Step 10j (Python side): cache contract + --reextract command

Implements the two-tier contract from docs/CACHE_CONTRACT.md (extension
repo, locked at step 9):

  cache_schema       on-disk shape; mismatch -> force rebuild
  id_rules           bumps when extraction rules change
  id_rules_signature sha256 over canonical rule text; catches drift
                     when the integer bump is forgotten

New constants in rcjav/cache.py:

  CACHE_SCHEMA_VERSION = 1
  ID_RULES_VERSION = 1     (the legacy "version: 3" cache reads as
                            id_rules: 0 after in-place migration)

New helpers:

  rcjav.ids.current_rules_signature()
      Sha256 over the canonical text of every rule that influences
      a jav_id: built-in regexes, BUILTIN_PART_RES, PART_RES (which
      captures user-added part patterns), FC2 handling.

  rcjav.cache.load_cache(signature=None)
      Reads cache.json. Legacy `version: 3` headers get an in-place
      header upgrade with no forced rescan; the cache is stamped as
      `id_rules: 0` + signature "legacy" so it surfaces as
      "stale by rules" in cache_state. Schema mismatch on the new
      header still forces a rebuild.

  rcjav.cache.cache_state(cache, signature)
      Classifies a cache as "fresh" / "stale_by_rules" /
      "schema_mismatch". Drives the three-state extension UX.

  rcjav.cache.stamp_current_rules(cache, signature)
      Updates id_rules and id_rules_signature in place. Called after
      a successful full scan or --reextract.

New CLI command:

  rc-jav.py --reextract

Walks `cache["remotes"][r]["files"]` against the live rule set and
updates `jav_id` in place. No rclone calls — fast path (seconds on
a 7k-file cache). Reports changed/unchanged/dropped per remote.
Stamps current rules into the saved cache.

--scan (full, no --scan-since) now also stamps current rules.
--scan --scan-since deliberately does NOT stamp: it only re-walks
recently-modified files, so older entries may still carry jav_ids
from previous rules; cache stays "stale by rules" until a full scan
or --reextract.

Verified:
  - python rc-jav.py --reextract --format json on the live 7124-file
    cache → 0 changes (existing IDs already canonical), cache.json
    rewritten with new header
  - cache_state on the post-migration cache → "fresh"
  - tests + fixtures + --help all pass

Extension-side (host's cache_status response + options-cache.js
three-state UX + Re-extract IDs button) ships in a separate commit
in the extension repo.

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
This commit is contained in:
admin
2026-05-22 22:07:13 +02:00
parent 1cc2c38128
commit 33c495ad57
4 changed files with 233 additions and 21 deletions
+5
View File
@@ -82,11 +82,15 @@ from rcjav.dupes import ( # noqa: F401
from rcjav.cache import ( # noqa: F401
CACHE_PATH,
CACHE_VERSION,
CACHE_SCHEMA_VERSION,
ID_RULES_VERSION,
CACHE_STALE_HOURS,
load_cache,
save_cache,
cache_age_hours,
fmt_age,
cache_state,
stamp_current_rules,
)
from rcjav.ids import ( # noqa: F401
PRIMARY_ID_RE,
@@ -103,4 +107,5 @@ from rcjav.ids import ( # noqa: F401
normalize_id,
describe_id_match,
expand_range,
current_rules_signature,
)
+108 -20
View File
@@ -2,10 +2,20 @@
This module owns the on-disk cache contract: where the file lives,
what the header looks like, and how mismatches are handled. The
current shape predates the two-tier `cache_schema` + `id_rules` split
documented in docs/CACHE_CONTRACT.md (extension repo) — step 10j
implements that contract; until then this is the legacy
`version: 3` reader.
contract is the two-tier `cache_schema` + `id_rules` model from
docs/CACHE_CONTRACT.md (extension repo).
cache_schema on-disk shape. Mismatch -> force rebuild.
id_rules integer; bumps when extraction rules change.
Mismatch -> mark stale, allow lazy re-extract.
id_rules_signature sha256 over canonical rule text (see
rcjav.ids.current_rules_signature). Belt-and-
braces drift check that catches a forgotten
`id_rules` bump.
Legacy users on `version: 3` get an in-place header upgrade with no
forced rescan; the cache is marked as `id_rules: 0` so it shows up
as "stale by rules" until they Re-extract IDs.
"""
from __future__ import annotations
@@ -19,29 +29,107 @@ from pathlib import Path
# Lives next to rc-jav.py at the repo root.
CACHE_PATH = Path(__file__).resolve().parents[1] / "cache.json"
CACHE_VERSION = 3 # bumped: extract_id handles bracket-wrapped IDs + no-hyphen fallback
CACHE_STALE_HOURS = 24
# Two-tier version contract (see docs/CACHE_CONTRACT.md):
CACHE_SCHEMA_VERSION = 1 # on-disk shape; bump = force rebuild
ID_RULES_VERSION = 1 # extraction rules; bump = mark stale (lazy re-extract)
# Legacy alias preserved for any external caller that still imports it.
# Maps to CACHE_SCHEMA_VERSION + ID_RULES_VERSION under the new contract.
CACHE_VERSION = 3
def _fresh_cache(signature: str = "unknown") -> dict:
return {
"cache_schema": CACHE_SCHEMA_VERSION,
"id_rules": ID_RULES_VERSION,
"id_rules_signature": signature,
"remotes": {},
}
def _migrate_legacy_v3(data: dict) -> dict:
"""Translate a legacy `version: 3` cache to the new header in place.
Sets `id_rules: 0` so the cache reads as "stale by rules" — user
sees the new amber state and can opt into a fast Re-extract without
a rclone re-scan.
"""
return {
"cache_schema": CACHE_SCHEMA_VERSION,
"id_rules": 0,
"id_rules_signature": "legacy",
"remotes": data.get("remotes", {}),
}
def load_cache(current_signature: str | None = None) -> dict:
"""Read and (if necessary) migrate cache.json.
`current_signature` is the value of `rcjav.ids.current_rules_signature()`
captured by the caller. It's only stamped into the header when this
function has to mint a *fresh* cache; when migrating legacy data we
deliberately stamp `"legacy"` so the cache reads as stale-by-rules.
"""
fresh_sig = current_signature or "unknown"
def load_cache() -> dict:
if not CACHE_PATH.exists():
return {"version": CACHE_VERSION, "remotes": {}}
return _fresh_cache(fresh_sig)
try:
data = json.loads(CACHE_PATH.read_text(encoding="utf-8"))
if (
not isinstance(data, dict)
or data.get("version") != CACHE_VERSION
or not isinstance(data.get("remotes"), dict)
):
if isinstance(data, dict) and "version" in data and data["version"] != CACHE_VERSION:
sys.stderr.write(
f"[warn] cache version mismatch (got {data['version']}, "
f"expected {CACHE_VERSION}); forcing full rescan.\n"
)
return {"version": CACHE_VERSION, "remotes": {}}
return data
except (json.JSONDecodeError, OSError):
return {"version": CACHE_VERSION, "remotes": {}}
return _fresh_cache(fresh_sig)
if not isinstance(data, dict) or not isinstance(data.get("remotes"), dict):
return _fresh_cache(fresh_sig)
# Legacy header: { "version": 3, "remotes": {...} } — migrate in place.
if "version" in data and "cache_schema" not in data:
if data.get("version") == 3:
return _migrate_legacy_v3(data)
sys.stderr.write(
f"[warn] unknown legacy cache version {data.get('version')!r}; "
f"rebuilding.\n"
)
return _fresh_cache(fresh_sig)
# New header: validate schema. Mismatch = force rebuild (per contract).
if data.get("cache_schema") != CACHE_SCHEMA_VERSION:
sys.stderr.write(
f"[warn] cache_schema mismatch (got {data.get('cache_schema')!r}, "
f"expected {CACHE_SCHEMA_VERSION}); forcing full rescan.\n"
)
return _fresh_cache(fresh_sig)
return data
def cache_state(cache: dict, current_signature: str) -> str:
"""Classify a cache dict against the live rule set.
Returns one of: "fresh", "stale_by_rules", "schema_mismatch".
"schema_mismatch" should normally never reach the caller — load_cache
already rebuilds. It's reported for diagnostics flows that read
cache.json directly without going through load_cache.
"""
if cache.get("cache_schema") != CACHE_SCHEMA_VERSION:
return "schema_mismatch"
rules_match = cache.get("id_rules") == ID_RULES_VERSION
sig_match = cache.get("id_rules_signature") == current_signature
return "fresh" if (rules_match and sig_match) else "stale_by_rules"
def stamp_current_rules(cache: dict, current_signature: str) -> None:
"""Stamp `id_rules` and `id_rules_signature` to current values in place.
Use after a successful re-extract or full scan completes against the
live rule set.
"""
cache["id_rules"] = ID_RULES_VERSION
cache["id_rules_signature"] = current_signature
def save_cache(cache: dict) -> None:
+88 -1
View File
@@ -404,6 +404,11 @@ def main():
help="Relative path of the file to rename (within --remote).")
ap.add_argument("--new-path", metavar="PATH",
help="New relative path after rename (within --remote).")
ap.add_argument("--reextract", action="store_true",
help="Walk cache.json and recompute jav_id on every file entry "
"using the current ID extraction rules. No rclone calls — "
"fast path for picking up rule changes without re-scanning. "
"Outputs JSON when --format json, plain otherwise.")
ap.add_argument("--basic", action="store_true",
help="Plain text output, no rich tables/panels/progress bars. "
"Useful for piping or low-bandwidth terminals.")
@@ -480,6 +485,79 @@ def main():
args.catalog = list(DEFAULT_CATALOG)
# --library-issues: read-only cache scan for non-canonical filenames.
# --reextract: rebuild jav_id values from current rules without re-scanning.
if args.reextract:
from rcjav.ids import current_rules_signature
from rcjav.cache import stamp_current_rules
sig = current_rules_signature()
cache = load_cache(sig)
changed = 0
unchanged = 0
dropped = 0
per_remote = []
for remote, entry in (cache.get("remotes") or {}).items():
r_changed = 0
r_unchanged = 0
r_dropped = 0
files = entry.get("files") or []
for f in files:
old_id = f.get("jav_id") or ""
new_id = extract_id(Path(f.get("path", "")).name)
if new_id is None:
if old_id:
f["jav_id"] = ""
r_dropped += 1
continue
if new_id != old_id:
f["jav_id"] = new_id
r_changed += 1
else:
r_unchanged += 1
changed += r_changed
unchanged += r_unchanged
dropped += r_dropped
per_remote.append({
"remote": remote,
"changed": r_changed,
"unchanged": r_unchanged,
"dropped": r_dropped,
"files": len(files),
})
stamp_current_rules(cache, sig)
save_cache(cache)
summary = {
"ok": True,
"changed": changed,
"unchanged": unchanged,
"dropped": dropped,
"total": changed + unchanged + dropped,
"id_rules_signature": sig,
"remotes": per_remote,
}
if args.format == "json" or BASIC:
print(json.dumps(summary))
else:
console.print(Panel(
f"[bold]Re-extracted IDs against current rules[/]\n"
f" changed: [yellow]{changed:,}[/]\n"
f" unchanged: [dim]{unchanged:,}[/]\n"
f" dropped: [red]{dropped:,}[/]\n"
f" total: {summary['total']:,}",
title="Re-extract", border_style="green"))
if per_remote:
from rich.table import Table as _Tbl
t = _Tbl(title="Per-remote", show_lines=False)
t.add_column("Remote", style="cyan")
t.add_column("Changed", justify="right", style="yellow")
t.add_column("Unchanged", justify="right", style="dim")
t.add_column("Dropped", justify="right", style="red")
t.add_column("Files", justify="right")
for r in per_remote:
t.add_row(r["remote"], f"{r['changed']:,}", f"{r['unchanged']:,}",
f"{r['dropped']:,}", f"{r['files']:,}")
console.print(t)
sys.exit(0)
if args.library_issues:
cache = load_cache()
issues = find_library_issues(cache)
@@ -546,7 +624,10 @@ def main():
console.print(f"[red]invalid --scan-since value: {args.scan_since!r} "
f"(expected e.g. 24h, 7d, 30m, 90s)[/]")
sys.exit(2)
cache = load_cache()
from rcjav.ids import current_rules_signature
from rcjav.cache import stamp_current_rules
_scan_sig = current_rules_signature()
cache = load_cache(_scan_sig)
cache_meta: dict[str, dict] = {}
skipped: list[tuple[str, str]] = []
t0 = time.perf_counter()
@@ -566,6 +647,12 @@ def main():
use_cache=not args.no_cache, force_update=True,
cache_meta=cache_meta, scan_since=scan_since))
if not args.no_cache:
# Stamp current rules only on a FULL scan. An incremental
# (--scan-since) only re-walked some files; older files in the
# cache may still have jav_ids from the previous rule set, so the
# cache remains "stale by rules" until a full scan or --reextract.
if not scan_since:
stamp_current_rules(cache, _scan_sig)
save_cache(cache)
elapsed = time.perf_counter() - t0
if BASIC:
+32
View File
@@ -231,6 +231,38 @@ def describe_id_match(display_query: str, matched_query: str, matched_id: str,
}
def current_rules_signature() -> str:
"""Sha256 over the canonical text of every rule that influences a jav_id.
Includes built-in regex sources, BUILTIN_PART_RES sources, and PART_RES
(which captures user-added part patterns applied by
`configure_part_patterns`). Output prefixed with `sha256:` so callers can
sniff the algorithm without re-deriving it.
Stable across invocations: dict is dumped with sort_keys=True. Bumping a
regex changes the digest; reordering BUILTIN_PART_RES also changes it
(order is part of the contract because part-detection short-circuits).
"""
import hashlib
import json as _json
data = {
"schema": 1, # bump when this signature schema itself changes
"primary": PRIMARY_ID_RE.pattern,
"compound": COMPOUND_ID_RE.pattern,
"fallback": FALLBACK_ID_RE.pattern,
"nohyphen": _NOHYPHEN_ID_RE.pattern,
"bracket": _BRACKET_ID_RE.pattern,
"variant": _VARIANT_SUFFIX_RE.pattern,
"xofy": _XOFY_PRIORITY_RE.pattern,
"resolution_tag": _RESOLUTION_TAG_RE.pattern,
"builtin_part_res": [r.pattern for r in BUILTIN_PART_RES],
"part_res": [r.pattern for r in PART_RES],
"fc2_handling": "fc2_to_ppv",
}
text = _json.dumps(data, sort_keys=True, ensure_ascii=False)
return "sha256:" + hashlib.sha256(text.encode("utf-8")).hexdigest()
def expand_range(raw: str) -> list[str] | None:
"""Expand a bracket range like 'IPZZ-[820-860]' into individual ID strings.
Returns None if no range marker present."""