content-hash-cache-pattern
Scannednpx machina-cli add skill shimo4228/claude-code-learned-skills/content-hash-cache-pattern --openclawContent-Hash File Cache Pattern
コンテンツハッシュキャッシュパターン
Extracted / 抽出日: 2026-02-10 Context / コンテキスト: ファイル処理結果をSHA-256ハッシュでキャッシュし、サービス層でラップするパターン
Problem / 課題
ファイル処理(PDF解析、テキスト抽出等)は時間がかかるが、同じファイルの再処理は無駄:
# WRONG: 毎回フルパイプライン実行
def process_file(path: Path) -> Result:
return expensive_extraction(path) # Always re-runs
# WRONG: パスベースキャッシュ(ファイル移動で無効化)
cache = {"/path/to/file.pdf": result} # Path changes → cache miss
# WRONG: 既存関数にキャッシュパラメータ追加(SRP違反)
def extract_text(path, *, cache_enabled=False, cache_dir=None):
if cache_enabled: # Extraction function now has cache responsibility
...
Solution / 解決策
1. Content-Hash Based Cache Key
ファイルパスではなくファイル内容のSHA-256ハッシュをキーに使う:
import hashlib
from pathlib import Path
_HASH_CHUNK_SIZE = 65536 # 64KB chunks for large files
def compute_file_hash(path: Path) -> str:
"""SHA-256 of file contents (chunked for large files)."""
if not path.is_file():
raise FileNotFoundError(f"File not found: {path}")
sha256 = hashlib.sha256()
with open(path, "rb") as f:
while True:
chunk = f.read(_HASH_CHUNK_SIZE)
if not chunk:
break
sha256.update(chunk)
return sha256.hexdigest()
利点: ファイル移動・リネームでもキャッシュヒット、内容変更で自動無効化
2. Frozen Dataclass for Cache Entry
from dataclasses import dataclass
@dataclass(frozen=True, slots=True)
class CacheEntry:
file_hash: str
source_path: str
document: ExtractedDocument # The cached result
3. JSON Serialization of Frozen Dataclasses
dataclasses.asdict() はネストしたfrozen dataclassで問題が起きるため、手動マッピング:
import json
from typing import Any
def _serialize_entry(entry: CacheEntry) -> dict[str, Any]:
"""Manual mapping for full control over serialized format."""
doc = entry.document
return {
"file_hash": entry.file_hash,
"source_path": entry.source_path,
"document": {
"text": doc.text,
"chunks": list(doc.chunks), # tuple → list for JSON
"file_type": doc.file_type,
# ... other fields
},
}
def _deserialize_entry(data: dict[str, Any]) -> CacheEntry:
doc_data = data["document"]
document = ExtractedDocument(
text=doc_data["text"],
chunks=tuple(doc_data["chunks"]), # list → tuple
file_type=doc_data["file_type"],
)
return CacheEntry(
file_hash=data["file_hash"],
source_path=data["source_path"],
document=document,
)
4. Service Layer Wrapper (SRP)
純粋な処理関数を変更せず、サービス層でキャッシュロジックをラップ:
# service.py — cache wrapper
def extract_with_cache(file_path: Path, *, config: AppConfig) -> ExtractedDocument:
"""Service layer: cache check → extraction → cache write."""
if not config.cache_enabled:
return extract_text(file_path) # Pure function, no cache knowledge
cache_dir = Path(config.cache_dir)
file_hash = compute_file_hash(file_path)
# Check cache
cached = read_cache(cache_dir, file_hash)
if cached is not None:
logger.info("Cache hit: %s (hash=%s)", file_path.name, file_hash[:12])
return cached.document
# Cache miss → extract → store
logger.info("Cache miss: %s (hash=%s)", file_path.name, file_hash[:12])
doc = extract_text(file_path)
entry = CacheEntry(file_hash=file_hash, source_path=str(file_path), document=doc)
write_cache(cache_dir, entry)
return doc
5. Graceful Corruption Handling
def read_cache(cache_dir: Path, file_hash: str) -> CacheEntry | None:
cache_file = cache_dir / f"{file_hash}.json"
if not cache_file.is_file():
return None
try:
raw = cache_file.read_text(encoding="utf-8")
data = json.loads(raw)
return _deserialize_entry(data)
except (json.JSONDecodeError, ValueError, KeyError):
logger.warning("Corrupted cache entry: %s", cache_file)
return None # Treat corruption as cache miss
Key Design Choices / 設計上のポイント
| Choice / 選択 | Reason / 理由 |
|---|---|
| SHA-256 content hash | Path-independent, auto-invalidates on content change |
{hash}.json file naming | O(1) lookup, no index file needed |
| Service layer wrapper | SRP: extraction stays pure, cache is separate concern |
| Manual JSON serialization | Full control over frozen dataclass serialization |
| Corruption → None | Graceful degradation, re-extracts on next run |
cache_dir.mkdir(parents=True) | Lazy directory creation on first write |
When to Use / 使用すべき場面
- ファイル処理パイプライン(PDF解析、画像処理、テキスト抽出)
- 処理コストが高く、同一ファイルの再処理が頻繁な場合
- CLI ツールで
--cache/--no-cacheオプションが必要な場合 - 既存の純粋関数にキャッシュを追加する場合(SRP維持)
When NOT to Use / 使用すべきでない場面
- リアルタイム更新が必要なデータ(常に最新が必要)
- キャッシュエントリが非常に大きい場合(メモリ/ディスク圧迫)
- 処理結果がファイル内容以外のパラメータに依存する場合(設定変更でキャッシュ無効化が必要)
Related Patterns / 関連パターン
python-immutable-accumulator.md— frozen dataclass + slotsパターンbackward-compatible-frozen-extension.md— frozen dataclass拡張cost-aware-llm-pipeline.md— LLMパイプラインでのキャッシュ活用
Source
git clone https://github.com/shimo4228/claude-code-learned-skills/blob/main/skills/content-hash-cache-pattern/SKILL.mdView on GitHub Overview
This pattern caches expensive file processing results by hashing file contents with SHA-256 and storing a frozen CacheEntry. A service layer wrapper keeps cache logic out of pure processing functions, enabling fast cache hits even after moves or renames.
How This Skill Works
It chunk-hashes the file contents (64KB chunks) to produce a stable SHA-256 key, checks a cache for a matching CacheEntry, and returns the cached document on a hit. If there’s a miss, it performs the extraction, builds a frozen CacheEntry (file_hash, source_path, document), and writes a serialized cache entry back to disk for future calls.
When to Use It
- Caching expensive file processing results (e.g., PDF parsing, text extraction, OCR) to avoid rework on repeated runs
- You want cache hits to survive file moves or renames since the key is file content, not path
- You prefer to keep cache logic out of the pure processing function to adhere to SRP
- You need robust cache serialization for frozen dataclasses with explicit control over format
- Graceful corruption handling is important to avoid failing reads when cache is invalid
Quick Start
- Step 1: Compute the SHA-256 hash of the file contents using chunked reading (e.g., 64KB blocks)
- Step 2: In the service layer, check your cache_dir for a JSON file named by the hash; on hit, return the cached document
- Step 3: On cache miss, run the extraction, build a CacheEntry (file_hash, source_path, document), serialize it, write to disk, and return the document
Best Practices
- Compute the SHA-256 hash using chunked reads (e.g., 64KB chunks) to handle large files efficiently
- Use a frozen dataclass for CacheEntry to ensure immutability and safe reuse as a cache record
- Serialize CacheEntry with a manual mapper instead of dataclasses.asdict() to handle nested frozen dataclasses
- Wrap the cache logic in a service layer so the extraction function remains pure and cache-agnostic
- Implement graceful corruption handling when reading the cache to avoid cascading failures
Example Use Cases
- Cache results of PDF parsing and text extraction to serve repeated analyses quickly
- Reuse extracted text from identical content even if the file path changes via moves or renames
- Cache results of OCR on large scanned PDFs to prevent repeated heavy processing
- Cache extraction of table data or metadata from large documents in data pipelines
- Maintain a content-hash based cache in CI/CD/content pipelines to accelerate repeated reports