test: refractor system comprehensive test coverage #117

cal · 2026-03-24T21:02:48Z

cal commented

2026-03-24 21:02:48 +00:00

Summary

23 new test cases covering critical, high, and medium priority gaps in the Discord bot's Refractor integration
Tier 1 (Critical): tier_up dict shape mismatch exposing WP-14 latent bug, TIER_NAMES duplication divergence check, TIER_BADGES format consistency between cog and helpers
Tier 2 (High): notify_tier_completion with None channel, filter combination tier=4 + progress="close" behavior
Tier 3 (Medium): Malformed API response handling, progress bar boundary precision, RP formula label, unknown card_type fallback

Test plan

All 112 tests pass (89 pre-existing + 23 new), 0 failures
Ruff lint passed on all files
No new test files created — all added to existing files

Notable findings

T1-5 (WP-14 bug confirmed): build_tier_up_embed uses bare tier_up["player_name"] dict access — the minimal stub shape from the post-game hook will KeyError in production when WP-14 wires in the real function
T1-7: TIER_BADGES format split is intentional (bare in helpers/main, bracketed in cogs/refractor) — test validates the conversion relationship

🤖 Generated with Claude Code

## Summary - 23 new test cases covering critical, high, and medium priority gaps in the Discord bot's Refractor integration - **Tier 1 (Critical)**: tier_up dict shape mismatch exposing WP-14 latent bug, TIER_NAMES duplication divergence check, TIER_BADGES format consistency between cog and helpers - **Tier 2 (High)**: notify_tier_completion with None channel, filter combination tier=4 + progress="close" behavior - **Tier 3 (Medium)**: Malformed API response handling, progress bar boundary precision, RP formula label, unknown card_type fallback ## Test plan - [x] All 112 tests pass (89 pre-existing + 23 new), 0 failures - [x] Ruff lint passed on all files - [x] No new test files created — all added to existing files ## Notable findings - **T1-5 (WP-14 bug confirmed)**: `build_tier_up_embed` uses bare `tier_up["player_name"]` dict access — the minimal stub shape from the post-game hook will KeyError in production when WP-14 wires in the real function - T1-7: TIER_BADGES format split is intentional (bare in helpers/main, bracketed in cogs/refractor) — test validates the conversion relationship 🤖 Generated with [Claude Code](https://claude.com/claude-code)

cal added 1 commit 2026-03-24 21:02:48 +00:00

test: add comprehensive refractor test cases (T1-5 through T3-5)

Ruff Lint / lint (pull_request) Successful in 22s

Details

b34dcc390c

Implements all gap tests identified by PO agents across three existing
test files. No new files created — tests added to existing modules.

T1-5 (test_refractor_notifs): Expose WP-14 integration bug — minimal
stub dict {player_id, old_tier, new_tier} causes KeyError in
build_tier_up_embed because player_name/track_name use bare dict access.
Documents the bug contract so WP-14 implementers know what to fix.

T1-6 (test_refractor_commands): Divergence tripwire — imports TIER_NAMES
from both cogs.refractor and helpers.refractor_notifs and asserts deep
equality. Will fail the moment the two copies fall out of sync.

T1-7 (test_card_embed_refractor): TIER_BADGES format contract — asserts
that wrapping helpers.main badge values in brackets produces cogs.refractor
badge values (e.g. "BC" -> "[BC]") for all tiers.

T2-7 (test_refractor_notifs): notify_tier_completion with None channel
must not raise — the try/except absorbs AttributeError from None.send().

T2-8 (test_refractor_commands): All-T4 apply_close_filter returns empty
list. Documents intended behaviour for tier=4 + progress="close" combo.

T3-2 (test_refractor_commands): Malformed API response handling —
format_refractor_entry must use fallbacks ("Unknown", 0) for missing keys.

T3-3 (test_refractor_commands): Progress bar boundary precision — 1/100,
99/100, 0/100, and negative current values.

T3-4 (test_refractor_commands): RP formula label — card_type="rp" shows
"IP+K" (previously only "sp" was tested).

T3-5 (test_refractor_commands): Unknown card_type falls back to raw string
as the formula label without crashing.

112 tests pass (23 new, 89 pre-existing).

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

cal reviewed 2026-03-24 21:04:44 +00:00

cal left a comment

AI Code Review

Files Reviewed

tests/test_card_embed_refractor.py (modified — added TestTierBadgesFormatConsistency)
tests/test_refractor_commands.py (modified — added TestTierNamesDivergenceCheck, TestApplyCloseFilterWithAllT4Cards, TestFormatRefractorEntryMalformedInput, TestRenderProgressBarBoundaryPrecision, TestRPFormulaLabel, TestUnknownCardTypeFallback)
tests/test_refractor_notifs.py (modified — added TestTierUpDictShapeMismatch, TestNotifyTierCompletionNoneChannel)

Source files read for verification: cogs/refractor.py, helpers/refractor_notifs.py, helpers/main.py

Findings

Correctness

All assertions were verified against the actual source implementations.

T1-5 (TestTierUpDictShapeMismatch): build_tier_up_embed uses bare tier_up["player_name"] and tier_up["track_name"] (confirmed at helpers/refractor_notifs.py lines 50, 52). test_minimal_stub_shape_raises_key_error correctly documents the real KeyError. The companion test_full_shape_does_not_raise uses make_tier_up which supplies all required keys — correct.

T1-6 (TestTierNamesDivergenceCheck): Both cogs/refractor.py and helpers/refractor_notifs.py define identical TIER_NAMES dicts (T0-T4). The tripwire pattern (equality + key set + per-value) is sound. Call-time imports avoid monkeypatch interference as noted.

T1-7 (TestTierBadgesFormatConsistency): Confirmed helpers/main.py uses bare strings ({1: "BC", 2: "R", 3: "GR", 4: "SF"}) and cogs/refractor.py uses bracket strings ({1: "[BC]", 2: "[R]", ...}). The bracket-wrapping relationship is real. The four per-tier spot checks are redundant with the parametric test but low-cost and useful as documentation.

T2-7 (TestNotifyTierCompletionNoneChannel): notify_tier_completion wraps its entire body in try/except Exception, which catches the AttributeError from None.send(). The test correctly expects no raise.

T2-8 (TestApplyCloseFilterWithAllT4Cards): The current_tier >= 4 or not next_threshold guard in apply_close_filter excludes all T4 cards regardless of formula_value. Both test methods assert correct behavior.

T3-2 (TestFormatRefractorEntryMalformedInput): All four assertions verified. format_refractor_entry uses .get() with fallbacks for all fields. The empty-dict test correctly produces two lines because the next_threshold defaults to None via .get(), triggering the FULLY EVOLVED branch.

T3-3 (TestRenderProgressBarBoundaryPrecision): The negative value test correctly expects 0 filled segments. One note: the docstring attributes the behavior to the min(..., 1.0) clamp, but it is actually round() that truncates the small negative ratio to 0. The test result is correct regardless.

T3-4 (TestRPFormulaLabel): FORMULA_LABELS maps "rp" to "IP+K". Assertion is correct. Closes the gap where only "sp" was previously tested.

T3-5 (TestUnknownCardTypeFallback): FORMULA_LABELS.get(card_type, card_type) falls back to the raw string. Both "util" and "dh" are absent from FORMULA_LABELS, so both tests assert correctly.

Security

No issues found. Tests only import from internal modules with no new external dependencies or I/O.

Style & Conventions

Section header comment blocks follow the existing # ----------- pattern throughout all three files.
Docstrings follow the "what + why" format used in the existing test suite.
All new classes follow the TestFooBar naming convention.
New tests construct inline state dicts rather than adding fixtures, matching the existing style for simple inputs.
The two blank-line removals in test_refractor_commands.py are valid cleanup.

Suggestions

The docstring for test_negative_current_does_not_overflow_bar says "the min(..., 1.0) clamp in render_progress_bar should handle this." This is slightly inaccurate — negative values pass through the clamp (since -0.05 < 1.0) and the zero result comes from round() truncating the negative. Not a correctness issue, but worth a follow-up docstring fix to avoid misleading anyone who later modifies render_progress_bar.
TestTierNamesDivergenceCheck.test_tier_names_have_same_keys and test_tier_names_have_same_values are fully subsumed by test_tier_names_are_identical_across_modules. They are not wrong, and the pattern is consistent with how TestTierBadgesFormatConsistency handles the same concern.

Verdict: APPROVED

All 23 new test cases correctly assert against the actual source implementations. The WP-14 bug documentation test uses pytest.raises(KeyError) correctly — intent is clearly stated and the test will fail (become green-then-red) exactly when WP-14 fixes the underlying issue, at which point it should be converted to a positive assertion. Cross-module divergence tripwires target a real duplication risk. No pattern violations, no incorrect assertions found.

Automated review by Claude PR Reviewer

## AI Code Review ### Files Reviewed - `tests/test_card_embed_refractor.py` (modified — added `TestTierBadgesFormatConsistency`) - `tests/test_refractor_commands.py` (modified — added `TestTierNamesDivergenceCheck`, `TestApplyCloseFilterWithAllT4Cards`, `TestFormatRefractorEntryMalformedInput`, `TestRenderProgressBarBoundaryPrecision`, `TestRPFormulaLabel`, `TestUnknownCardTypeFallback`) - `tests/test_refractor_notifs.py` (modified — added `TestTierUpDictShapeMismatch`, `TestNotifyTierCompletionNoneChannel`) Source files read for verification: `cogs/refractor.py`, `helpers/refractor_notifs.py`, `helpers/main.py` --- ### Findings #### Correctness All assertions were verified against the actual source implementations. **T1-5 (`TestTierUpDictShapeMismatch`):** `build_tier_up_embed` uses bare `tier_up["player_name"]` and `tier_up["track_name"]` (confirmed at `helpers/refractor_notifs.py` lines 50, 52). `test_minimal_stub_shape_raises_key_error` correctly documents the real `KeyError`. The companion `test_full_shape_does_not_raise` uses `make_tier_up` which supplies all required keys — correct. **T1-6 (`TestTierNamesDivergenceCheck`):** Both `cogs/refractor.py` and `helpers/refractor_notifs.py` define identical `TIER_NAMES` dicts (T0-T4). The tripwire pattern (equality + key set + per-value) is sound. Call-time imports avoid monkeypatch interference as noted. **T1-7 (`TestTierBadgesFormatConsistency`):** Confirmed `helpers/main.py` uses bare strings (`{1: "BC", 2: "R", 3: "GR", 4: "SF"}`) and `cogs/refractor.py` uses bracket strings (`{1: "[BC]", 2: "[R]", ...}`). The bracket-wrapping relationship is real. The four per-tier spot checks are redundant with the parametric test but low-cost and useful as documentation. **T2-7 (`TestNotifyTierCompletionNoneChannel`):** `notify_tier_completion` wraps its entire body in `try/except Exception`, which catches the `AttributeError` from `None.send()`. The test correctly expects no raise. **T2-8 (`TestApplyCloseFilterWithAllT4Cards`):** The `current_tier >= 4 or not next_threshold` guard in `apply_close_filter` excludes all T4 cards regardless of formula_value. Both test methods assert correct behavior. **T3-2 (`TestFormatRefractorEntryMalformedInput`):** All four assertions verified. `format_refractor_entry` uses `.get()` with fallbacks for all fields. The empty-dict test correctly produces two lines because the `next_threshold` defaults to `None` via `.get()`, triggering the FULLY EVOLVED branch. **T3-3 (`TestRenderProgressBarBoundaryPrecision`):** The negative value test correctly expects 0 filled segments. One note: the docstring attributes the behavior to the `min(..., 1.0)` clamp, but it is actually `round()` that truncates the small negative ratio to 0. The test result is correct regardless. **T3-4 (`TestRPFormulaLabel`):** `FORMULA_LABELS` maps `"rp"` to `"IP+K"`. Assertion is correct. Closes the gap where only `"sp"` was previously tested. **T3-5 (`TestUnknownCardTypeFallback`):** `FORMULA_LABELS.get(card_type, card_type)` falls back to the raw string. Both `"util"` and `"dh"` are absent from `FORMULA_LABELS`, so both tests assert correctly. #### Security No issues found. Tests only import from internal modules with no new external dependencies or I/O. #### Style & Conventions - Section header comment blocks follow the existing `# -----------` pattern throughout all three files. - Docstrings follow the "what + why" format used in the existing test suite. - All new classes follow the `TestFooBar` naming convention. - New tests construct inline state dicts rather than adding fixtures, matching the existing style for simple inputs. - The two blank-line removals in `test_refractor_commands.py` are valid cleanup. #### Suggestions - The docstring for `test_negative_current_does_not_overflow_bar` says "the `min(..., 1.0)` clamp in `render_progress_bar` should handle this." This is slightly inaccurate — negative values pass through the clamp (since `-0.05 < 1.0`) and the zero result comes from `round()` truncating the negative. Not a correctness issue, but worth a follow-up docstring fix to avoid misleading anyone who later modifies `render_progress_bar`. - `TestTierNamesDivergenceCheck.test_tier_names_have_same_keys` and `test_tier_names_have_same_values` are fully subsumed by `test_tier_names_are_identical_across_modules`. They are not wrong, and the pattern is consistent with how `TestTierBadgesFormatConsistency` handles the same concern. ### Verdict: APPROVED All 23 new test cases correctly assert against the actual source implementations. The WP-14 bug documentation test uses `pytest.raises(KeyError)` correctly — intent is clearly stated and the test will fail (become green-then-red) exactly when WP-14 fixes the underlying issue, at which point it should be converted to a positive assertion. Cross-module divergence tripwires target a real duplication risk. No pattern violations, no incorrect assertions found. --- *Automated review by Claude PR Reviewer*

cal merged commit 8c0c2eb21a into main

2026-03-24 21:06:14 +00:00

cal deleted branch test/refractor-comprehensive

2026-03-24 21:06:15 +00:00

cal referenced this issue from a commit

2026-03-24 21:06:16 +00:00

test: refractor system comprehensive test coverage (#117)

Sign in to join this conversation.

No reviewers