pi-ceo-docparser

What problem does it solve?

It removes the manual burden of converting PDFs, DOCX files, or plain text into consistent, page-cited, structured content that downstream workflows can reliably consume.

Core Features & Use Cases

• Deterministic extraction into ParsedDoc: Produces a ParsedDoc containing full text, page-separated content, tables (DOCX), and metadata such as title, without any LLM calls.
• Failed-soft dependency handling: Tries PyMuPDF first for PDFs, falls back to pypdf if needed, supports DOCX via python-docx, and always supports TXT with stdlib; failures populate doc.error instead of raising.
• Page-number citation preservation: Keeps page indices so later research or marketing pipelines can cite sources like “p. 3” precisely.

Quick Start

Ask your pipeline to parse the document at a local path like /path/to/customer-interview.pdf into a ParsedDoc and then read doc.text and doc.pages for downstream analysis.