AutoRef - Academic Paper BibTeX Retrieval Tool

A powerful command-line utility for automatically retrieving BibTeX entries for academic papers using DBLP and other open APIs. Streamline your bibliography management workflow with one command.

Features

Core Capabilities

• Automatic BibTeX Retrieval: Search academic papers by title and instantly get properly formatted BibTeX entries
• Multi-Source Support: Extensible architecture supporting multiple retrieval engines (currently DBLP, easily expandable)
• Batch Processing: Process entire reading lists by providing a text file with paper titles
• Smart Filtering: Automatically filter and prioritize peer-reviewed conference/journal versions over preprints
• Intelligent Key Generation: Auto-generate Google Scholar-style citation keys (LastnameYearFirstword)

Advanced Features

• Automatic Field Cleaning: Remove redundant metadata (timestamps, URLs for OpenReview entries)
• Flexible Output Options: Print to terminal or save directly to BibTeX files
• Graceful Error Handling: Continue processing on failures with detailed error reporting
• Verbose Logging: Optional detailed execution logs for debugging and optimization
• Author-aware Processing: Proper handling of multi-author papers and various naming conventions

Installation

Prerequisites

• Python 3.6+
• requests library

Setup

# Clone the repository
git clone <repository-url>
cd autoRef

# Install dependencies
pip install requests

# Make executable (optional)
chmod +x auto_ref.py

Usage

Basic Usage - Single Paper

Retrieve BibTeX for a single paper by title:

python auto_ref.py "Attention Is All You Need"

Output to terminal (formatted BibTeX display):

----------------------------------------
@inproceedings{vaswani2017attention,
  author = {Ashish Vaswani and Noam Shazeer and ...},
  title = {Attention Is All You Need},
  year = {2017},
  booktitle = {Advances in Neural Information Processing Systems}
}
----------------------------------------

Save to File

Append BibTeX entry to a bibliography file:

python auto_ref.py "Attention Is All You Need" -o references.bib

Batch Processing - Multiple Papers

Create a text file with paper titles (one per line):

# papers.txt
Attention Is All You Need
BERT: Pre-training of Deep Bidirectional Transformers
ImageNet-21K Pretraining for Semantic Segmentation

Process all titles in batch:

python auto_ref.py -f papers.txt

This generates ref.bib with all entries and provides a summary:

Found 3 titles, starting batch retrieval...

[1/3] Retrieving: Attention Is All You Need
  Successfully retrieved

[2/3] Retrieving: BERT: Pre-training of Deep Bidirectional Transformers
  Successfully retrieved

[3/3] Retrieving: ImageNet-21K Pretraining for Semantic Segmentation
  BibTeX not found

==================================================
Retrieval completed
Success: 2/3
Failed: 1/3

Failed titles:
  - ImageNet-21K Pretraining for Semantic Segmentation (Not found)

Save Batch Results to Custom File

python auto_ref.py -f papers.txt -o my_references.bib

Include Preprint Versions

By default, AutoRef filters out preprint versions from arXiv/CoRR, prioritizing formal conference/journal publications. Allow preprints:

python auto_ref.py "Paper Title" --allow-preprint

Verbose Logging

Display detailed execution process and API interactions:

python auto_ref.py "Paper Title" -v

Output with verbose flag:

08:23:45 - [INFO] - Retrieving via DBLP engine: Paper Title
08:23:46 - [INFO] - Found target version (Venue: NeurIPS), downloading BibTeX...
08:23:47 - [INFO] - Successfully retrieved BibTeX data

Command Reference

usage: auto_ref.py [-h] [-f FILE] [-o FILE] [--allow-preprint] [-v] [title]

positional arguments:
  title                 Full title of the paper to retrieve (wrap in quotes).
                        Or use -f to specify a text file with multiple titles.

optional arguments:
  -h, --help            Show this help message and exit
  
  -f FILE, --file FILE  Text file containing paper titles, one per line.
                        Will retrieve and generate ref.bib file.
                        
  -o FILE, --output FILE
                        Save retrieved BibTeX to specified file (example: refs.bib).
                        If not specified, print to terminal.
                        
  --allow-preprint      Allow retrieving preprint versions from arXiv/CoRR
                        (default: filter preprints, require formal conference/journal versions)
                        
  -v, --verbose         Show detailed execution process and logging information

Examples

Workflow Example 1: Building a Bibliography

# Create papers list
echo "Attention Is All You Need" > my_papers.txt
echo "BERT: Pre-training of Deep Bidirectional Transformers" >> my_papers.txt
echo "GPT-3: Language Models are Few-Shot Learners" >> my_papers.txt

# Retrieve all BibTeX entries
python auto_ref.py -f my_papers.txt -o references.bib -v

# Now references.bib contains properly formatted entries ready for LaTeX/Markdown

Workflow Example 2: Quick Terminal Reference

# Quickly lookup a paper without saving
python auto_ref.py "Deep Residual Learning for Image Recognition"

# Copy the output and paste directly into your BibTeX file

Workflow Example 3: Research Import from List

# Import bibliography from research template
python auto_ref.py -f related_work.txt -o chapter2_references.bib --allow-preprint

Architecture

Class Structure

BaseBibtexEngine (Abstract Base)
    └── DblpEngine

BibtexManager
    ├── register_engine()
    ├── get_bibtex()
    ├── _clean_and_format_bibtex()
    └── _generate_google_scholar_key()

Data Flow

User Input (title) 
    ↓
BibtexManager.get_bibtex()
    ↓
DblpEngine.fetch_bibtex() [API call]
    ↓
BibtexManager._clean_and_format_bibtex()
    ↓
BibtexManager._generate_google_scholar_key()
    ↓
Formatted BibTeX Output

Extensibility

The tool is designed for easy extension with new retrieval engines:

class CustomEngine(BaseBibtexEngine):
    def fetch_bibtex(self, title: str) -> Optional[str]:
        # Your implementation
        pass

# Register in main()
manager = BibtexManager()
manager.register_engine(CustomEngine())

BibTeX Processing

AutoRef performs intelligent post-processing on retrieved BibTeX:

1. Field Normalization: Removes redundant fields (timestamp, biburl, bibsource)
2. Key Generation: Creates consistent citation keys from author + year + title
3. Author Parsing: Intelligently handles multi-author papers and naming conventions
4. Metadata Cleanup: Removes OpenReview publisher entries when appropriate
5. Whitespace Normalization: Cleans extra spaces and line breaks

Error Handling

Common Issues and Solutions

Issue	Cause	Solution
BibTeX not found	Paper not in DBLP database	Check spelling, try full author names, use --allow-preprint
DBLP matched no results	Title mismatch	Use partial title, try different keywords
File write failed	Permission denied	Check file permissions, use different output path
No retrieval engines registered	Configuration error	Ensure DblpEngine is properly registered in main()

Performance

• Single Paper Retrieval: ~1-3 seconds (depends on network)
• Batch Processing: ~2-5 seconds per paper (with API rate limits)
• Memory Usage: Minimal (< 10MB)

Limitations

• API Dependency: Requires network connection and DBLP API availability
• Coverage: Limited to papers indexed in DBLP (primarily Computer Science)
• Accuracy: Title must be reasonably close to official paper title
• Rate Limiting: Subject to DBLP API rate limits (respects delays)

Contributing

Contributions are welcome! Areas for improvement:

• Additional retrieval engines
• Enhanced error recovery
• Performance optimization
• Test coverage expansion

License

This project is provided as-is for academic use.

Support

For issues, questions, or feature requests, please check the documentation or create an issue in the repository.

---

Happy researching! Let AutoRef handle your bibliography while you focus on the science.