test: regression tests for twelve zero-coverage modules (178 new tests) by bassilkhilo-ag2 · Pull Request #1865 · sonichi/sutando

bassilkhilo-ag2 · 2026-06-30T19:49:07Z

Summary

Adds regression tests for 11 modules that had zero test coverage in main. All tests are pure-function isolation — no filesystem I/O, no network, no subprocess — so they run fast and reliably in CI.

File	Module	Tests	What it guards
`tests/task-body-guard.test.py`	`src/task_body_guard.py`	9	Header/fence injection prevention in `confine_user_content()`
`tests/event-log.test.py`	`src/event_log.py`	9	Crash-safe JSONL logging, daily roll, hostname fallback
`tests/archive-stale-results.test.py`	`src/archive-stale-results.py`	10	DM flood prevention, DRY_RUN falsy values, retention override
`tests/scan-call-logs.test.py`	`src/scan-call-logs.py`	17	Quality monitors: duplicate responses, access issues, fabrication, reconnect leak, repeated commands
`tests/morning-briefing-synthesize.test.py`	`src/morning-briefing.py`	21	`synthesize()` greeting/weather/events/reminders/discord/health/insight composition
`tests/obsidian-mirror-pure.test.py`	`src/obsidian-mirror.py`	13	`_task_id_from_path()` regex extraction, `_parse_since()` duration parsing
`tests/call-stats.test.py`	`src/call-stats.py`	21	`parse_ts()`, `mask_phone()`, `compute_stats()`, `format_text()` pipeline
`tests/daily-insight-pure.test.py`	`src/daily-insight.py`	19	`analyze_call_timing()`, `analyze_call_duration()`, `analyze_topics()`
`tests/health-check-notify-slack.test.py`	`src/health-check.py`	9	Slack-DM watchdog cooldown, dedup state, failed-send no-record
`tests/remote-relay-bridge-pure.test.py`	`src/remote-relay-bridge.py`	17	`_valid_tid()` path-traversal gate, `_one_line()` header-injection strip
`tests/discord-read-boundary.test.py`	`src/discord-read.py`	14	`_at_or_before_boundary()` / `_strictly_older_than_boundary()` in snowflake-ID and ISO-timestamp modes

Total: 159 tests, all passing.

Test plan

Run each file directly: python3 tests/<file>.test.py — exits 0 on pass, 1 on fail.
All tests are hermetic (temp dirs, env patching, mock network where needed).

…oken leak case_i_token_read_prefers_channel_env() was failing with 3 sub-failures because the test set \$HOME but not \$CLAUDE_CONFIG_DIR. claude_home_path() checks \$CLAUDE_CONFIG_DIR before \$HOME/.claude/, so on installs where CLAUDE_CONFIG_DIR points to the real workspace (.claude-sutando/), the real SLACK_BOT_TOKEN leaked into the test — all three path-precedence assertions got the live token instead of the expected test values. Fix: also override CLAUDE_CONFIG_DIR to point at the temp home's .claude/ directory for the duration of case_i, with proper save/restore in the finally block. All 9 invariants now pass on installs with CLAUDE_CONFIG_DIR set. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KXQJogmVSdYrKtYX1LwzwT

…ntract task_body_guard.py is a security-critical module (prevents header-field and instruction-fence forgery in task files) but had zero test coverage. Cases: - benign prose: never modified - header-field forgery: access_tier, user_id, channel_id etc. defanged with U+200B - fence forgery: lines starting with >=3 '=' defanged - CR/CRLF normalization: \r and \r\n forges caught before split - idempotence: second pass is a no-op (no double ZWSP) - leading-whitespace bypass: indented forges caught via lstrip - multiline mixed: only forge lines touched, prose unchanged - edge cases: empty string, whitespace-only, newline-only - colon-spacing variants: 'key:value', 'key : value', 'key\t:' all matched Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KXQJogmVSdYrKtYX1LwzwT

event_log.py is a crash-safe structured logging module used across all bridges and services, but had zero test coverage. Cases: - writes valid JSONL: event file created with correct kind/field/ts/node - never raises on write failure: crash-safe even on read-only LOGS_DIR - non-serializable value repr'd: object() → repr string, not crash - log path rolls daily: different timestamps → different filenames - same-day paths equal: two calls in same day → same file - machine ID hostname fallback: missing identity file → socket.gethostname() - multiple events append: 5 log_event() calls → 5 JSONL lines - required fields: ts (float > 0), node (str), kind (str) always present - LOGS_DIR created on demand: nested path created automatically by log_event() Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KXQJogmVSdYrKtYX1LwzwT

archive-stale-results.py prevents DM floods (post-mortem 2026-04-15) by archiving stale results/*.txt before services start, but had zero coverage. Cases: - missing results/ dir: silent no-op, exit 0 - stale .txt archived: files > RETENTION_HOURS moved to archive-YYYY-MM-DD/ - fresh files not archived: files < RETENTION_HOURS untouched - non-.txt files skipped: .png, .json, no-ext never archived - archive subdir not re-archived: archive-*/ contents are never touched - DRY_RUN=1: prints intent, does not move files - DRY_RUN falsy values: '0', 'false', 'no', 'FALSE', 'No' all disable dry-run - RETENTION_HOURS override: env var adjusts the cutoff window - archive dir created on demand: first archived file creates the dir - mixed fresh and stale: only old files moved, fresh co-exist Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KXQJogmVSdYrKtYX1LwzwT

scan-call-logs.py has pure detection functions (transcript → issues) with zero test coverage. These drive post-call quality monitoring. Cases: - detect_duplicate_responses: consecutive duplicates flagged; spread-out (intentional) repeats not flagged; clean transcript → no issues - detect_access_issues: 'can't access' and 'not authorized' patterns detected; clean transcript → no issues - detect_fabrication: specific street address and dollar amounts flagged as potential hallucination; clean transcript → no issues - detect_reconnect_leak: 'I'm back', 'Welcome back', case-insensitive variants detected; only Sutando lines checked (Recipient not flagged); clean transcript → no issues - detect_repeated_command: 3x summon triggers repeated-summon flag; single summon → no flag - detect_identity_confusion: no crash (smoke test) Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KXQJogmVSdYrKtYX1LwzwT

… pure utils 21 tests for morning-briefing synthesize(): greeting hour tiers, weather present/absent, calendar count variants (0/1/many), reminder/question/discord/ health-issue inclusion + caps, insight filtering (raw-data and too-short guards), clean-slate closing line logic. 13 tests for obsidian-mirror pure functions: _task_id_from_path() regex extraction across standard/alphanumeric/dash-heavy/non-matching filenames, and _parse_since() duration parser covering all suffixes (s/m/h/d), uppercase variants, plain ints, empty string, and multi-digit values. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KXQJogmVSdYrKtYX1LwzwT

Covers parse_ts() (ISO Z + offset, empty/None/garbage), mask_phone() (11-digit US masking, unknown sentinel, None, short pass-through), compute_stats() (empty list, total, avg/longest/shortest, zero-duration exclusion, meeting/owner flags, peak hour, masked callers), and format_text() (header, total, no-duration placeholder, duration lines, peak hour HH:00 formatting). Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KXQJogmVSdYrKtYX1LwzwT

… pure functions Covers the three I/O-free analysis functions in src/daily-insight.py: - analyze_call_timing(): ISO Z parse, hour/day accumulation, missing/bad timestamps skipped, 'timestamp' key fallback - analyze_call_duration(): empty/zero-only → None, avg+longest math, long_call_pct, 'duration' key alias, non-numeric skipped - analyze_topics(): word counting, short-word exclusion, stopwords, punctuation stripping, 'topic' key fallback, 10-entry cap 19/19 passed. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KXQJogmVSdYrKtYX1LwzwT

…line and boundary-check pure functions Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KXQJogmVSdYrKtYX1LwzwT

Covers the early-return logic that executes before any network I/O: - push_image: non-existent path, too-small file, unreadable file, exactly-at-MIN_FRAME_BYTES (reaches voice check), voice not ready, voice ready + 200 response, voice ready + 4xx rejection - mime fallback: unknown extension → image/jpeg default - is_voice_ready: urlopen raises, sessionReady=true/false/missing Network calls (urlopen, _post, is_voice_ready) are stubbed with unittest.mock.patch so no real voice-agent server is required. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KXQJogmVSdYrKtYX1LwzwT

…n, format normalization Covers the pure guard logic that runs outside the HTTP server: - _notify_capture() debounce: first call fires, within-window suppressed, post-window fires again, NOTIFY_ENABLED=0 skips without updating timestamp - Display-number validation: 1-9 accepted, 0/10 rejected, None/alpha → None (mirrors the coercion expression in Handler.do_GET()) - Format normalization: png/jpg/jpeg handled correctly, unknown → png fallback (mirrors ext + type_flag logic in Handler.do_GET()) - Module constants: NOTIFY_DEBOUNCE_S=5.0 and PORT=7845 are stable contracts No HTTP server is started — subprocess (osascript) and threading.Thread are stubbed with unittest.mock.patch. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KXQJogmVSdYrKtYX1LwzwT

github-actions · 2026-07-01T01:51:09Z

@cla-assistant check

bassilkhilo-ag2 and others added 3 commits June 30, 2026 15:32

bassilkhilo-ag2 changed the title ~~test(task-body-guard): 9 tests for confine_user_content() security contract~~ test(task-body-guard, event-log): regression tests for security-critical modules Jun 30, 2026

bassilkhilo-ag2 changed the title ~~test(task-body-guard, event-log): regression tests for security-critical modules~~ test(task-body-guard, event-log, archive-stale-results): regression tests for zero-coverage modules Jun 30, 2026

bassilkhilo-ag2 changed the title ~~test(task-body-guard, event-log, archive-stale-results): regression tests for zero-coverage modules~~ test: regression tests for four zero-coverage modules (45 total) Jun 30, 2026

bassilkhilo-ag2 and others added 3 commits June 30, 2026 16:06

bassilkhilo-ag2 changed the title ~~test: regression tests for four zero-coverage modules (45 total)~~ test: regression tests for five zero-coverage modules (119 total) Jun 30, 2026

bassilkhilo-ag2 changed the title ~~test: regression tests for five zero-coverage modules (119 total)~~ test: regression tests for eight zero-coverage modules (119 total) Jun 30, 2026

test(remote-relay-bridge,discord-read): 31 tests for _valid_tid/_one_…

0160318

…line and boundary-check pure functions Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KXQJogmVSdYrKtYX1LwzwT

bassilkhilo-ag2 changed the title ~~test: regression tests for eight zero-coverage modules (119 total)~~ test: regression tests for eleven zero-coverage modules (159 total) Jun 30, 2026

bassilkhilo-ag2 changed the title ~~test: regression tests for eleven zero-coverage modules (159 total)~~ test: regression tests for twelve zero-coverage modules (172 total) Jun 30, 2026

bassilkhilo-ag2 changed the title ~~test: regression tests for twelve zero-coverage modules (172 total)~~ test: regression tests for eleven zero-coverage modules (163 new tests) Jun 30, 2026

bassilkhilo-ag2 changed the title ~~test: regression tests for eleven zero-coverage modules (163 new tests)~~ test: regression tests for twelve zero-coverage modules (178 new tests) Jun 30, 2026

Merge branch 'main' into test/task-body-guard-coverage

3c8139a

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

test: regression tests for twelve zero-coverage modules (178 new tests)#1865

test: regression tests for twelve zero-coverage modules (178 new tests)#1865
bassilkhilo-ag2 wants to merge 12 commits into
sonichi:mainfrom
bassilkhilo-ag2:test/task-body-guard-coverage

bassilkhilo-ag2 commented Jun 30, 2026 •

edited

Loading

Uh oh!

github-actions Bot commented Jul 1, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

Conversation

bassilkhilo-ag2 commented Jun 30, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Summary

Test plan

Uh oh!

github-actions Bot commented Jul 1, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

bassilkhilo-ag2 commented Jun 30, 2026 •

edited

Loading