fix(import): fail the metadata import on over-length free text

[jacobvjk]

Closes #990.

The metadata importer used to drop most free text that was over its character limit, and the only sign was a line in the report. Now every limit is an authoring error, so nothing is written until the workbook is fixed. The limits are the ones the cookbook now states in RMI/tpr-tpc#37 (decision 0028).

Tab › column	Limit	Before	Now
draft_core_drivers › all 7 drivers	500	note, value left null	error
draft_dependencies › all 10 descriptions	500	note, dependency skipped	error
draft_data_availability › Scope limitations	500	error	error (counted the same way as the others)
draft_metadata › Pathway Overview	2,500	note, existing value kept	error
draft_metadata › Description	100	note, template description used	error
draft_metadata › Name of Pathway	200 whole cell, 20 inside the brackets	not checked	error
draft_metadata › Name of publication	200	not checked	error

Worth knowing

• New pathways only: Description, Name of Pathway and Name of publication are checked only for new pathways. An update keeps the name, description and publication already in its JSON and never reads those cells. The cookbook validator (R2.5) still checks them upstream.
• How characters are counted: in code points, the way the cookbook and JSON-schema maxLength count them. CO₂ is 3 characters and an emoji is 1. JavaScript's .length would count the emoji as 2.
• What is measured: the text as written, after the importer adds a final period. A 500-character cell with no period becomes 501 and fails, which schema:check would reject anyway. This is one character stricter than the cookbook validator, which measures the raw cell.
• v1 fallback unchanged: a pathwayDescription taken from an old v1 expertOverview still only gets a note when it is over 2,500. That text isn't in the workbook, so there's no cell to point the author to.
• First import after merging may fail: the next import can stop on cells that used to be dropped silently. Each error names the cell to shorten.

Verification

• npx vitest run scripts/import-pathway-data.test.ts: 35 passing. 9 are new and cover each limit at exactly the limit and one over, the added period, code-point counting, and an update with over-length cells that must not error.
• eslint and prettier are clean on both files. tsc -p tsconfig.scripts.json reports 5 errors in these files, all on unchanged lines and all present on epic/v2 before this change.
• @jacobvjk ran a dry run against the current pathway_data_prepared.xlsx and it behaved as expected.

🤖 Generated with Claude Code

[github-actions]

Expected version change and release notes

🚨 WARNING: This PR is not expected to trigger a new version

To trigger a version bump, use at least one conventional commit message in this branch. See: https://www.conventionalcommits.org/en/v1.0.0/

[github-actions]

Azure Static Web Apps: Your stage site is ready! Visit it here: https://proud-glacier-0f640931e-992.westus2.2.azurestaticapps.net

[Copilot]

🟡 Changes recommended

Target filtering can skip over-length pathway abbreviations before validation, allowing the import to succeed.

1 open finding

• Validate supported row names before TARGETS lookup · New

What changed in this PR

Makes over-length workbook text block metadata imports instead of being silently dropped, aligning limits with cookbook decision 0028.

Changes:

• Adds shared code-point counting and length validation.
• Applies limits across metadata fields while preserving documented exceptions.
• Adds boundary, punctuation, Unicode, and update-exemption tests.

File	Description
scripts/​import-pathway-data.ts	Reports over-length imported text as blocking errors.
scripts/​import-pathway-data.test.ts	Tests character limits and validation exceptions.

🧠 Review effort: Balanced

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.