reincarnate

Appearance

RAGS (Rich Authoring and Generation System) ​

Status: Planned — No implementation started. rags2html provides a working reference implementation.

Background ​

RAGS is a Windows-only interactive fiction authoring tool popular in the NSFW IF community. Games are click-and-read affairs with an inventory system, branching narrative, image display, and timed events. Development of the engine stopped years ago; the game format is closed and proprietary.

Format ​

RAGS uses two distinct binary formats depending on the engine version:

Pre-1.7 (NRBF):

  • .NET Remoting Binary Format — a serialization of the engine's internal C# classes using BinaryFormatter
  • Encrypted with AES-256-CBC using a hardcoded key and IV
  • Game data after decryption is a graph of C# object instances: rooms, characters, items, timers, and media references

Post-1.7 (SDF):

  • Microsoft SQL Server Compact Edition (SQL CE) database
  • Each 4096-byte page is encrypted with AES-128-CBC; per-page keys are derived from a global key and a page checksum
  • The SDF format is otherwise undocumented

Both formats store: rooms, characters, items, inventory, timers, variables, player state, and media file references (images, audio). Game logic is data-driven — there is no Turing-complete scripting language. Conditions are stored as structured predicates, not code.

Existing Tool: rags2html ​

rags2html is an open-source tool that:

  1. Decrypts both the NRBF and SDF formats using the known keys
  2. Parses the game data
  3. Generates a self-contained playable HTML file using a reimplemented "Regalia" JavaScript runtime
  4. Has been tested against ~100 games

This is the definitive reference for the format. The reincarnate approach would build on this work.

Lifting Strategy ​

Data-driven extraction (Tier 2). Since game logic is stored as structured predicates rather than code:

  1. Decrypt the .rags file (keys are known from rags2html)
  2. Deserialize the NRBF or SDF structure
  3. Translate room/item/character/event data to IR or directly to TypeScript
  4. Emit a replacement runtime for the "Regalia" engine UI (text display, inventory, images, choices)

The absence of a scripting language makes this one of the simpler engines to lift — the entire game state machine is in the data, not code.

What Needs Building ​

Format Parser (new crate: reincarnate-frontend-rags) ​

  • [ ] AES-256-CBC decryption (NRBF variant) — key/IV already documented in rags2html
  • [ ] NRBF deserializer (pre-1.7): parse .NET BinaryFormatter stream → game object graph
  • [ ] SQL CE SDF parser (post-1.7): page-level AES decryption + SQL CE B-tree page format → game object graph
  • [ ] Game data model: rooms, exits, items, characters, variables, timers, conditions, actions, media references

IR / Code Generation ​

Since the game logic is data-driven (not a scripting language), the "IR" is more of a data model translation:

  • [ ] Rooms → TypeScript objects with text, image, exit list, item list
  • [ ] Conditions → TypeScript boolean expressions over game state
  • [ ] Actions → TypeScript functions (set variable, move item, print text, play audio, etc.)
  • [ ] Event system → triggered actions on condition match

Replacement Runtime (runtime/rags/ts/) ​

  • [ ] Text output area with formatted HTML
  • [ ] Image display panel (room/character images)
  • [ ] Inventory display
  • [ ] Choice/button UI (exits, item interactions, conversation options)
  • [ ] Variable state management
  • [ ] Timer system (timed events)
  • [ ] Audio playback (background music, sound effects)
  • [ ] Save/load

References ​