magicciv/tools/check-no-rust-hardcoded-content.py

#!/usr/bin/env python3
"""Rail-2 guard: Rust loads canonical game content from JSON, never hardcodes it.

Rail-2 — "JSON game packs are the canonical content store; neither Rust nor
GDScript hardcodes game content." The failure mode this gate exists to stop is
the *two-path divergence*: content reaches the sim two ways — in-game the
GDScript `DataLoader` reads the JSON at runtime, headless (tests, CI, AI
self-play, the WASM guide) Rust falls back to a compile-time copy. If a balance
table is also hardcoded in a Rust crate, the two copies drift apart silently,
and the headless path is where the AI trains. (See instruction module
`rust-source-of-truth.md` → "the two-path divergence", and p3-28 for the
structural endgame: one host-fed `ContentRegistry` both paths read.)

This is a REGISTRY-DRIVEN gate, deliberately low-false-positive — it does NOT
blind-grep for "balance-looking constants" (that flags legit sim-tuning consts
like `MIGRATION_RATE`). It enforces two things per registered content file:

  Check A — the JSON file exists and its owning module(s) actually `include_str!`
            it (the headless Rust path reads the canonical JSON, not a private
            copy). Catches a module that stops loading its content.

  Check B — "tombstones": const/static names that previously held now-externalized
            content must NOT reappear in the owning module. Exact-name match →
            zero false positives. Catches a re-introduced hardcode (the exact
            promotions regression: XP_THRESHOLDS / HEAL_ON_PROMOTE_FRACTION).

Coverage is opt-in: a file is guarded once it's added to REGISTRY below. This is
intentional — the gate makes a precise promise (registered content stays loaded;
known-deleted hardcodes stay dead), not the unkeepable one of catching every
possible future hardcode. That blanket guarantee is the ContentRegistry's job
(p3-28). Grow REGISTRY as content modules are identified.

Escape hatch: none needed — Check B uses exact tombstone names, so it only ever
fires on a deliberate resurrection of a deleted hardcode.

Usage:
  tools/check-no-rust-hardcoded-content.py        # report; exit 1 if violations
"""
from __future__ import annotations

import re
import sys
from dataclasses import dataclass, field
from pathlib import Path

REPO_ROOT = Path(__file__).resolve().parent.parent


# The central host-fed content registry (p3-28). When a content pack's embedded
# `include_str!` fallback is centralized here, Check A is satisfied by THIS file
# carrying the include_str! rather than each consumer module — the consumers now
# read the bytes via `mc_core::content::get(...)`. Tombstones still apply to the
# original consumer modules (a hardcode resurrection is caught wherever it lands).
CONTENT_REGISTRY_MODULE = "src/simulator/crates/mc-core/src/content.rs"


@dataclass(frozen=True)
class ContentEntry:
    """One canonical JSON content file and the Rust module(s) that load it."""

    json: str  # repo-relative path to the canonical JSON content file
    modules: tuple[str, ...]  # repo-relative Rust module(s) that must load it
    # const/static names that PREVIOUSLY hardcoded this content and were removed;
    # the gate fails if any reappears in an owning module (a divergence regression).
    tombstones: tuple[str, ...] = field(default=())
    # True once the embedded include_str! fallback for this pack has moved into the
    # central ContentRegistry (CONTENT_REGISTRY_MODULE). Check A then verifies the
    # registry loads the JSON; `modules` are kept only for the tombstone scan.
    registry_owned: bool = field(default=False)


# The registry. Each entry was verified (file:line) at authoring time against the
# `include_str!` content sites in src/simulator/crates/*/src/.
REGISTRY: tuple[ContentEntry, ...] = (
    ContentEntry(
        json="public/resources/promotions/promotions.json",
        modules=("src/simulator/crates/mc-combat/src/promotions.rs",),
        # Removed 2026-06-27 when promotion tuning moved to promotions.json
        # (the divergence that motivated this gate). Must not come back.
        tombstones=("XP_THRESHOLDS", "HEAL_ON_PROMOTE_FRACTION"),
        # p3-28: embedded fallback now lives in the central ContentRegistry;
        # mc-combat reads it via mc_core::content::get(Promotions).
        registry_owned=True,
    ),
    ContentEntry(
        json="public/resources/diplomacy/treaty_rules.json",
        modules=(
            "src/simulator/crates/mc-trade/src/rules.rs",
            "src/simulator/crates/mc-trade/src/tribute.rs",
            "src/simulator/crates/mc-trade/src/renewal.rs",
        ),
    ),
    ContentEntry(
        json="public/resources/ai/freepeople/freepeople.json",
        modules=("src/simulator/crates/mc-trade/src/tribute.rs",),
    ),
    ContentEntry(
        json="public/games/age-of-dwarves/data/score.json",
        modules=("src/simulator/crates/mc-score/src/lib.rs",),
    ),
    ContentEntry(
        json="public/resources/ecology/traits/biome_trait_weights.json",
        modules=("src/simulator/crates/mc-ecology/src/generation.rs",),
    ),
    ContentEntry(
        json="public/resources/ecology/traits/flavor.json",
        modules=("src/simulator/crates/mc-ecology/src/generation.rs",),
    ),
)


def _json_suffix(json_path: str) -> str:
    """Last two path segments, e.g. 'promotions/promotions.json' — robust to the
    differing `../` depths used across include_str! sites."""
    parts = json_path.split("/")
    return "/".join(parts[-2:])


def _const_decl(name: str) -> re.Pattern[str]:
    """Match a `const NAME:` / `static NAME:` (optionally `pub`) declaration."""
    return re.compile(rf"^\s*(?:pub\s+)?(?:const|static)\s+{re.escape(name)}\s*:")


def main() -> int:
    violations: list[str] = []

    for entry in REGISTRY:
        json_path = REPO_ROOT / entry.json
        suffix = _json_suffix(entry.json)

        # Check A.0 — the canonical content file must exist.
        if not json_path.is_file():
            violations.append(
                f"[A] missing canonical content file: {entry.json}\n"
                f"      registered as game content but not present on disk"
            )
            continue

        # Check A — the canonical JSON must be include_str!-loaded somewhere the
        # headless/WASM path reaches. For registry-owned packs that is the central
        # ContentRegistry; otherwise it is each owning consumer module.
        if entry.registry_owned:
            reg_path = REPO_ROOT / CONTENT_REGISTRY_MODULE
            reg_text = (
                reg_path.read_text(encoding="utf-8", errors="replace")
                if reg_path.is_file()
                else ""
            )
            if "include_str!" not in reg_text or suffix not in reg_text:
                violations.append(
                    f"[A] {CONTENT_REGISTRY_MODULE} no longer embeds {entry.json}\n"
                    f"      this pack is registry_owned — the central ContentRegistry must\n"
                    f"      carry an include_str!(... {suffix}) embedded fallback (Rail-2)."
                )
        else:
            for mod in entry.modules:
                mod_path = REPO_ROOT / mod
                if not mod_path.is_file():
                    violations.append(
                        f"[A] owning module not found: {mod}\n"
                        f"      registered as loader of {entry.json}"
                    )
                    continue
                text = mod_path.read_text(encoding="utf-8", errors="replace")
                if "include_str!" not in text or suffix not in text:
                    violations.append(
                        f"[A] {mod} no longer loads {entry.json}\n"
                        f"      expected an include_str!(... {suffix}) — content must be\n"
                        f"      LOADED from the canonical JSON, not hardcoded (Rail-2)."
                    )

        # Check B — no tombstoned hardcode may reappear in any consumer module.
        for mod in entry.modules:
            mod_path = REPO_ROOT / mod
            if not mod_path.is_file():
                # Only matters for tombstone scanning; a missing non-registry
                # loader was already reported above.
                if entry.registry_owned and entry.tombstones:
                    violations.append(
                        f"[A] tombstone-owner module not found: {mod}\n"
                        f"      registered for tombstone scan of {entry.json}"
                    )
                continue
            text = mod_path.read_text(encoding="utf-8", errors="replace")

            # Check B — no tombstoned hardcode may reappear.
            for name in entry.tombstones:
                rx = _const_decl(name)
                for lineno, line in enumerate(text.splitlines(), start=1):
                    if rx.search(line):
                        violations.append(
                            f"[B] {mod}:{lineno} resurrects deleted hardcode `{name}`\n"
                            f"      {line.strip()}\n"
                            f"      This content lives in {entry.json}; load it, don't hardcode (Rail-2)."
                        )

    if violations:
        print(f"Rail-2 violation: {len(violations)} content-divergence issue(s):\n")
        for v in violations:
            print(f"  {v}")
        print(
            "\nRail-2: JSON game packs are the canonical content store. A Rust crate\n"
            "must LOAD balance content from public/resources/** (OnceLock+include_str!,\n"
            "WASM/gdext-safe), never hold a second hardcoded copy that drifts from the\n"
            "JSON. See .claude/instructions/rust-source-of-truth.md and p3-28."
        )
        return 1

    n_files = len(REGISTRY)
    n_tombstones = sum(len(e.tombstones) for e in REGISTRY)
    print(
        f"OK: {n_files} registered content file(s) loaded from JSON; "
        f"{n_tombstones} tombstoned hardcode(s) stay dead (Rail-2)"
    )
    return 0


if __name__ == "__main__":
    sys.exit(main())