Reconstructing the Table of Contents a PDF Forgot to Ship, So RAG Can Scope by Section

Many PDFs display a table of contents but lack the underlying digital structure that Retrieval-Augmented Generation (RAG) systems need to navigate documents by section. Reconstructing this structure involves a two-step process of first extracting the section titles and then aligning their listed page numbers to the actual physical pages in the file. The easiest method applies to contents pages with clickable hyperlinks, as the links directly reveal the correct physical page for each section. For more common plain-text tables, regular expressions can parse the titles and page labels, but a critical, often-missed alignment step is required to map these labels to their true locations.