LibAudio: Detect and read FLAC metadata

FLAC uses the very simple vorbis comment metadata format, which we can
now read most standard and non-standard fields from.

5 files changed, 160 insertions, 1 deletions

diff --git a/Userland/Libraries/LibAudio/CMakeLists.txt b/Userland/Libraries/LibAudio/CMakeLists.txt
index b77d0bedb7..8f9f665acb 100644
--- a/Userland/Libraries/LibAudio/CMakeLists.txt
+++ b/Userland/Libraries/LibAudio/CMakeLists.txt
@@ -9,6 +9,7 @@ set(SOURCES
     QOALoader.cpp
     QOATypes.cpp
     UserSampleQueue.cpp
+    VorbisComment.cpp
 )
 
 if (SERENITYOS)
@@ -20,4 +21,4 @@ if (SERENITYOS)
 endif()
 
 serenity_lib(LibAudio audio)
-target_link_libraries(LibAudio PRIVATE LibCore LibIPC LibThreading)
+target_link_libraries(LibAudio PRIVATE LibCore LibIPC LibThreading LibUnicode)
diff --git a/Userland/Libraries/LibAudio/FlacLoader.cpp b/Userland/Libraries/LibAudio/FlacLoader.cpp
index 35206b949c..1f44634fdb 100644
--- a/Userland/Libraries/LibAudio/FlacLoader.cpp
+++ b/Userland/Libraries/LibAudio/FlacLoader.cpp
@@ -21,6 +21,7 @@
 #include <LibAudio/FlacTypes.h>
 #include <LibAudio/LoaderError.h>
 #include <LibAudio/Resampler.h>
+#include <LibAudio/VorbisComment.h>
 #include <LibCore/File.h>
 
 namespace Audio {
@@ -131,6 +132,10 @@ MaybeLoaderError FlacLoaderPlugin::parse_header()
             [[fallthrough]];
         case FlacMetadataBlockType::PADDING:
             // Note: A padding block is empty and does not need any treatment.
+            break;
+        case FlacMetadataBlockType::VORBIS_COMMENT:
+            load_vorbis_comment(block);
+            break;
         default:
             // TODO: Parse the remaining metadata block types.
             break;
@@ -180,6 +185,17 @@ MaybeLoaderError FlacLoaderPlugin::load_picture(FlacRawMetadataBlock& block)
     return {};
 }
 
+// 11.15. METADATA_BLOCK_VORBIS_COMMENT
+void FlacLoaderPlugin::load_vorbis_comment(FlacRawMetadataBlock& block)
+{
+    auto metadata_or_error = Audio::load_vorbis_comment(block.data);
+    if (metadata_or_error.is_error()) {
+        dbgln("FLAC Warning: Vorbis comment invalid, error: {}", metadata_or_error.release_error());
+        return;
+    }
+    m_metadata = metadata_or_error.release_value();
+}
+
 // 11.13. METADATA_BLOCK_SEEKTABLE
 MaybeLoaderError FlacLoaderPlugin::load_seektable(FlacRawMetadataBlock& block)
 {
diff --git a/Userland/Libraries/LibAudio/FlacLoader.h b/Userland/Libraries/LibAudio/FlacLoader.h
index 043ab58435..3bac59af78 100644
--- a/Userland/Libraries/LibAudio/FlacLoader.h
+++ b/Userland/Libraries/LibAudio/FlacLoader.h
@@ -78,6 +78,8 @@ private:
     // decode a single rice partition that has its own rice parameter
     ALWAYS_INLINE ErrorOr<Vector<i32>, LoaderError> decode_rice_partition(u8 partition_type, u32 partitions, u32 partition_index, FlacSubframeHeader& subframe, BigEndianInputBitStream& bit_input);
     MaybeLoaderError load_seektable(FlacRawMetadataBlock&);
+    // Note that failing to read a Vorbis comment block is not treated as an error of the FLAC loader, since metadata is optional.
+    void load_vorbis_comment(FlacRawMetadataBlock&);
     MaybeLoaderError load_picture(FlacRawMetadataBlock&);
 
     // Converters for special coding used in frame headers
diff --git a/Userland/Libraries/LibAudio/VorbisComment.cpp b/Userland/Libraries/LibAudio/VorbisComment.cpp
new file mode 100644
index 0000000000..78e211d9e4
--- /dev/null
+++ b/Userland/Libraries/LibAudio/VorbisComment.cpp
@@ -0,0 +1,122 @@
+/*
+ * Copyright (c) 2023, kleines Filmröllchen <filmroellchen@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#include "VorbisComment.h"
+#include <AK/Endian.h>
+#include <AK/MemoryStream.h>
+#include <AK/Span.h>
+#include <AK/String.h>
+
+namespace Audio {
+
+// "Content vector format"
+static ErrorOr<void> read_vorbis_field(Metadata& metadata_to_write_into, String const& unparsed_user_comment)
+{
+    // Technically the field name has to be ASCII, but we just accept all UTF-8.
+    auto field_name_and_contents = TRY(unparsed_user_comment.split_limit('=', 2));
+
+    if (field_name_and_contents.size() != 2)
+        return Error::from_string_view("User comment does not contain '='"sv);
+    auto contents = field_name_and_contents.take_last();
+    auto field_name = TRY(field_name_and_contents.take_first().to_uppercase());
+
+    // Some of these are taken from https://age.hobba.nl/audio/tag_frame_reference.html
+    if (field_name == "TITLE"sv) {
+        if (metadata_to_write_into.title.has_value())
+            TRY(metadata_to_write_into.add_miscellaneous(field_name, contents));
+        else
+            metadata_to_write_into.title = contents;
+    } else if (field_name == "VERSION"sv) {
+        if (metadata_to_write_into.subtitle.has_value())
+            TRY(metadata_to_write_into.add_miscellaneous(field_name, contents));
+        else
+            metadata_to_write_into.subtitle = contents;
+    } else if (field_name == "ALBUM"sv) {
+        if (metadata_to_write_into.album.has_value())
+            TRY(metadata_to_write_into.add_miscellaneous(field_name, contents));
+        else
+            metadata_to_write_into.album = contents;
+    } else if (field_name == "COPYRIGHT"sv) {
+        if (metadata_to_write_into.copyright.has_value())
+            TRY(metadata_to_write_into.add_miscellaneous(field_name, contents));
+        else
+            metadata_to_write_into.copyright = contents;
+    } else if (field_name == "ISRC"sv) {
+        if (metadata_to_write_into.isrc.has_value())
+            TRY(metadata_to_write_into.add_miscellaneous(field_name, contents));
+        else
+            metadata_to_write_into.isrc = contents;
+    } else if (field_name == "GENRE"sv) {
+        if (metadata_to_write_into.genre.has_value())
+            TRY(metadata_to_write_into.add_miscellaneous(field_name, contents));
+        else
+            metadata_to_write_into.genre = contents;
+    } else if (field_name == "COMMENT"sv) {
+        if (metadata_to_write_into.comment.has_value())
+            TRY(metadata_to_write_into.add_miscellaneous(field_name, contents));
+        else
+            metadata_to_write_into.comment = contents;
+    } else if (field_name == "TRACKNUMBER"sv) {
+        if (metadata_to_write_into.track_number.has_value())
+            TRY(metadata_to_write_into.add_miscellaneous(field_name, contents));
+        else if (auto maybe_number = contents.to_number<unsigned>(); maybe_number.has_value())
+            metadata_to_write_into.track_number = maybe_number.release_value();
+        else
+            TRY(metadata_to_write_into.add_miscellaneous(field_name, contents));
+    } else if (field_name == "DATE"sv) {
+        if (metadata_to_write_into.unparsed_time.has_value())
+            TRY(metadata_to_write_into.add_miscellaneous(field_name, contents));
+        else
+            metadata_to_write_into.unparsed_time = contents;
+    } else if (field_name == "PERFORMER"sv) {
+        TRY(metadata_to_write_into.add_person(Person::Role::Performer, contents));
+    } else if (field_name == "ARTIST"sv) {
+        TRY(metadata_to_write_into.add_person(Person::Role::Artist, contents));
+    } else if (field_name == "COMPOSER"sv) {
+        TRY(metadata_to_write_into.add_person(Person::Role::Composer, contents));
+    } else if (field_name == "CONDUCTOR"sv) {
+        TRY(metadata_to_write_into.add_person(Person::Role::Conductor, contents));
+    } else if (field_name == "LYRICIST"sv) {
+        TRY(metadata_to_write_into.add_person(Person::Role::Lyricist, contents));
+    } else if (field_name == "ORGANIZATION"sv) {
+        TRY(metadata_to_write_into.add_person(Person::Role::Publisher, contents));
+    } else if (field_name == "PUBLISHER"sv) {
+        TRY(metadata_to_write_into.add_person(Person::Role::Publisher, contents));
+    } else if (field_name == "ENCODED-BY"sv) {
+        TRY(metadata_to_write_into.add_person(Person::Role::Engineer, contents));
+    } else {
+        TRY(metadata_to_write_into.add_miscellaneous(field_name, contents));
+    }
+
+    return {};
+}
+
+ErrorOr<Metadata, LoaderError> load_vorbis_comment(ByteBuffer const& vorbis_comment)
+{
+    FixedMemoryStream stream { vorbis_comment };
+    auto vendor_length = TRY(stream.read_value<LittleEndian<u32>>());
+    Vector<u8> raw_vendor_string;
+    TRY(raw_vendor_string.try_resize(vendor_length));
+    TRY(stream.read_entire_buffer(raw_vendor_string));
+    auto vendor_string = TRY(String::from_utf8(StringView { raw_vendor_string.span() }));
+
+    Metadata metadata;
+    metadata.encoder = move(vendor_string);
+
+    auto user_comment_count = TRY(stream.read_value<LittleEndian<u32>>());
+    for (size_t i = 0; i < user_comment_count; ++i) {
+        auto user_comment_length = TRY(stream.read_value<LittleEndian<u32>>());
+        Vector<u8> raw_user_comment;
+        TRY(raw_user_comment.try_resize(user_comment_length));
+        TRY(stream.read_entire_buffer(raw_user_comment));
+        auto unparsed_user_comment = TRY(String::from_utf8(StringView { raw_user_comment.span() }));
+        TRY(read_vorbis_field(metadata, unparsed_user_comment));
+    }
+
+    return metadata;
+}
+
+}
diff --git a/Userland/Libraries/LibAudio/VorbisComment.h b/Userland/Libraries/LibAudio/VorbisComment.h
new file mode 100644
index 0000000000..f505dc315d
--- /dev/null
+++ b/Userland/Libraries/LibAudio/VorbisComment.h
@@ -0,0 +1,18 @@
+/*
+ * Copyright (c) 2023, kleines Filmröllchen <filmroellchen@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include <AK/ByteBuffer.h>
+#include <LibAudio/LoaderError.h>
+#include <LibAudio/Metadata.h>
+
+namespace Audio {
+
+// https://www.xiph.org/vorbis/doc/v-comment.html
+ErrorOr<Metadata, LoaderError> load_vorbis_comment(ByteBuffer const& vorbis_comment);
+
+}