This project reconstructs the original Japanese TH09 version 1.50a executable. It starts from an independently versioned, hash-attested target and tracks boundary, origin, source presence, codegen exactness, native-product closure, runtime semantics, and portability as separate facts.
Place your own legal copy at the ignored canonical path
resources/th09.exe. Repository tools use that path by default. An explicit
path or TH09_TARGET_PATH is supported only for deliberate local overrides;
normal Factory work must use the canonical path.
| Property | Required value |
|---|---|
| Version | original Japanese 1.50a |
| Size | 685,056 bytes |
| SHA-256 | 10350095bcf95edb59e03bee9849a2dc8a7714b4927ad5909c569c550fce6822 |
| MD5 | cf634df46e05552e104fa97a971aaac0 |
| Image base | 0x00400000 |
| Entry point | 0x0047D45F |
The pinned thcrap version database
classifies this exact SHA-256 and size as th09, v1.50a, (original) and
lists the English-patched executable under a different digest. The
official 1.50a release post is dated
2005-10-08 JST, consistent with the target's PE timestamp.
python3 scripts/verify-target.py
python3 scripts/validate-tracking.py --require-target
python3 scripts/report-reconstruction-status.pyIDA attestation is environment-specific: local Codex calls get_metadata with
{} through the direct provider and runs scripts/check-ida-mcp.py for mapped-byte
attestation, while Factory/Web must call
get_metadata with {} through the attested th09-ida provider. See
AGENTS.md and docs/TOOLS.md before analysis.
Original executables, game data, IDA databases, decompiler output, and downloaded tools are private and never committed.
The current operator targets 95% exact reviewed-authored code bytes, with
both large functions (especially RunEcl, Enemy/Etama, TitleScreen and ExAttack)
and small leaves in scope. Factory/Web and local Codex use their respective
attested IDA routes and the pinned VC7.1 compiler Oracles. The ledger-derived
counts are in docs/PROGRESS.md; the replace-in-place
docs/RE_HANDOFF.md is the live routing and restart guide.
Historical experiments in the knowledge base are not current codegen status.
This measures confirmed authored bytes, not whole-image or runnable-product
coverage.
Every new session must attest the target and live IDA database. Current
exact/non-exact state comes from config/functions.csv,
config/matches.csv, config/match-units.toml, and a fresh status report.
docs/KNOWLEDGE_BASE.md indexes durable evidence and bounded investigation
notes; historical measurements do not replace the live ledgers.
The PE and Rich header identify Microsoft Visual C++ .NET 2003 build 3077.
Some runtime-library contributions and compiler-generated helpers are proven,
but the complete translation-unit graph, resources, data owners, compiler
profiles, and link order remain open. config/build.toml records that state
instead of pretending the project is buildable.
Run python3 scripts/progress.py to regenerate the status card,
python3 scripts/validate-docs.py to catch stale handoff totals, percentage,
95% gap, non-exact routing and accepted-owner manifest/receipt entries, and
python3 scripts/ci.py for the target-independent validation set. The ignored
build/ directory is deliberately cleared before handoff; replay a function
by running python3 scripts/build-match-unit.py --unit NAME before its
corresponding comparison or non-crediting inspector. Preview disposable-cache
cleanup with python3 scripts/clean-generated-artifacts.py and explicitly
apply it using --apply; original game data and historical probe source are
not cleanup targets.
- Preserve the reviewed boundary/origin inventory and recover remaining exact VC7.1 authored units.
- Close and exercise the faithful Windows i386 product, including static data ownership, initializers, resources, libraries, and runtime scenarios.
- Perform semantic reconstruction with two independent runtime Oracles: the original target and the reconstructed native i386 product.
- Start a portability branch only after the native semantic baseline is trustworthy.
Exact functions do not imply a complete or runnable product. A modern port can hide missing original data owners, ABI behavior, and initialization, so it is not an acceptable shortcut around the native-product stage.
AGENTS.md— mandatory session, evidence, and checkpoint rules.config/target.toml— exact target and observed PE/toolchain facts.config/build.toml— whole-product graph and explicit unknowns.docs/RE_WORKFLOW.md— bounded reconstruction loop and stage gates.docs/ORACLES.md— what can falsify or accept each claim.docs/RE_HANDOFF.md— current state, retained evidence, and restart checks.docs/KNOWLEDGE_BASE.md— durable TH09-only facts.docs/TOOLS.md— local and Factory tool routing.docs/PROGRESS.md— generated ledger totals.- Touhou Reconstruction Factory — shared ontology, workflow, and cross-game knowledge.
Repository-authored code and documentation are provided under the MIT License. This does not grant rights to the original game or its assets.