Fix lint_param_docs parser bugs and deduplicate CI source lint

[sbryngelson]

Summary

Follow-up to #1327. Fixes bugs found by Claude CI review and deduplicates CI workflow.

Fixes

• Docstring: Updated to reflect all 3 tiers are blocking (was incorrectly saying Tier 2 is "warnings")
• Case-sensitivity bug: _parse_namelist_params used stripped.lower() for detection but stripped.split("/user_inputs/") for extraction — fails on uppercase Fortran. Now uses index-based slicing from the lowercase match.
• Single-line namelist bug: Parser read one line past a single-line namelist declaration. Now checks for continuation & immediately.
• Multiple namelist blocks: Uses += for accumulator so subsequent namelist /user_inputs/ blocks are captured.
• CI DRY violation: Replaced 3 hardcoded grep steps in test.yml with single python3 toolchain/mfc/lint_source.py call. Added missing lint_param_docs.py step. CI now matches precheck exactly.

Test plan

• python3 toolchain/mfc/lint_param_docs.py exits 0
• ./mfc.sh precheck -j 8 passes all 6/6

🤖 Generated with Claude Code

[github-actions]

Claude Code Review

Head SHA: 1a5d523

Files changed:

• 2
• .github/workflows/test.yml
• toolchain/mfc/lint_param_docs.py

Findings:

toolchain/mfc/lint_param_docs.py — _parse_namelist_params: accum += drops parameters at statement boundaries when multiple namelist /user_inputs/ declarations appear in one file

The new code replaces accum = stripped.split("/user_inputs/", 1)[1] (reset per statement) with accum += " " + stripped[idx:] (accumulate across all statements). When a file contains two or more namelist /user_inputs/ statements (which is valid Fortran — repeated namelist groups merge at runtime), the boundary between them produces a single comma-less token like "beta gamma" which fails the re.match(r"^[a-zA-Z_]\w*$", name) guard and is silently dropped.

Minimal reproduction:

text = "  namelist /user_inputs/ alpha, beta\n  namelist /user_inputs/ gamma, delta\n"
# New code returns: {'alpha', 'delta'}   # 'beta' and 'gamma' are silently dropped
# Expected:         {'alpha', 'beta', 'gamma', 'delta'}

All three current m_start_up.fpp files each have exactly one namelist /user_inputs/ statement, so this does not affect CI today. However, if a developer splits a growing namelist into two statements (common as Fortran line-continuation limits are hit), parameters at the statement boundary will be silently excluded from Tier 2 checking, producing false negatives rather than errors.

The fix is to insert a comma separator rather than a space when appending a new namelist statement's content to accum (e.g. accum += ", " + stripped[idx:]), or to reset accum to "" at the start of each new statement and flush collected params into the output set before resetting.

toolchain/mfc/lint_param_docs.py — in_namelist continuation detection tests the full accumulated buffer rather than the current line

In the new code:

accum += " " + stripped[idx:]
in_namelist = accum.rstrip().endswith("&")

in_namelist should reflect whether the current namelist statement has an open continuation, but the endswith("&") test operates on the entire accum string (all previously accumulated content plus the current line). If earlier content in accum happens to end with & (e.g. from a prior continuation that was mid-flight when a second namelist /user_inputs/ line was encountered), in_namelist would be set True incorrectly. The correct test is stripped[idx:].rstrip().endswith("&"), checking only the portion of the current line that follows /user_inputs/.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: 423aab7b-8ed8-4cd7-a4ca-f36aabddc045

📥 Commits

Reviewing files that changed from the base of the PR and between 2162f7a and 1a5d523.

📒 Files selected for processing (2)

• .github/workflows/test.yml
• toolchain/mfc/lint_param_docs.py

---

📝 Walkthrough

Walkthrough

The pull request consolidates three separate grep-based "Lint Source" checks in the CI/CD workflow into a single Python script (lint_source.py), which validates source code against disallowed raw directives, double-precision intrinsics, and junk code markers. Additionally, the lint_param_docs.py script is enhanced to improve detection of multi-line Fortran namelist parameter blocks in source files by using incremental content accumulation and checking for continuation markers. The Tier 2 validation rule for namelist parameter synchronization is updated to be blocking.

📝 Coding Plan

• Generate coding plan for human review comments

---

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.}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 45.01%. Comparing base (2ffec9a) to head (1a5d523).
⚠️ Report is 3 commits behind head on master.

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #1329   +/-   ##
=======================================
  Coverage   45.01%   45.01%           
=======================================
  Files          70       70           
  Lines       20562    20562           
  Branches     1962     1962           
=======================================
  Hits         9255     9255           
  Misses      10179    10179           
  Partials     1128     1128           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.