53 lines
1.1 KiB
JSON
53 lines
1.1 KiB
JSON
|
{
|
||
|
|
"id": "config_chunking_prepend_heading_context",
|
||
|
|
"description": "Tests markdown chunker prepends heading hierarchy to chunk content",
|
||
|
|
"tags": [
|
||
|
|
"contract",
|
||
|
|
"config",
|
||
|
|
"chunking",
|
||
|
|
"heading-context"
|
||
|
|
],
|
||
|
|
"call": "extract_file_sync",
|
||
|
|
"input": {
|
||
|
|
"path": "markdown/extraction_test.md",
|
||
|
|
"config": {
|
||
|
|
"chunking": {
|
||
|
|
"chunker_type": "markdown",
|
||
|
|
"max_chars": 300,
|
||
|
|
"max_overlap": 50,
|
||
|
|
"prepend_heading_context": true
|
||
|
|
}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"assertions": [
|
||
|
|
{
|
||
|
|
"type": "min_length",
|
||
|
|
"field": "content",
|
||
|
|
"value": 10
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"type": "count_min",
|
||
|
|
"field": "chunks",
|
||
|
|
"value": 2
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"type": "is_true",
|
||
|
|
"field": "chunks_have_content"
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"type": "is_true",
|
||
|
|
"field": "chunks_have_heading_context"
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"type": "is_true",
|
||
|
|
"field": "first_chunk_starts_with_heading"
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"skip": {
|
||
|
|
"languages": [
|
||
|
|
"wasm"
|
||
|
|
],
|
||
|
|
"reason": "WASM cannot access filesystem; use extractBytes with file content instead"
|
||
|
|
}
|
||
|
|
}
|