High‑precision PDF text extraction and intelligent chunking utility for building Retrieval-Augmented Generation (RAG), semantic search, summarization, and embedding pipelines in OutSystems Developer Cloud.

Overview

Transforms raw PDF binaries into normalized UTF‑8 text and structured, overlap-aware chunks with page mapping, hash fingerprinting, and lightweight token estimation. Eliminates boilerplate parsing and prepares content for vector storage or AI enrichment.

Parameter Guidance

• chunkSizeChars: 800–1500 (balance context vs. token cost)
• overlapSizeChars: 10–20% of chunk size
• normalizeWhitespace: True for embedding scenarios
• collapseRepeatedNewlines: Reduces sparsity
• includePageNumberPrefix: Enable if cross-referencing back to the source is needed
• maxTotalChars: Set >0 for guardrails on very large PDFs