Check a segmenter against the segmenter contract

Cuts the bundled example document (gr_example()) with one registered segmenter and reports which rules of the segmenter contract it breaks. Nothing is refused: a segmenter that fails a rule stays registered and runs as before. Use it on a segmenter of your own after gr_register_segmenter().

Usage

gr_check_segmenter(segmenter, client = NULL)

Arguments

segmenter: A registered segmenter's name; see gr_segmenters().
client: A gr_client to read with. NULL (the default) uses a gr_mock_client() whose reply to a request with a JSON schema is an object of that schema's shape, and a fixed sentence otherwise.

Value

An hr_conformance data frame; see hr_check_module().

Details

The spec is gr_segment_spec(method = segmenter, max_tokens = 150, context_source = "llm"), so contextual writes its context lines with the model. A client you pass is used whatever the segmenter declares, so calls_declared sees every call it makes with one. With none, a segmenter that needs a client, or declares calls, is given the offline one.

Rules

is_function, runs: The segmenter's fn is a function, and gr_segment() with it finishes without an error.
chunks_class: It returns a gr_chunks whose chunks is a data frame.
chunk_columns: chunks has the columns gr_chunks documents: chunk_id, text, tokens, chars, page, section and block_id.
chunk_text: Every chunk has text that is not blank.
chunk_ids: Chunk ids are unique and not NA.
within_cap: No chunk the segmenter returned was over max_tokens. gr_segment() splits an oversized chunk and records it in extra$cap_enforced, and the segmenter is reported for each one it had to split.
calls_declared: The label of every model call in the trace is covered by a stem the segmenter declared in calls, by the rule an effort map is checked with.

Examples

gr_check_segmenter("paragraph")
#> Using cached ingestion for this document + settings.
#> Segmenting with 'paragraph' (cap 150 tokens, overlap 0).
#> <hr_conformance> segmenter 'paragraph', contract 0.1
#>   passes all 8 rules
#>   ok    is_function
#>   ok    runs
#>   ok    chunks_class
#>   ok    chunk_columns
#>   ok    chunk_text
#>   ok    chunk_ids
#>   ok    within_cap
#>   ok    calls_declared: no model calls