15 lines
401 B
JSON
15 lines
401 B
JSON
|
{
|
||
|
|
"document": "../../../test_documents/archives/book_war_and_peace_1p.txt.gz",
|
||
|
|
"file_type": "gz",
|
||
|
|
"file_size": 1630,
|
||
|
|
"expected_frameworks": ["kreuzberg", "tika"],
|
||
|
|
"metadata": {
|
||
|
|
"description": "Gzip compressed text file",
|
||
|
|
"category": "archive"
|
||
|
|
},
|
||
|
|
"ground_truth": {
|
||
|
|
"text_file": "../../../test_documents/ground_truth/gz/book_war_and_peace_1p.txt",
|
||
|
|
"source": "manual"
|
||
|
|
}
|
||
|
|
}
|