Merge pull request #749 from kagoromo/dev

Chaika plugin adds more tags
Difegue · Feb 14, 2023 · ee003e3 · ee003e3
2 parents 0b4e94b + db23e95
commit ee003e3
Show file tree

Hide file tree

Showing 4 changed files with 129 additions and 49 deletions.
diff --git a/lib/LANraragi/Plugin/Metadata/Chaika.pm b/lib/LANraragi/Plugin/Metadata/Chaika.pm
@@ -19,11 +19,16 @@ sub plugin_info {
         type        => "metadata",
         namespace   => "trabant",
         author      => "Difegue",
-        version     => "2.2",
+        version     => "2.3",
         description => "Searches chaika.moe for tags matching your archive. This will try to use the thumbnail first, and fallback to a default text search.",
         icon =>
           "data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAABQAAAAUCAYAAACNiR0NAAAACXBIWXMAAAsTAAALEwEAmpwYAAAA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\nSUVORK5CYII=",
-        parameters  => [ { type => "bool", desc => "Save archive title" } ],
+        parameters  => [
+            { type => "bool", desc => "Save archive title" },
+            { type => "bool", desc => "Add the following tags if available: download URL, gallery ID, category, timestamp" },
+            { type => "bool", desc => "Add tags without a namespace to the 'other:' namespace instead, mirroring E-H's behavior of namespacing everything" },
+            { type => "string", desc => "Add a custom 'source:' tag to your archive. Example: chaika. Will NOT add a tag if blank" }
+        ],
         oneshot_arg => "Chaika Gallery or Archive URL (Will attach matching tags to your archive)"
     );
 
@@ -34,7 +39,7 @@ sub get_tags {
 
     shift;
     my $lrr_info = shift;    # Global info hash
-    my ($savetitle) = @_;    # Plugin parameters
+    my ( $savetitle, $addextra, $addother, $addsource ) = @_;    # Plugin parameters
 
     my $logger   = get_plugin_logger();
     my $newtags  = "";
@@ -43,25 +48,24 @@ sub get_tags {
     # Parse the given link to see if we can extract type and ID
     my $oneshotarg = $lrr_info->{oneshot_param};
     if ( $oneshotarg =~ /https?:\/\/panda\.chaika\.moe\/(gallery|archive)\/([0-9]*)\/?.*/ ) {
-        ( $newtags, $newtitle ) = tags_from_chaika_id( $1, $2 );
+        ( $newtags, $newtitle ) = tags_from_chaika_id( $1, $2, $addextra, $addother, $addsource );
     } else {
 
         # Try SHA-1 reverse search first
         $logger->info("Using thumbnail hash " . $lrr_info->{thumbnail_hash});
-        ( $newtags, $newtitle ) = tags_from_sha1( $lrr_info->{thumbnail_hash} );
+        ( $newtags, $newtitle ) = tags_from_sha1( $lrr_info->{thumbnail_hash}, $addextra, $addother, $addsource );
 
         # Try text search if it fails
         if ( $newtags eq "" ) {
             $logger->info("No results, falling back to text search.");
-            ( $newtags, $newtitle ) = search_for_archive( $lrr_info->{archive_title}, $lrr_info->{existing_tags} );
+            ( $newtags, $newtitle ) = search_for_archive( $lrr_info->{archive_title}, $lrr_info->{existing_tags}, $addextra, $addother, $addsource );
         }
     }
 
     if ( $newtags eq "" ) {
         $logger->info("No matching Chaika Archive Found!");
         return ( error => "No matching Chaika Archive Found!" );
     } else {
-
         $logger->info("Sending the following tags to LRR: $newtags");
         #Return a hash containing the new metadata
         if ( $savetitle && $newtags ne "" ) { return ( tags => $newtags, title => $newtitle ); }
@@ -79,8 +83,7 @@ sub get_tags {
 sub search_for_archive {
 
     my $logger = get_plugin_logger();
-    my $title  = $_[0];
-    my $tags   = $_[1];
+    my ( $title, $tags, $addextra, $addother, $addsource ) = @_;
 
     #Auto-lowercase the title for better results
     $title = lc($title);
@@ -103,33 +106,33 @@ sub search_for_archive {
     my $textrep = $res->body;
     $logger->debug("Chaika API returned this JSON: $textrep");
 
-    my ( $chaitags, $chaititle ) = parse_chaika_json( $res->json->{"galleries"}->[0] );
+    my ( $chaitags, $chaititle ) = parse_chaika_json( $res->json->{"galleries"}->[0], $addextra, $addother, $addsource );
 
     return ( $chaitags, $chaititle );
 }
 
 # Uses the jsearch API to get the best json for a file.
 sub tags_from_chaika_id {
 
-    my ( $type, $ID ) = @_;
+    my ( $type, $ID, $addextra, $addother, $addsource ) = @_;
 
     my $json = get_json_from_chaika( $type, $ID );
-    return parse_chaika_json( $json );
+    return parse_chaika_json( $json, $addextra, $addother, $addsource );
 }
 
 # tags_from_sha1
 # Uses chaika's SHA-1 search with the first page hash we have.
 sub tags_from_sha1 {
 
-    my ( $sha1 ) = @_;
+    my ( $sha1, $addextra, $addother, $addsource ) = @_;
 
     my $logger = get_plugin_logger();
 
     # The jsearch API immediately returns a JSON.
     # Said JSON is an array containing multiple archive objects.
     # We just take the first one.
     my $json_by_sha1 = get_json_from_chaika( 'sha1', $sha1 );
-    return parse_chaika_json( $json_by_sha1->[0] );
+    return parse_chaika_json( $json_by_sha1->[0], $addextra, $addother, $addsource );
 }
 
 # Calls chaika's API
@@ -154,15 +157,47 @@ sub get_json_from_chaika {
 # Parses the JSON obtained from the Chaika API to get the tags.
 sub parse_chaika_json {
 
-    my ( $json ) = @_;
+    my ( $json, $addextra, $addother, $addsource ) = @_;
 
     my $tags = $json->{"tags"} || ();
     foreach my $tag (@$tags) {
         #Replace underscores with spaces
         $tag =~ s/_/ /g;
+
+        #Add 'other' namespace if none
+        if ($addother && index($tag, ":") == -1) {
+            $tag = "other:" . $tag;
+        }
     }
 
-    return ( join( ', ', @$tags ), $json->{"title"} );
+    my $category = lc $json->{"category"};
+    my $download = $json->{"download"} ? $json->{"download"} : $json->{"archives"}->[0]->{"link"};
+    my $gallery = $json->{"gallery"} ? $json->{"gallery"} : $json->{"id"};
+    my $timestamp = $json->{"posted"};
+    if ($tags && $addextra) {
+        if ($category ne "") {
+            push(@$tags, "category:" . $category);
+        }
+        if ($download ne "") {
+            push(@$tags, "download:" . $download);
+        }
+        if ($gallery ne "") {
+            push(@$tags, "gallery:" . $gallery);
+        }
+        if ($timestamp ne "") {
+            push(@$tags, "timestamp:" . $timestamp);
+        }
+    }
+
+    if ($gallery && $gallery ne "") {
+        # add custom source, but only if having found gallery
+        if ($addsource && $addsource ne "") {
+            push(@$tags, "source:" . $addsource);
+        }
+        return ( join( ', ', @$tags ), $json->{"title"} );
+    } else {
+        return "";
+    }
 }
 
 1;
diff --git a/tests/LANraragi/Plugin/Metadata/Chaika.t b/tests/LANraragi/Plugin/Metadata/Chaika.t
@@ -15,9 +15,15 @@ my $SAMPLES = "$cwd/tests/samples";
 require "$cwd/tests/mocks.pl";
 
 my @tags_list= (
-    'female:sole female', 'male:sole male', 'artist:kemuri haku', 'full censorship',
-    'male:shotacon', 'female:defloration', 'female:nakadashi', 'female:big breasts',
-    'language:translated', 'language:english'
+    'full censorship', 'female:sole female', 'male:sole male', 'artist:kemuri haku', 'female:tall girl', 
+    'female:cunnilingus', 'male:shotacon', 'female:defloration', 'female:nakadashi', 'female:x-ray',
+    'female:big breasts', 'language:translated', 'language:english'
+);
+my @tags_list_extra= (
+    'other:full censorship', 'female:sole female', 'male:sole male', 'artist:kemuri haku', 'female:tall girl', 
+    'female:cunnilingus', 'male:shotacon', 'female:defloration', 'female:nakadashi', 'female:x-ray',
+    'female:big breasts', 'language:translated', 'language:english', 'category:manga', 'download:/archive/27240/download/',
+    'gallery:23532', 'timestamp:1521357552', 'source:chaika'
 );
 
 use_ok('LANraragi::Plugin::Metadata::Chaika');
@@ -58,4 +64,34 @@ note ( 'testing retrieving tags by SHA1 ...' );
     cmp_deeply( \@type_params, [ 'sha1' ], 'API call sequence');
 }
 
+note ( 'testing retrieving tags by SHA1 with additional tags ...' );
+
+{
+    my $json_by_sha1 = decode_json( Mojo::File->new("$SAMPLES/chaika/002_sha1_response.json")->slurp );
+    my $json = decode_json( Mojo::File->new("$SAMPLES/chaika/001_gid_27240.json")->slurp );
+    my @type_params = ();
+
+    no warnings 'once', 'redefine';
+    local *LANraragi::Plugin::Metadata::Chaika::get_plugin_logger = sub { return get_logger_mock(); };
+    local *LANraragi::Plugin::Metadata::Chaika::get_json_from_chaika = sub {
+        my ( $type, $value ) = @_;
+        push( @type_params, $type );
+        return ( $type eq 'sha1' ) ? $json_by_sha1 : $json;
+    };
+
+    my $addextra = 1;
+    my $addother = 1;
+    my $addsource = 'chaika';
+    my ( $tags, $title ) = LANraragi::Plugin::Metadata::Chaika::tags_from_sha1(
+        "my-hash",
+        $addextra,
+        $addother,
+        $addsource
+    );
+
+    is($title, $json->{title}, 'gallery title');
+    cmp_bag( [ split( ', ', $tags ) ] , \@tags_list_extra, 'gallery tag list');
+    cmp_deeply( \@type_params, [ 'sha1' ], 'API call sequence');
+}
+
 done_testing();
diff --git a/tests/samples/chaika/001_gid_27240.json b/tests/samples/chaika/001_gid_27240.json
@@ -6,21 +6,24 @@
     "filesize": 63418139,
     "fjord": true,
     "gallery": 23532,
-    "posted": 1521353952,
-    "rating": 4.45,
+    "posted": 1521357552,
+    "rating": 4.49,
     "tags": [
+        "full_censorship",
         "female:sole_female",
         "male:sole_male",
         "artist:kemuri_haku",
-        "full_censorship",
+        "female:tall_girl",
+        "female:cunnilingus",
         "male:shotacon",
         "female:defloration",
         "female:nakadashi",
+        "female:x-ray",
         "female:big_breasts",
         "language:translated",
         "language:english"
     ],
     "title": "[Kemuri Haku] Zettai Seikou Keikaku | Absolute Intercourse Plan (COMIC Shitsurakuten 2016-03) [English] [Redlantern]",
     "title_jpn": "[煙ハク] 絶対セイコウ計画 (COMIC 失楽天 2016年3月号) [英訳]",
     "uploader": "dankestdungeon"
-}
+}
diff --git a/tests/samples/chaika/002_sha1_response.json b/tests/samples/chaika/002_sha1_response.json
@@ -1,26 +1,32 @@
-[{
-    "category": "Manga",
-    "download": "/archive/27240/download/",
-    "expunged": false,
-    "filecount": 19,
-    "filesize": 63418139,
-    "fjord": true,
-    "gallery": 23532,
-    "posted": 1521353952,
-    "rating": 4.45,
-    "tags": [
-        "female:sole_female",
-        "male:sole_male",
-        "artist:kemuri_haku",
-        "full_censorship",
-        "male:shotacon",
-        "female:defloration",
-        "female:nakadashi",
-        "female:big_breasts",
-        "language:translated",
-        "language:english"
-    ],
-    "title": "[Kemuri Haku] Zettai Seikou Keikaku | Absolute Intercourse Plan (COMIC Shitsurakuten 2016-03) [English] [Redlantern]",
-    "title_jpn": "[煙ハク] 絶対セイコウ計画 (COMIC 失楽天 2016年3月号) [英訳]",
-    "uploader": "dankestdungeon"
-}]
+[
+    {
+        "category": "Manga",
+        "download": "/archive/27240/download/",
+        "expunged": false,
+        "filecount": 19,
+        "filesize": 63418139,
+        "fjord": true,
+        "gallery": 23532,
+        "id": 27240,
+        "posted": 1521357552,
+        "rating": 4.49,
+        "tags": [
+            "full_censorship",
+            "female:sole_female",
+            "male:sole_male",
+            "artist:kemuri_haku",
+            "female:tall_girl",
+            "female:cunnilingus",
+            "male:shotacon",
+            "female:defloration",
+            "female:nakadashi",
+            "female:x-ray",
+            "female:big_breasts",
+            "language:translated",
+            "language:english"
+        ],
+        "title": "[Kemuri Haku] Zettai Seikou Keikaku | Absolute Intercourse Plan (COMIC Shitsurakuten 2016-03) [English] [Redlantern]",
+        "title_jpn": "[煙ハク] 絶対セイコウ計画 (COMIC 失楽天 2016年3月号) [英訳]",
+        "uploader": "dankestdungeon"
+    }
+]