This commit is contained in:
16
tools/benchmark-harness/fixtures/latex/basic_sections.json
Normal file
16
tools/benchmark-harness/fixtures/latex/basic_sections.json
Normal file
@@ -0,0 +1,16 @@
|
||||
{
|
||||
"document": "../../../../test_documents/latex/basic_sections.tex",
|
||||
"file_type": "latex",
|
||||
"file_size": 474,
|
||||
"expected_frameworks": ["kreuzberg"],
|
||||
"metadata": {
|
||||
"description": "LaTeX document with sections, subsections, and subsubsections including title/author/date metadata",
|
||||
"source": "handcrafted",
|
||||
"size_category": "tiny"
|
||||
},
|
||||
"ground_truth": {
|
||||
"text_file": "../../../../test_documents/ground_truth/latex/basic_sections.txt",
|
||||
"markdown_file": "../../../../test_documents/ground_truth/latex/basic_sections.md",
|
||||
"source": "pandoc"
|
||||
}
|
||||
}
|
||||
16
tools/benchmark-harness/fixtures/latex/formatting.json
Normal file
16
tools/benchmark-harness/fixtures/latex/formatting.json
Normal file
@@ -0,0 +1,16 @@
|
||||
{
|
||||
"document": "../../../../test_documents/latex/formatting.tex",
|
||||
"file_type": "latex",
|
||||
"file_size": 340,
|
||||
"expected_frameworks": ["kreuzberg"],
|
||||
"metadata": {
|
||||
"description": "LaTeX document with bold, italic, underline, emphasis, monospace, and combined formatting",
|
||||
"source": "handcrafted",
|
||||
"size_category": "tiny"
|
||||
},
|
||||
"ground_truth": {
|
||||
"text_file": "../../../../test_documents/ground_truth/latex/formatting.txt",
|
||||
"markdown_file": "../../../../test_documents/ground_truth/latex/formatting.md",
|
||||
"source": "pandoc"
|
||||
}
|
||||
}
|
||||
16
tools/benchmark-harness/fixtures/latex/latex-reader.json
Normal file
16
tools/benchmark-harness/fixtures/latex/latex-reader.json
Normal file
@@ -0,0 +1,16 @@
|
||||
{
|
||||
"document": "../../../../test_documents/latex/latex-reader.latex",
|
||||
"file_type": "latex",
|
||||
"file_size": 12572,
|
||||
"expected_frameworks": ["kreuzberg"],
|
||||
"metadata": {
|
||||
"description": "latex test: latex-reader",
|
||||
"source": "pandoc-generated",
|
||||
"size_category": "small"
|
||||
},
|
||||
"ground_truth": {
|
||||
"markdown_file": "../../../../test_documents/ground_truth/latex/latex-reader.md",
|
||||
"source": "pandoc",
|
||||
"text_file": "../../../../test_documents/ground_truth/latex/latex-reader.txt"
|
||||
}
|
||||
}
|
||||
16
tools/benchmark-harness/fixtures/latex/latex_document.json
Normal file
16
tools/benchmark-harness/fixtures/latex/latex_document.json
Normal file
@@ -0,0 +1,16 @@
|
||||
{
|
||||
"document": "../../../../test_documents/latex/latex_document.tex",
|
||||
"file_type": "latex",
|
||||
"file_size": 1475,
|
||||
"expected_frameworks": ["kreuzberg"],
|
||||
"metadata": {
|
||||
"description": "Comprehensive LaTeX document with abstract, math equations, lists, tables, and bibliography reference",
|
||||
"source": "handcrafted",
|
||||
"size_category": "small"
|
||||
},
|
||||
"ground_truth": {
|
||||
"text_file": "../../../../test_documents/ground_truth/latex/latex_document.txt",
|
||||
"markdown_file": "../../../../test_documents/ground_truth/latex/latex_document.md",
|
||||
"source": "pandoc"
|
||||
}
|
||||
}
|
||||
16
tools/benchmark-harness/fixtures/latex/lists.json
Normal file
16
tools/benchmark-harness/fixtures/latex/lists.json
Normal file
@@ -0,0 +1,16 @@
|
||||
{
|
||||
"document": "../../../../test_documents/latex/lists.tex",
|
||||
"file_type": "latex",
|
||||
"file_size": 566,
|
||||
"expected_frameworks": ["kreuzberg"],
|
||||
"metadata": {
|
||||
"description": "LaTeX document with itemize, enumerate, nested lists, and description lists",
|
||||
"source": "handcrafted",
|
||||
"size_category": "tiny"
|
||||
},
|
||||
"ground_truth": {
|
||||
"text_file": "../../../../test_documents/ground_truth/latex/lists.txt",
|
||||
"markdown_file": "../../../../test_documents/ground_truth/latex/lists.md",
|
||||
"source": "pandoc"
|
||||
}
|
||||
}
|
||||
16
tools/benchmark-harness/fixtures/latex/math.json
Normal file
16
tools/benchmark-harness/fixtures/latex/math.json
Normal file
@@ -0,0 +1,16 @@
|
||||
{
|
||||
"document": "../../../../test_documents/latex/math.tex",
|
||||
"file_type": "latex",
|
||||
"file_size": 501,
|
||||
"expected_frameworks": ["kreuzberg"],
|
||||
"metadata": {
|
||||
"description": "LaTeX document with inline math, display math, equation and align environments",
|
||||
"source": "handcrafted",
|
||||
"size_category": "tiny"
|
||||
},
|
||||
"ground_truth": {
|
||||
"text_file": "../../../../test_documents/ground_truth/latex/math.txt",
|
||||
"markdown_file": "../../../../test_documents/ground_truth/latex/math.md",
|
||||
"source": "pandoc"
|
||||
}
|
||||
}
|
||||
16
tools/benchmark-harness/fixtures/latex/minimal.json
Normal file
16
tools/benchmark-harness/fixtures/latex/minimal.json
Normal file
@@ -0,0 +1,16 @@
|
||||
{
|
||||
"document": "../../../../test_documents/latex/minimal.tex",
|
||||
"file_type": "latex",
|
||||
"file_size": 80,
|
||||
"expected_frameworks": ["kreuzberg"],
|
||||
"metadata": {
|
||||
"description": "Minimal LaTeX document with single line of text - smallest possible valid LaTeX",
|
||||
"source": "handcrafted",
|
||||
"size_category": "tiny"
|
||||
},
|
||||
"ground_truth": {
|
||||
"text_file": "../../../../test_documents/ground_truth/latex/minimal.txt",
|
||||
"markdown_file": "../../../../test_documents/ground_truth/latex/minimal.md",
|
||||
"source": "pandoc"
|
||||
}
|
||||
}
|
||||
16
tools/benchmark-harness/fixtures/latex/tables.json
Normal file
16
tools/benchmark-harness/fixtures/latex/tables.json
Normal file
@@ -0,0 +1,16 @@
|
||||
{
|
||||
"document": "../../../../test_documents/latex/tables.tex",
|
||||
"file_type": "latex",
|
||||
"file_size": 471,
|
||||
"expected_frameworks": ["kreuzberg"],
|
||||
"metadata": {
|
||||
"description": "LaTeX document with simple tabular and table environment with caption",
|
||||
"source": "handcrafted",
|
||||
"size_category": "tiny"
|
||||
},
|
||||
"ground_truth": {
|
||||
"text_file": "../../../../test_documents/ground_truth/latex/tables.txt",
|
||||
"markdown_file": "../../../../test_documents/ground_truth/latex/tables.md",
|
||||
"source": "pandoc"
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user