aeat.adapters.inbound.justificante._parsers package

Private text-backend dispatch for the inbound justificante adapter.

Callers outside adapters.inbound.justificante must never import from here. Backends expose a common extract_text entry point that returns the raw concatenated text of a justificante PDF; all field-level extraction happens in adapters.inbound.justificante._extract.

The dispatch is keyed on domain.justificante.JustificanteParserBackend, and parse failures are normalised as domain.justificante.JustificanteParseError. The path entry point caches concatenated text by source digest, backend, size, and mtime; the bytes entry point stays uncached so secure-storage callers do not need to materialise or persist plaintext files.

extract_text(pdf_path, backend)[source]

Extract concatenated text from pdf_path using backend.

The path result is cached for stable file revisions so repeated corpus parses do not reopen the same PDF.

Parameters:

  • pdf_path (Path) – Absolute path to the PDF to read.

  • backend (JustificanteParserBackend) – Which backend to dispatch to.

Return type:

str

Returns:

The concatenated text of every page in the PDF, joined by newlines.

Raises:

JustificanteParseError – If the path cannot be read or the selected backend is not implemented.

extract_text_from_bytes(pdf_bytes, backend)[source]

Extract concatenated text from in-memory PDF bytes using backend.

Bytes are sent directly to the selected backend and are not cached, which keeps secure-storage and live-capture parsing free of plaintext filesystem artefacts.

Return type:

str

Parameters:

Submodules