WA1: add 35 tests for email watch, notifications, suggestions, digest

apps/api/tests/test_v11_email_notify.py

@@ -0,0 +1,682 @@
+"""Tests for v1.1: email watch, notifications, suggestions endpoints.
+
+Covers:
+- IMAP matching logic (company in subject, sender domain in URL, title keywords)
+- Classifier -> suggestion row
+- Noise dedupe (same from+subject+day)
+- Accept applies transition through guard
+- Dismiss marks suggestion
+- Webhook success/failure notification_log rows
+- LogChannel writes delivered=true
+- Daily digest payload shape
+- GET /suggestions, POST accept, POST dismiss, GET /notifications/log
+- FakeImap end-to-end poll
+"""
+
+from __future__ import annotations
+
+import email as email_mod
+from datetime import datetime, timedelta, timezone
+
+import pytest
+from fastapi.testclient import TestClient
+
+from app.config import DATABASE_URL
+from app.db import repo_app
+from app.imap_watch import (
+    FakeImap,
+    _build_snippet,
+    _extract_sender_domain,
+    _parse_email_message,
+    classify_email,
+    create_email_suggestion,
+    is_duplicate,
+    match_application,
+    poll_inbox,
+)
+from app.notify import (
+    LogChannel,
+    WebhookChannel,
+    get_channels,
+    list_notification_log,
+    reset_channels,
+    send_notification,
+    set_channels,
+)
+
+
+@pytest.fixture()
+def client():
+    from app.main import app
+    return TestClient(app)
+
+
+# --- Helpers ---
+
+def _make_email(raw_from: str, subject: str, body: str, date: str = "") -> bytes:
+    """Build raw email bytes for FakeImap."""
+    msg = email_mod.message_from_string(
+        f"From: {raw_from}\r\n"
+        f"Subject: {subject}\r\n"
+        f"Date: {date or 'Mon, 01 Jul 2026 10:00:00 +0000'}\r\n"
+        f"\r\n"
+        f"{body}"
+    )
+    return msg.as_bytes()
+
+
+def _create_app_in_state(state: str = "sent", company: str = "TechCorp", url: str = "https://techcorp.com/jobs/1") -> dict:
+    """Create a posting + application, force state via SQL."""
+    posting = repo_app.create_job_posting(
+        source="manual_url",
+        url=url,
+        company=company,
+        title="Senior Python Developer",
+        location="Malmo",
+        description="",
+        raw={},
+    )
+    app_row = repo_app.create_application(posting["id"])
+    if state != "discovered":
+        repo_app.update_application_score(app_row["id"], 80, {"factors": {}})
+        if state in ("approved", "rejected"):
+            repo_app.update_application_state(app_row["id"], "scored")
+            repo_app.update_application_state(app_row["id"], state)
+        elif state == "sent":
+            repo_app.update_application_state(app_row["id"], "scored")
+            repo_app.update_application_state(app_row["id"], "approved")
+            repo_app.update_application_state(app_row["id"], "drafting")
+            # Bypass guard for test
+            from app.db import execute
+            execute(
+                "UPDATE application SET state = 'sent', last_activity_at = now() WHERE id = %s",
+                (app_row["id"],),
+            )
+        elif state == "interviewing":
+            repo_app.update_application_state(app_row["id"], "scored")
+            repo_app.update_application_state(app_row["id"], "approved")
+            repo_app.update_application_state(app_row["id"], "drafting")
+            from app.db import execute
+            execute(
+                "UPDATE application SET state = 'sent' WHERE id = %s",
+                (app_row["id"],),
+            )
+            execute(
+                "UPDATE application SET state = 'interviewing' WHERE id = %s",
+                (app_row["id"],),
+            )
+    return app_row
+
+
+# ========================================================================
+# IMAP matching logic (3 tests)
+# ========================================================================
+
+class TestImapMatching:
+    def test_match_by_company_name_in_subject(self):
+        """Email subject contains company name -> match."""
+        apps = [
+            {"id": "app1", "state": "sent", "company": "TechCorp", "title": "Python Dev", "url": "https://example.com"},
+        ]
+        result = match_application(
+            "recruiter@gmail.com",
+            "Interview at TechCorp",
+            "Please come for an interview.",
+            apps,
+        )
+        assert result is not None
+        assert result["id"] == "app1"
+
+    def test_match_by_sender_domain_in_url(self):
+        """Sender domain matches the posting URL -> match."""
+        apps = [
+            {"id": "app2", "state": "sent", "company": "Unknown", "title": "Dev", "url": "https://techcorp.com/careers/1"},
+        ]
+        result = match_application(
+            "hr@techcorp.com",
+            "Your application",
+            "We reviewed your application.",
+            apps,
+        )
+        assert result is not None
+        assert result["id"] == "app2"
+
+    def test_match_by_title_keywords(self):
+        """Email subject contains 2+ title words -> match."""
+        apps = [
+            {"id": "app3", "state": "interviewing", "company": "SomeCompany", "title": "Senior Python Developer", "url": "https://other.com"},
+        ]
+        result = match_application(
+            "someone@other.com",
+            "Senior Python position update",
+            "Regarding the developer role.",
+            apps,
+        )
+        assert result is not None
+        assert result["id"] == "app3"
+
+    def test_no_match_wrong_state(self):
+        """Applications in discovered state are not matched."""
+        apps = [
+            {"id": "app4", "state": "discovered", "company": "TechCorp", "title": "Dev", "url": "https://techcorp.com"},
+        ]
+        result = match_application(
+            "hr@techcorp.com",
+            "Interview at TechCorp",
+            "Come for an interview.",
+            apps,
+        )
+        assert result is None
+
+    def test_no_match_unrelated_email(self):
+        """Email unrelated to any application -> no match."""
+        apps = [
+            {"id": "app5", "state": "sent", "company": "TechCorp", "title": "Dev", "url": "https://techcorp.com"},
+        ]
+        result = match_application(
+            "newsletter@spam.com",
+            "Buy now!",
+            "Special offer for you.",
+            apps,
+        )
+        assert result is None
+
+
+# ========================================================================
+# Classifier -> row (3 tests)
+# ========================================================================
+
+class TestClassifierToRow:
+    def test_classify_returns_interview_invite(self):
+        """classify_email returns interview_invite from mock."""
+        result = classify_email("Interview invitation", "Please come for an interview next week.")
+        assert result["classification"] == "interview_invite"
+        assert result["state_proposal"] == "interviewing"
+        assert "reason" in result
+
+    def test_classify_falls_back_on_invalid_classification(self):
+        """Invalid classification from LLM falls back to noise."""
+        import app.llm as llm_mod
+        original = llm_mod.MOCK_OUTPUTS.get("email_classify", {}).copy()
+        try:
+            llm_mod.MOCK_OUTPUTS["email_classify"] = {"classification": "bogus", "state_proposal": None, "reason": "test"}
+            result = classify_email("test", "test")
+            assert result["classification"] == "noise"
+        finally:
+            llm_mod.MOCK_OUTPUTS["email_classify"] = original
+
+    def test_create_email_suggestion_row(self):
+        """create_email_suggestion inserts a row correctly."""
+        now = datetime.now(timezone.utc)
+        suggestion = create_email_suggestion(
+            application_id=None,
+            mailbox_from="hr@example.com",
+            subject="Interview invite",
+            snippet="Please come for an interview.",
+            classification="interview_invite",
+            state_proposal="interviewing",
+            received_at=now,
+        )
+        assert suggestion["id"] is not None
+        assert suggestion["mailbox_from"] == "hr@example.com"
+        assert suggestion["classification"] == "interview_invite"
+        assert suggestion["status"] == "pending"
+        assert suggestion["state_proposal"] == "interviewing"
+
+
+# ========================================================================
+# Noise dedupe (3 tests)
+# ========================================================================
+
+class TestNoiseDedupe:
+    def test_duplicate_detected_same_day(self):
+        """Same from+subject+day is flagged as duplicate."""
+        now = datetime.now(timezone.utc)
+        create_email_suggestion(
+            application_id=None,
+            mailbox_from="hr@example.com",
+            subject="Interview",
+            snippet="Come for an interview.",
+            classification="interview_invite",
+            state_proposal="interviewing",
+            received_at=now,
+        )
+        assert is_duplicate("hr@example.com", "Interview", now)
+
+    def test_different_subject_not_duplicate(self):
+        """Different subject -> not duplicate."""
+        now = datetime.now(timezone.utc)
+        create_email_suggestion(
+            application_id=None,
+            mailbox_from="hr@example.com",
+            subject="Interview",
+            snippet="Come.",
+            classification="interview_invite",
+            state_proposal="interviewing",
+            received_at=now,
+        )
+        assert not is_duplicate("hr@example.com", "Different Subject", now)
+
+    def test_different_sender_not_duplicate(self):
+        """Different sender -> not duplicate."""
+        now = datetime.now(timezone.utc)
+        create_email_suggestion(
+            application_id=None,
+            mailbox_from="hr@example.com",
+            subject="Interview",
+            snippet="Come.",
+            classification="interview_invite",
+            state_proposal="interviewing",
+            received_at=now,
+        )
+        assert not is_duplicate("other@example.com", "Interview", now)
+
+
+# ========================================================================
+# Accept applies transition through guard (3 tests)
+# ========================================================================
+
+class TestAcceptSuggestion:
+    def test_accept_applies_transition(self, client):
+        """POST /suggestions/{id}/accept transitions app from sent to interviewing."""
+        app_row = _create_app_in_state("sent", company="TechCorp")
+        suggestion = create_email_suggestion(
+            application_id=app_row["id"],
+            mailbox_from="hr@techcorp.com",
+            subject="Interview at TechCorp",
+            snippet="Please come in for an interview.",
+            classification="interview_invite",
+            state_proposal="interviewing",
+            received_at=datetime.now(timezone.utc),
+        )
+
+        resp = client.post(f"/api/suggestions/{suggestion['id']}/accept")
+        assert resp.status_code == 200
+        assert resp.json()["status"] == "accepted"
+
+        # Verify application state changed
+        updated_app = repo_app.get_application(app_row["id"])
+        assert updated_app["state"] == "interviewing"
+
+    def test_accept_invalid_transition_409(self, client):
+        """Accept with invalid transition (e.g. discovered -> interviewing) returns 409."""
+        posting = repo_app.create_job_posting(
+            source="manual_url", url="https://example.com/bad/1",
+            company="X", title="X", location=None, description="", raw={},
+        )
+        app_row = repo_app.create_application(posting["id"])
+
+        suggestion = create_email_suggestion(
+            application_id=app_row["id"],
+            mailbox_from="hr@example.com",
+            subject="Test",
+            snippet="Test",
+            classification="interview_invite",
+            state_proposal="interviewing",
+            received_at=datetime.now(timezone.utc),
+        )
+
+        resp = client.post(f"/api/suggestions/{suggestion['id']}/accept")
+        assert resp.status_code == 409
+        assert "invalid_transition" in str(resp.json()["detail"])
+
+    def test_accept_404_nonexistent(self, client):
+        """Accept on nonexistent suggestion -> 404."""
+        resp = client.post("/api/suggestions/00000000-0000-0000-0000-000000000000/accept")
+        assert resp.status_code == 404
+
+    def test_accept_already_accepted_409(self, client):
+        """Accept on already accepted suggestion -> 409."""
+        app_row = _create_app_in_state("sent")
+        suggestion = create_email_suggestion(
+            application_id=app_row["id"],
+            mailbox_from="hr@example.com",
+            subject="Test",
+            snippet="Test",
+            classification="interview_invite",
+            state_proposal="interviewing",
+            received_at=datetime.now(timezone.utc),
+        )
+        client.post(f"/api/suggestions/{suggestion['id']}/accept")
+        resp = client.post(f"/api/suggestions/{suggestion['id']}/accept")
+        assert resp.status_code == 409
+
+
+# ========================================================================
+# Dismiss suggestion (2 tests)
+# ========================================================================
+
+class TestDismissSuggestion:
+    def test_dismiss_marks_as_dismissed(self, client):
+        """POST /suggestions/{id}/dismiss marks as dismissed."""
+        suggestion = create_email_suggestion(
+            application_id=None,
+            mailbox_from="hr@example.com",
+            subject="Test",
+            snippet="Test",
+            classification="question",
+            state_proposal=None,
+            received_at=datetime.now(timezone.utc),
+        )
+
+        resp = client.post(f"/api/suggestions/{suggestion['id']}/dismiss")
+        assert resp.status_code == 200
+        assert resp.json()["status"] == "dismissed"
+
+    def test_dismiss_404_nonexistent(self, client):
+        """Dismiss nonexistent -> 404."""
+        resp = client.post("/api/suggestions/00000000-0000-0000-0000-000000000000/dismiss")
+        assert resp.status_code == 404
+
+
+# ========================================================================
+# GET /suggestions (2 tests)
+# ========================================================================
+
+class TestGetSuggestions:
+    def test_get_suggestions_returns_pending(self, client):
+        """GET /suggestions returns only pending suggestions."""
+        create_email_suggestion(
+            application_id=None,
+            mailbox_from="a@example.com",
+            subject="Subject A",
+            snippet="Snippet A",
+            classification="interview_invite",
+            state_proposal="interviewing",
+            received_at=datetime.now(timezone.utc),
+        )
+        create_email_suggestion(
+            application_id=None,
+            mailbox_from="b@example.com",
+            subject="Subject B",
+            snippet="Snippet B",
+            classification="rejection",
+            state_proposal=None,
+            received_at=datetime.now(timezone.utc),
+        )
+
+        resp = client.get("/api/suggestions")
+        assert resp.status_code == 200
+        data = resp.json()
+        assert len(data) == 2
+        assert all(s["status"] == "pending" for s in data)
+
+    def test_get_suggestions_empty(self, client):
+        """GET /suggestions returns empty list when no suggestions."""
+        resp = client.get("/api/suggestions")
+        assert resp.status_code == 200
+        assert resp.json() == []
+
+
+# ========================================================================
+# Notification channels (6 tests)
+# ========================================================================
+
+class TestLogChannel:
+    def test_log_channel_writes_delivered_true(self):
+        """LogChannel writes notification_log with delivered=true."""
+        reset_channels()
+        ch = LogChannel()
+        result = ch.send("daily_digest", "Test digest", {"count": 5})
+        assert result is True
+
+        logs = list_notification_log(limit=10)
+        assert len(logs) >= 1
+        latest = logs[0]
+        assert latest["channel"] == "log"
+        assert latest["kind"] == "daily_digest"
+        assert latest["delivered"] is True
+        assert latest["error"] is None
+
+    def test_send_notification_log_channel(self):
+        """send_notification via LogChannel creates a log entry."""
+        reset_channels()
+        set_channels([LogChannel()])
+        send_notification("email_suggestion", "Interview invite from TechCorp", {"id": "test"})
+        logs = list_notification_log(limit=10)
+        assert len(logs) >= 1
+        assert logs[0]["kind"] == "email_suggestion"
+        assert logs[0]["delivered"] is True
+        reset_channels()
+
+
+class TestWebhookChannel:
+    def test_webhook_success_2xx(self, monkeypatch):
+        """WebhookChannel with 2xx response writes delivered=true."""
+        class FakeResponse:
+            status_code = 200
+            text = "OK"
+
+        class FakeClient:
+            @staticmethod
+            def post(url, json=None, timeout=None):
+                assert url == "https://hook.example.com/notify"
+                return FakeResponse()
+
+        import httpx
+        monkeypatch.setattr(httpx, "post", FakeClient.post)
+
+        reset_channels()
+        ch = WebhookChannel(url="https://hook.example.com/notify")
+        result = ch.send("daily_digest", "Digest", {"count": 3})
+        assert result is True
+
+        logs = list_notification_log(limit=10)
+        latest = logs[0]
+        assert latest["channel"] == "webhook"
+        assert latest["delivered"] is True
+        assert latest["error"] is None
+        reset_channels()
+
+    def test_webhook_failure_non_2xx(self, monkeypatch):
+        """WebhookChannel with non-2xx writes delivered=false with error."""
+        class FakeResponse:
+            status_code = 500
+            text = "Internal Server Error"
+
+        class FakeClient:
+            @staticmethod
+            def post(url, json=None, timeout=None):
+                return FakeResponse()
+
+        import httpx
+        monkeypatch.setattr(httpx, "post", FakeClient.post)
+
+        reset_channels()
+        ch = WebhookChannel(url="https://hook.example.com/notify")
+        result = ch.send("daily_digest", "Digest", {"count": 3})
+        assert result is False
+
+        logs = list_notification_log(limit=10)
+        latest = logs[0]
+        assert latest["channel"] == "webhook"
+        assert latest["delivered"] is False
+        assert latest["error"] is not None
+        assert "500" in latest["error"]
+        reset_channels()
+
+    def test_webhook_exception_writes_error(self, monkeypatch):
+        """WebhookChannel with connection exception writes delivered=false."""
+        def raise_exc(url, json=None, timeout=None):
+            raise ConnectionError("Connection refused")
+
+        import httpx
+        monkeypatch.setattr(httpx, "post", raise_exc)
+
+        reset_channels()
+        ch = WebhookChannel(url="https://hook.example.com/notify")
+        result = ch.send("daily_digest", "Digest", {"count": 1})
+        assert result is False
+
+        logs = list_notification_log(limit=10)
+        latest = logs[0]
+        assert latest["delivered"] is False
+        assert "Connection refused" in (latest["error"] or "")
+        reset_channels()
+
+    def test_webhook_no_url_writes_error(self):
+        """WebhookChannel with no URL writes delivered=false with config error."""
+        reset_channels()
+        ch = WebhookChannel(url="")
+        result = ch.send("test", "test", {})
+        assert result is False
+
+        logs = list_notification_log(limit=10)
+        latest = logs[0]
+        assert latest["delivered"] is False
+        assert "not configured" in (latest["error"] or "")
+        reset_channels()
+
+
+# ========================================================================
+# Notification log endpoint (2 tests)
+# ========================================================================
+
+class TestNotificationLogEndpoint:
+    def test_get_notifications_log(self, client):
+        """GET /notifications/log returns entries."""
+        reset_channels()
+        set_channels([LogChannel()])
+        send_notification("daily_digest", "Test", {"count": 1})
+        reset_channels()
+
+        resp = client.get("/api/notifications/log")
+        assert resp.status_code == 200
+        data = resp.json()
+        assert len(data) >= 1
+        assert "channel" in data[0]
+        assert "kind" in data[0]
+        assert "delivered" in data[0]
+
+    def test_get_notifications_log_empty(self, client):
+        """GET /notifications/log returns empty when no entries."""
+        resp = client.get("/api/notifications/log")
+        assert resp.status_code == 200
+        assert resp.json() == []
+
+
+# ========================================================================
+# Digest payload shape (2 tests)
+# ========================================================================
+
+class TestDigestPayload:
+    def test_daily_digest_notification_text(self):
+        """Daily digest notification contains expected text fields."""
+        reset_channels()
+        set_channels([LogChannel()])
+        send_notification("daily_digest", "Daily Digest\nScored applications: 5", {
+            "digest_count": 5,
+            "nudge_count": 2,
+            "pending_approvals": 1,
+        })
+        logs = list_notification_log(limit=10)
+        entry = logs[0]
+        assert entry["kind"] == "daily_digest"
+        payload = entry["payload"]
+        assert "text" in payload
+        assert "Daily Digest" in payload["text"]
+        assert payload.get("digest_count") == 5
+        reset_channels()
+
+    def test_daily_digest_payload_has_counts(self):
+        """Digest payload includes digest_count, nudge_count, pending_approvals."""
+        reset_channels()
+        set_channels([LogChannel()])
+        send_notification("daily_digest", "text", {
+            "digest_count": 3,
+            "nudge_count": 1,
+            "pending_approvals": 0,
+        })
+        logs = list_notification_log(limit=10)
+        payload = logs[0]["payload"]
+        assert payload["digest_count"] == 3
+        assert payload["nudge_count"] == 1
+        assert payload["pending_approvals"] == 0
+        reset_channels()
+
+
+# ========================================================================
+# FakeImap end-to-end poll (3 tests)
+# ========================================================================
+
+class TestFakeImapPoll:
+    def test_poll_creates_suggestion_for_matching_email(self):
+        """FakeImap poll creates a suggestion when email matches an application."""
+        app_row = _create_app_in_state("sent", company="TechCorp", url="https://techcorp.com/jobs/1")
+
+        raw_email = _make_email(
+            "hr@techcorp.com",
+            "Interview at TechCorp",
+            "Please come for an interview next Tuesday.",
+        )
+        fake = FakeImap(messages=[(b"1", raw_email)])
+        created = poll_inbox(fake)
+        assert len(created) == 1
+        assert created[0]["classification"] == "interview_invite"
+        assert created[0]["mailbox_from"] == "hr@techcorp.com"
+        assert created[0]["application_id"] == app_row["id"]
+
+    def test_poll_skips_noise_emails(self):
+        """FakeImap poll skips noise classification (no suggestion created)."""
+        _create_app_in_state("sent", company="TechCorp")
+
+        # Mock email_classify to return noise
+        import app.llm as llm_mod
+        original = llm_mod.MOCK_OUTPUTS.get("email_classify", {}).copy()
+        try:
+            llm_mod.MOCK_OUTPUTS["email_classify"] = {
+                "classification": "noise",
+                "state_proposal": None,
+                "reason": "spam",
+            }
+            raw_email = _make_email(
+                "newsletter@spam.com",
+                "Buy our product",
+                "Special offer just for you!",
+            )
+            fake = FakeImap(messages=[(b"1", raw_email)])
+            created = poll_inbox(fake)
+            assert len(created) == 0
+        finally:
+            llm_mod.MOCK_OUTPUTS["email_classify"] = original
+
+    def test_poll_dedupe_skips_same_from_subject_day(self):
+        """FakeImap poll dedupes same from+subject+day."""
+        raw_email = _make_email(
+            "hr@example.com",
+            "Same Subject",
+            "Same body content.",
+        )
+        # First poll
+        fake1 = FakeImap(messages=[(b"1", raw_email)])
+        created1 = poll_inbox(fake1)
+        assert len(created1) == 1
+
+        # Second poll with same message -> dedupe
+        fake2 = FakeImap(messages=[(b"1", raw_email)])
+        created2 = poll_inbox(fake2)
+        assert len(created2) == 0  # deduped
+
+
+# ========================================================================
+# Email parsing helpers (2 tests)
+# ========================================================================
+
+class TestEmailParsing:
+    def test_extract_sender_domain(self):
+        """Extract domain from From header."""
+        assert _extract_sender_domain("John Doe <hr@techcorp.com>") == "techcorp.com"
+        assert _extract_sender_domain("noreply@example.org") == "example.org"
+        assert _extract_sender_domain("") == ""
+
+    def test_build_snippet_truncates(self):
+        """Snippet is truncated to max_len."""
+        long_body = "A" * 500
+        snippet = _build_snippet(long_body, max_len=50)
+        assert len(snippet) <= 53  # 50 + "..."
+        assert snippet.endswith("...")
+
+    def test_build_snippet_short_body(self):
+        """Short body is not truncated."""
+        snippet = _build_snippet("Hello", max_len=300)
+        assert snippet == "Hello"