pkgs/profpatsch/youtube2audiopodcast: youtube playlist to rss

Initial code that fetches a youtube playlist (from ID) and converts it
to an rss feed.

2 files changed, 201 insertions, 4 deletions

diff --git a/pkgs/profpatsch/youtube2audiopodcast/Main.hs b/pkgs/profpatsch/youtube2audiopodcast/Main.hs
new file mode 100755
index 00000000..98be3867
--- /dev/null
+++ b/pkgs/profpatsch/youtube2audiopodcast/Main.hs
@@ -0,0 +1,111 @@
+{-# language OverloadedStrings #-}
+{-# language RecordWildCards #-}
+{-# language NamedFieldPuns #-}
+{-# language DeriveGeneric #-}
+module Main where
+
+import Data.Text (Text)
+import Text.RSS.Syntax
+import Text.Feed.Types (Feed(RSSFeed))
+import Text.Feed.Export (textFeed)
+
+import qualified Data.Aeson as Json
+import Data.Aeson (FromJSON)
+import GHC.Generics (Generic)
+import qualified Data.ByteString.Lazy as BS
+
+-- | Info from the config file
+data Config = Config
+  { channelName :: Text
+  , channelURL :: Text
+  }
+  deriving (Show, Generic)
+instance FromJSON Config where
+
+data Channel = Channel
+  { channelInfo :: ChannelInfo
+  , channelItems :: [ItemInfo]
+  }
+  deriving (Show, Generic)
+instance FromJSON Channel where
+
+-- | Info fetched from the channel
+data ChannelInfo = ChannelInfo
+
+  { channelDescription :: Text
+  , channelLastUpdate :: DateString -- TODO
+  , channelImage :: Maybe () --RSSImage
+  }
+  deriving (Show, Generic)
+instance FromJSON ChannelInfo where
+
+-- | Info of each channel item
+data ItemInfo = ItemInfo
+  { itemTitle :: Text
+  , itemDescription :: Text
+  , itemYoutubeLink :: Text
+  , itemCategory :: Text
+  , itemTags :: [Text]
+  , itemURL :: Text
+  , itemSizeBytes :: Integer
+  , itemHash :: Text
+  }
+  deriving (Show, Generic)
+instance FromJSON ItemInfo where
+
+-- main = print $ textFeed $ RSSFeed $ toRSS
+--   (ChannelInfo
+--     { channelDescription = "description"
+--     , channelLastUpdate = "some date"
+--     , channelImage = nullImage "imageURL" "imageTitle" "imageLink"
+--     })
+--   []
+main :: IO ()
+main = do
+  input <- BS.getContents
+  let (Right rss) = Json.eitherDecode input
+  let exConfig = (Config
+        { channelName = "channel name"
+        , channelURL = "channel url"
+        })
+  print $ textFeed $ RSSFeed $ toRSS exConfig rss
+
+toRSS :: Config -> Channel -> RSS
+toRSS Config{..} Channel{channelInfo, channelItems}  =
+  let ChannelInfo{..} = channelInfo in
+  (nullRSS channelName channelURL)
+  { rssChannel = (nullChannel channelName channelURL)
+      { rssDescription = channelDescription
+      , rssLastUpdate = Just channelLastUpdate
+      , rssImage = Nothing --channelImage TODO
+      , rssGenerator = Just "youtube2audiopodcast"
+      , rssItems = map rssItem channelItems
+      }
+  }
+
+  where
+    rssItem ItemInfo{..} = (nullItem itemTitle)
+      { rssItemLink = Just itemYoutubeLink
+      , rssItemDescription = Just itemDescription
+      -- rssItemAuthor =
+      , rssItemCategories =
+        map (\name -> RSSCategory
+              { rssCategoryDomain = Nothing
+              , rssCategoryAttrs = []
+              , rssCategoryValue = name
+              }) $ itemCategory : itemTags
+      , rssItemEnclosure = Just $ RSSEnclosure
+          { rssEnclosureURL = itemURL
+          , rssEnclosureLength = Just itemSizeBytes
+          , rssEnclosureType = "audio/opus"
+          , rssEnclosureAttrs = []
+          }
+      , rssItemGuid = Just $ RSSGuid
+          { rssGuidPermanentURL = Just False
+          , rssGuidAttrs = []
+          , rssGuidValue = itemHash
+          }
+      -- TODO: date conversion
+      -- https://tools.ietf.org/html/rfc822#section-5.1
+      -- , rssItemPubDate 
+      }
diff --git a/pkgs/profpatsch/youtube2audiopodcast/default.nix b/pkgs/profpatsch/youtube2audiopodcast/default.nix
index ed8ad967..f0c9189d 100644
--- a/pkgs/profpatsch/youtube2audiopodcast/default.nix
+++ b/pkgs/profpatsch/youtube2audiopodcast/default.nix
@@ -8,16 +8,48 @@ let
     // getBins pkgs.execline [ "fdmove" "backtick" "importas" "if" "redirfd" "pipeline" ]
     // getBins pkgs.s6-portable-utils [
          { use = "s6-cat"; as = "cat"; }
-       ];
+       ]
+    // getBins pkgs.jl [ "jl" ];
 
+  # fetch the audio of a youtube video to ./audio.opus, given video ID
   youtube-dl-audio = writeExecline "youtube-dl-audio" { readNArgs = 1; } [
     bins.youtube-dl
       "--verbose"
       "--extract-audio"
       "--audio-format" "opus"
-      "--output" "./audio.opus" "https://www.youtube.com/watch?v=\${1}"
+      # We have to give a specific filename (with the right extension).
+      # youtube-dl is really finicky with output filenames.
+      "--output" "./audio.opus"
+      "https://www.youtube.com/watch?v=\${1}"
   ];
 
+  # print youtube playlist information to stdout, given playlist ID
+  youtube-playlist-info = writeExecline "youtube-playlist-info" { readNArgs = 1; } [
+    bins.youtube-dl
+      "--verbose"
+      # don’t query detailed info of every video,
+      # which takes a lot of time
+      "--flat-playlist"
+      # print a single line of json to stdout
+      "--dump-single-json"
+      "--yes-playlist"
+      "https://www.youtube.com/playlist?list=\${1}"
+  ];
+
+  writeHaskellInterpret = nameOrPath: { withPackages ? lib.const [] }: content:
+    let ghc = pkgs.haskellPackages.ghcWithPackages withPackages; in
+    pkgs.writers.makeScriptWriter {
+      interpreter = "${ghc}/bin/runhaskell";
+      check = pkgs.writers.writeDash "ghc-typecheck" ''
+        ln -s "$1" ./Main.hs
+        ${ghc}/bin/ghc -fno-code -Wall ./Main.hs
+      '';
+    } nameOrPath content;
+
+  printFeed = writeHaskellInterpret "print-feed" {
+    withPackages = hps: [ hps.feed hps.aeson ];
+  } ./Main.hs;
+
   # minimal CGI request parser for use as UCSPI middleware
   yolo-cgi = pkgs.writers.writePython3 "yolo-cgi" {} ''
     import sys
@@ -43,6 +75,7 @@ let
     os.execlp(cmd, cmd, *args)
   '';
 
+  # print the contents of an envar to the stdin of $@
   envvar-to-stdin = writeExecline "envvar-to-stdin" { readNArgs = 1; } [
     "importas" "VAR" "$1"
     "pipeline" [ bins.printf "%s" "$VAR" ] "$@"
@@ -86,6 +119,59 @@ let
     serve-http-opus-file "./audio.opus"
   ];
 
+  example-config = pkgs.writeText "example-config.json" (lib.generators.toJSON {} {
+    channelName = "Lonely Rolling Star";
+    channelURL = "https://www.youtube.com/playlist?list=PLV9hywkogVcOuHJ8O121ulSfFDKUhJw66";
+  });
+
+  transform-flat-playlist-to-rss = hostUrl:
+    let
+      playlist-item-info-jl = ''
+        (\o ->
+          { itemTitle: o.title
+          , itemYoutubeLink: append "https://youtube.com/watch?v=" o.id
+          ${/*TODO how to add the url here nicely?*/""}
+          , itemURL: append "${hostUrl}/" o.id
+
+          ${/*# TODO*/""}
+          , itemDescription: ""
+          , itemCategory: ""
+          , itemTags: []
+          , itemSizeBytes: 0
+          , itemHash: ""
+          })
+     '';
+     playlist-info-jl = ''
+       \pl ->
+         { channelInfo:
+           { channelDescription: pl.title
+
+           ${/*# TODO*/""}
+           , channelLastUpdate: "000"
+           , channelImage: null
+           }
+         , channelItems:
+           map
+             ${playlist-item-info-jl}
+             pl.entries
+         }
+     '';
+   in writeExecline "youtube-dl-playlist-json-to-rss-json" {} [
+    bins.jl playlist-info-jl
+  ];
+
+  print-feed-json = writeExecline "ex2" {} [
+    "pipeline" [
+      youtube-playlist-info "PLV9hywkogVcOuHJ8O121ulSfFDKUhJw66"
+    ] (transform-flat-playlist-to-rss "localhost:8888")
+  ];
+
+  print-example-feed = writeExecline "ex" {} [
+    "pipeline" [ print-feed-json ]
+    printFeed
+  ];
+
+
 # in printFeed
-in serve-audio
-# in youtube-dl-audio
+# in serve-audio
+in print-example-feed