fix(workflows): split expression operators across any whitespace

[huiq777]

Description

If a workflow condition wraps onto a second line after or, and, in, not in or not, it evaluates to false, with no error.

The evaluator finds word operators by their surrounding spaces (" or ", " and ", " in ", " not in ", expr.startswith("not ")). An operator next to a newline or tab is never split on. The whole expression is then looked up as one dot path and resolves to None.

YAML keeps that line break in the usual way of wrapping a long condition. A | block keeps every break, and > keeps the break before a more-indented continuation line:

steps:
  - id: route
    type: if
    condition: >-
      {{ inputs.skip_review or
         inputs.scope == 'docs' }}
    then:
      - id: fast-path
        type: shell
        run: "echo took the then branch"
    else:
      - id: review
        type: shell
        run: "echo took the else branch"

With skip_review: false and scope: docs the condition is true, but main runs the else branch. specify workflow run validates the file without a warning:

$ specify workflow run wf.yml          # main
  ▸ [route] if …
  ▸ [review] shell …
Status: completed

$ specify workflow run wf.yml          # this branch
  ▸ [route] if …
  ▸ [fast-path] shell …
Status: completed

The same thing happens to while / do-while conditions and to any other {{ }} expression wrapped this way.

Fix: _evaluate_simple_expression now collapses each run of whitespace outside a quoted string to a single space before parsing, as Jinja2 treats any whitespace between tokens alike.

• Quoted operands keep their text exactly, so inputs.title == 'two spaces' still compares against two spaces.
• Every entry point goes through this function: the typed fast path, interpolation, filter arguments and the validator's probes. They all see the same normalized text.

The condition gate had pinned the old behaviour. For the bare condition inputs.x == 1\nand inputs.name == 'abc', it withheld the "Wrap the expression" correction, because the wrapped form read False (see the comment in MULTI_POSITION_UNFIXABLE). Wrapping now repairs it, so that case moves to a new test. The test checks that the correction is offered and that the wrapped form evaluates True.

This is a sibling of #4417 (a parenthesised group read as a dot path and resolving to None). The failure mode is the same; the trigger is different.

Testing

• Tested locally with uv run specify --help
• Ran existing tests with uv sync && uv run pytest
• Tested with a sample project (if applicable)

Details:

• New tests in tests/test_workflows.py::TestExpressions:
  • test_operators_separated_by_any_whitespace: 9 cases covering or / and at the end of a line, a |-style break before or, a tab, not, in, not in split both ways, and CRLF inside a group.
  • test_condition_wrapped_across_lines_in_yaml: the folded scalar above, through yaml.safe_load and evaluate_condition.
  • test_whitespace_inside_quoted_operand_is_kept: a guard for quoted text.
• New in tests/unit/test_condition_expression_block.py: test_a_condition_wrapped_across_lines_gets_a_paste_ready_correction (If/While/DoWhile).
• Red on main, green here: with expressions.py reverted, the new cases fail (10 in test_workflows.py and 3 in the gate test). The quoted-text guard passes on both, as intended.
• The end-to-end run above used the workflow shown.
• Full suite: 8984 passed, 249 skipped, 14 failed on the first run.
  • 3 were the gate cases above, now updated.
  • 11 were *_python_parity template tests (test_resolve_template_python_parity.py, test_check_prerequisites_python_parity.py, and others). They pass when rerun on their own, both on this branch and on main, and they don't go through the workflow evaluator.
  • After the update, tests/test_workflows.py and tests/unit/test_condition_expression_block.py pass: 1074 tests.
• uvx ruff@0.15.0 check src tests (the CI pin): all checks passed.

AI Disclosure

• I did not use AI assistance for this contribution
• I did use AI assistance (fill in the disclosure below)

AI disclosure: Opened on behalf of @huiq777 with Claude Code (model: Claude Opus 5.5, default settings, autonomous). Claude Code found the bug while probing the expression evaluator for siblings of #4416, #4417 and #4572. It wrote the fix, the tests and this description, and ran the reproduction and the test suite shown above. The commit carries an Assisted-by: trailer.

🤖 Generated with Claude Code

[Copilot]

Copilot review overview

🟢 Approval recommended

The focused implementation preserves quoted content and has comprehensive regression coverage.

Review effort: Balanced
Findings: None

What changed in this PR

Normalizes unquoted whitespace so multiline workflow expressions parse consistently.

Changes:

• Collapse whitespace outside quoted strings.
• Add operator, YAML, quote-preservation, and remediation regressions.
• Static review only; tests were not independently rerun.

File	Description
src/​specify_cli/​workflows/​expressions.py	Adds quote-aware whitespace normalization.
tests/​test_workflows.py	Covers multiline operators and quoted whitespace.
tests/​unit/​test_condition_expression_block.py	Verifies multiline-condition remediation.

---

💡 Configure MCP servers for context-aware, tailored reviews. Learn more in the docs.