test gap: multi-paragraph counter accumulation for canonical 5-run SEQ never tested (#104 follow-up)

[kiki830621]

Problem

From verification of PsychQuant/che-word-mcp#104 (verify report: comment-4346442620):

「Multi-paragraph counter accumulation never tested. Issue expects {Figure: 12, Table: 7} from 19-paragraph reproducer; tests only verify {Figure: 1} from single paragraph.」

— Source: team:requirements + codex (gpt-5.5)

#104's reproducer is 19 SEQ paragraphs scattered throughout the body (12 Figure + 7 Table). The acceptance criterion in the issue body is update_all_fields returning {"Figure": 12, "Table": 7}. Current tests only verify {"Figure": 1} from a single-paragraph fixture, so:

• The counter accumulation logic across multiple paragraphs is unverified for the canonical-form path
• The case where SEQ Figure paragraph N+1 follows SEQ Table paragraph N (interleaved identifiers) is unverified
• Heading-reset semantics interacting with canonical-form SEQ are unverified

Type

test gap (P3)

Acceptance criteria

• Add a test that builds a WordDocument with N=12 SEQ Figure paragraphs + M=7 SEQ Table paragraphs in canonical form (mixed order), runs updateAllFields(), asserts the returned dict equals ["Figure": 12, "Table": 7].
• Optionally: add a heading-reset variant (Heading 1 between groups) and verify per-chapter resets work for canonical form (parallel to existing baked-form heading-reset coverage).

Why low priority

Counter accumulation logic is shared between baked and canonical forms (the per-identifier [String: Int] dict updates the same way regardless of detection path). Existing baked-form tests cover counter accumulation. Risk of regression is low, but the issue's acceptance criterion ({Figure: 12, Table: 7}) is literally not asserted anywhere — surfaced by #104's verify discipline.

Related: PsychQuant/che-word-mcp#104

---

Current Status

Phase: implemented
Last updated: 2026-05-02 by idd-implement

Key Decisions

• Implemented the required 19-paragraph canonical accumulation coverage in PR Cover canonical sequence field edge cases #37.
• Asserted both returned counter totals and cached-run rawXML rewrite consistency.

Scope Changes

• Optional heading-reset variant was left out because existing baked-form tests already pin reset behavior.

Blocking

• Awaiting IDD verify / PR review.

Commits

• 11860b5 test: cover canonical sequence field edge cases

[kiki830621]

Implementation Plan

Scope

Pin canonical 5-run SEQ counter accumulation across many body paragraphs and mixed identifiers.

Plan

• Add a WordDocument fixture with 12 canonical SEQ Figure paragraphs and 7 canonical SEQ Table paragraphs in mixed order.
• Run updateAllFields() and assert the returned counter dictionary is ["Figure": 12, "Table": 7].
• Assert representative cached run rewrites prove both identifiers increment independently.

[kiki830621]

Implementation Complete

Checklist (synced from TaskList)

• Build a 19-paragraph canonical SEQ document with 12 Figure and 7 Table fields -> commit 11860b5
• Assert updateAllFields() returns ["Figure": 12, "Table": 7] -> commit 11860b5
• Assert each cached run text and rawXML are rewritten in sequence -> commit 11860b5

Changes

• 11860b5: added mixed-order canonical counter accumulation coverage in Issue104FieldParserCanonicalFormTests.
• PR: Cover canonical sequence field edge cases #37

Files Changed

Sources/OOXMLSwift/Parsing/FieldParser.swift       |   3 +-
Tests/OOXMLSwiftTests/Issue104FieldParserCanonicalFormTests.swift | 164 +++++++++++++++++++++

Verification

• swift test --filter Issue104FieldParserCanonicalFormTests -> passed
• swift test -> 834 tests, 1 skipped, 0 failures

Scope Compliance

Covered the required 12 Figure / 7 Table canonical accumulation path. The optional heading-reset variant was left out because existing baked-form coverage already pins reset behavior and this issue marks that variant optional.

Next: verification pending