dataset.yaml

name: Bluemoon
description: "Bluemoon 1-1 Fandom RP+ERP cleaned and formatted for Fastchat"
build:
  conversationBatch: 100
  cleanDirectory: [ 'out/*' ]
  verifyRequirements: false
  verifyArtifacts: true
  sendPackedFiles: true
  gzipRemotePackedFiles: true
  gzipLocalPackedFiles: false
  #concurrency:
  #  local: 16

_input_steps:
  - step: &bluemoon
      type: CsvInput
      sync: main
      config:
        path: "original/rentry/bluemoon/bluemoon-fanbased-roleplays-1x1.csv"
        fields:
          conversation: 'thread_href'
          from: 'message_username'
          message: 'message'
  - step: &bluemoon_cleaned
      type: FastChatInput
      sync: main
      config:
        path: "out/data/full/bluemoon.train.json"

_output_steps:
  - step: &bluemoon_fastchat
      type: FastChatOutput
      sync: main
      config:
        path: "out/data/full/bluemoon.train.json"
        indent: 2
        #includeMessageId: true

_steps:
  - step: &html_linebreaks
      type: ExactReplace
      description: "Convert html <br/> line breaks to \n"
      config:
        replacements: [ [ '<br/>', "\n" ] ]

  - step: &html_strip
      type: HtmlStrip
      description: "Strip HTML"
      config:
        stripAnchorPatterns: [ '/', '.jpg', '.JPG', '.png', '.PNG', '.gif', '.GIF' ]
        stripDom: [ 'blockquote.bbCodeBlock' ]

  - step: &ooc_strip
      type: RegexReplace
      description: "Strip OOC content and some HTML errors."
      config:
        dotAll: true
        multiLine: true
        replacements: [
          # This might be a bit overzealous.
          [ '\{.*\}', "" ], [ '\[.*?\]+', "" ], [ '\(\(.*?(\)|\W)\)+', "" ],
          # Match for (...) but only at the beginning and end of messages.
          [ '^\(.*?\)+', "" ],  [ '\([^\(]+\)+$', "" ],
          [ '(OOC|ooc|OoC|OCC|ooc|Ooc)(:| |-|;)\n?.*?(\n|$)', "" ],
          [ '(\n\nSee more|See more\n\n)', "" ], [ '^(RE|Re):.*?\n', "" ], [ '^NSFW:.*?\n$', "" ],
          # More author's notes and the like
          [ '^Note to everyone.*\)\)\n', "" ], [ '_*Author.*\n*_', "" ],
          [ 'x{3,}.*TEST?x*', "" ], [ '^- Let me.*-', "" ], [ '^.*\^\^;.*\){1}', "" ],
          [ '^As the title suggests.*Littleton.', "" ],
          [ '^.*\)\)\n', "" ]
        ]

  - step: &links
      type: RegexReplace
      description: "Remove all links"
      config:
        replacements: [
          [ '\[(img|video).*\].*\[(\/img|video)\]', "" ],
          [ '[hH]ttp[s]?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]|[!*\(\),]|(?:%[0-9a-fA-F][0-9a-fA-F]))+', "" ]
        ]

  - step: &reduce_whitespace
      type: ExactReplace
      description: "Reduce standalone tabs that do not affect formatting, and newlines to maximum of two consecutive."
      config:
        recursive: true
        replacements: [
          [ "\t\n", "\n" ], [ "\n\n\n", "\n\n" ], [ '  ', ' ' ], [ " \n", "\n" ]
        ]

  - step: &reduce_separators
      type: ExactReplace
      description: "Reduce nonsensical separators to a maximum of 3 consecutive."
      config:
        recursive: true
        replacements: [ [ "~~~~", "~~~" ], [ "----", "---" ], [ "<>", "*" ], [ "****", "***" ], [ ".....", "...." ] ]

  - step: &reduce_punctuation
      type: ExactReplace
      description: "Reduce excessive punctuation"
      config:
        recursive: true
        replacements: [ [ "!!!!", "!!!" ], [ "????", "???" ], [ "?!?!?!", "?!?!" ], [ "!?!?!?", "!?!?" ] ]

  - step: &trim
      type: Trim
      description: "Trim leading and trailing whitespace."

  - step: &ooc_prune
      type: FullMatch
      description: "Prune fully OOC messages with known content."
      config:
        action: drop
        caseSensitive: false
        patterns: [
          "BUMP", "-BUMP-", "~BUMP~", "Closed...", "Closed", "Closed\n\n...", "This thread is closed", "bump.", "*DELETED*",
          "...", "Okay... so you didn't make a starting post??", "---", "~~~",
          # Remnants of malformed OOC
          ")", "\n(", "))", "ﾉ ))", "ᕗ ))", " ))", "」I dunno!))", "ง ))", "ﾉﾞ ))", "_ ))", "ᕗ))", "＿ ))", "ノ ))",
          "The End ^^\nshortest RP ever XDD", "the end^^", "i think so^^", "Fin^^", "Fin ^^", "FIN​", "-Placeholder-", "Molly was",
          "Hahahaha, Yeah, I was a little confused",
          "Lol do you want to delted this thread possibly and start a new one I'm sorry",
          "Sure, what ever you want!", "How about a demon/human what do you want the plot to be and if you could start it",
          "Okay, do you want to be the demon, or do you want me to be the demon?", "You be the demon", "Okay, do you want to start today or tomorrow?",
          "Today if possible and what's the plot", "Hmmmmm I have no idea hahahaha you?", "Hmmm demon falls for a human even though he ain't supposed to",
          "Ok, sounds good! Anything else you want in?",
          "Not really maybe demon fight\n\nAnd that the demon has to shroud him self in human form and reveals himself to her",
          "Okay, I'll make it a bit later.", "Okay can't wait",
          "THIS RP IS OFFICIALLY CLOSED NO FURTHER POSTS WILL BE POSTED AS I HAVE LOST INTEREST IN THE PLOT SORRY FOR THE LOSSES",
          # TODO Do some whitespace reduction before post.
          "This is a Supernatural Fandom Romance RP between Dean Winchester and Castiel the angel. \n\nWARNING! CONTAINS HOMOSEXUAL CONTENT! IF YOU'RE HOMOPHOBIC, GO AWAY!",
          "\n\nThis will be a one x one between myself and Iikaitlynii based off the AMC television series The Walking Dead​",
          "Well I doubt he is going to be able to contact me if he is banned....can somebody please tell me why he got banned? I was just rping with him and he role plays well and then I saw he got banned for Ban-Evasion but what exactly was he doing to get banned?",
          "title or something idk", "Geralt = \n\nDandelion = ",
          "(Are we still rping this?"
        ]

  - step: &strip_unicode
      type: Encoding
      description: "Strip unicode sequences not present in ascii. This doesn't affect output encoding."
      config:
        invalid: ""
        codec: us-ascii

  - step: &strip_nonwords
      type: RegexReplace
      description: "Consecutive non-word characters"
      config:
        dotAll: true
        multiLine: true
        replacements: [
          [ '^(-|,|\||\.|=|_|])\W*(s|\W)+\W(\b|\n)', "" ],
          [ '^([w_=x]{5,}|~\+\+~)(\b|\n)', "" ],
          [ '^(_|])+(\b|\n)', "" ], [ '^(\s|\.|,|\|)+', "" ],
          [ '([^a-zA-Z0-9 !?\.])\1{4,}', "" ],
        ]

  - step: &spelling
      type: ExactReplace
      description: "Spelling mistakes and typos."
      pack: [ 'resources/spelling.csv' ]
      config:
        caseSensitive: false
        packedReplacements: [ 'resources/spelling.csv' ]

  - step: &empty
      type: FullMatch
      description: "Drop empty messages."
      config:
        action: drop
        patterns: [ "" ]

  - step: &two_alternating_participants
      type: Participants
      description: "Require exactly 2 alternating participants"
      config:
        min: 2
        max: 2
        alternating: true

  - step: &rename_participants
      type: RenameParticipants
      description: "Rename participants to their conversation roles."
      config:
        names: [ "human", "gpt" ]

  - step: &edge_cases
      type: ExactReplace
      description: "Remove a few edge cases."
      pack: [ 'resources/edge_cases.csv' ]
      config:
        packedReplacements: [ 'resources/edge_cases.csv' ]

  - step: &stats_output
      type: StatsOutput
      description: "Output statistics"
      sync: main
      config:
        path: 'out/stats/full/train/stats.json'
        vocabPath: 'out/stats/full/train/vocabulary.json.gz'
        gZipVocabulary: true
        conversationsPath: 'out/stats/full/train/conversations.json.gz'
        gZipConversations: true
_passes:
  - pass: &clean
      required:
        - path: "original/rentry/bluemoon/bluemoon-fanbased-roleplays-1x1.csv"
          sha512: f31d6bd278bc4211736d6aace3917cd0c1d0143bec9bf9d07054f5f9b32060e17399b6ea0935774b2271ac45a88309a60ef8eec8c4ac5283d1b353a255529cc5
      steps: [ *bluemoon,
               *html_linebreaks, *html_strip, *links, *trim, *reduce_whitespace,
               *ooc_strip, *reduce_separators, *reduce_punctuation, *spelling, *ooc_prune,
               *empty,
               *strip_unicode,
               *reduce_whitespace, *strip_nonwords, *edge_cases, *trim, *empty,
               *two_alternating_participants, *rename_participants,
               *bluemoon_fastchat, *stats_output
      ]
      artifacts:
        - file: 'out/data/full/bluemoon.train.json'
          sha512: 70ddecc11f8b280a5d641ce502e91a2392aa9dc12e0704617edb1599e4c235b6e8015987f109a09c83746f717cec715fcad04587cc066897583eaf2bd439616b
  - pass: &test_sample
      required:
        - path: "out/data/full/bluemoon.train.json"
      steps:
        - *bluemoon_cleaned
        - type: RawSampleOutput
          sync: main
          config:
            ratio: 0.01
            forwardFilter: true
            path: "out/data/full/bluemoon.test.raw"
        - type: StatsOutput
          description: "Output statistics"
          sync: main
          config:
            path: 'out/stats/full/test/stats.json'
            vocabPath: 'out/stats/full/test/vocabulary.json.gz'
            gZipVocabulary: true
            conversationsPath: 'out/stats/full/test/conversations.json.gz'
            gZipConversations: true
      artifacts:
        - file: 'out/data/full/bluemoon.test.raw'
          sha512: 4a01cc53de82c36dc1fa30ffb41c1c8f1a9bfa0a51f632f9fc9d8dbe57a2200008a93562521a23bfbc8779f039116c0604b6cc511dcaf907ba701bfa80620b48
passes: [ *clean, *test_sample ]