Dataset Viewer

Duplicate

Split (1)

test

The dataset viewer is not available for this split.

Cannot extract the features (columns) for the split 'test' of the config 'default' of the dataset.

Error code:   FeaturesError
Exception:    ArrowInvalid
Message:      Schema at index 1 was different: 
schema_name: string
version: string
name: string
origin: struct<mimetype: string, binary_hash: double, filename: string>
furniture: struct<self_ref: string, children: list<item: null>, content_layer: string, name: string, label: string>
body: struct<self_ref: string, children: list<item: struct<$ref: string>>, content_layer: string, name: string, label: string>
groups: list<item: struct<self_ref: string, parent: struct<$ref: string>, children: list<item: struct<$ref: string>>, content_layer: string, name: string, label: string>>
texts: list<item: struct<self_ref: string, parent: struct<$ref: string>, children: list<item: null>, content_layer: string, label: string, prov: list<item: struct<page_no: int64, bbox: struct<l: double, t: double, r: double, b: double, coord_origin: string>, charspan: list<item: int64>>>, orig: string, text: string, level: int64, enumerated: bool, marker: string>>
pictures: list<item: struct<self_ref: string, parent: struct<$ref: string>, children: list<item: struct<$ref: string>>, content_layer: string, label: string, prov: list<item: struct<page_no: int64, bbox: struct<l: double, t: double, r: double, b: double, coord_origin: string>, charspan: list<item: int64>>>, captions: list<item: null>, references: list<item: null>, footnotes: list<item: null>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, annotations: list<item: null>>>
tables: list<item: struct<self_ref: string, parent: struct<$ref: string>, children: list<item: null>, content_layer: string, label: string, prov: list<item: struct<page_no: int64, bbox: struct<l: double, t: double, r: double, b: double, coord_origin: string>, charspan: list<item: int64>>>, captions: list<item: null>, references: list<item: null>, footnotes: list<item: null>, data: struct<table_cells: list<item: struct<bbox: struct<l: double, t: double, r: double, b: double, coord_origin: string>, row_span: int64, col_span: int64, start_row_offset_idx: int64, end_row_offset_idx: int64, start_col_offset_idx: int64, end_col_offset_idx: int64, text: string, column_header: bool, row_header: bool, row_section: bool>>, num_rows: int64, num_cols: int64, grid: list<item: list<item: struct<bbox: struct<l: double, t: double, r: double, b: double, coord_origin: string>, row_span: int64, col_span: int64, start_row_offset_idx: int64, end_row_offset_idx: int64, start_col_offset_idx: int64, end_col_offset_idx: int64, text: string, column_header: bool, row_header: bool, row_section: bool>>>>>>
key_value_items: list<item: null>
form_items: list<item: null>
pages: struct<1: struct<size: struct<width: double, height: double>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, page_no: int64>, 2: struct<size: struct<width: double, height: double>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, page_no: int64>, 3: struct<size: struct<width: double, height: double>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, page_no: int64>, 4: struct<size: struct<width: double, height: double>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, page_no: int64>, 5: struct<size: struct<width: double, height: double>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, page_no: int64>, 6: struct<size: struct<width: double, height: double>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, page_no: int64>, 7: struct<size: struct<width: double, height: double>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, page_no: int64>>
vs
id: string
title: string
content: string
contents: string
Traceback:    Traceback (most recent call last):
                File "/src/services/worker/src/worker/job_runners/split/first_rows.py", line 243, in compute_first_rows_from_streaming_response
                  iterable_dataset = iterable_dataset._resolve_features()
                                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 3608, in _resolve_features
                  features = _infer_features_from_batch(self.with_format(None)._head())
                                                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 2368, in _head
                  return next(iter(self.iter(batch_size=n)))
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 2573, in iter
                  for key, example in iterator:
                                      ^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 2060, in __iter__
                  for key, pa_table in self._iter_arrow():
                                       ^^^^^^^^^^^^^^^^^^
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 2082, in _iter_arrow
                  yield from self.ex_iterable._iter_arrow()
                File "/usr/local/lib/python3.12/site-packages/datasets/iterable_dataset.py", line 588, in _iter_arrow
                  yield new_key, pa.Table.from_batches(chunks_buffer)
                                 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                File "pyarrow/table.pxi", line 5039, in pyarrow.lib.Table.from_batches
                File "pyarrow/error.pxi", line 155, in pyarrow.lib.pyarrow_internal_check_status
                File "pyarrow/error.pxi", line 92, in pyarrow.lib.check_status
              pyarrow.lib.ArrowInvalid: Schema at index 1 was different: 
              schema_name: string
              version: string
              name: string
              origin: struct<mimetype: string, binary_hash: double, filename: string>
              furniture: struct<self_ref: string, children: list<item: null>, content_layer: string, name: string, label: string>
              body: struct<self_ref: string, children: list<item: struct<$ref: string>>, content_layer: string, name: string, label: string>
              groups: list<item: struct<self_ref: string, parent: struct<$ref: string>, children: list<item: struct<$ref: string>>, content_layer: string, name: string, label: string>>
              texts: list<item: struct<self_ref: string, parent: struct<$ref: string>, children: list<item: null>, content_layer: string, label: string, prov: list<item: struct<page_no: int64, bbox: struct<l: double, t: double, r: double, b: double, coord_origin: string>, charspan: list<item: int64>>>, orig: string, text: string, level: int64, enumerated: bool, marker: string>>
              pictures: list<item: struct<self_ref: string, parent: struct<$ref: string>, children: list<item: struct<$ref: string>>, content_layer: string, label: string, prov: list<item: struct<page_no: int64, bbox: struct<l: double, t: double, r: double, b: double, coord_origin: string>, charspan: list<item: int64>>>, captions: list<item: null>, references: list<item: null>, footnotes: list<item: null>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, annotations: list<item: null>>>
              tables: list<item: struct<self_ref: string, parent: struct<$ref: string>, children: list<item: null>, content_layer: string, label: string, prov: list<item: struct<page_no: int64, bbox: struct<l: double, t: double, r: double, b: double, coord_origin: string>, charspan: list<item: int64>>>, captions: list<item: null>, references: list<item: null>, footnotes: list<item: null>, data: struct<table_cells: list<item: struct<bbox: struct<l: double, t: double, r: double, b: double, coord_origin: string>, row_span: int64, col_span: int64, start_row_offset_idx: int64, end_row_offset_idx: int64, start_col_offset_idx: int64, end_col_offset_idx: int64, text: string, column_header: bool, row_header: bool, row_section: bool>>, num_rows: int64, num_cols: int64, grid: list<item: list<item: struct<bbox: struct<l: double, t: double, r: double, b: double, coord_origin: string>, row_span: int64, col_span: int64, start_row_offset_idx: int64, end_row_offset_idx: int64, start_col_offset_idx: int64, end_col_offset_idx: int64, text: string, column_header: bool, row_header: bool, row_section: bool>>>>>>
              key_value_items: list<item: null>
              form_items: list<item: null>
              pages: struct<1: struct<size: struct<width: double, height: double>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, page_no: int64>, 2: struct<size: struct<width: double, height: double>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, page_no: int64>, 3: struct<size: struct<width: double, height: double>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, page_no: int64>, 4: struct<size: struct<width: double, height: double>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, page_no: int64>, 5: struct<size: struct<width: double, height: double>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, page_no: int64>, 6: struct<size: struct<width: double, height: double>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, page_no: int64>, 7: struct<size: struct<width: double, height: double>, image: struct<mimetype: string, dpi: int64, size: struct<width: double, height: double>, uri: string>, page_no: int64>>
              vs
              id: string
              title: string
              content: string
              contents: string

Need help to make the dataset viewer work? Make sure to review how to configure the dataset viewer, and open a discussion for direct support.

YAML Metadata Warning: empty or missing yaml metadata in repo card (https://huggingface.co/docs/hub/datasets-cards)

Indexing Fixtures Dataset

This repository serves as a comprehensive test dataset for validating indexing functionality across diverse file types and edge cases. It ensures that our indexing pipeline can handle real-world document scenarios and maintain consistent performance across all supported formats.

📁 Dataset Structure

The dataset is organized into directories based on file types, specific issues, testing purposes, and problem topics:

File Type Categories

pdf/ - PDF documents (208 files) - Primary document format testing
word/ - Microsoft Word documents (.doc, .docx) - Word processing files
excel/ - Excel spreadsheets (.xls, .xlsx) - Structured data files
powerpoint/ - PowerPoint presentations (.pptx) - Presentation files
csv/ - Comma-separated values files - Tabular data
tsv/ - Tab-separated values files - Alternative tabular format
images/ - Image files (.jpg, .png, .webp) - Visual content
htms/ - HTML and XML files - Web content formats
zipped/ - Compressed archives (.zip, .7z) - Archive handling

Issue-Based Categories

taishin-problems/ - Specific client issues and edge cases
- Files with parsing errors (#209, #215)
- Text duplication problems
- Chunk explosion scenarios
- Unprocessable entity errors (422)
no-chunks/ - Image-based PDFs that produce no chunks with docling parser
- Government forms and official documents (scanned/image format)
- Passport application materials (require VLM processing)
- Solution: Convert PDF pages to PNG and extract information using Vision Language Models
sensitive/ - Privacy-sensitive test files
- Identity documents
- Personal information samples

Purpose-Based Categories

QA/ - Quality assurance test files (74 files)
- Mixed format validation
- Corporate reports and multimedia
paper/ - Academic papers and research documents
- Scientific publication formats
- Citation and reference handling
passport/ - Government document processing
- Official forms and procedures
- Multi-language content
donghua/ - University-specific documents
- Academic administrative files
- Course regulations and procedures

Special Categories

excluded/ - Files intentionally excluded from processing
- Alternative formats (mht/)
- JSON test data
benchmark_results/ - Performance testing data
- CSV files with benchmark metrics
- Model evaluation records

🎯 Naming Convention

Directory and file names follow a systematic approach based on:

Issue Tracking: Files prefixed with issue numbers (e.g., #209_ERROR_, #215_PDF解析的文字不斷重複_)
Topic Classification: Grouped by subject matter (passport, donghua, taishin)
File Type: Organized by format for systematic testing
Purpose: Categorized by intended use case (QA, benchmarks, problems)

🔧 Technical Details

Git LFS Configuration

Large files are managed through Git LFS, including:

All binary formats (PDF, Office documents, images, audio, video)
Compressed archives
Model files and data exports

File Coverage

Total Files: 400+ test files
Format Coverage: 20+ file types
Size Range: From small forms to large multimedia files
Language Support: Multi-language content (Chinese, English)

🚀 Usage

This dataset is designed for:

Continuous Integration: Automated testing of indexing pipelines
Regression Testing: Ensuring fixes don't break existing functionality
Performance Benchmarking: Measuring processing speed and accuracy
Edge Case Validation: Testing problematic file scenarios
Format Compatibility: Verifying support across all file types

📊 Quality Assurance

Each directory serves specific testing purposes:

Positive Tests: Standard files that should process successfully
Negative Tests: Files that should fail gracefully (excluded/)
Edge Cases: Problematic files for robustness testing (taishin-problems/)
VLM Processing Tests: Image-based PDFs requiring vision model extraction (no-chunks/)
Performance Tests: Large files for scalability validation

🔍 Problem Categories

The dataset includes known challenging scenarios:

Text duplication in PDF parsing
Chunk explosion (excessive segmentation)
Encoding issues with special characters
Unprocessable entity errors
Large file handling
Multi-format archives

📝 Contributing

When adding new test files:

Place in appropriate directory based on type/purpose
Use descriptive naming with issue references if applicable
Update this README if adding new categories
Ensure proper Git LFS tracking for binary files

🏷️ Tags

dataset indexing testing qa document-processing nlp huggingface

Downloads last month: 16