What it does
This server extracts text, images, and metadata from PDF files with automatic parallel processing that achieves 5-10x speedup over sequential extraction. It preserves document layout through Y-coordinate-based content ordering and supports both absolute and relative paths on Windows and Unix systems. Built in TypeScript with 94% test coverage and minimal dependencies, it extracts full text at 5,575 ops/sec and handles 50-page PDFs in seconds.
Who it's for
Document analysts extracting text from scanned reports, software engineers integrating PDF processing into AI agent workflows, and data teams batch-processing large document collections.
Common use cases
- Extract full text from a PDF for document analysis and indexing
- Process specific page ranges from multi-page documents without loading the entire file
- Retrieve PDF metadata (author, title, creation date) for cataloging without full extraction
- Process multiple PDFs in parallel for comparative analysis or batch operations
- Handle PDFs referenced by relative or absolute paths across different development and production environments
Setup pitfalls
- Requires filesystem read and write access — verify permissions and consider sandboxing if processing untrusted PDFs
- Windows paths (like
C:\Users\...\file.pdf) need proper JSON escaping; v1.3.0+ handles normalization automatically - Parallel processing scales with available CPU cores; heavily constrain it on resource-limited systems to avoid memory exhaustion
- Node.js must have write permissions to the output directory for operations that extract or write data