Format:Ren'Py RPA archive
| On this wiki | Facts · Code · Notes · Topics |
|---|
Ren'Py's archive format (.rpa, and .rpi for version 1): version 3 starts with the ASCII line RPA-3.0 plus a 16-digit hex index offset and an 8-digit hex key; the index is a zlib-compressed pickle mapping file names to lists of (offset, length) tuples that are XORed with the key.
Headers
The handlers in the engine are tried by file header:
RPA-3.0followed by 16 hex digits (offset of the index), a space, 8 hex digits (key) and a newline; the loader reads 40 bytes.RPA-2.0followed by 16 hex digits (index offset) and a newline; 24 bytes. No key.- Version 1 is an index file
.rpithat starts with a zlib stream (78 9C) and has no header.
Index
Seek to the offset, zlib-decompress to the end of the file, and unpickle: a dict from a path to a list of tuples. In v3 each (offset, length) is XORed with the key, and a three-element tuple (offset, length, prefix) carries a short prefix of bytes (the first bytes of the file) that go before the data.
Related
Library:Ren'Py, Format:Ren'Py RPYC, Format:Zlib stream
Unpickling is code execution: parse the pickle with a restricted unpickler.
Code & repositories
- https://github.com/renpy/renpy @
6202e7fb43fd23594ab866af83bf928525a06410· license: MIT (declared by the contributor; GitHub's detector did not recognize a license file)
Ren'Py engine source: renpy/loader.py (RPA handlers) and renpy/script.py (rpyc reading and writing) are the format references.
Tested on: source read, not run: source read only · checked 2026-10-18