From 59a157ebdecf3705f5705749b53009842307c6de Mon Sep 17 00:00:00 2001
From: Colin B <lanzaa@1101b.com>
Date: Mon, 13 Mar 2023 11:58:59 -0700
Subject: [PATCH] feat: restore support for nonstandard `path.utf-8` (#3454)

---
 libtransmission/torrent-metainfo.cc           |  26 ++++++++++++++----
 tests/utils/CMakeLists.txt                    |  11 ++++----
 .../utils/assets/bitcomet.utf-8.special.show  |  26 ++++++++++++++++++
 .../assets/bitcomet.utf-8.special.torrent     | Bin 0 -> 638 bytes
 tests/utils/run_transmission_show.cmake       |   4 +++
 5 files changed, 56 insertions(+), 11 deletions(-)
 create mode 100644 tests/utils/assets/bitcomet.utf-8.special.show
 create mode 100644 tests/utils/assets/bitcomet.utf-8.special.torrent

diff --git a/libtransmission/torrent-metainfo.cc b/libtransmission/torrent-metainfo.cc
index 19209b160..fe3ca41f0 100644
--- a/libtransmission/torrent-metainfo.cc
+++ b/libtransmission/torrent-metainfo.cc
@@ -171,6 +171,11 @@ struct MetainfoHandler final : public transmission::benc::BasicHandler<MaxBencDe
             file_subpath_.clear();
             file_length_ = 0;
         }
+        else if (pathStartsWith(InfoKey, FilesKey, ""sv, PathUtf8Key))
+        {
+            // torrent has a utf8 path, drop the other one due to probable non-utf8 encoding
+            file_subpath_.clear();
+        }
 
         return BasicHandler::StartArray(context);
     }
@@ -251,7 +256,8 @@ struct MetainfoHandler final : public transmission::benc::BasicHandler<MaxBencDe
             pathStartsWith(InfoKey, FileDurationKey) || //
             pathStartsWith(InfoKey, FileMediaKey) || //
             pathStartsWith(InfoKey, ProfilesKey) || //
-            pathStartsWith(LibtorrentResumeKey))
+            pathStartsWith(LibtorrentResumeKey) || //
+            pathStartsWith(NodesKey))
         {
             // unused by Transmission
         }
@@ -292,7 +298,7 @@ struct MetainfoHandler final : public transmission::benc::BasicHandler<MaxBencDe
         }
         else if (state_ == State::Files)
         {
-            if (curdepth > 1 && key(curdepth - 1) == PathKey)
+            if (curdepth > 1 && (key(curdepth - 1) == PathKey || key(curdepth - 1) == PathUtf8Key))
             {
                 if (!std::empty(file_subpath_))
                 {
@@ -312,8 +318,7 @@ struct MetainfoHandler final : public transmission::benc::BasicHandler<MaxBencDe
                 pathIs(InfoKey, FilesKey, ""sv, Md5Key) || //
                 pathIs(InfoKey, FilesKey, ""sv, Md5sumKey) || //
                 pathIs(InfoKey, FilesKey, ""sv, MtimeKey) || // (why a string?)
-                pathIs(InfoKey, FilesKey, ""sv, Sha1Key) || //
-                pathStartsWith(InfoKey, FilesKey, ""sv, PathUtf8Key))
+                pathIs(InfoKey, FilesKey, ""sv, Sha1Key))
             {
                 // unused by Transmission
             }
@@ -330,7 +335,10 @@ struct MetainfoHandler final : public transmission::benc::BasicHandler<MaxBencDe
         {
             tm_.creator_ = tr_strv_replace_invalid(value);
         }
-        else if (pathIs(SourceKey) || pathIs(InfoKey, SourceKey) || pathIs(PublisherKey) || pathIs(InfoKey, PublisherKey))
+        else if (
+            pathIs(SourceKey) || pathIs(InfoKey, SourceKey) || //
+            pathIs(PublisherKey) || pathIs(InfoKey, PublisherKey) || //
+            pathIs(PublisherUtf8Key) || pathIs(InfoKey, PublisherUtf8Key))
         {
             // “publisher” is rare, but used by BitComet and appears
             // to have the same use as the 'source' key
@@ -404,12 +412,14 @@ struct MetainfoHandler final : public transmission::benc::BasicHandler<MaxBencDe
             pathIs(InfoKey, EntropyKey) || //
             pathIs(InfoKey, Md5sumKey) || //
             pathIs(InfoKey, PublisherUrlKey) || //
+            pathIs(InfoKey, PublisherUrlUtf8Key) || //
             pathIs(InfoKey, Sha1Key) || //
             pathIs(InfoKey, UniqueKey) || //
             pathIs(InfoKey, XCrossSeedKey) || //
             pathIs(LocaleKey) || //
             pathIs(LogCallbackKey) || //
             pathIs(PublisherUrlKey) || //
+            pathIs(PublisherUrlUtf8Key) || //
             pathIs(TitleKey) || //
             pathStartsWith(AzureusPrivatePropertiesKey) || //
             pathStartsWith(AzureusPropertiesKey) || //
@@ -417,7 +427,8 @@ struct MetainfoHandler final : public transmission::benc::BasicHandler<MaxBencDe
             pathStartsWith(InfoKey, FileDurationKey) || //
             pathStartsWith(InfoKey, ProfilesKey) || //
             pathStartsWith(LibtorrentResumeKey) || //
-            pathStartsWith(MagnetInfoKey))
+            pathStartsWith(MagnetInfoKey) || //
+            pathStartsWith(NodesKey))
         {
             // unused by Transmission
         }
@@ -594,6 +605,7 @@ private:
     static constexpr std::string_view MtimeKey = "mtime"sv;
     static constexpr std::string_view NameKey = "name"sv;
     static constexpr std::string_view NameUtf8Key = "name.utf-8"sv;
+    static constexpr std::string_view NodesKey = "nodes"sv;
     static constexpr std::string_view PathKey = "path"sv;
     static constexpr std::string_view PathUtf8Key = "path.utf-8"sv;
     static constexpr std::string_view PieceLayersKey = "piece layers"sv;
@@ -603,7 +615,9 @@ private:
     static constexpr std::string_view PrivateKey = "private"sv;
     static constexpr std::string_view ProfilesKey = "profiles"sv;
     static constexpr std::string_view PublisherKey = "publisher"sv;
+    static constexpr std::string_view PublisherUtf8Key = "publisher.utf-8"sv;
     static constexpr std::string_view PublisherUrlKey = "publisher-url"sv;
+    static constexpr std::string_view PublisherUrlUtf8Key = "publisher-url.utf-8"sv;
     static constexpr std::string_view Sha1Key = "sha1"sv;
     static constexpr std::string_view SourceKey = "source"sv;
     static constexpr std::string_view TitleKey = "title"sv;
diff --git a/tests/utils/CMakeLists.txt b/tests/utils/CMakeLists.txt
index 146ebe573..ea7995e7d 100644
--- a/tests/utils/CMakeLists.txt
+++ b/tests/utils/CMakeLists.txt
@@ -11,8 +11,9 @@ function(AddShowTest name file_basename)
 endfunction()
 
 # disabled AddShowTest(transmission-show-bittorrent-v2 bittorrent-v2-test)
-addshowtest(transmission-show-bittorrent-v2-hybrid-test bittorrent-v2-hybrid-test)
-addshowtest(transmission-show-inner-sanctum Inner_Sanctum_movie_archive)
-addshowtest(transmission-show-thor Thor_and_the_Amazon_Women.avi)
-addshowtest(transmission-show-ubuntu ubuntu-20.04.3-desktop-amd64.iso)
-addshowtest(transmission-show-ubuntu-hybrid hybrid-single-ubuntu-20.04.3-desktop-amd64.iso)
+AddShowTest(transmission-show-bittorrent-v2-hybrid-test bittorrent-v2-hybrid-test)
+AddShowTest(transmission-show-inner-sanctum Inner_Sanctum_movie_archive)
+AddShowTest(transmission-show-thor Thor_and_the_Amazon_Women.avi)
+AddShowTest(transmission-show-ubuntu ubuntu-20.04.3-desktop-amd64.iso)
+AddShowTest(transmission-show-ubuntu-hybrid hybrid-single-ubuntu-20.04.3-desktop-amd64.iso)
+AddShowTest(transmission-show-bitcomet-path.utf8 bitcomet.utf-8.special)
diff --git a/tests/utils/assets/bitcomet.utf-8.special.show b/tests/utils/assets/bitcomet.utf-8.special.show
new file mode 100644
index 000000000..fb61f96f7
--- /dev/null
+++ b/tests/utils/assets/bitcomet.utf-8.special.show
@@ -0,0 +1,26 @@
+Name: modified test-utf8
+File: assets/bitcomet.utf-8.special.torrent
+
+GENERAL
+
+  Name: modified test-utf8
+  Hash v1: 52bb66ab192b38cad7ce20fb8ee2dfe1d870bea8
+  Created by: Hand with love
+  Created on: Sat Jul 16 21:13:04 2022
+
+  Source: 私たちの世界
+  Piece Count: 1
+  Piece Size: 1.00 MiB
+  Total Size: 0.01 kB
+  Privacy: Public torrent
+
+TRACKERS
+
+  Tier #1
+  http://fake.example.com:7070/announce
+
+FILES
+
+  modified test-utf8/別の世界f1.txt (0.01 kB)
+  modified test-utf8/別の世界f2.txt (0.01 kB)
+
diff --git a/tests/utils/assets/bitcomet.utf-8.special.torrent b/tests/utils/assets/bitcomet.utf-8.special.torrent
new file mode 100644
index 0000000000000000000000000000000000000000..d5091028eff688b456ebab045524d3c62c050611
GIT binary patch
literal 638
zcma)3Jx{|h5M^fTKga-M6Fbd^dkm3SK;lo3CYQ#-v7^L+GB89e3`k6{A_m09%>OB<
z3qJyfwkaSX)gG*S_wMOEU&<hsQZ+KcQ3$!#B{)tNpP`MjxF`kMi7Fr@A#v6`I_gJ;
zg4bGzZ6^^afuzD%V`@z<sRzfgOsxy9b4#dmG|JyyuB4S35~qRB2=T|vLk1`lm2x>9
z!O`J~2V7<<^&#T|Ymo*ZP)>Eu1N5MbbuK6aK9#CMtD{)d>d?+W_u7rlhOFa!+&s_Q
z*Y~%F#r<tYZ9UV78@2EsHFp~|cd2G1WL$LN3gbWHCcDO38z&vSbYhJBY7UJ*MqMa5
zCTOjqQQ|Q_3@}*2wM*ct*@*_t;2?@3j6SdL_V!PsF_g{3EGEZ_hU-7*p~(2?i(yj<
zQ?J%8R1yn^{+RFjD}#So{zj#buy|?b?c=<C>VFh{kSa9`5@VolXcKcOXM_<nXZH(&
C{^qy<

literal 0
HcmV?d00001

diff --git a/tests/utils/run_transmission_show.cmake b/tests/utils/run_transmission_show.cmake
index 00148a024..e3c526cf4 100644
--- a/tests/utils/run_transmission_show.cmake
+++ b/tests/utils/run_transmission_show.cmake
@@ -14,9 +14,13 @@ else()
     message(STATUS "      output_file ${output_file}")
     message(STATUS "   reference_file ${reference_file}")
 
+    # We want UTF-8
+    set(ENV{LC_ALL} "en_US.UTF-8")
+
     # The app's output includes timestamps, so fake our TZ to ensure
     # the test doesn't depend on the physical TZ of the test machine
     set(ENV{TZ} "UTC")
+
     execute_process(
         COMMAND ${transmission_show} ${torrent_file}
         OUTPUT_FILE ${output_file})