mkvpropedit: add option for normalizing IETF BCP 47 language tags

Part of the implementation of #3307.

src/propedit/propedit_cli_parser.cpp

@@ -214,6 +214,19 @@ propedit_cli_parser_c::disable_language_ietf() {
   mtx::bcp47::language_c::disable();
 }
 
+void
+propedit_cli_parser_c::set_language_ietf_normalization_mode() {
+  if (m_next_arg.empty())
+    mxerror(fmt::format(Y("'{0}' lacks its argument.\n"), "--normalize-language-ietf"));
+
+  if ((m_next_arg != "canonical"s) && (m_next_arg != "extlang"s))
+    mxerror(fmt::format(Y("'{0}' is not a valid language normalization mode.\n"), m_next_arg));
+
+  auto mode = m_next_arg  == "canonical"s ? mtx::bcp47::normalization_mode_e::canonical : mtx::bcp47::normalization_mode_e::extlang;
+
+  mtx::bcp47::language_c::set_normalization_mode(mode);
+}
+
 void
 propedit_cli_parser_c::enable_legacy_font_mime_types() {
   g_use_legacy_font_mime_types = true;
@@ -252,7 +265,8 @@ propedit_cli_parser_c::init_parser() {
   add_option("attachment-uid=<uid>",                              std::bind(&propedit_cli_parser_c::set_attachment_uid,         this), YT("Set the UID to use for the following '--add-attachment', '--replace-attachment' or '--update-attachment' option"));
 
   add_section_header(YT("Other options"));
-  add_option("disable-language-ietf", std::bind(&propedit_cli_parser_c::disable_language_ietf, this), YT("Do not change LanguageIETF track header elements when the 'language' property is changed."));
+  add_option("disable-language-ietf",          std::bind(&propedit_cli_parser_c::disable_language_ietf,                this), YT("Do not change LanguageIETF track header elements when the 'language' property is changed."));
+  add_option("normalize-language-ietf=<mode>", std::bind(&propedit_cli_parser_c::set_language_ietf_normalization_mode, this), YT("Normalize all IETF BCP 47 language tags of changed elements to either their canonical form (mode 'canonical') or their extended language subtags form (mode 'extlang')"));
   add_common_options();
 
   add_separator();
@@ -281,6 +295,8 @@ propedit_cli_parser_c::init_parser() {
   add_information(YT("3. Either 'name:<value>' or 'mime-type:<value>' in which case the selector applies to all attachments whose name or MIME type respectively equals <value>."), 2);
 
   add_hook(mtx::cli::parser_c::ht_unknown_option, std::bind(&propedit_cli_parser_c::set_file_name, this));
+
+  set_to_parse_first("--normalize-language-ietf");
 }
 
 void

src/propedit/propedit_cli_parser.h

@@ -40,6 +40,7 @@ protected:
   void set_file_name();
   void disable_language_ietf();
   void enable_legacy_font_mime_types();
+  void set_language_ietf_normalization_mode();
 
   void set_attachment_name();
   void set_attachment_description();

tests/results.txt

@@ -582,3 +582,6 @@ T_0734legacy_language_code_fallback_via_extlang_prefix:27e262aae722c5306f6ffcfbe
 T_0735normalize_language_ietf_tracks:fc2fbd7d3978298f5e0a91b4cc222075-ok-46cd0d3e23f2293952e01cd72758ce25-ok-536dc6a75677c959a8fcccc0a389d651-ok:passed:20220327-103026:0.315131033
 T_0736normalize_language_ietf_chapters:60865f8a77e2ba933357027ced7dda3b-ok-336926f4fed38eaaa240d260df7789fc-ok-a0ee62bbac0b3cdc4d8d285b9571eefe-ok:passed:20220327-103029:0.173048279
 T_0737normalize_language_ietf_tags:b23b6fb1821313e506780642ba35435e-ok-b477e432c6701cbdbcdf4734bc524d91-ok-ed6c2ba405d9e276aca9a5cf0c540406-ok:passed:20220327-103814:0.178980995
+T_0738propedit_normalize_language_ietf_tracks:8c830d72272956b13cc467192241c52d-f727657a1cc3e8898503676c058de2a8-ok-8c830d72272956b13cc467192241c52d-667ca2882b82259f833bf4b4a4375e06-ok-8c830d72272956b13cc467192241c52d-99381838032c53bc1d4230c85bdaf9c0-ok:passed:20220327-114045:0.365823403
+T_0739propedit_normalize_language_ietf_chapters:fd026c48afdf123a9d500ad456f751d4-eb76e15349d1cbe0de47a82f7bcd8250-ok-fd026c48afdf123a9d500ad456f751d4-284c1493450e4a988846d6a2ced87cb8-ok-fd026c48afdf123a9d500ad456f751d4-f55da6687176bbda38205705d6f27d27-ok:passed:20220327-114610:0.353790859
+T_0740propedit_normalize_language_ietf_tags:f45e6f666f85b477a0d8c2296bea9291-66edf4dcad3528ff3506ae287979780c-ok-f45e6f666f85b477a0d8c2296bea9291-1f75e9469442a031d4602181b1151dbb-ok-f45e6f666f85b477a0d8c2296bea9291-f6b69443dab50658f393ce212103b347-ok:passed:20220327-115423:0.36762433

tests/test-0738propedit_normalize_language_ietf_tracks.rb

@@ -0,0 +1,23 @@
+#!/usr/bin/ruby -w
+
+# T_738propedit_normalize_language_ietf_tracks
+describe "mkvpropedit / normalize language IETF in tracks"
+
+def test0738 expected_languages, normalization_mode = nil
+  src = "data/subtitles/srt/ven.srt"
+
+  test_merge src, :args => "#{src} #{src}", :keep_tmp => true
+
+  test "propedit #{expected_languages}" do
+    normalization_mode = normalization_mode ? "--normalize-language-ietf #{normalization_mode}" : ""
+
+    propedit tmp, "--edit track:1 --set language=fr-fx --edit track:2 --set language=zh-yue --edit track:3 --set language=yue #{normalization_mode}"
+    hash_tmp false
+  end
+
+  compare_languages_tracks(*expected_languages)
+end
+
+test0738 [ [ "fre", "fr-FX" ], [ "chi", "zh-yue" ], [ "chi", "yue" ] ]
+test0738 [ [ "fre", "fr-FR" ], [ "chi", "yue" ],    [ "chi", "yue" ] ],    :canonical
+test0738 [ [ "fre", "fr-FR" ], [ "chi", "zh-yue" ], [ "chi", "zh-yue" ] ], :extlang

tests/test-0739propedit_normalize_language_ietf_chapters.rb

@@ -0,0 +1,24 @@
+#!/usr/bin/ruby -w
+
+# T_739propedit_normalize_language_ietf_chapters
+describe "mkvpropedit / normalize language IETF in chapters"
+
+def test0739 expected_languages, normalization_mode = nil
+  src1 = "data/subtitles/srt/ven.srt"
+  src2 = "data/chapters/ietf-normalization-test.xml"
+
+  test_merge src1, :keep_tmp => true
+
+  test "propedit #{expected_languages}" do
+    normalization_mode = normalization_mode ? "--normalize-language-ietf #{normalization_mode}" : ""
+
+    propedit tmp, "--chapters #{src2} #{normalization_mode}"
+    hash_tmp false
+  end
+
+  compare_languages_chapters(*expected_languages)
+end
+
+test0739 [ [ "fre", "fr-FX" ], [ "chi", "zh-yue" ], [ "chi", "yue" ] ]
+test0739 [ [ "fre", "fr-FR" ], [ "chi", "yue" ],    [ "chi", "yue" ] ],    :canonical
+test0739 [ [ "fre", "fr-FR" ], [ "chi", "zh-yue" ], [ "chi", "zh-yue" ] ], :extlang

tests/test-0740propedit_normalize_language_ietf_tags.rb

@@ -0,0 +1,25 @@
+#!/usr/bin/ruby -w
+
+# T_740propedit_normalize_language_ietf_tags
+describe "mkvpropedit / normalize language IETF in tags"
+
+def test0740 expected_languages, normalization_mode = nil
+  src1 = "data/subtitles/srt/ven.srt"
+  src2 = "data/tags/one-tag-fr-FX.xml"
+  src3 = "data/tags/ietf-normalization-test.xml"
+
+  test_merge src1, :args => "--tags 0:#{src2}", :keep_tmp => true
+
+  test "propedit #{expected_languages}" do
+    normalization_mode = normalization_mode ? "--normalize-language-ietf #{normalization_mode}" : ""
+
+    propedit tmp, "--tags global:#{src3} #{normalization_mode}"
+    hash_tmp false
+  end
+
+  compare_languages_tags(*expected_languages)
+end
+
+test0740 [ [ "fre", "fr-FX" ], [ "fre", "fr-FX" ], [ "chi", "zh-yue" ], [ "chi", "yue" ] ]
+test0740 [ [ "fre", "fr-FX" ], [ "fre", "fr-FR" ], [ "chi", "yue" ],    [ "chi", "yue" ] ],    :canonical
+test0740 [ [ "fre", "fr-FX" ], [ "fre", "fr-FR" ], [ "chi", "zh-yue" ], [ "chi", "zh-yue" ] ], :extlang