v1.2.1: Fix Extractous API Usage

🐛 Critical Bug Fix

Extractous API Correction

• Fixed API Usage: Corrected extract_file_to_string() call to handle tuple return value
• Proper Text Extraction: Now extracts full text content instead of just 2 characters
• Performance Validation: Confirmed Extractous working correctly with proper extraction

📊 Performance Update

Local Benchmark Results

• Text File: Kreuzberg 0.001s vs Extractous 0.046s
• PDF File: Kreuzberg 0.033s vs Extractous 0.123s
• Kreuzberg Speed: 3.8x - 50x faster than Extractous on test files

Key Findings

• Kreuzberg uses pypdfium2 (Google's PDFium library) - highly optimized
• Extractous 18x speed claims may be context-dependent or for different document types
• Both frameworks now working correctly for comprehensive benchmarking

🚀 What's Fixed

• ✅ Extractous properly extracts full text content
• ✅ API compatibility issues resolved
• ✅ Ready for comprehensive benchmarking
• ✅ Performance baselines established

🔬 Next Steps

This release triggers a new comprehensive benchmark run with the corrected Extractous implementation. Results will show accurate performance comparisons across all frameworks and document types.

The benchmark pipeline will now provide reliable data comparing:

• Kreuzberg (pypdfium2-based)
• Extractous (Rust-based)
• Unstructured, MarkItDown, Docling

🚦 Testing

All frameworks validated with proper text extraction and performance measurement.