diff --git a/README.md b/README.md --- a/README.md +++ b/README.md @@ -15,15 +15,15 @@ $ cabal install nicovideo-translator -Aliasing `msg.nicovideo.jp` to localhost ----------------------------------------- +Aliasing `nmsg.nicovideo.jp` to localhost +----------------------------------------- To make the translator to intercept comments from Nico Nico comment server, -you have to alias Nico Nico comment server domain (`msg.nicovideo.jp`) to +you have to alias Nico Nico comment server domain (`nmsg.nicovideo.jp`) to your localhost (`127.0.0.1`). Open your [hosts file][2] using text editor (you probably need administrator permission), and then add the following line: - 127.0.0.1 msg.nicovideo.jp + 127.0.0.1 nmsg.nicovideo.jp [2]: http://en.wikipedia.org/wiki/Hosts_%28file%29 @@ -35,30 +35,34 @@ you watch Nico Nico videos. You can invoke the proxy server using CLI (you probably need administrator permission to listen 80 port): - $ nicovideo-translator + $ nicovideo-translator EiP3NSgLid81OjSwpOMkgV0rzD9SLHRqwqUwx2r Running on http://0.0.0.0:80/ (Press ^C to quit) - Upstream: msg.nicovideo.jp (202.248.110.173) + Upstream: nmsg.nicovideo.jp (202.248.110.173) +Note that it takes a [Google Translate API][3] key as its first argument. + You can terminate the server by pressing Ctrl-C. It can optionally take the target language which is a two-letter -e.g. `en`, `ko` through `-l`/`--language` option: +e.g. `en`, `ko` through `-l`/`--lang`/`--language` option: - $ nicovideo-translator --language ko + $ nicovideo-translator -l ko +[3]: https://cloud.google.com/translate/ + Open source ----------- -It's written by [Hong Minhee][3], and distributed under [AGPLv3][]. -You can find the source code from the [Git repository][4]: +It's written by [Hong Minhee][4], and distributed under [AGPLv3][]. +You can find the source code from the [Git repository][5]: $ git clone git://github.com/dahlia/nicovideo-translator -Please report bugs to the [issue tracker][5] if you find. +Please report bugs to the [issue tracker][6] if you find. Pull requests welcome! -[3]: http://hongminhee.org/ -[4]: https://github.com/dahlia/nicovideo-translator -[5]: https://github.com/dahlia/nicovideo-translator/issues +[4]: https://hongminhee.org/ +[5]: https://github.com/dahlia/nicovideo-translator +[6]: https://github.com/dahlia/nicovideo-translator/issues [AGPLv3]: http://www.gnu.org/licenses/agpl-3.0.html diff --git a/lib/NicovideoTranslator/Main.hs b/lib/NicovideoTranslator/Main.hs --- a/lib/NicovideoTranslator/Main.hs +++ b/lib/NicovideoTranslator/Main.hs @@ -18,13 +18,31 @@ import Network.DNS.Resolver (defaultResolvConf, makeResolvSeed, withResolver) import Network.DNS.Lookup (lookupA) import Network.Wai.Handler.Warp (Port, run) -import System.Console.CmdArgs (cmdArgs, help, typ, (&=)) +import System.Console.CmdArgs ( argPos + , cmdArgs + , def + , details + , explicit + , help + , name + , program + , summary + , typ + , (&=) + ) import System.Locale.SetLocale (Category(LC_CTYPE), setLocale) -import NicovideoTranslator.Proxy (ProxyConfiguration(ProxyConfiguration), app) +import NicovideoTranslator.Proxy ( ProxyConfiguration ( ProxyConfiguration + , apiKey + , language + , upstreamHost + ) + , app + ) data Translator = Translator { port :: Port - , language :: String + , language' :: String + , apiKey' :: String } deriving (Show, Data, Typeable) formatIoError :: Params ps => Format -> ps -> IO a @@ -45,21 +63,30 @@ Nothing -> formatIoError "locale is not set" () defaultUpstreamHost :: T.Text -defaultUpstreamHost = "msg.nicovideo.jp" +defaultUpstreamHost = "nmsg.nicovideo.jp" translateCmdArgs :: String -> Translator translateCmdArgs lang = - Translator { language = lang &= typ "LANG" - &= help "Target language to translate to [en]" + Translator { language' = lang &= explicit + &= name "language" + &= name "lang" + &= name "l" + &= typ "LANG" + &= help "Target language to translate to [en]" , port = 80 &= typ "PORT" &= help "Port number to listen [80]" + , apiKey' = def &= argPos 0 &= typ "API_KEY" } + &= program "nicovideo-translator" + &= summary "Nico Nico Douga (ニコニコ動画) Comment Translator" + &= details + ["It takes a Google Translate API key as its first argument."] main :: IO () main = do currentLang <- catchIOError currentLanguage (\_ -> return L.EN) opts <- cmdArgs $ translateCmdArgs $ L.language currentLang - lang <- readLanguageCode $ language opts + lang <- readLanguageCode $ language' opts let portNum = port opts hPutStrLn stdout $ LT.toStrict $ format "Running on http://0.0.0.0:{}/ (Press ^C to quit)" (Only portNum) @@ -68,13 +95,17 @@ \resolver -> lookupA resolver (encodeUtf8 defaultUpstreamHost) case resolution of Right (resolvedHost:_) -> - let upstreamHost = T.pack $ show resolvedHost + let upstreamHost' = T.pack $ show resolvedHost in do hPutStrLn stdout $ T.concat ["Upstream: " , defaultUpstreamHost - , " (", upstreamHost, ")" + , " (", upstreamHost', ")" ] - run portNum $ app (ProxyConfiguration lang upstreamHost) + let conf = ProxyConfiguration { language = lang + , upstreamHost = upstreamHost' + , apiKey = T.pack $ apiKey' opts + } + run portNum $ app conf _ -> hPutStrLn stderr $ T.concat [ "error: failed to resolve " , defaultUpstreamHost ] diff --git a/lib/NicovideoTranslator/Proxy.hs b/lib/NicovideoTranslator/Proxy.hs --- a/lib/NicovideoTranslator/Proxy.hs +++ b/lib/NicovideoTranslator/Proxy.hs @@ -1,9 +1,11 @@ {-# LANGUAGE OverloadedStrings #-} module NicovideoTranslator.Proxy - ( ProxyConfiguration(ProxyConfiguration) + ( ProxyConfiguration ( ProxyConfiguration + , apiKey + , language + , upstreamHost + ) , app - , language - , upstreamHost ) where import Data.List (find) @@ -29,11 +31,13 @@ def, elementNodes, parseLBS, renderLBS) import Text.XML.Cursor (content, element, fromDocument, node, ($//), (&//)) -import NicovideoTranslator.Translate (translateMultiple) +import NicovideoTranslator.Translate (ApiKey, translate) -data ProxyConfiguration = ProxyConfiguration { language :: ISO639_1 - , upstreamHost :: Text - } +data ProxyConfiguration = + ProxyConfiguration { language :: ISO639_1 + , upstreamHost :: Text + , apiKey :: ApiKey + } app :: ProxyConfiguration -> Application app config req respond = @@ -70,9 +74,17 @@ rHeaders = (response ^. responseHeaders) toBeTranslated = method == "POST" && mimetype == "text/xml" translated <- if toBeTranslated - then translateResponse (language config) rBody + then translateResponse (apiKey config) + (language config) + rBody else return rBody - respond $ responseLBS rStatus rHeaders translated + let headers = [ (name, value) + | (name, value) <- rHeaders + , name /= "content-length" + ] + -- Content-Length becomes invalid since the translated text doesn't + -- have the same length to its source text + respond $ responseLBS rStatus headers translated where method = requestMethod req headerList = requestHeaders req @@ -89,15 +101,15 @@ request "DELETE" = \options url _ -> deleteWith options url request _ = \_ _ _ -> ioError $ userError $ "unsupported method" -translateResponse :: ISO639_1 -> LB.ByteString -> IO LB.ByteString -translateResponse lang response = case parseLBS def response of +translateResponse :: ApiKey -> ISO639_1 -> LB.ByteString -> IO LB.ByteString +translateResponse apiKey' lang response = case parseLBS def response of Left _ -> return response - Right doc -> do translatedDoc <- translateXml lang doc + Right doc -> do translatedDoc <- translateXml apiKey' lang doc return $ renderLBS def translatedDoc -translateXml :: ISO639_1 -> Document -> IO Document -translateXml lang doc = do - translatedTexts <- translateMultiple lang texts +translateXml :: ApiKey -> ISO639_1 -> Document -> IO Document +translateXml apiKey' lang doc = do + translatedTexts <- translate apiKey' lang texts let translatedElems = [ (el, el { elementNodes = [NodeContent text] }) | (el, text) <- zip elems translatedTexts ] return $ transformXml doc translatedElems diff --git a/lib/NicovideoTranslator/Translate.hs b/lib/NicovideoTranslator/Translate.hs --- a/lib/NicovideoTranslator/Translate.hs +++ b/lib/NicovideoTranslator/Translate.hs @@ -1,16 +1,57 @@ -{-# LANGUAGE OverloadedStrings #-} -module NicovideoTranslator.Translate - ( - translate - , translateMultiple - ) where +{-# LANGUAGE OverloadedLists, OverloadedStrings #-} +module NicovideoTranslator.Translate (ApiKey, translate) where -import qualified Language.Translate.Naver as Naver -import Data.LanguageCodes (ISO639_1(JA)) -import Data.Text (Text) +import GHC.Exts (IsList (toList)) +import Prelude hiding (lookup) -translate :: ISO639_1 -> Text -> IO Text -translate = Naver.translate JA +import Control.Concurrent.Async (concurrently) +import Control.Lens ((&), (.~), (^.)) +import Data.Aeson.Types (Value (Array, Bool, Object, String), toJSON) +import Data.HashMap.Strict (lookup) +import Data.LanguageCodes (ISO639_1) +import Data.Text (Text, pack, toLower, unpack) +import Network.Wreq ( Options + , Response + , asJSON + , defaults + , param + , postWith + , responseBody + ) -translateMultiple :: ISO639_1 -> [Text] -> IO [Text] -translateMultiple = Naver.translateMultiple JA +type ApiKey = Text + +apiUrl :: String +apiUrl = "https://translation.googleapis.com/language/translate/v2" + +translate :: ApiKey -> ISO639_1 -> [Text] -> IO [Text] +translate apiKey target texts = + case splitAt 128 texts of + ([], _) -> return [] + (head, []) -> translate' apiKey target head + (head, tail) -> do + let trans = translate apiKey target + (headResult, tailResult) <- concurrently (trans head) (trans tail) + return $ headResult ++ tailResult + +translate' :: ApiKey -> ISO639_1 -> [Text] -> IO [Text] +translate' apiKey target texts = do + response <- (asJSON =<< postWith query apiUrl params) :: IO (Response Value) + let Object body = response ^. responseBody + Just (Object data') = lookup "data" body + Just (Array translations) = lookup "translations" data' + return $ [ s + | Object r <- toList translations + , Just (String s) <- [lookup "translatedText" r] + ] + where + query :: Options + query = defaults & param "key" .~ [apiKey] + params :: Value + params = Object + [ ("target", String $ toLower . pack . show $ target) + , ("source", String "ja") + , ("prettyprint", Bool False) + , ("format", String "text") + , ("q", toJSON texts) + ] diff --git a/nicovideo-translator.cabal b/nicovideo-translator.cabal --- a/nicovideo-translator.cabal +++ b/nicovideo-translator.cabal @@ -1,49 +1,62 @@ name: nicovideo-translator -version: 0.1.0.1 +version: 0.2.0.0 synopsis: Nico Nico Douga (ニコニコ動画) Comment Translator +description: Translate comments (written in Japanese) on + Nico Nico Douga (ニコニコ動画) to your language. + . + See also README.md for more details. homepage: https://github.com/dahlia/nicovideo-translator bug-reports: https://github.com/dahlia/nicovideo-translator/issues license: AGPL-3 license-file: LICENSE author: Hong Minhee -maintainer: hongminhee@member.fsf.org -copyright: (c) 2015 Hong Minhee +maintainer: hong.minhee@gmail.com +copyright: (c) 2015–2016 Hong Minhee stability: alpha +category: Translation build-type: Simple extra-source-files: README.md cabal-version: >=1.10 +source-repository head + type: git + location: git://github.com/dahlia/nicovideo-translator.git + library hs-source-dirs: lib exposed-modules: NicovideoTranslator.Main, NicovideoTranslator.Proxy, NicovideoTranslator.Translate - build-depends: base >=4.5.0.0 && <4.8.0.0, + build-depends: aeson >=0.11.2.1 && <1.1.0.0, + async >=2.1.0 && <2.2.0, + base >=4.9.0.0 && <4.10.0.0, bytestring >=0.10.4.0 && <0.11.0.0, case-insensitive >=1.2.0.0 && <1.3.0.0, cmdargs >=0.10.12 && <0.11.0, containers, - dns >=1.4.4 && <1.5.0, + dns >=2.0.0 && <2.1.0, http-client >=0.4.9 && <0.5.0, - http-types >=0.8.6 && <0.9.0, + http-types >=0.9.0 && <0.10.0, iso639, - lens >=4.8 && <4.9, - naver-translate >=0.1.0.0 && <0.2.0.0, + lens >=4.14 && <4.15, + lens-aeson >=1.0.0.5 && <1.1.0.0, setlocale >=1.0.0.2 && <1.1.0.0, - text >=1.1.0.0 && <1.2.0.0, + text >=1.1.0.0 && <1.3.0.0, text-format >=0.3.1.1 && <0.4.0.0, - wai >=3.0.2.3 && <3.1.0.0, - warp >=3.0.10 && <3.1.0, - wreq >=0.3.0.1 && <0.4.0.0, - xml-conduit >=1.2.3.3 && <1.3.0.0 + unordered-containers, + wai >=3.2.1.1 && <3.3.0.0, + warp >=3.2.8 && <3.3.0, + wreq >=0.4.1.0 && <0.5.0.0, + xml-conduit >=1.3.5 && <1.4.0 default-language: Haskell2010 - ghc-options: -Wall -fwarn-incomplete-uni-patterns -threaded + ghc-options: -Wall -fwarn-incomplete-uni-patterns executable nicovideo-translator hs-source-dirs: src main-is: Main.hs - build-depends: base >=4.5.0.0 && <4.8.0.0, + build-depends: base >=4.9.0.0 && <4.10.0.0, nicovideo-translator default-language: Haskell2010 - ghc-options: -Wall -fwarn-incomplete-uni-patterns -threaded + ghc-options: -Wall -fwarn-incomplete-uni-patterns + -threaded -with-rtsopts=-N