Merge pull request JDimproved#1277 from JDimproved/add-guess-encoding…

…-from-text-option add guess encoding from text option
ma8ma · Oct 21, 2023 · 7e23cb3 · 7e23cb3
2 parents 058a6eb + 42db53c
commit 7e23cb3
Show file tree

Hide file tree

Showing 4 changed files with 46 additions and 16 deletions.
diff --git a/src/board/preference.cpp b/src/board/preference.cpp
@@ -211,25 +211,33 @@ Preferences::Preferences( Gtk::Window* parent, const std::string& url, const std
         m_radio_encoding_http_header.set_group( m_group_encoding );
         m_radio_encoding_http_header.set_label( "HTTPヘッダーのエンコーディング情報を使う" );
 
+        m_radio_encoding_guess.set_group( m_group_encoding );
+        m_radio_encoding_guess.set_label( "テキストからエンコーディングを推測する" );
+
         m_vbox_encoding_analysis_method.set_margin_start( 30 );
         m_vbox_encoding_analysis_method.set_margin_end( 30 );
         m_vbox_encoding_analysis_method.set_margin_top( 10 );
         m_vbox_encoding_analysis_method.set_margin_bottom( 10 );
         m_vbox_encoding_analysis_method.pack_start( m_label_encoding_analysis_method, Gtk::PACK_SHRINK );
         m_vbox_encoding_analysis_method.pack_start( m_radio_encoding_default, Gtk::PACK_SHRINK );
         m_vbox_encoding_analysis_method.pack_start( m_radio_encoding_http_header, Gtk::PACK_SHRINK );
+        m_vbox_encoding_analysis_method.pack_start( m_radio_encoding_guess, Gtk::PACK_SHRINK );
 
         m_revealer_encoding.add( m_vbox_encoding_analysis_method );
 
         m_vbox_encoding.pack_start( m_hbox_encoding, Gtk::PACK_SHRINK );
         m_vbox_encoding.pack_start( m_revealer_encoding, Gtk::PACK_SHRINK );
 
-        if( const int method = DBTREE::board_encoding_analysis_method( get_url() );
-                method == EncodingAnalysisMethod::http_header ) {
-            m_radio_encoding_http_header.set_active( true );
-        }
-        else {
-            m_radio_encoding_default.set_active( true );
+        switch( DBTREE::board_encoding_analysis_method( get_url() ) ) {
+            case EncodingAnalysisMethod::http_header:
+                m_radio_encoding_http_header.set_active( true );
+                break;
+            case EncodingAnalysisMethod::guess:
+                m_radio_encoding_guess.set_active( true );
+                break;
+            default:
+                m_radio_encoding_default.set_active( true );
+                break;
         }
     }
 
@@ -624,6 +632,9 @@ void Preferences::slot_ok_clicked()
     if( m_radio_encoding_http_header.get_active() ) {
         DBTREE::board_set_encoding_analysis_method( get_url(), EncodingAnalysisMethod::http_header );
     }
+    else if( m_radio_encoding_guess.get_active() ) {
+        DBTREE::board_set_encoding_analysis_method( get_url(), EncodingAnalysisMethod::guess );
+    }
     else {
         DBTREE::board_set_encoding_analysis_method( get_url(), EncodingAnalysisMethod::use_default );
     }

diff --git a/src/board/preference.h b/src/board/preference.h
@@ -98,6 +98,7 @@ namespace BOARD
         Gtk::RadioButtonGroup m_group_encoding;
         Gtk::RadioButton m_radio_encoding_default;
         Gtk::RadioButton m_radio_encoding_http_header;
+        Gtk::RadioButton m_radio_encoding_guess;
 
         // ネットワーク設定
         Gtk::Box m_vbox_network;

diff --git a/src/jdencoding.h b/src/jdencoding.h
@@ -30,7 +30,8 @@ struct EncodingAnalysisMethod
 {
     static constexpr const int use_default = 0; ///< デフォルト設定を使う
     static constexpr const int http_header = 1; ///< HTTPヘッダーのエンコーディング情報を使う
-    static constexpr const int max = http_header;
+    static constexpr const int guess = 2; ///< テキストからエンコーディングを推測する
+    static constexpr const int max = guess;
 };
 
 #endif
diff --git a/src/skeleton/loadable.cpp b/src/skeleton/loadable.cpp
@@ -20,8 +20,8 @@ enum class Loadable::CharsetDetection
 {
     parse_header, ///< HTTP header を解析する
     parse_meta,   ///< HTML meta 要素を解析する
-    finished,     ///< 検出を終えた
-    use_default,  ///< デフォルト設定を使う
+    finished,     ///< 検出を終えた、またはデフォルト設定を使う
+    guess,        ///< テキストからエンコーディングを推測する
 };
 
 
@@ -113,12 +113,17 @@ bool Loadable::start_load( const JDLIB::LOADERDATA& data )
 
     // 情報初期化
     // m_date_modified, m_cookie は初期化しない
-    if( data.encoding_analysis_method == EncodingAnalysisMethod::http_header ) {
-        m_charset_det = CharsetDetection::parse_header;
-    }
-    else {
-        m_charset_det = CharsetDetection::use_default;
-        set_encoding( m_default_encoding );
+    switch( data.encoding_analysis_method ) {
+        case EncodingAnalysisMethod::http_header:
+            m_charset_det = CharsetDetection::parse_header;
+            break;
+        case EncodingAnalysisMethod::guess:
+            m_charset_det = CharsetDetection::guess;
+            break;
+        default:
+            m_charset_det = CharsetDetection::finished;
+            set_encoding( m_default_encoding );
+            break;
     }
     m_code = HTTP_INIT;
     m_str_code = std::string();
@@ -156,7 +161,19 @@ void Loadable::receive( const char* data, size_t size )
     if( ! m_total_length && m_code != HTTP_INIT ) m_total_length = get_loader_length();
     m_current_length += size;
 
-    if( m_charset_det == CharsetDetection::parse_header ) {
+    if( m_charset_det == CharsetDetection::finished ) {
+        receive_data( std::string_view{ data, size } );
+        return;
+    }
+
+    if( m_charset_det == CharsetDetection::guess ) {
+        const Encoding enc = MISC::detect_encoding( std::string_view{ data, size } );
+        if( enc != Encoding::unknown ) {
+            set_encoding( enc );
+            m_charset_det = CharsetDetection::finished;
+        }
+    }
+    else if( m_charset_det == CharsetDetection::parse_header ) {
         const Encoding enc = get_loader_content_charset();
         if( enc != Encoding::unknown ) {
             set_encoding( enc );