Squid’s Cache Is Not a Permanent PDF Archive
Squid 7's disk cache stores reusable objects under cache and eviction rules; enabling it cannot guarantee a complete durable PDF collection.
For a retained PDF collection, record and validate completed downloads through an explicit application workflow, with their URLs and required metadata. Decide how redirects and byte ranges contribute to each final document.
Squid's cache serves another goal: reusing eligible HTTP responses. For its Squid 7 cache_dir mechanism, a writable directory is only a prerequisite; object-size limits, HTTP policy and replacement still decide storage and eviction.
The backend may not expose ordinary files. Rock, for example, stores objects in database slots, so searching for .pdf filenames is not a general export interface.
Inside an ordinary HTTPS CONNECT tunnel, Squid cannot see the inner PDF for object caching at all. Neither cache setup nor tunnel relaying supplies the document-retention guarantee.