fix(sqleditor): skip read-only columns when saving inserted rows

[asheshv]

Summary

Fixes #9939.

Query Tool result sets can include expression/alias columns (e.g. first_name || ' ' || last_name AS the_name) that don't exist on the underlying table. The editability detection in is_query_resultset_updatable.py already marks them is_editable=False, so the column header correctly shows the lock icon. But when a user adds a new row, the frontend populated every column key on the new row object and shipped the whole thing to /sqleditor/save. save_changed_data accepted those keys verbatim, and the rendered INSERT referenced the alias — Postgres replied with column "the_name" does not exist in relation "some_table".

Filter on both sides:

• Backend — get_columns_types now propagates is_editable into the column metadata stored on the session (is_editable=True defaulted for View Data Tool, where columns are always real table columns).
• Backend — save_changed_data drops non-editable keys from the added-row data before rendering insert.sql. Defense-in-depth: protects against stale/buggy clients.
• Frontend — ResultSet.jsx::triggerSaveData omits cells whose column has can_edit === false from the added-row payload.

Test plan

• python regression/runtests.py --pkg tools.sqleditor.utils.tests.test_save_changed_data — 13/13 pass, including the new TestSaveAddedRowSkipsNonEditableColumn scenario that runs the issue's exact first_name || ' ' || last_name AS the_name SELECT and verifies the INSERT succeeds.
• python regression/runtests.py --pkg tools.sqleditor.utils.tests.test_is_query_resultset_updatable — 10/10 pass (3 OID-only scenarios skipped on PG12+, as before).
• python regression/runtests.py --pkg tools.sqleditor.utils.tests.test_save_changed_uuid_data — 3/3 pass (the other consumer of the changed helpers).
• make check-pep8, yarn run linter clean.
• Manual smoke in Query Tool: run the issue's queries, add a row, F5 — INSERT should succeed and only write real columns.

Summary by CodeRabbit

Release Notes

• Bug Fixes
  • SQL Editor now correctly saves newly-added rows when query results include non-editable columns (such as expression or alias columns), preventing save failures and invalid INSERT statements.

[coderabbitai]

Walkthrough

The PR fixes Query Editor failures when saving new rows in result sets containing non-editable columns (aliases, expressions). The fix filters such columns on both frontend and backend: metadata propagation adds is_editable flags, the frontend filters before sending, the backend filters during INSERT construction, and a regression test validates the scenario.

Changes

Non-editable column filtering in Query Editor saves

Layer / File(s)	Summary
Non-editable column metadata propagation web/pgadmin/tools/sqleditor/utils/get_column_types.py	The get_columns_types function adds an is_editable flag to column type descriptors, extracting it from column metadata and defaulting to True.
Frontend filtering of non-editable columns web/pgadmin/tools/sqleditor/static/js/components/sections/ResultSet.jsx	The triggerSaveData function builds a nonEditableKeys set from columns where can_edit === false and omits those keys from each newly-added row's data before sending to the save API.
Backend filtering during insert construction web/pgadmin/tools/sqleditor/utils/save_changed_data.py	The save_changed_data function filters row data to retain only editable columns from columns_info before rendering INSERT SQL, preventing non-existent table columns from being referenced.
Regression test for non-editable alias columns web/pgadmin/tools/sqleditor/utils/tests/test_save_changed_data.py	A new test class TestSaveAddedRowSkipsNonEditableColumn validates the insert flow with an aliased expression column (`first_name

🎯 3 (Moderate) | ⏱️ ~20 minutes

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 20.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title 'fix(sqleditor): skip read-only columns when saving inserted rows' is concise, specific, and accurately describes the main change across frontend, backend, and test modifications.
Linked Issues check	✅ Passed	The pull request fully addresses issue #9939 by implementing frontend filtering, backend filtering, and metadata propagation to prevent non-editable expression/alias columns from being referenced in INSERT statements.
Out of Scope Changes check	✅ Passed	All changes are directly scoped to the objective of skipping non-editable columns during insert operations; no unrelated modifications to unrelated features or infrastructure.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

🧪 Generate unit tests (beta)

• Create PR with unit tests

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 1

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@web/pgadmin/tools/sqleditor/utils/save_changed_data.py`:
- Around line 121-129: The filtering of incoming row data currently treats
unknown column keys as editable because the predicate uses columns_info.get(k,
{}).get('is_editable', True); change the default to False so unknown keys are
rejected: update the comprehension that builds data (the dict comprehension that
references columns_info and 'is_editable') to use .get('is_editable', False) so
only known editable columns are retained before INSERT rendering.

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: ce5ec567-c303-4937-8f91-14fa9791425d

📥 Commits

Reviewing files that changed from the base of the PR and between 291a55e and 0f8a14b.

📒 Files selected for processing (4)

• web/pgadmin/tools/sqleditor/static/js/components/sections/ResultSet.jsx
• web/pgadmin/tools/sqleditor/utils/get_column_types.py
• web/pgadmin/tools/sqleditor/utils/save_changed_data.py
• web/pgadmin/tools/sqleditor/utils/tests/test_save_changed_data.py

[dpage]

Merged to master as f9b9bc9 (squashed, preserving @asheshv as author). I folded in @coderabbitai's suggestion — the added-row filter now requires k in columns_info so unknown client keys are dropped rather than defaulted to editable — and added a release-notes entry for #9939. The squash rewrites the SHA so GitHub won't auto-close this — closing manually. Thanks @asheshv!

[coderabbitai]

Tip

For best results, initiate chat on the files or code changes.

@dpage Thanks for merging and for folding in the k in columns_info guard — that extra safety net means the backend will correctly reject any unknown client-supplied keys even if the frontend sends something unexpected in the future. Great work by @asheshv on the fix! 🎉

(ﾉ◕ヮ◕)ﾉ*:･ﾟ✧ 🐇

[hiteshjambhale]

Hi @asheshv
This is fixed for add rows.

1. Update rows still fails with same error, We can implement this for update rows as well.
2. I would suggest resolving this on frontend. The feedback on adding a new row with non editable column is wrong. It initially appeared that the wrong value was inserted(null or added value), and it stays until we re-execute the select query