README.md

metadata

dataset_info:
  features:
    - name: file_name
      dtype: string
    - name: url_id
      dtype: int64
    - name: country
      dtype: string
    - name: lang
      dtype: string
    - name: pdf_bytes_base64
      dtype: binary
    - name: hosts
      struct:
        - name: country
          dtype: string
        - name: host
          dtype: string
        - name: ip_address
          dtype: string
        - name: latitude
          dtype: float64
        - name: longitude
          dtype: float64
        - name: tld
          dtype: string
    - name: provenance
      struct:
        - name: cc_detected_mime
          dtype: string
        - name: cc_digest
          dtype: string
        - name: cc_http_mime
          dtype: string
        - name: cc_truncated
          dtype: string
        - name: cc_warc_end
          dtype: int64
        - name: cc_warc_file_name
          dtype: string
        - name: cc_warc_start
          dtype: int64
        - name: fetched_digest
          dtype: string
        - name: fetched_length
          dtype: int64
        - name: fetched_status
          dtype: string
        - name: url
          dtype: string
    - name: info
      struct:
        - name: created
          dtype: string
        - name: creator
          dtype: string
        - name: custom_metadata
          dtype: string
        - name: exit_value
          dtype: int64
        - name: form
          dtype: string
        - name: javascript
          dtype: string
        - name: metadata_stream
          dtype: string
        - name: modified
          dtype: string
        - name: optimized
          dtype: string
        - name: page_rotation
          dtype: float64
        - name: page_size
          dtype: string
        - name: pages
          dtype: float64
        - name: parse_time_millis
          dtype: int64
        - name: pdf_version
          dtype: float64
        - name: producer
          dtype: string
        - name: stderr
          dtype: string
        - name: tagged
          dtype: string
        - name: timeout
          dtype: string
        - name: user_properties
          dtype: string
    - name: tika
      struct:
        - name: attachment_count
          dtype: float64
        - name: container_exception
          dtype: string
        - name: created
          dtype: string
        - name: encrypted
          dtype: string
        - name: has_collection
          dtype: string
        - name: has_marked_content
          dtype: string
        - name: has_signature
          dtype: string
        - name: has_xfa
          dtype: string
        - name: has_xmp
          dtype: string
        - name: location
          dtype: string
        - name: macro_count
          dtype: float64
        - name: mime
          dtype: string
        - name: modified
          dtype: string
        - name: num_pages
          dtype: float64
        - name: parse_status
          dtype: string
        - name: parse_time_millis
          dtype: int64
        - name: pdf_contains_damaged_font
          dtype: string
        - name: pdf_contains_non_embedded_font
          dtype: string
        - name: pdf_has_acroform_fields
          dtype: string
        - name: pdf_incremental_updates
          dtype: float64
        - name: pdf_num_3d_annotations
          dtype: float64
        - name: pdf_overall_unmapped_unicode_chars
          dtype: float64
        - name: pdf_producer
          dtype: string
        - name: pdf_version
          dtype: float64
        - name: pdfa_version
          dtype: string
        - name: pdfuaid_part
          dtype: float64
        - name: pdfvt_version
          dtype: float64
        - name: pdfx_conformance
          dtype: string
        - name: pdfx_version
          dtype: string
        - name: pdfxid_version
          dtype: string
        - name: tika_eval_lang
          dtype: string
        - name: tika_eval_num_alpha_tokens
          dtype: float64
        - name: tika_eval_num_tokens
          dtype: float64
        - name: tika_eval_oov
          dtype: float64
        - name: xmp_creator_tool
          dtype: string
    - name: tika_attachments
      list:
        - name: container_exception
          dtype: string
        - name: created
          dtype: string
        - name: emb_depth
          dtype: int64
        - name: embedded_exception
          dtype: float64
        - name: embedded_id
          dtype: float64
        - name: embedded_id_path
          dtype: string
        - name: embedded_resource_type
          dtype: string
        - name: encrypted
          dtype: string
        - name: has_collection
          dtype: string
        - name: has_marked_content
          dtype: string
        - name: has_signature
          dtype: string
        - name: has_xfa
          dtype: string
        - name: has_xmp
          dtype: string
        - name: location
          dtype: string
        - name: mime
          dtype: string
        - name: modified
          dtype: string
        - name: num_pages
          dtype: float64
        - name: parse_status
          dtype: string
        - name: parse_time_millis
          dtype: int64
        - name: pdf_contains_damaged_font
          dtype: string
        - name: pdf_contains_non_embedded_font
          dtype: string
        - name: pdf_has_acroform_fields
          dtype: string
        - name: pdf_incremental_updates
          dtype: float64
        - name: pdf_num_3d_annotations
          dtype: float64
        - name: pdf_overall_unmapped_unicode_chars
          dtype: float64
        - name: pdf_producer
          dtype: string
        - name: pdf_version
          dtype: float64
        - name: pdfa_version
          dtype: string
        - name: pdfuaid_part
          dtype: float64
        - name: pdfvt_version
          dtype: float64
        - name: pdfx_conformance
          dtype: string
        - name: pdfx_version
          dtype: string
        - name: pdfxid_version
          dtype: string
        - name: tika_eval_lang
          dtype: string
        - name: tika_eval_num_alpha_tokens
          dtype: float64
        - name: tika_eval_num_tokens
          dtype: float64
        - name: tika_eval_oov
          dtype: float64
        - name: xmp_creator_tool
          dtype: string
  splits:
    - name: train
      num_bytes: 1840299303
      num_examples: 1000
  download_size: 1756624272
  dataset_size: 1840299303
configs:
  - config_name: default
    data_files:
      - split: train
        path: data/train-*
tags:
  - PDF
size_categories:
  - 1K<n<10K

Dataset Card for SAFEDOCS 1k PDF

This dataset is a subset (first 1,000 PDFs) of the original SAFEDOCS (CC-MAIN-2021-31-PDF-UNTRUNCATED) merged with the corresponding metadata.

The PDFs in this dataset are found in the 0000.zip file available here. The metadata files are available here.

Each file corresponds to a feature in the dataset:

• hosts: cc-hosts-20230324-1k.csv
• provenance: cc-provenance-20230324-1k.csv
• info: pdfinfo-20230324-1k.csv
• tika: tika-20230714-1k.csv
• tika_attachments: tika-with-attachments-20230714-1k.csv

The PDFs were encoded in base64 and added as the pdf_bytes_base64 feature.

The keys country (from hosts) and lang (from tika, as tika_eval_lang) were replicated as features to make it easier to filter the data based on origin and language.

You can easily and quickly load it:

dataset = load_dataset("dvgodoy/SAFEDOCS_1k")

Dataset({
    features: ['file_name', 'url_id', 'country', 'lang', 'pdf_bytes_base64', 'hosts', 'provenance', 'info', 'tika', 'tika_attachments'],
    num_rows: 1000
})

Decoding PDFs

To handle the PDFs, you will need to decode the pdf_bytes_base64 feature and load it into a PDF object using your favorite PDF library (e.g. pdfplumber):

import io
import pdfplumber

# load the bytes into your favorite PDF library e.g., `pdfplumber`
for encoded in mini_batch['pdf_bytes_base64']:
    bytes_content = io.BytesIO(base64.b64decode(encoded))
    pdf_obj = pdfplumber.open(bytes_content)
    # process the PDFs
    # ...
    # CLOSE the objects after you've used them
    bytes_content.close()
    pdf_obj.close()

You can use any other library/package to load the PDF, just make sure it can open a PDF from bytes.

Table of Contents

• Dataset Description
  • Dataset Summary
• Dataset Structure
  • Data Instances
• Dataset Creation
  • Curation Rationale
  • Source Data
• Additional Information
  • Credits
  • Licensing Information

Dataset Description

• Homepage: SAFEDOCS (CC-MAIN-2021-31-PDF-UNTRUNCATED)
• Repository:
• Paper: Making more sense of PDF structures in the wild at scale
• Leaderboard:
• Point of Contact:

Dataset Summary

The original SAFEDOCS (CC-MAIN-2021-31-PDF-UNTRUNCATED) consists of nearly 8,000,000 PDFs gathered from across the web in July/August of 2021.

This "1k" version contains only the first 1,000 PDFs from the original dataset.

Dataset Structure

Data Instances

A sample from the training set is provided below :

{
    'file_name': '0000000.pdf',
    'url_id': 6368476,
    'country': 'US',
    'lang': 'eng',
    'pdf_bytes_base64': b'JVBERi0xLjcKJe...',
    'hosts': {
        'host': 'augustaarchives.com',
        'tld': 'com',
        'ip_address': '66.228.60.224',
        'country': 'US',
        'latitude': 33.7485,
        'longitude': -84.3871
    },
    'provenance': {
        'url': 'http://augustaarchives.com/cgi-bin/content/view.php?data=ib_math_sl_practice_questions_and_answers&filetype=pdf&id=14310d60ac64f9023c7dbc458f0f7b38',
        'cc_digest': 'H66IGSGFXBJLJIKKN7LSHQSM3YTX4TDF',
        'cc_http_mime': 'application/pdf',
        'cc_detected_mime': 'application/pdf',
        'cc_warc_file_name': 'crawl-data/CC-MAIN-2021-31/segments/1627046154042.23/warc/CC-MAIN-20210731011529-20210731041529-00143.warc.gz',
        'cc_warc_start': 3724499,
        'cc_warc_end': 3742341,
        'cc_truncated': '',
        'fetched_status': 'ADDED_TO_REPOSITORY',
        'fetched_digest': '00000836a41510292765a86e57a80f1e182360ca5c6ff0841482b31c8f25ab84',
        'fetched_length': 29690
    },
    'info': {
        'parse_time_millis': 385,
        'exit_value': 0,
        'timeout': 'f',
        'stderr': '',
        'pdf_version': 1.7,
        'creator': 'Acrobat 9.0.0',
        'producer': 'Adobe InDesign CC 2015 (Macintosh)|x|Adobe PDF Library 15.0; modified using iText 2.1.7 by 1T3XT',
        'created': '2021-07-31T03:13:33Z',
        'modified': '2021-07-31T03:13:33Z',
        'custom_metadata': 'no',
        'metadata_stream': 'yes',
        'tagged': 'no',
        'user_properties': 'no',
        'form': 'none',
        'javascript': 'no',
        'pages': 12.0,
        'page_size': '297.638 x 419.528 pts',
        'page_rotation': 0.0,
        'optimized': 'no'
    },
    'tika': {
        'parse_status': 'OK',
        'parse_time_millis': 1430,
        'mime': 'application/pdf',
        'macro_count': nan,
        'attachment_count': nan,
        'created': '2021-07-31 03:13:33',
        'modified': '2021-07-31 03:13:33',
        'encrypted': 'f',
        'has_xfa': 'f',
        'has_xmp': 't',
        'has_collection': 'f',
        'has_marked_content': 'f',
        'num_pages': 12.0,
        'xmp_creator_tool': 'Acrobat 9.0.0',
        'pdf_producer': 'Microsoft Word|x|DocuCom PDF Driver 4.61 for NT',
        'pdf_version': 1.7,
        'pdfa_version': '',
        'pdfuaid_part': nan,
        'pdfx_conformance': '',
        'pdfx_version': '',
        'pdfxid_version': '',
        'pdfvt_version': nan,
        'pdf_num_3d_annotations': 0.0,
        'pdf_has_acroform_fields': '',
        'pdf_incremental_updates': 0.0,
        'pdf_overall_unmapped_unicode_chars': 0.0,
        'pdf_contains_damaged_font': 'f',
        'pdf_contains_non_embedded_font': 't',
        'has_signature': '',
        'location': '',
        'tika_eval_num_tokens': 1489.0,
        'tika_eval_num_alpha_tokens': 1344.0,
        'tika_eval_lang': 'eng',
        'tika_eval_oov': 0.2961309552192688,
        'container_exception': ''
    },
    'tika_attachments': [{
        'parse_status': 'OK',
        'parse_time_millis': 1430,
        'mime': 'application/pdf',
        'emb_depth': 0,
        'embedded_id': nan,
        'embedded_id_path': '',
        'embedded_resource_type': '',
        'created': '2021-07-31 03:13:33',
        'modified': '2021-07-31 03:13:33',
        'encrypted': 'f',
        'has_xfa': 'f',
        'has_xmp': 't',
        'has_collection': 'f',
        'has_marked_content': 'f',
        'num_pages': 12.0,
        'xmp_creator_tool': 'Acrobat 9.0.0',
        'pdf_producer': 'Microsoft Word|x|DocuCom PDF Driver 4.61 for NT',
        'pdf_version': 1.7,
        'pdfa_version': '',
        'pdfuaid_part': nan,
        'pdfx_conformance': '',
        'pdfx_version': '',
        'pdfxid_version': '',
        'pdfvt_version': nan,
        'pdf_num_3d_annotations': 0.0,
        'pdf_has_acroform_fields': '',
        'pdf_incremental_updates': 0.0,
        'pdf_overall_unmapped_unicode_chars': 0.0,
        'pdf_contains_damaged_font': 'f',
        'pdf_contains_non_embedded_font': 't',
        'has_signature': '',
        'location': '',
        'tika_eval_num_tokens': 1489.0,
        'tika_eval_num_alpha_tokens': 1344.0,
        'tika_eval_lang': 'eng',
        'tika_eval_oov': 0.2961309552192688,
        'container_exception': '',
        'embedded_exception': nan
    }]
}

Dataset Creation

Curation Rationale

PDF is a ubiquitous format and is used across many industrial and research domains. Many existing corpora focusing on extant data (such as GovDocs1) are now quite old and no longer reflect current changes and trends in both PDF itself (as a file format) or in PDF-creating and authoring applications. With advances in machine learning technology, the need for larger data sets is also in high demand. This corpus is thus helpful for:

• PDF technology and software testing, assessment, and evaluation
• Information privacy research
• Document understanding, text extraction, table identification, OCR/ICR, formula identification, document recognition and analysis, and related document engineering domains
• Malware and cyber-security research
• ML/AI applications and research (document classification, document content, text extraction, etc)
• Preservation and archival research
• Usability and accessibility research
• Software engineering research (parsing, formal methods, etc.)

Source Data

Initial Data Collection and Normalization

The PDF files were initially identified by Common Crawl as part of their July/August 2021 crawl (identified as CC-MAIN-2021-31) and subsequently updated and collated as part of the DARPA SafeDocs program.

Additional Information

Credits

The original dataset was gathered by a team at NASA’s Jet Propulsion Laboratory (JPL), California Institute of Technology while supporting the Defense Advance Research Project Agency (DARPA)’s SafeDocs Program. The JPL team included Chris Mattmann (PI), Wayne Burke, Dustin Graf, Tim Allison, Ryan Stonebraker, Mike Milano, Philip Southam and Anastasia Menshikova.

The JPL team collaborated with Peter Wyatt, the Chief Technology Officer of the PDF Association and PI on the SafeDocs program, in the design and documentation of this corpus.

The JPL team and PDF Association would like to thank Simson Garfinkel and Digital Corpora for taking ownership of this dataset and publishing it. Our thanks are extended to the Amazon Open Data Sponsorship Program for enabling this large corpus to be free and publicly available as part of Digital Corpora initiative.

Reference herein to any specific commercial product, process, or service by trade name, trademark, manufacturer, or otherwise, does not constitute or imply its endorsement by the United States Government or the Jet Propulsion Laboratory, California Institute of Technology.

The research was carried out at the NASA (National Aeronautics and Space Administration) Jet Propulsion Laboratory, California Institute of Technology under a contract with the Defense Advanced Research Projects Agency (DARPA) SafeDocs program. Government sponsorship acknowledged.

Licensing Information

• Common Crawl
• Digital Corpora