31 lines
821 B
JSON
31 lines
821 B
JSON
|
{
|
||
|
|
"document": "../../../test_documents/pdf/copy_protected.pdf",
|
||
|
|
"file_type": "pdf",
|
||
|
|
"file_size": 270117,
|
||
|
|
"expected_frameworks": [
|
||
|
|
"kreuzberg",
|
||
|
|
"docling",
|
||
|
|
"markitdown",
|
||
|
|
"mineru",
|
||
|
|
"pdfminer",
|
||
|
|
"pdfplumber",
|
||
|
|
"pdftotext",
|
||
|
|
"pymupdf4llm",
|
||
|
|
"pypdf",
|
||
|
|
"tika",
|
||
|
|
"unstructured"
|
||
|
|
],
|
||
|
|
"metadata": {
|
||
|
|
"description": "Copy-protected PDF document - 263KB document with copy protection flags to test extraction from restricted documents",
|
||
|
|
"category": "pdf-protected",
|
||
|
|
"size_class": "medium",
|
||
|
|
"has_protection": true,
|
||
|
|
"special_handling": "copy_protected"
|
||
|
|
},
|
||
|
|
"ground_truth": {
|
||
|
|
"text_file": "../../../test_documents/ground_truth/pdf/pdf_protected.txt",
|
||
|
|
"source": "vision",
|
||
|
|
"markdown_file": "../../../test_documents/ground_truth/pdf/pdf_protected.md"
|
||
|
|
}
|
||
|
|
}
|