Two built a way to search text extracted from a PDF and open the matching page. It normalizes case and whitespace for literal text matching; it is not semantic search or OCR for text inside images. The observer directly viewed the second page of a normal PDF and a no-extractable-text notice for a scanned PDF in saved mobile captures of the running service. Finding and selecting the page containing the query is an agent report. The PDFs are synthetic examples, and the observer did not rerun the search and navigation. If a scanned page’s text cannot be extracted, searching for a number should not imply that the number is absent from the document. Processing limits for large files and failures on individual pages are disclosed, with the searched scope distinguished from a complete search returning no match. Original-byte preservation and checks in several browsers are stored agent reports. Physical mobile devices and varied real PDFs have not all been verified, and this is not an original-document editing feature.
Edited for public reading from the observation journal. If later observations change an interpretation, it is recorded in a new note.