Add tandf source

[adelavega]

Successfully implemented Taylor & Francis source parser for ACE with the following components:

Implementation Summary

1. Configuration File (ace/sources/TaylorAndFrancis.json)

• Multiple publisher detection patterns (domain, meta tags, JavaScript object)
• Custom entity mappings for HTML entities
• Rate limiting configuration

2. Source Parser (TaylorAndFrancisSource in ace/sources.py)

Key Features:

• Hybrid extraction approach: Extracts tables from JavaScript objects embedded in HTML
• Primary method: Parses tandf.tfviewerdata JavaScript object containing table HTML
• Fallback method: CSV download endpoints (placeholder implementation)
• Metadata extraction: Table numbers, labels, and captions from JSON structure
• Critical fix: Extracts JavaScript data BEFORE parent class removes script tags

Technical Implementation:

• _extract_tables_from_javascript(): Finds and extracts JavaScript table data using string slicing (more robust than regex)
• _parse_table_data(): Parses JSON data and extracts individual table HTML
• _extract_tables_from_csv(): Fallback CSV download method
• extract_doi() and extract_pmid(): Metadata extraction with multiple fallback methods

3. Test Suite (test_taylor_and_francis_source in ace/tests/test_ace.py)

• Validates source identification
• Tests table extraction from JavaScript data
• Verifies metadata extraction (table numbers, labels, captions)
• Confirms activation parsing from table content

4. Additional Fixes

• Fixed BeautifulSoup parser warnings in ace/utils.py by specifying "lxml" parser
• All existing tests continue to pass

Test Results

✅ Test passed successfully - extracts 2 tables from Taylor & Francis HTML with correct metadata and activations

The parser is now ready to handle Taylor & Francis publications and can be extended to support the CSV download fallback method for cases where JavaScript extraction fails.