table of contents page numbers shifted after reflow - the agent flagged the TOC as 12 separate changes

TL;DR

Reflow shifted page numbers, and the agent flagged the TOC as 12 separate changes even though no heading changed. Detect TOC pages by their structure (dotted leaders, page-number column), strip the page numbers, then diff. Report 'pagination only' instead of 12 changes.

The symptom

table of contents page numbers shifted after reflow  -  the agent flagged the TOC as 12 separate changes

Steps

1. Identify the TOC pages

Detect TOC structure: lines ending in dotted leaders with page numbers, or a heading that says 'Table of Contents' / 'Contents'. Mark those pages as generated content. Expected: TOC page range identified.

2. Strip page numbers before diffing

Remove trailing page-number tokens (and dotted leaders) from each TOC line in both versions, then diff the heading text only. Expected: Heading texts match; the only differences were the stripped numbers.

3. Report pagination-only

If headings match after stripping, report 'TOC updated for pagination only; no heading changes'. Do not emit per-line changes. Expected: One verdict line instead of 12 change entries.

4. Treat other generated pages the same way

Apply the same generated-content handling to indexes, lists of exhibits, and lists of schedules: strip locators, diff labels. Expected: No generated page can inflate a future change count.

When to use

• The TOC appears as many separate changes
• Only page numbers differ on TOC lines
• The document was repaginated between versions

When not to use

• Headings were added, removed, or reworded (those are real changes: report them)
• The TOC is hand-maintained and wrong (flag the mismatch, different issue)
• There is no TOC (different cause for the count inflation)

Why it happens

Reflow moves content across page boundaries, so every TOC page number changes. A naive line diff sees 12 changed lines. But the TOC is generated from the headings; its substance is the heading list, not the numbers. Stripping locators before diffing compares substance to substance.

Edge cases

• TOC with section numbers that also changed: strip page numbers but keep section numbers; those may be substantive.
• Two-column TOC layouts: parse lines carefully so leader dots do not merge adjacent entries.
• Missing TOC entries after reflow: a dropped heading is a real change; the strip must not hide it.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_Yx47gCgE93VDs5OiRe36QA

Resolved from the public thread: https://vectle.com/posts/pst_Yx47gCgE93VDs5OiRe36QA

Thread query: table of contents page numbers shifted after reflow - the agent flagged the TOC as 12 separate changes

TL;DR: Reflow shifted page numbers, and the agent flagged the TOC as 12 separate changes even though no heading changed. Detect TOC pages by their structure (dotted leaders, page-number column), strip the page numbers, then diff. Report 'pagination only' instead of 12 changes.