.ui
   ${CMAKE_CURRENT_SOURCE_DIR}/include/highlightersdialog.ui
@@ -91,6 +94,8 @@ add_library(
   ${CMAKE_CURRENT_SOURCE_DIR}/src/downloader.cpp
   ${CMAKE_CURRENT_SOURCE_DIR}/src/decompressor.cpp
   ${CMAKE_CURRENT_SOURCE_DIR}/src/colorlabelsmanager.cpp
+  ${CMAKE_CURRENT_SOURCE_DIR}/src/SGRParser.cpp
+  ${CMAKE_CURRENT_SOURCE_DIR}/src/ColorfulTextParser.cpp
 )
 
 set_target_properties(klogg_ui PROPERTIES AUTOUIC ON)

src/ui/include/ANSI.h

@@ -0,0 +1,94 @@
+//
+// Created by marvin on 23-7-18.
+//
+#pragma once
+
+#include <cstdint>
+
+namespace ANSI {
+
+enum SequenceStartCnt {
+    HEAD_CNT = 2,
+};
+
+// reference: https://www.ecma-international.org/publications-and-standards/standards/ecma-48/
+enum SequenceFirst {
+    EXC = 0x1B, // Escape（转义）
+};
+
+// reference: https://www.ecma-international.org/publications-and-standards/standards/ecma-48/
+enum SequenceSecond {
+    PAD = 0x40, // 0x80  Padding Character（填充字符）
+    HOP = 0x41, // 0x81  High Octet Preset（高字节前置）
+    BPH = 0x42, // 0x82  Break Permitted Here（此处允许中断）
+    NBH = 0x43, // 0x83  No Break Here（此处禁止中断）
+    IND = 0x44, // 0x84  Index（索引）
+    NEL = 0x45, // 0x85  Next Line（下一行）
+    SSA = 0x46, // 0x86  Start of Selected Area（选择区域开始）
+    ESA = 0x47, // 0x87  End of Selected Area（选择区域结束）
+    HTS = 0x48, // 0x88  Horizontal Tab Set（水平制表设置）
+    HTJ = 0x49, // 0x89  Horizontal Tab Justified（水平制表调整）
+    VTS = 0x4A, // 0x8A  Vertical Tab Set（垂直制表设置）
+    PLD = 0x4B, // 0x8B  Partial Line Forward（部分行前移）
+    PLU = 0x4C, // 0x8C  Partial Line Backward（部分行后移）
+    RI  = 0x4D, // 0x8D  Reverse Line Feed（逆向馈行）
+    SS2 = 0x4E, // 0x8E  Single-Shift 2（单个移动2）
+    SS3 = 0x4F, // 0x8F  Single-Shift 3（单个移动3）
+    DCS = 0x50, // 0x90  Device Control String（设备控制串）
+    PU1 = 0x51, // 0x91  Private Use 1（私用1）
+    PU2 = 0x52, // 0x92  Private Use 2（私用2）
+    STS = 0x53, // 0x93  Set Transmit State（发送规则设置）
+    CCH = 0x54, // 0x94  Cancel Character（取消字符）
+    MW  = 0x55, // 0x95  Message Waiting（消息等待）
+    SPA = 0x56, // 0x96  Start of Protected Area（保护区域开始）
+    EPA = 0x57, // 0x97  End of Protected Area（保护区域结束）
+    SOS = 0x58, // 0x98  Start of String（串开始）
+    SGC = 0x59, // 0x99  Single Graphic Char Intro（单个图形字符描述）
+    SCI = 0x5A, // 0x9A  Single Char Intro（单个字符描述）
+    CSI = 0x5B, // 0x9B  Control Sequence Intro（控制顺序描述）
+    ST  = 0x5C, // 0x9C  String Terminator（串终止）
+    OSC = 0x5D, // 0x9D  OS Command（操作系统指令）
+    PM  = 0x5E, // 0x9E  Private Message（私讯）
+    APC = 0x5F, // 0x9F  App Program Command（应用程序命令）
+
+    SECOND_BYTE_BEGIN = PAD,
+    SECOND_BYTE_END   = APC,
+};
+
+// reference: https://www.ecma-international.org/publications-and-standards/standards/ecma-48/
+enum CSIParameterBytes {
+    // ASCII: 0–9:;<=>?
+    NUM_BEGIN                  = 0x30,
+    NUM_END                    = 0x39,
+    SUB_PARA_SEPARATOR         = 0x3A,
+    PARA_SEPARATOR             = 0x3B,
+    STANDARDIZATION_KEEP_BEGIN = 0x3C,
+    STANDARDIZATION_KEEP_END   = 0x3F,
+
+    CSI_PARAMETER_BEGIN = 0x30,
+    CSI_PARAMETER_END   = 0x3F,
+};
+
+enum CSIIntermediateBytes {
+    // ASCII: Space、!"#$%&'()*+,-./
+    CSIIntermediateBegin = 0x20,
+    CSIIntermediateEnd   = 0x2F,
+};
+
+enum CSIFinalBytes {
+    // ASCII: @A–Z[\]^_`a–z{|}~
+    SGR = 0x6D, // m
+
+    CSIFinalBegin = 0x40,
+    CSIFinalEnd   = 0x7E,
+
+    CSIFinalExperimentalBegin = 0x70,
+    CSIFinalExperimentalEnd   = CSIFinalEnd,
+};
+
+enum class Return {
+    PARSE_ERROR = -1,
+    PARSE_SUCC  = 0,
+};
+
+} // namespace ANSI

src/ui/include/ColorfulTextParser.h

@@ -0,0 +1,73 @@
+//
+// Created by marvin on 23-7-24.
+//
+
+#pragma once
+
+#include <vector>
+
+#include <QString>
+
+#include "SGRParser.h"
+#include "highlightedmatch.h"
+
+struct TextColorAttr {
+    ANSI::Color color;
+    size_t start;
+    size_t len;
+    operator HighlightedMatch();
+};
+
+struct ColorfulText {
+    std::string text;
+    std::vector<TextColorAttr> color;
+};
+
+class CSIFilter {
+  public:
+    CSIFilter() = default;
+    ~CSIFilter() = default;
+
+    // ANSI: start, data
+    using SGRSequence = std::pair<size_t, std::string>;
+
+    static std::vector<SGRSequence> filter( QString& stdText );
+    static std::vector<SGRSequence> filter( std::string& stdText );
+};
+
+class ColorfulTextParser {
+  public:
+    enum class Mode {
+        MARKED_TEXT,
+        ALL_TEXT,
+    };
+
+  public:
+    explicit ColorfulTextParser( const ANSI::TextAttribute& defaultAttr,
+                                 const ANSI::TextAttribute& currentAttr );
+
+    // QString
+    ColorfulText parse( QString strings, Mode mode = Mode::ALL_TEXT );
+
+    std::vector<ColorfulText> parse( const std::vector<QString>& strings,
+                                     Mode mode = Mode::ALL_TEXT );
+
+    // std::string
+    ColorfulText parse( std::string strings, Mode mode = Mode::ALL_TEXT );
+
+    std::vector<ColorfulText> parse( const std::vector<std::string>& strings,
+                                     Mode mode = Mode::ALL_TEXT );
+
+  private:
+    // std::string
+    void markedStringToText( std::vector<ColorfulText>& textList,
+                             const std::vector<CSIFilter::SGRSequence>& sgrSeqs,
+                             std::string&& string );
+    void allStringToText( std::vector<ColorfulText>& textList,
+                          const std::vector<CSIFilter::SGRSequence>& sgrSeqs,
+                          std::string&& string );
+
+  private:
+    ANSI::TextAttribute currentTextAttr_;
+    ANSI::SGRParser sgrParser_;
+};

src/ui/include/SGRParser.h

@@ -0,0 +1,206 @@
+//
+// Created by marvin on 23-7-19.
+//
+#pragma once
+
+#include <map>
+#include <string>
+#include <utility>
+
+#include "ANSI.h"
+
+namespace ANSI {
+
+struct RGB {
+    uint8_t r, g, b;
+};
+
+struct Color {
+    RGB front;
+    RGB back;
+};
+
+struct TextAttribute {
+    enum class State {
+        DEFAULT,
+        CUSTOM,
+    };
+
+    State state;
+    Color color;
+};
+
+class SGRParser {
+  public:
+    using SGRParseReturn = std::pair<Return, TextAttribute>;
+
+  public:
+    explicit SGRParser( const TextAttribute& defaultTextAttr );
+    ~SGRParser() = default;
+
+    SGRParser( const SGRParser& ) = delete;
+    SGRParser( SGRParser&& ) = delete;
+    SGRParser& operator=( const SGRParser& ) = delete;
+    SGRParser& operator=( SGRParser&& ) = delete;
+
+    /*
+     * @param currentTextAttr   Properties of the current text
+     * @param sequence          SGR sequence, example: "\033[31m"
+     * @return                  {return value, parsed text attribute}
+     *
+     * If the return value is ERROR, the parsed value is still guaranteed to be valid.
+     */
+    SGRParseReturn parseSGRSequence( const TextAttribute& currentTextAttr,
+                                     const std::string& sequence );
+
+  private:
+    TextAttribute defaultTextAttr_;
+};
+
+class ColorTable;
+
+class SGRParseCore {
+    friend class ColorTable;
+
+  public:
+    enum class ReturnVal {
+        RETURN_SUCCESS_BREAK,
+        RETURN_SUCCESS_CONTINUE,
+
+        RETURN_ERROR_BREAK,
+        RETURN_ERROR_CONTINUE,
+    };
+
+    enum class ParseResult {
+        RESULT_UNSUPPORTED_ATTR,
+        RESULT_FRONT_COLOR,
+        RESULT_BACK_COLOR,
+        RESULT_DEFAULT_FRONT_COLOR,
+        RESULT_DEFAULT_BACK_COLOR,
+        RESULT_DEFAULT_TEXT_ATTR,
+        RESULT_CURRENT_TEXT_ATTR,
+    };
+
+  private:
+    enum class ColorVersion : uint8_t {
+        BIT_8 = 5,
+        BIT_24 = 2,
+    };
+
+    enum class ParseState {
+        STATE_WAIT_FIRST_PARAMETER,
+        STATE_WAIT_VERSION,
+        STATE_WAIT_BIT_8_ARGS,
+        STATE_WAIT_BIT_24_ARGS_R,
+        STATE_WAIT_BIT_24_ARGS_G,
+        STATE_WAIT_BIT_24_ARGS_B,
+    };
+
+  public:
+    SGRParseCore();
+    ~SGRParseCore() = default;
+
+    SGRParseCore( const SGRParseCore& ) = default;
+    SGRParseCore( SGRParseCore&& ) = default;
+    SGRParseCore& operator=( const SGRParseCore& ) = default;
+    SGRParseCore& operator=( SGRParseCore&& ) = default;
+
+    ReturnVal parse( std::string_view& seqs );
+
+    inline void reset()
+    {
+        new ( this ) SGRParseCore();
+    }
+
+    inline ParseResult result()
+    {
+        return result_;
+    }
+
+    inline RGB color()
+    {
+        return color_;
+    }
+
+  private:
+    SGRParseCore( ParseResult result, RGB rgb,
+                  ParseState s = ParseState::STATE_WAIT_FIRST_PARAMETER );
+
+    ReturnVal stringToParameter( const std::string_view& in, uint8_t& out );
+
+    ReturnVal setFirstParameter( const std::string_view& num );
+    ReturnVal setColorVersion( const std::string_view& num );
+    ReturnVal setBit8Color( const std::string_view& num );
+    ReturnVal setBit24Color( const std::string_view& num );
+
+  private:
+    ParseResult result_;
+    ParseState state_;
+    RGB color_;
+    bool bit24Valid_;
+};
+
+class ColorTable {
+  public:
+    enum ColorIndex : uint8_t {
+        RESET_DEFAULT = 0,
+
+        // 3/4-bit front color
+        F_BLACK = 30,
+        F_RED = 31,
+        F_GREEN = 32,
+        F_YELLOW = 33,
+        F_BLUE = 34,
+        F_MAGENTA = 35,
+        F_CYAN = 36,
+        F_WHITE = 37,
+
+        // custom front color
+        F_CUSTOM_COLOR = 38,
+        // default front color
+        F_DEFAULT_COLOR = 39,
+
+        // 3/4-bit back color
+        B_BLACK = 40,
+        B_RED = 41,
+        B_GREEN = 42,
+        B_YELLOW = 43,
+        B_BLUE = 44,
+        B_MAGENTA = 45,
+        B_CYAN = 46,
+        B_WHITE = 47,
+
+        // custom back color
+        B_CUSTOM_COLOR = 48,
+        // default back color
+        B_DEFAULT_COLOR = 49,
+
+        // 3/4-bit front bright color
+        F_BRIGHT_BLACK = 90,
+        F_BRIGHT_RED = 91,
+        F_BRIGHT_GREEN = 92,
+        F_BRIGHT_YELLOW = 93,
+        F_BRIGHT_BLUE = 94,
+        F_BRIGHT_MAGENTA = 95,
+        F_BRIGHT_CYAN = 96,
+        F_BRIGHT_WHITE = 97,
+
+        // 3/4-bit back bright color
+        B_BRIGHT_BLACK = 100,
+        B_BRIGHT_RED = 101,
+        B_BRIGHT_GREEN = 102,
+        B_BRIGHT_YELLOW = 103,
+        B_BRIGHT_BLUE = 104,
+        B_BRIGHT_MAGENTA = 105,
+        B_BRIGHT_CYAN = 106,
+        B_BRIGHT_WHITE = 107,
+    };
+
+  public:
+    static SGRParseCore index( ColorIndex num );
+
+  private:
+    static std::map<ColorIndex, SGRParseCore> colorTable;
+};
+
+} // namespace ANSI