fix(test): remove conditional guards from SSR and shortcode tests

[zeroedin]

Summary

Remove if err != nil / else conditional guards that let tests pass without the behavior being implemented. Tests must assert one path directly.

SSR skip behavior test (build_test.go): cat is a valid SSR command that reads stdin and writes stdout — it must succeed. Changed from if err != nil { check error } else { check success } to direct Expect(err).NotTo(HaveOccurred()).

Pipeline shortcode bridging test (plugin_template_test.go): An unregistered shortcode ({% greeting %}) must cause a build error. Changed from if err != nil { check error } else { check result } to direct Expect(err).To(HaveOccurred()).

Test plan

• No remaining if err != nil or if err == nil guards in test files (after spec(ssr): resolve conflicting error expectations in SSR tests #167 merges)
• No Skip() calls in test files

🤖 Generated with Claude Code

[zeroedin]

PR #169 Review — Remove Conditional Guards from SSR and Shortcode Tests

Scope: build_test.go (+6/-15), plugin_template_test.go (+11/-16). Removes if err != nil / if err == nil branching from two tests, making assertions unconditional.

Changes

1. Build test — "Build() runs Phase 2 when SSR config is present"

Old: if err != nil { check error references SSR } else { check result } — both branches pass, test never fails.

New: unconditional Expect(err).NotTo(HaveOccurred()) + result assertions. This is correct — cat is a valid stdin→stdout pass-through, so Build() with Command: "cat" should succeed. The old conditional guard was left over from a time when cat might have failed (the CLI-arg model where cat would try to open a file).

2. Plugin template test — "BuildWithContent renders plugin shortcode"

Old: if err != nil { check error msg } else { check result } — both branches pass.

New: unconditional Expect(err).To(HaveOccurred()) + error message assertion. The test uses {% greeting "World" %} without registering the shortcode — the build MUST fail with an unknown tag error. The old conditional was hedging against the possibility that the build might silently pass through, but the correct spec behavior is that unregistered tags are errors, not silent pass-throughs.

Verification

Both changes align with established patterns:

• cat as SSR command succeeds (proven in scanner tests and PR #150's per-page model)
• Unregistered shortcode in Liquid template is a parse/render error (Liquid doesn't silently ignore unknown tags)

Verdict

Clean cleanup. Conditional test guards that allow both outcomes to pass are effectively untestable — they can never fail. Both tests now have a single expected outcome with unconditional assertions. No issues.