Keep two independent reports in one PDF distinct

Find report boundaries before identifying rows

Keep the report boundary with each row so the workbook does not merge evidence from different sources.

For multipage invoices, use ParseMill to review extracted invoice data beside the source and export separate summary and line-item sheets. Include a page break in your trial and check that every required line made it through. Other report types need their own fit assessment.

Useful boundary evidence includes a new cover, report number, issue date, issuer block or explicitly restarted report heading. A blank page or a numbering restart is a clue, not sufficient proof by itself.

Create a segment manifest

These are invented identifiers. The two Sample 1 records remain distinct because their report identities differ. Segment C is deliberately unresolved: attaching it to B solely because it follows B would invent a relationship. Keep its source available while the owner clarifies its role.

Adobe: Renumber pages in PDFs explains page labels. Maintain physical ranges in the container as well as each report’s printed labels to make segment references reproducible.

Assign rows to segments

1. Inventory covers and report identifiers through the entire PDF. Note boundaries where identity evidence changes and ranges where it remains unclear.
2. Create a segment identifier local to this received file. Record the source evidence supporting each report identity and revision.
3. Attach the segment identifier to every extracted row before combining page outputs. Keep the original sample identifier unchanged in its own field.
4. Check the first and last row of each segment and every row near an uncertain boundary. Reconcile counts separately for each report before producing a combined total.

W3C: PROV Data Model models source entities separately from activities. A container file and the reports it carries can therefore have distinct recorded identities in your own provenance manifest.

Handle revisions and attachments carefully

Two segments with the same report number may be different revisions, duplicate copies or separate attachments. Compare revision markings and content before choosing a treatment. Do not automatically select the later physical segment as the current report. The delivery order may reflect scanning convenience rather than chronology.

A report-level result or date should be propagated only within a verified segment and only when its source meaning applies to those rows. Preserve any unresolved context as missing or uncertain rather than borrowing it from the preceding cover.

Check the consolidated register

For this fixture, a reliable row key includes segment A or B as well as Sample 1. The attachment remains outside either report’s observations until clarified. The recipient should be able to trace each row back through the segment manifest to the physical PDF range.

For the invoice trial, use a PDF containing invoices with distinct printed invoice numbers and check the record boundaries. The sample-report segmentation illustrated above is a requirement to discuss separately for a broader document workflow; the invoice workspace does not promise general laboratory-report segmentation.