47 lines
845 B
JSON
47 lines
845 B
JSON
|
{
|
||
|
|
"id": "config_keywords",
|
||
|
|
"description": "Tests keyword extraction via YAKE algorithm",
|
||
|
|
"tags": [
|
||
|
|
"contract",
|
||
|
|
"config",
|
||
|
|
"keywords"
|
||
|
|
],
|
||
|
|
"call": "extract_file_sync",
|
||
|
|
"input": {
|
||
|
|
"path": "pdf/fake_memo.pdf",
|
||
|
|
"config": {
|
||
|
|
"keywords": {
|
||
|
|
"algorithm": "yake",
|
||
|
|
"max_keywords": 10
|
||
|
|
}
|
||
|
|
}
|
||
|
|
},
|
||
|
|
"assertions": [
|
||
|
|
{
|
||
|
|
"type": "equals",
|
||
|
|
"field": "mime_type",
|
||
|
|
"value": "application/pdf"
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"type": "min_length",
|
||
|
|
"field": "content",
|
||
|
|
"value": 10
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"type": "not_empty",
|
||
|
|
"field": "keywords"
|
||
|
|
},
|
||
|
|
{
|
||
|
|
"type": "count_min",
|
||
|
|
"field": "keywords",
|
||
|
|
"value": 1
|
||
|
|
}
|
||
|
|
],
|
||
|
|
"skip": {
|
||
|
|
"languages": [
|
||
|
|
"wasm"
|
||
|
|
],
|
||
|
|
"reason": "WASM cannot access filesystem; use extractBytes with file content instead"
|
||
|
|
}
|
||
|
|
}
|