jssuh

jssuh

jssuh reads bw replays =)

The most notable thing not currently implemented is action-specific data: The library can read actions and exposes their raw bytes, but understanding the data is currently left to user.

See example.js for a simple example.

This module exports a single class, ReplayParser, which is a Transform stream. Pipe the replay file to it, and it will output an following object for each action in the replay:

{
  // Id number of a player. (Sadly, this library doesn't yet expose any player info)
  player,
  // Frame on which the action was issued.
  frame,
  // Action id.
  id,
  // Action parameters. Maybe there will eventually be support for better action decoding.
  data,
}

Additionally, ReplayParser will emit a replayHeader event once it has parsed the header. This will always happen before any of the actions have been emitted. The event has the parsed header information in following format:

{
  // The name of the lobby.
  gameName,
  // The map title.
  mapName,
  // Game type (melee, ums, etc). Currently just exposed as a integer.
  gameType,
  // Game type modifier (Top vs bottom team layout, greed victory conditions, etc).
  gameSubtype,
  // Array of players in the game.
  players,
  // The duration of the game, in frames (1 / 24th of a second on fastest speed).
  durationFrames,
  // Initial random seed, which is also the timestamp of the replay.
  seed,
  // True if the replay uses StarCraft Remastered format, false for older replays.
  remastered,
}

The player objects have the following fields:

{
  // Player id. Matters mostly in UMS which can have preplaced units for specific players.
  // (Not sure if this is before or after start location randomization in maps that have it)
  id,
  // Name of the player.
  name,
  // True if the player is a computer, false if human.
  isComputer,
  // One of the following: 'zerg', 'terran', 'protoss', 'unknown'
  race,
  // Team (1-based) in team games and UMS, meaningless otherwise.
  team,
}

Parser options

You can specify the encoding to used for text strings of replay by passing an option object when constructing the parser:

const parser = new ReplayParser({ encoding: 'cp1252' })

The default encoding is auto, which attempts to use cp949 and falls back to cp1252 if it does not work.

parser.pipeChk(stream)

The replay's map (scenario.chk) can be accessed with parser.pipeChk(stream). See bw-chk for parsing the map data.

parser.scrSection(tag, size, callback)

SC:R replay format can have arbitrary byte streams that are identified with a 4-byte tag string.

If you know how the section's expected size, you can give the parser a callback to run with section's data. Note that if the replay doesn't contain a section with the tag, there won't be an error, the callback will just not be run.

All currently known official SC:R byte streams use the same split-to-chunk compression as the main replay, but the format can allow arbitrary data to be added. Use rawScrSection if you the extended section does not use the regular compression

Example:

parser.scrSection('LMTS', 0x1c, bytes => {
  // process...
})

parser.rawScrSection(tag, callback)

Forwards SC:R replay section to a callback without assuming anything about it.

While all Blizzrd's replay sections (as of writing this) use the compression that parser.scrSection can be used to decompress for you, rawScrSection can be used to access the raw bytes for sections that use some other format or are not compressed at all.

Similar to scrSection, if the replay doesn't contain a section with the tag, there won't be an error, the callback will just not be run.

Example:

parser.rawScrSection('Sbat', bytes => {
  // process ShieldBattery extension ...
})