Doc2X Docs

English

简体中文

English

简体中文

Theme

Parsing Features Introduction ​

Basic Functions ​

  • Parse text/tables/formulas/layout from PDFs and restore them to Markdown, LaTeX, and Word formats (Word does not include layout restoration)
  • Use cases: Provide higher quality data for large language model training and RAG
  • Core scenarios: Including but not limited to Chinese/English papers, financial reports/annual reports, middle school science test papers, various books, etc.

Features ​

Remove Headers and Footers from PDFs ​

  • Such as page numbers, journal names, authors that repeatedly appear at the top/bottom of paper pages

Universal Table Recognition ​

  • Recognizes tables in HTML format (markdown tables don't support merged cell syntax)
  • No specific table type limitations, performs well in general scenarios
  • Supports recognition of rotated tables on pages (both left and right rotated tables)
  • Supports recognition of formulas/images/paragraphs within tables
  • Supports merging of cross-page tables, removing continuation table text, merging cross-page cells and removing duplicate headers
  • Does not support recognition of nested tables

Formula Recognition ​

  • Supports mixed text and formula recognition as well as Chinese formula recognition
  • Supports most formulas except extremely large equation systems and matrices

Layout Restoration ​

  • Restores complex layout documents to single-column text flow
  • Supports most layouts except newspaper-style multi-column layouts
  • Currently supporting multi-level headings (h1-h5)
  • Partial support for code block indentation

Supported Languages ​

  • Supported languages: Chinese (Simplified/Traditional), English, Western European languages, Japanese
  • Future support planned: Russian, Hindi, Arabic

Handwriting Recognition ​

  • Handwritten text/formula recognition is continuously supported

Parsing Tutorial ​

Step 1: Upload Document ​

  • Click the "Start Parsing File" button or directly drag and drop PDF files to the upload area
  • Supports single file upload, maximum 300MB PDF documents supported

parse_step1

Step 2: Start Parsing ​

  • Page Range: Select the page range to parse (All/Specific pages/Specific range)
  • Click the "Confirm Processing" button, system begins processing the document
  • Processing progress is displayed in real-time
  • After parsing is complete, you can preview results and download files

parse_step2

Step 3: Preview Parsing Results ​

  • Parsing Results: View document elements identified by the system, such as titles, paragraphs, tables, images, etc.
  • Operation Menu:
    • Copy parsing results as Markdown
    • Export as Markdown, Word, and other formats
    • Single-column/double-column toggle

parse_step3

Step 4: Download Parsing Results ​

  • Click the "Export" icon, select the file format to save (Markdown, Word, etc.)
  • You can then save the parsing results locally

parse_step4