Endnote Citation Extractor

Extract endnotes, hyperlinks, page references, and context sentences from Microsoft Word documents for fact-checking and citation analysis.

Features

• Extracts endnote numbers and full citation text from Word .docx files
• Captures context sentences - the sentence immediately before each citation reference
• Extracts hyperlinks using 4 different methods to handle various Word formatting:
  • <w:hyperlink> tags with relationship resolution
  • Simple field codes (HYPERLINK instructions)
  • Complex field codes (<w:instrText>)
  • Plain text URLs pasted directly into endnotes
• Parses page numbers from citations (supports "p." and "pp." formats)
• Exports to UTF-8 CSV for analysis in Excel, R, or other tools

Installation

Required R Packages

# Install required packages
install.packages(c(
  "xml2",      # XML parsing
  "dplyr",     # Data manipulation
  "tidyr",     # Data tidying
  "purrr",     # Functional programming
  "stringr",   # String operations
  "readr",     # CSV export
  "here",      # Path management
  "tools"      # File utilities
))

Download the Code

Clone or download this repository:

git clone https://github.com/marisam99/endnote_extractor_MM.git

Usage

Running the Extractor

1. Open extractor_main.R in RStudio or your R environment
2. Source the entire script (Ctrl/Cmd + Shift + S in RStudio, or run all lines)
3. A file chooser dialog will appear - select your .docx file
4. The script will process the document and automatically save the results

Output Location

The CSV file will be saved in the same directory as your input file with the prefix extracted_:

Example:

• Input: /Documents/Projects/report.docx
• Output: /Documents/Projects/extracted_report.csv

Output Format

The exported CSV contains 5 columns:

Column	Description	Example
number	Endnote number	1, 2, 3
context_sentence	Sentence immediately before the citation	"This finding was later confirmed."
source_link	Extracted URL/hyperlink from the endnote	https://example.com/article
page_ref	Page numbers referenced	p.12, pp.45-50, p.12, pp.45-50
full_endnote	Complete endnote text	"Smith, J. (2020). Title. Journal, 10(2), 45-50."

Note: If an endnote contains multiple citations with different URLs, each will appear on a separate row with the same endnote number.

How It Works

1. Unzips the .docx file - Word documents are ZIP archives containing XML files
2. Parses XML structure:
   • word/document.xml - Main document content and endnote references
   • word/endnotes.xml - Endnote definitions and text
   • word/_rels/endnotes.xml.rels - Relationship mappings for hyperlinks
3. Extracts context - Finds the sentence immediately before each <w:endnoteReference> tag
4. Extracts hyperlinks - Uses 4 different extraction methods to capture URLs regardless of how they're formatted
5. Parses page numbers - Intelligently splits multi-citation endnotes and extracts page references using regex
6. Exports to CSV - Combines all data into a structured table with UTF-8 encoding

Requirements

• R version: 4.0 or higher recommended
• Input format: Microsoft Word .docx files only (not .doc)
• Page number format: Must use "p." or "pp." prefix for extraction to work
  • ✅ Works: p. 12, pp. 45-50, p. S12, pp. iv-x
  • ❌ Won't extract: page 12, 12-15, random numbers without prefix

Known Limitations

• Endnotes only - Does not process footnotes (different XML structure)
• Page number detection - Requires "p." or "pp." prefix; won't extract page numbers without this delimiter
• Multi-citation endnotes - Page numbers are matched to URLs within the same endnote; matching may be imperfect for complex citations
• Context extraction - Captures the last sentence before the citation marker; may not work perfectly with complex paragraph structures

Project Structure

endnote_extractor_MM/
├── extractor_main.R       # Main script (current version with all features)
├── helpers.R              # Helper functions for hyperlinks and page parsing
├── extractor_hyperlinks.R # Standalone version (older, no modular helpers)
├── extractor_original.R   # Biko's original version (basic functionality)
└── README.md              # This file

Version History

• Current (extractor_main.R): Full-featured with hyperlink extraction, page parsing, and modular architecture
• extractor_hyperlinks.R: Added hyperlink extraction (all functions inline)
• extractor_original.R: Original version by Biko - basic endnote and context extraction

Acknowledgments

• Original script by Biko
• Enhanced and refactored by Marisa Mission