Arwing's goal is to be a querying engine for finding interactions in a set of SSBM slippi replays.

   ./arwing -p PLAYER_CHARACTER -o OPPONENT_CHARACTER -d DIRECTORY_TO_PARSE -i INTERACTION -e JSON_EXPORT_LOCATION

• PLAYER_CHARACTER and OPPONENT_CHARACTER are all-caps character names with space delimited by underscores, ex: GAME_AND_WATCH

• DIRECTORY_TO_PARSE parses recursively through each folder and finds each .slp file to analyze. Symlinks are ignored.

• INTERACTION is a set of three data fields repeated any amount of times:

  • Character Performing Action (In the same format as PLAYER_CHARACTER)
  • ACTION_STATE of the action (Action State IDs, Character-Specific Action State IDs)
  • Amount of frames between previous action and this action

• EXPORT: If the -e flag is specified a json file will be exported for use in in-engine replay with Clippi (loaded in the same way as any clippi combo.json)

• What do top foxes do after drilling pikachu?
• How do players DI out of chaingrabs?
• What does a particular opponent like to do out of laser?

• Missing L-cancels
• Not getting enough galint on ledgedash
• Getting stuck in shine
• Not extending combos as far as possible
• And much much more!

• Parse data stream instead of in-memory

• Evaluate using concurrency/parallelism

• Evaluate using arrow

• Cache Data

• Refine query system

  • Select stages
  • Parse by Player Name
  • Add Meta-Actions i.e. From-ledge, Wavedash, Retreat, Advance

• Create GUI App

  • Likely Tauri
  • Represent user flow