Merge pull request #19 from titsuki/neologd

Introduce mecab-ipadic-neologd
titsuki · Feb 12, 2018 · 9b0edde · 9b0edde
2 parents 6d71eb1 + 3952783
commit 9b0edde
Show file tree

Hide file tree

Showing 4 changed files with 148 additions and 10 deletions.
diff --git a/README.md b/README.md
@@ -86,6 +86,46 @@ MeCab depends on the following:
 
 Once the build starts, it automatically downloads `mecab-0.996` and `mecab-ipadic-2.7.0-20070801` with `wget` and installs these stuffs under the `$HOME/.p6mecab` directory, where `$HOME` is your home directory.
 
+Use 3rd-party dictionary
+========================
+
+mecab-ipadic-neologd
+--------------------
+
+  * Step1: download and install neologd
+
+Example:
+
+    $ git clone --depth 1 https://github.com/neologd/mecab-ipadic-neologd.git
+    $ cd mecab-ipadic-neologd
+    $ export PATH=$HOME/.p6mecab/bin:$PATH
+    $ ./bin/install-mecab-ipadic-neologd -p $HOME/.p6mecab/lib/mecab/dic/ipadic-neologd
+
+  * Step2: Use .new(:dicdir(PATH_TO_THE_DIR))
+
+Example:
+
+    use MeCab;
+    use MeCab::Tagger;
+
+    my Str $text = "トランプ大統領 ワシントンで大規模軍事パレードを指示";
+    my $mecab-tagger = MeCab::Tagger.new(:dicdir("$*HOME/.p6mecab/lib/mecab/dic/ipadic-neologd"));
+    loop ( my MeCab::Node $node = $mecab-tagger.parse-tonode($text); $node; $node = $node.next ) {
+           say ($node.surface, $node.feature).join("\t");
+    }
+
+    # OUTPUT«
+    #         BOS/EOS,*,*,*,*,*,*,*,*
+    # トランプ大統領  名詞,固有名詞,人名,一般,*,*,ドナルド・トランプ,トランプダイトウリョウ,トランプダイトウリョー
+    # ワシントン      名詞,固有名詞,地域,一般,*,*,ワシントン,ワシントン,ワシントン
+    # で      助詞,格助詞,一般,*,*,*,で,デ,デ
+    # 大規模  名詞,一般,*,*,*,*,大規模,ダイキボ,ダイキボ
+    # 軍事パレード    名詞,固有名詞,一般,*,*,*,軍事パレード,グンジパレード,グンジパレード
+    # を      助詞,格助詞,一般,*,*,*,を,ヲ,ヲ
+    # 指示    名詞,サ変接続,*,*,*,*,指示,シジ,シジ
+    #         BOS/EOS,*,*,*,*,*,*,*,*
+    # »
+
 AUTHOR
 ======
 
@@ -99,3 +139,4 @@ Copyright 2016 titsuki
 libmecab ( http://taku910.github.io/mecab/ ) by Taku Kudo is licensed under the GPL, LGPL or BSD Licenses.
 
 This library is free software; you can redistribute it and/or modify it under the Artistic License 2.0.
+
diff --git a/lib/MeCab.pm6 b/lib/MeCab.pm6
@@ -141,6 +141,45 @@ MeCab depends on the following:
 
 Once the build starts, it automatically downloads C<mecab-0.996> and C<mecab-ipadic-2.7.0-20070801> with C<wget> and installs these stuffs under the C<$HOME/.p6mecab> directory, where C<$HOME> is your home directory.
 
+=head1 Use 3rd-party dictionary
+
+=head2 mecab-ipadic-neologd
+
+=item1 Step1: download and install neologd
+
+Example:
+
+       $ git clone --depth 1 https://github.com/neologd/mecab-ipadic-neologd.git
+       $ cd mecab-ipadic-neologd
+       $ export PATH=$HOME/.p6mecab/bin:$PATH
+       $ ./bin/install-mecab-ipadic-neologd -p $HOME/.p6mecab/lib/mecab/dic/ipadic-neologd
+
+
+=item1 Step2: Use .new(:dicdir(PATH_TO_THE_DIR))
+
+Example:
+
+       use MeCab;
+       use MeCab::Tagger;
+       
+       my Str $text = "トランプ大統領 ワシントンで大規模軍事パレードを指示";
+       my $mecab-tagger = MeCab::Tagger.new(:dicdir("$*HOME/.p6mecab/lib/mecab/dic/ipadic-neologd"));
+       loop ( my MeCab::Node $node = $mecab-tagger.parse-tonode($text); $node; $node = $node.next ) {
+              say ($node.surface, $node.feature).join("\t");
+       }
+       
+       # OUTPUT«
+       #         BOS/EOS,*,*,*,*,*,*,*,*
+       # トランプ大統領  名詞,固有名詞,人名,一般,*,*,ドナルド・トランプ,トランプダイトウリョウ,トランプダイトウリョー
+       # ワシントン      名詞,固有名詞,地域,一般,*,*,ワシントン,ワシントン,ワシントン
+       # で      助詞,格助詞,一般,*,*,*,で,デ,デ
+       # 大規模  名詞,一般,*,*,*,*,大規模,ダイキボ,ダイキボ
+       # 軍事パレード    名詞,固有名詞,一般,*,*,*,軍事パレード,グンジパレード,グンジパレード
+       # を      助詞,格助詞,一般,*,*,*,を,ヲ,ヲ
+       # 指示    名詞,サ変接続,*,*,*,*,指示,シジ,シジ
+       #         BOS/EOS,*,*,*,*,*,*,*,*
+       # »
+
 =head1 AUTHOR
 
 titsuki <titsuki@cpan.org>

diff --git a/lib/MeCab/Model.pm6 b/lib/MeCab/Model.pm6
@@ -13,14 +13,42 @@ my sub mecab_model_new2(Str) returns MeCab::Model is native($library) { * }
 my sub mecab_model_new_tagger(MeCab::Model) returns MeCab::Tagger is native($library) { * }
 my sub mecab_model_new_lattice(MeCab::Model) returns MeCab::Lattice is native($library) { * }
 
-multi method new {
-    my Str $argv = "-C";
-    mecab_model_new2($argv)
+multi submethod new {
+    mecab_model_new2("-C")
 }
 
-multi method new(Str $extra-argv) {
-    my Str $argv = "-C " ~ $extra-argv;
-    mecab_model_new2($argv)
+multi submethod new(Str $argv) {
+    mecab_model_new2($argv);
+}
+
+multi submethod new(
+    Str :$rcfile,
+    Str :$dicdir,
+    Str :$userdic,
+) {
+    my @args;
+    @args.push('-C'); # allocate-sentence
+
+    if $rcfile.defined {
+        $rcfile.IO.f or die "$rcfile doesn't exist.";
+        $dicdir.defined or die ":rcfile requires :dicdir.";
+        $dicdir.IO.d or die "$dicdir doesn't exist.";
+
+        @args.push(sprintf('-r %s', $rcfile));
+    }
+
+    if $dicdir.defined {
+        $dicdir.IO.d or die "$dicdir doesn't exist.";
+
+        @args.push(sprintf('-d %s', $dicdir));
+    }
+
+    if $userdic.defined {
+        $userdic.IO.f or die "$userdic doesn't exist.";
+
+        @args.push(sprintf('-u %s', $userdic))
+    }
+    mecab_model_new2(@args.join(' '));
 }
 
 method create-tagger {

diff --git a/lib/MeCab/Tagger.pm6 b/lib/MeCab/Tagger.pm6
@@ -23,12 +23,42 @@ my sub mecab_sparse_tostr3(MeCab::Tagger, size_t, Str, size_t, CArray[int8], siz
 my sub mecab_dictionary_info(MeCab::Tagger) returns MeCab::DictionaryInfo is native($library) { * }
 my sub mecab_strerror(MeCab::Tagger) returns Str is native($library) { * }
 
-multi method new(Str $arg) {
-    mecab_new2($arg);
+multi submethod new {
+    mecab_new2("-C");
 }
 
-multi method new {
-    mecab_new2("-C");
+multi submethod new(Str $argv) {
+    mecab_new2($argv);
+}
+
+multi submethod new(
+    Str :$rcfile,
+    Str :$dicdir,
+    Str :$userdic,
+) {
+    my @args;
+    @args.push('-C'); # allocate-sentence
+
+    if $rcfile.defined {
+        $rcfile.IO.f or die "$rcfile doesn't exist.";
+        $dicdir.defined or die ":rcfile requires :dicdir.";
+        $dicdir.IO.d or die "$dicdir doesn't exist.";
+
+        @args.push(sprintf('-r %s', $rcfile));
+    }
+
+    if $dicdir.defined {
+        $dicdir.IO.d or die "$dicdir doesn't exist.";
+
+        @args.push(sprintf('-d %s', $dicdir));
+    }
+
+    if $userdic.defined {
+        $userdic.IO.f or die "$userdic doesn't exist.";
+
+        @args.push(sprintf('-u %s', $userdic));
+    }
+    mecab_new2(@args.join(' '));
 }
 
 method version {