content-hash-cache-pattern

Path-based file caches break because they use file locations as cache keys, so moving a file invalidates the cache. Content-hash cache keys use SHA-256 hashes instead, remaining valid after moves and auto-invalidating when content changes.

You can cache batch PDF parsing without modifying pure processing functions by using a separate service layer wrapper. This wrapper handles SHA-256 content hash cache keys and file-based storage, keeping your extraction logic clean and single-responsibility.

The best way to skip redundant OCR and text extraction is implementing content-hash caching. It uses SHA-256 file hashes as O(1) lookup keys in JSON files, automatically skipping identical content processed in previous runs.

You can use content-hash caching for image analysis pipelines, PDF parsing, OCR, and text extraction. It applies to any file processing pipeline where identical files are processed repeatedly, using SHA-256 hashes for path-independent cache validation.

File-based content-hash caches handle corruption gracefully by storing entries as individual JSON files named by their content hash. This eliminates the need for a separate index file, isolating corruption to individual entries rather than breaking the entire cache.

You should avoid the content-hash cache pattern when processing files that are only read once or when the computational cost of generating SHA-256 hashes exceeds the processing time saved. It is best suited for repeated batch processing of identical files.