fix: align command condition expansion with shell commands

[yottahmd]

Summary

• align command-form condition evaluation with shell command expansion semantics
• share runtime command eval options between the command executor and condition execution paths
• add regression coverage for home-relative DAG env vars in preconditions and repeat-policy conditions

Root Cause

Command-form conditions were evaluated with eval.OnlyReplaceVars(), which deferred scoped variables to the shell. That diverged from shell command: steps, where Dagu expands the command text before execution. Values like ~/... therefore worked in shell commands but failed in conditions because tilde expansion does not happen after shell parameter expansion.

Testing

• go test ./internal/runtime -run "TestEvalConditions|TestRunner_DAGPreconditions|TestRunner_RepeatPolicy" -count=1
• go test ./internal/intg -run "TestPreconditionWithDAGEnvVars|TestDAGLevelPreconditionWithEnvVars|TestPreconditionWithHomeRelativeDAGEnvVar|TestRepeatPolicy" -count=1
• go test ./internal/runtime/builtin/command -run TestCommandExecutor_GetEvalOptions -count=1

Closes #2057

Summary by CodeRabbit

• Documentation

  • Updated OpenAPI schema documentation for condition execution semantics, clarifying behavior when the expected parameter is omitted versus specified.

• Tests

  • Added integration tests for condition evaluation with environment variable expansion and repeat policy.
  • Enhanced unit tests with improved capability registration lifecycle management.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 9daf5c11-c4eb-4914-a6b0-4816e3e73128

📥 Commits

Reviewing files that changed from the base of the PR and between 02bf754 and 0a771e2.

📒 Files selected for processing (2)

• internal/core/capabilities_test.go
• internal/runtime/builtin/command/command.go

✅ Files skipped from review due to trivial changes (1)

• internal/core/capabilities_test.go

---

📝 Walkthrough

Walkthrough

The pull request fixes variable expansion inconsistencies in preconditions by centralizing command evaluation options logic. It updates OpenAPI schema documentation for Condition properties, introduces a new CommandEvalOptions function to standardize shell-backed command string evaluation, refactors condition evaluation to use this function, and adds integration and unit tests to verify home-relative path expansion in preconditions and repeat policies.

Changes

Cohort / File(s)	Summary
API Schema Documentation api/v1/api.yaml	Updated Condition.condition and Condition.expected property descriptions to clarify execution semantics: command checks via shell variable expansion when expected is omitted; string comparison when expected is set.
Runtime Evaluation Logic internal/runtime/eval_options.go	New CommandEvalOptions(shell []string) function that computes eval options based on shell type: enables OS expansion for direct/empty shells; uses WithoutDollarEscape with conditional WithoutExpandEnv for Unix-like and nix shells; preserves env expansion for other shells (e.g., PowerShell).
Command/Condition Evaluation internal/runtime/builtin/command/command.go, internal/runtime/condition.go	Refactored to delegate shell-to-eval-option logic to CommandEvalOptions(shell) instead of local decisions; removed inline shell expansion rules and eval.OnlyReplaceVars() usage.
Integration Tests internal/intg/precondition_env_test.go, internal/intg/repeat_test.go	Added POSIX-specific tests (skipped on Windows) verifying home-relative paths (~/...) in environment variables are correctly expanded in precondition command checks and repeat policy conditions.
Unit Tests internal/runtime/condition_test.go, internal/core/capabilities_test.go	New unit test for EvalConditions verifying home-relative path expansion in command-form preconditions; test cleanup adding per-subtest capability registration teardown to prevent cross-test leakage.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~25 minutes

Possibly related PRs

• dagucloud/dagu#2072: Modifies evaluation option construction and usage to control substitution/expansion semantics in variable handling.
• dagucloud/dagu#2075: Introduces field-specific eval-option hooks (CommandEvalOptions/ScriptEvalOptions) and refactors multiple call sites to determine command/script evaluation behavior.

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 26.67% 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 accurately describes the main change: aligning command condition expansion behavior with shell commands to ensure consistent variable and tilde expansion semantics.
Linked Issues check	✅ Passed	The PR successfully addresses all objectives from issue #2057: eliminates expansion inconsistency via CommandEvalOptions, ensures home-relative paths work in preconditions/repeat-policy, and adds regression test coverage for the exact user scenario.
Out of Scope Changes check	✅ Passed	All changes are directly related to aligning condition expansion semantics with command expansion: schema documentation updates, new eval_options logic, condition evaluation refactoring, and regression tests for the specific use case.

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

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix-precondition-command-expansion

---

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

_{Review rate limit: 7/8 reviews remaining, refill in 7 minutes and 30 seconds.}

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

[yottahmd]

@coderabbitai review

[coderabbitai]

✅ Actions performed

Review triggered.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.

[coderabbitai]

🧹 Nitpick comments (1)

internal/core/capabilities_test.go (1)

93-99: ⚡ Quick win

Add cleanup for globally registered capabilities in subtests.

Each subtest registers into a global registry without unregistering. Adding t.Cleanup(...) after each registration keeps tests isolated and prevents cross-test state bleed.

Suggested patch

 		RegisterExecutorCapabilities("command-eval-opts-test", ExecutorCapabilities{
 			Command: true,
 			GetCommandEvalOptions: func(_ context.Context, _ Step) []eval.Option {
 				return []eval.Option{eval.WithoutExpandShell()}
 			},
 		})
+		t.Cleanup(func() { UnregisterExecutorCapabilities("command-eval-opts-test") })

 		RegisterExecutorCapabilities("script-eval-opts-test", ExecutorCapabilities{
 			Command: true,
 			Script:  true,
 			GetScriptEvalOptions: func(_ context.Context, _ Step) []eval.Option {
 				return []eval.Option{eval.WithNoExpansion()}
 			},
 		})
+		t.Cleanup(func() { UnregisterExecutorCapabilities("script-eval-opts-test") })

 		RegisterExecutorCapabilities("config-eval-opts-test", ExecutorCapabilities{
 			Command: true,
 		})
+		t.Cleanup(func() { UnregisterExecutorCapabilities("config-eval-opts-test") })

 		RegisterExecutorCapabilities("legacy-eval-opts-test", ExecutorCapabilities{
 			Command: true,
 			Script:  true,
 			GetEvalOptions: func(_ context.Context, _ Step) []eval.Option {
 				return []eval.Option{eval.WithoutExpandShell()}
 			},
 		})
+		t.Cleanup(func() { UnregisterExecutorCapabilities("legacy-eval-opts-test") })

Also applies to: 107-114, 122-125, 132-139

🤖 Prompt for AI Agents

Verify each finding against the current code and only fix it if needed.

In `@internal/core/capabilities_test.go` around lines 93 - 99, Each subtest calls
RegisterExecutorCapabilities (e.g.,
RegisterExecutorCapabilities("command-eval-opts-test",
ExecutorCapabilities{...})) but never unregisters, so add t.Cleanup calls
immediately after each RegisterExecutorCapabilities call to remove the entry
when the subtest ends; specifically, after each
RegisterExecutorCapabilities("...") add t.Cleanup(func(){
UnregisterExecutorCapabilities("...") }) (use the same registration key string)
so tests don’t leak global state.

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Nitpick comments:
In `@internal/core/capabilities_test.go`:
- Around line 93-99: Each subtest calls RegisterExecutorCapabilities (e.g.,
RegisterExecutorCapabilities("command-eval-opts-test",
ExecutorCapabilities{...})) but never unregisters, so add t.Cleanup calls
immediately after each RegisterExecutorCapabilities call to remove the entry
when the subtest ends; specifically, after each
RegisterExecutorCapabilities("...") add t.Cleanup(func(){
UnregisterExecutorCapabilities("...") }) (use the same registration key string)
so tests don’t leak global state.

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 4a9ec84f-908c-41b4-bb44-0aa7251ba9f2

📥 Commits

Reviewing files that changed from the base of the PR and between 939a9d5 and 02bf754.

📒 Files selected for processing (3)

• internal/core/capabilities.go
• internal/core/capabilities_test.go
• internal/runtime/builtin/command/command.go

[yottahmd]

@coderabbitai review

[coderabbitai]

✅ Actions performed

Review triggered.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.