Initial deploy setup

tests/integration/test_applications.py

@@ -0,0 +1,191 @@
+from __future__ import annotations
+
+import csv
+import io
+import json
+import zipfile
+from datetime import timedelta
+from decimal import Decimal
+
+import pytest
+from django.contrib.auth import get_user_model
+from django.urls import reverse
+from django.utils import timezone
+
+from apps.jobs.models import Application, ApplicationTimelineEvent, JobSourceAlias, ScoreRun
+from apps.jobs.services.feedback import record_feedback
+
+
+@pytest.mark.integration
+@pytest.mark.django_db
+def test_applied_feedback_creates_sanitized_application_snapshot_and_timeline(job, profile, source, user):
+    JobSourceAlias.objects.create(
+        job=job,
+        source=source,
+        url=job.canonical_url,
+        canonical_url=job.canonical_url,
+        external_id="ext-applied-1",
+        source_title="Applied page",
+        source_employer=job.employer_name,
+        is_canonical=True,
+        extraction_method="manual-test",
+    )
+    record_feedback(user=user, job=job, action="applied")
+
+    application = Application.objects.get(user=user, job=job)
+    snapshot = application.snapshot
+    assert snapshot["schema_version"] == "1.0.0"
+    assert snapshot["snapshot_kind"] == "application"
+    assert snapshot["title"] == job.original_title
+    assert snapshot["job"]["id"] == str(job.pk)
+    assert snapshot["job"]["description_html"] == job.description_html_sanitized
+    assert snapshot["sources"][0]["url"] == job.canonical_url
+    assert snapshot["sources"][0]["is_canonical"] is True
+    assert snapshot["scores"] is None
+
+    event_types = set(ApplicationTimelineEvent.objects.filter(application=application).values_list("event_type", flat=True))
+    assert ApplicationTimelineEvent.EventType.CREATED in event_types
+    assert ApplicationTimelineEvent.EventType.SNAPSHOT_CAPTURED in event_types
+
+
+@pytest.mark.integration
+@pytest.mark.django_db
+def test_application_update_creates_timeline_events_for_status_notes_and_contact(client, job, profile, source, user):
+    record_feedback(user=user, job=job, action="applied")
+    application = Application.objects.get(user=user, job=job)
+
+    client.force_login(user)
+    response = client.post(
+        reverse("jobs:application-edit", kwargs={"pk": application.pk}),
+        {
+            "status": Application.Status.INTERVIEW,
+            "contact_name": "Sofie Van Hecke",
+            "contact_email": "recruiter@example.invalid",
+            "notes": "Eerste follow-up gepland voor vrijdag.",
+        },
+    )
+    assert response.status_code == 302
+
+    application.refresh_from_db()
+    assert application.status == Application.Status.INTERVIEW
+    assert application.contact_name == "Sofie Van Hecke"
+    assert application.contact_email == "recruiter@example.invalid"
+
+    event_types = set(ApplicationTimelineEvent.objects.filter(application=application).values_list("event_type", flat=True))
+    assert ApplicationTimelineEvent.EventType.STATUS_CHANGED in event_types
+    assert ApplicationTimelineEvent.EventType.NOTES_UPDATED in event_types
+    assert ApplicationTimelineEvent.EventType.CONTACT_UPDATED in event_types
+
+
+@pytest.mark.integration
+@pytest.mark.django_db
+def test_application_export_generates_sanitized_zip_payload(client, job, source, user, profile):
+    JobSourceAlias.objects.create(
+        job=job,
+        source=source,
+        url=job.canonical_url,
+        canonical_url=job.canonical_url,
+        external_id="ext-export-1",
+        source_title="Export page",
+        source_employer=job.employer_name,
+        is_canonical=True,
+        extraction_method="manual-test",
+    )
+    ScoreRun.objects.create(
+        job=job,
+        profile=profile,
+        profile_version=1,
+        score=Decimal("84.20"),
+        confidence=Decimal("0.71"),
+        recommendation=ScoreRun.Recommendation.POSSIBLE,
+        components={"content": 0.41},
+        positives=["python"],
+        concerns=["travel"],
+        hard_exclusions=[],
+        evidence={"notes": "Voorbeeldscore"},
+        model_version="testsuite",
+        prompt_version="1.0.0",
+    )
+    record_feedback(user=user, job=job, action="applied")
+    record_feedback(user=user, job=job, action="applied")
+    application = Application.objects.get(user=user, job=job)
+
+    application.notes = "Vaste follow-up notitie."
+    application.save(update_fields=["notes", "updated_at"])
+
+    client.force_login(user)
+    response = client.get(reverse("jobs:application-export", kwargs={"pk": application.pk}))
+    assert response.status_code == 200
+    assert response["Content-Type"] == "application/zip"
+    assert 'filename="' in response["Content-Disposition"]
+
+    filename = response["Content-Disposition"].partition("filename=")[2].strip().strip('"')
+    assert filename.startswith(f"application-{application.pk}-")
+    assert filename.endswith(".zip")
+    assert ".." not in filename and "/" not in filename and "\\" not in filename
+
+    archive = zipfile.ZipFile(io.BytesIO(response.content))
+    members = sorted(archive.namelist())
+    assert members == ["application.json", "application_print.html", "timeline.csv"]
+
+    application_payload = json.loads(archive.read("application.json").decode("utf-8"))
+    assert application_payload["application"]["id"] == str(application.pk)
+    assert application_payload["application"]["job_id"] == str(job.pk)
+    assert application_payload["application"]["status"] == Application.Status.APPLIED
+    assert application_payload["snapshot"]["scores"]["score"] == 84.2
+    assert application_payload["snapshot"]["sources"][0]["url"] == job.canonical_url
+
+    csv_rows = list(csv.DictReader(io.StringIO(archive.read("timeline.csv").decode("utf-8"))))
+    assert {row["type"] for row in csv_rows} >= {ApplicationTimelineEvent.EventType.CREATED, ApplicationTimelineEvent.EventType.SNAPSHOT_CAPTURED}
+
+    html = archive.read("application_print.html").decode("utf-8")
+    assert "<html" in html.lower()
+    assert "<link " not in html.lower()
+    assert "<script" not in html.lower()
+
+    assert "raw body" not in json.dumps(application_payload)
+
+    response_json = client.get(reverse("jobs:application-print", kwargs={"pk": application.pk}))
+    assert response_json.status_code == 200
+    assert response_json["Content-Type"] == "text/html; charset=utf-8"
+
+
+@pytest.mark.integration
+@pytest.mark.django_db
+def test_application_export_and_delete_are_scoped_and_idempotent(client, job, user, source):
+    other_user = get_user_model().objects.create_user(
+        username="other",
+        email="other@example.invalid",
+        password="correct-horse-battery-staple",
+    )
+
+    record_feedback(user=user, job=job, action="applied")
+    owner_application = Application.objects.get(user=user, job=job)
+    other_application = Application.objects.create(
+        user=other_user,
+        job=job,
+        status=Application.Status.APPLIED,
+        applied_at=timezone.now() - timedelta(days=1),
+        follow_up_date=timezone.now().date(),
+    )
+
+    client.force_login(user)
+    assert client.get(
+        reverse("jobs:application-export", kwargs={"pk": other_application.pk})
+    ).status_code == 404
+    assert client.post(
+        reverse("jobs:application-delete", kwargs={"pk": other_application.pk})
+    ).status_code == 404
+
+    delete_response = client.post(
+        reverse("jobs:application-delete", kwargs={"pk": owner_application.pk}),
+        follow=True,
+    )
+    assert delete_response.status_code == 200
+    assert not Application.objects.filter(pk=owner_application.pk).exists()
+
+    repeat_delete = client.post(
+        reverse("jobs:application-delete", kwargs={"pk": owner_application.pk}),
+        follow=True,
+    )
+    assert repeat_delete.status_code == 200

tests/integration/test_dedupe_lifecycle_feedback.py

@@ -0,0 +1,147 @@
+from datetime import timedelta
+
+import pytest
+from django.utils import timezone
+
+from apps.jobs.models import Application, Feedback, JobPosting, JobSourceAlias
+from apps.jobs.services.dedupe import candidate_similarity, find_existing_job, text_similarity
+from apps.jobs.services.feedback import record_feedback
+from apps.jobs.services.lifecycle import update_lifecycle
+from apps.jobs.services.normalization import CanonicalJobDraft
+from apps.sources.models import RawDocument
+
+
+def _draft(job, **overrides):
+    values = {
+        "source_url": job.canonical_url,
+        "canonical_url": job.canonical_url,
+        "external_id": "external-1",
+        "title": job.original_title,
+        "normalized_title": job.normalized_title,
+        "job_family": job.job_family,
+        "employer_name": job.employer_name,
+        "employer_domain": job.employer.domain,
+        "location_text": job.raw_location,
+        "region": job.region,
+        "municipality": job.municipality,
+        "postal_code": "3500",
+        "country": "BE",
+        "workplace_type": job.workplace_type,
+        "employment_types": job.employment_types,
+        "language": "nl",
+        "description_html": "",
+        "description_text": job.description_text,
+        "date_posted": None,
+        "valid_through": None,
+        "compensation": {},
+        "skills_required": job.skills_required,
+        "skills_preferred": job.skills_preferred,
+        "content_hash": job.content_hash,
+        "canonical_key": job.canonical_key,
+    }
+    values.update(overrides)
+    return CanonicalJobDraft(**values)
+
+
+@pytest.mark.django_db
+def test_dedupe_exact_fuzzy_and_new(job, source):
+    document = RawDocument.objects.create(
+        source=source,
+        kind=RawDocument.Kind.HTML,
+        content_hash="e" * 64,
+    )
+    JobSourceAlias.objects.create(
+        job=job,
+        source=source,
+        raw_document=document,
+        url=job.canonical_url,
+        canonical_url=job.canonical_url,
+        external_id="external-1",
+    )
+    exact = find_existing_job(_draft(job))
+    assert exact.job == job
+    assert exact.reason == "exact_external_id"
+
+    fuzzy_draft = _draft(
+        job,
+        external_id="",
+        canonical_url="https://jobs.example.org/other-id",
+        canonical_key="f" * 64,
+    )
+    fuzzy = find_existing_job(fuzzy_draft, threshold=0.8)
+    assert fuzzy.job == job
+    assert fuzzy.reason == "fuzzy_strong"
+    assert candidate_similarity(job, fuzzy_draft) >= 0.8
+    assert text_similarity("", "x") == 0
+
+    new = find_existing_job(
+        _draft(
+            job,
+            external_id="",
+            canonical_url="https://new.example.org/job",
+            canonical_key="1" * 64,
+            title="Chef kok",
+            normalized_title="chef kok",
+            employer_name="Restaurant",
+            employer_domain="restaurant.example",
+            description_text="Koken in een restaurant",
+        )
+    )
+    assert new.job is None
+    assert new.reason == "new"
+
+
+@pytest.mark.django_db
+def test_applied_feedback_creates_application_and_follow_up(user, profile, job):
+    feedback = record_feedback(user=user, job=job, action=Feedback.Action.APPLIED, reason="goed")
+    assert feedback.profile == profile
+    application = Application.objects.get(user=user, job=job)
+    assert application.status == Application.Status.APPLIED
+    assert application.snapshot["title"] == job.original_title
+    assert application.follow_up_date == timezone.localdate() + timedelta(days=7)
+
+    application.status = Application.Status.PREPARING
+    application.save(update_fields=["status"])
+    record_feedback(user=user, job=job, action=Feedback.Action.APPLIED)
+    application.refresh_from_db()
+    assert application.status == Application.Status.APPLIED
+
+
+@pytest.mark.django_db
+def test_lifecycle_transitions(employer):
+    now = timezone.now()
+    expired = JobPosting.objects.create(
+        employer=employer,
+        original_title="Expired",
+        normalized_title="expired",
+        canonical_key="2" * 64,
+        content_hash="3" * 64,
+        valid_through=now - timedelta(minutes=1),
+        status=JobPosting.Status.ACTIVE,
+    )
+    uncertain = JobPosting.objects.create(
+        employer=employer,
+        original_title="Uncertain",
+        normalized_title="uncertain",
+        canonical_key="4" * 64,
+        content_hash="5" * 64,
+        last_seen=now - timedelta(days=5),
+        status=JobPosting.Status.ACTIVE,
+    )
+    removed = JobPosting.objects.create(
+        employer=employer,
+        original_title="Removed",
+        normalized_title="removed",
+        canonical_key="6" * 64,
+        content_hash="7" * 64,
+        last_seen=now - timedelta(days=20),
+        status=JobPosting.Status.UNCERTAIN,
+    )
+    result = update_lifecycle()
+    assert result == {"expired": 1, "uncertain": 1, "removed": 1}
+    expired.refresh_from_db()
+    uncertain.refresh_from_db()
+    removed.refresh_from_db()
+    assert expired.status == JobPosting.Status.EXPIRED
+    assert uncertain.status == JobPosting.Status.UNCERTAIN
+    assert removed.status == JobPosting.Status.REMOVED

tests/integration/test_discovery_command.py

@@ -0,0 +1,86 @@
+from io import StringIO
+
+import pytest
+from django.core.management import call_command
+
+from apps.sources.models import Source
+
+
+@pytest.mark.django_db
+def test_discover_sources_command_runs_for_all_modes_and_is_replayable():
+    out = StringIO()
+    call_command(
+        "discover_sources",
+        "html",
+        "--input",
+        "fixtures/discovery/html_candidate_discovery.html",
+        "--base-url",
+        "https://example.org/",
+        stdout=out,
+    )
+    call_command(
+        "discover_sources",
+        "sitemap",
+        "--input",
+        "fixtures/discovery/sitemap_index.xml",
+        "--base-url",
+        "https://jobs.example.org/",
+        stdout=out,
+    )
+    call_command(
+        "discover_sources",
+        "feed",
+        "--input",
+        "fixtures/discovery/feed_jobs.xml",
+        "--base-url",
+        "https://jobs.example.org/feed/jobs",
+        stdout=out,
+    )
+    call_command(
+        "discover_sources",
+        "email",
+        "--input",
+        "fixtures/emails/sample_alert.eml",
+        stdout=out,
+    )
+
+    source_types = {
+        (source.domain, source.source_type)
+        for source in Source.objects.filter(status=Source.Status.CANDIDATE)
+    }
+    assert ("example.org", Source.Type.EMPLOYER) in source_types
+    assert ("jobs.example.org", Source.Type.EMPLOYER) in source_types
+    assert ("example.org", Source.Type.RSS) in source_types
+    assert ("jobs.example.org", Source.Type.SITEMAP) in source_types
+    assert ("jobs.example.org", Source.Type.RSS) in source_types
+    assert "Persist result" in out.getvalue()
+
+    out.seek(0)
+    out.truncate(0)
+    call_command(
+        "discover_sources",
+        "html",
+        "--input",
+        "fixtures/discovery/html_candidate_discovery.html",
+        "--base-url",
+        "https://example.org/",
+        stdout=out,
+    )
+    assert Source.objects.filter(status=Source.Status.CANDIDATE).count() == 5
+
+
+@pytest.mark.django_db
+def test_discover_sources_dry_run_keeps_db_empty():
+    out = StringIO()
+    call_command(
+        "discover_sources",
+        "feed",
+        "--input",
+        "fixtures/discovery/feed_jobs.xml",
+        "--base-url",
+        "https://jobs.example.org/feed/jobs",
+        "--dry-run",
+        stdout=out,
+    )
+    assert Source.objects.count() == 0
+    assert "Dry-run modus" in out.getvalue()

tests/integration/test_email_import_notifications.py

@@ -0,0 +1,72 @@
+from datetime import datetime
+from pathlib import Path
+from zoneinfo import ZoneInfo
+
+import pytest
+from django.core import mail
+from django.test import override_settings
+
+from apps.jobs.models import Feedback, ScoreRun
+from apps.jobs.services.scoring import score_and_save
+from apps.notifications.models import DigestOutbox
+from apps.notifications.services import build_digest_payload, create_daily_outbox, send_digest
+from apps.sources.models import EmailMessageRecord, RawDocument
+from apps.sources.services.email_import import ingest_email, message_identity
+
+
+@pytest.mark.django_db
+def test_email_import_is_idempotent(profile):
+    raw = Path("fixtures/emails/sample_alert.eml").read_bytes()
+    identity = message_identity(raw)
+    first = ingest_email(raw)
+    second = ingest_email(raw)
+    assert first.pk == second.pk
+    assert first.message_id == identity
+    assert EmailMessageRecord.objects.count() == 1
+    assert RawDocument.objects.filter(kind=RawDocument.Kind.EMAIL).count() == 1
+    assert len(first.links) == 2
+
+
+@pytest.mark.django_db
+def test_digest_payload_outbox_deduplication_and_send(profile, job):
+    score = score_and_save(job, profile)
+    score.recommendation = ScoreRun.Recommendation.STRONG
+    score.score = 92
+    score.save(update_fields=["recommendation", "score"])
+
+    local_tz = ZoneInfo(profile.timezone)
+    due = datetime(2026, 7, 20, 7, 35, tzinfo=local_tz)
+    payload = build_digest_payload(profile, local_date=due.date())
+    assert payload["strong_count"] == 1
+    assert payload["scores"][0]["title"] == job.original_title
+
+    with override_settings(DIGEST_RECIPIENT="digest@example.invalid"):
+        first = create_daily_outbox(profile, now=due)
+        second = create_daily_outbox(profile, now=due)
+    assert first is not None
+    assert second.pk == first.pk
+    assert first.status == DigestOutbox.Status.PENDING
+
+    send_digest(first)
+    first.refresh_from_db()
+    assert first.status == DigestOutbox.Status.SENT
+    assert first.sent_at is not None
+    assert len(mail.outbox) == 1
+    assert "Infrastructure Engineer" in mail.outbox[0].body
+
+
+@pytest.mark.django_db
+def test_digest_skips_outside_window_and_hidden_jobs(profile, job):
+    score_and_save(job, profile)
+    Feedback.objects.create(
+        user=profile.user,
+        profile=profile,
+        job=job,
+        action=Feedback.Action.HIDE,
+    )
+    payload = build_digest_payload(profile)
+    assert payload["scores"] == []
+
+    local_tz = ZoneInfo(profile.timezone)
+    too_early = datetime(2026, 7, 20, 6, 0, tzinfo=local_tz)
+    assert create_daily_outbox(profile, now=too_early) is None

tests/integration/test_import_geodata.py

@@ -0,0 +1,107 @@
+from __future__ import annotations
+
+from io import StringIO
+
+from django.core.management import call_command
+from django.core.management.base import CommandError
+import pytest
+
+from apps.jobs.models import ScoreRun
+from apps.jobs.services.scoring import score_and_save
+
+@pytest.mark.django_db
+def test_import_geodata_validate_only_and_import_command(tmp_path):
+    csv_path = tmp_path / "geodata.csv"
+    csv_path.write_text(
+        "postal_code,municipality,region,latitude,longitude\n"
+        "9000,Gent,Vlaams-Brabant,51.05,3.73\n"
+        "1040,Brussel,Brussels Hoofdstedelijk Gewest,50.85,4.35\n"
+    )
+    out = StringIO()
+    call_command(
+        "import_geodata",
+        str(csv_path),
+        "--source-name",
+        "local-test",
+        "--dataset-version",
+        "2026-01-01",
+        "--validate-only",
+        "--license-name",
+        "Test Dataset",
+        "--license-url",
+        "https://example.org/license",
+        stdout=out,
+    )
+    assert "Validatie geslaagd (2 rijen)." in out.getvalue()
+
+    out = StringIO()
+    call_command(
+        "import_geodata",
+        str(csv_path),
+        "--source-name",
+        "local-test",
+        "--dataset-version",
+        "2026-01-01",
+        "--license-name",
+        "Test Dataset",
+        "--license-url",
+        "https://example.org/license",
+        stdout=out,
+    )
+    assert "Geocodebron geimporteerd (4 lookuprijen)" in out.getvalue()
+
+
+@pytest.mark.django_db
+def test_import_geodata_rejects_half_license_metadata(tmp_path):
+    csv_path = tmp_path / "geodata.csv"
+    csv_path.write_text(
+        "postal_code,municipality,region,latitude,longitude\n9000,Gent,Vlaams-Brabant,51.05,3.73\n"
+    )
+    out = StringIO()
+        with pytest.raises(CommandError):
+            call_command(
+                "import_geodata",
+                str(csv_path),
+                "--source-name",
+            "local-test",
+            "--dataset-version",
+            "2026-01-01",
+                "--license-name",
+                "Test Dataset",
+                stdout=out,
+            )
+
+
+@pytest.mark.django_db
+def test_import_geodata_triggers_targeted_rescore(tmp_path, profile, job):
+    job.postal_code = "9000"
+    job.municipality = "Gent"
+    job.save(update_fields=["postal_code", "municipality"])
+
+    score_and_save(job, profile)
+    initial_scores = ScoreRun.objects.filter(profile=profile, job=job).count()
+
+    csv_path = tmp_path / "geodata.csv"
+    csv_path.write_text(
+        "postal_code,municipality,region,latitude,longitude\n9000,Gent,Vlaams-Brabant,51.05,3.73\n"
+    )
+    out = StringIO()
+    call_command(
+        "import_geodata",
+        str(csv_path),
+        "--source-name",
+        "local-test",
+        "--dataset-version",
+        "2026-01-01",
+        "--license-name",
+        "Test Dataset",
+        "--license-url",
+        "https://example.org/license",
+        stdout=out,
+    )
+    assert "scoreregels herberekend voor aangepaste geodata" in out.getvalue()
+    assert ScoreRun.objects.filter(profile=profile, job=job).count() == initial_scores + 1
+
+    latest = ScoreRun.objects.filter(profile=profile, job=job).order_by("-created_at").first()
+    assert latest is not None
+    assert latest.evidence["distance_km"] is not None

tests/integration/test_pipeline.py

@@ -0,0 +1,120 @@
+import hashlib
+from datetime import timedelta
+from pathlib import Path
+
+import pytest
+from django.utils import timezone
+
+from apps.jobs.models import Employer, JobPosting, JobSourceAlias, ScoreRun
+from apps.jobs.services.pipeline import process_raw_document
+from apps.sources.models import RawDocument, Source
+
+
+@pytest.mark.integration
+@pytest.mark.django_db
+def test_pipeline_is_idempotent_and_scores(source, profile):
+    content = Path("fixtures/pages/sample_jsonld_job.html").read_text(encoding="utf-8")
+    document = RawDocument.objects.create(
+        source=source,
+        url="https://jobs.example.org/vacatures/infrastructure-engineer",
+        final_url="https://jobs.example.org/vacatures/infrastructure-engineer",
+        kind=RawDocument.Kind.HTML,
+        content_type="text/html",
+        content_hash=hashlib.sha256(content.encode()).hexdigest(),
+        body_text=content,
+        byte_length=len(content.encode()),
+        retain_until=timezone.now() + timedelta(days=7),
+    )
+    first = process_raw_document(document)
+    second = process_raw_document(document)
+    assert first["created"] == 1
+    assert second["created"] == 0
+    assert JobPosting.objects.count() == 1
+    assert JobSourceAlias.objects.count() == 1
+    assert ScoreRun.objects.filter(profile=profile).count() == 2
+
+
+def test_pipeline_resolves_recruiter_alias_to_direct_employer():
+    source = Source.objects.create(
+        name="SmartRecruiters",
+        source_type=Source.Type.ATS,
+        base_url="https://jobs.smartrecruiters.com/example",
+        domain="jobs.smartrecruiters.com",
+        status=Source.Status.ACTIVE,
+        policy=Source.Policy.ALLOW,
+    )
+    employer = Employer.objects.create(
+        name="Example Public IT",
+        normalized_name="example public it",
+        domain="jobs.example.org",
+        is_direct_employer=True,
+        is_recruiter=False,
+        confidence=0.9,
+    )
+    direct_job = JobPosting.objects.create(
+        employer=employer,
+        original_title="Security Engineer",
+        normalized_title="security engineer",
+        canonical_url="https://jobs.example.org/vacatures/security-engineer",
+        canonical_key="8" * 64,
+        content_hash="e" * 64,
+        description_text="Beveiliger, hybride werk met support op locatie.",
+        raw_location="Antwerpen",
+        region="Antwerpen",
+        municipality="Antwerpen",
+        workplace_type=JobPosting.Workplace.HYBRID,
+        employment_types=["full_time"],
+        direct_employer=True,
+    )
+    content = """
+    <html lang="nl">
+      <head>
+        <meta charset="utf-8">
+        <title>Security Engineer | Example Public IT</title>
+        <link rel="canonical" href="https://jobs.smartrecruiters.com/example/security-engineer">
+        <script type="application/ld+json">
+        {
+          "@context": "https://schema.org",
+          "@type": "JobPosting",
+          "identifier": {"@type": "PropertyValue", "name": "Example Public IT", "value": "VR-DEMO-SEC-001"},
+          "title": "Security Engineer",
+          "description": "<p>Beveiliging, monitoring en hybride support.</p>",
+          "datePosted": "2026-07-20",
+          "validThrough": "2026-08-31T23:59:00+02:00",
+          "employmentType": ["FULL_TIME"],
+          "hiringOrganization": {"@type": "Organization", "name": "Example Public IT", "sameAs": "https://www.example.org"},
+          "jobLocation": {"@type": "Place", "address": {"@type": "PostalAddress", "streetAddress": "Voorbeeldstraat 1", "addressLocality": "Antwerpen", "addressRegion": "Antwerpen", "postalCode": "2000", "addressCountry": "BE"}},
+          "jobLocationType": "HYBRID",
+          "url": "/example/security-engineer"
+        }
+        </script>
+      </head>
+      <body><main><h1>Security Engineer</h1></main></body>
+    </html>
+    """.strip()
+    body_hash = hashlib.sha256(content.encode()).hexdigest()
+    document = RawDocument.objects.create(
+        source=source,
+        url="https://jobs.smartrecruiters.com/example/security-engineer",
+        final_url="https://jobs.smartrecruiters.com/example/security-engineer",
+        kind=RawDocument.Kind.HTML,
+        content_type="text/html",
+        content_hash=body_hash,
+        body_text=content,
+        byte_length=len(content.encode()),
+        retain_until=timezone.now() + timedelta(days=7),
+    )
+    result = process_raw_document(document)
+    direct_job.refresh_from_db()
+    assert direct_job.canonical_url == "https://jobs.example.org/vacatures/security-engineer"
+    assert result["created"] == 0
+    assert result["duplicates"] == 1
+    alias = JobSourceAlias.objects.get(job=direct_job, source=source)
+    payload = alias.payload.get("employer_resolution")
+    assert payload is not None
+    assert payload["reason"] == "resolved_direct_match"
+    assert payload["resolved_direct"] is True
+    assert (
+        payload["canonical_url"]
+        == "https://jobs.example.org/vacatures/security-engineer"
+    )

tests/integration/test_reminders.py

@@ -0,0 +1,132 @@
+from datetime import datetime, time
+from zoneinfo import ZoneInfo
+from decimal import Decimal
+
+import pytest
+from django.utils import timezone
+
+from apps.jobs.models import Application, Feedback, ScoreRun
+from apps.jobs.services.scoring import score_and_save
+from apps.notifications.models import ReminderOutbox
+from apps.notifications.services import (
+    create_closing_reminders,
+    create_follow_up_reminders,
+    create_top_match_reminders,
+)
+
+
+@pytest.mark.django_db
+def test_closing_reminder_is_idempotent_and_skips_hidden(profile, job):
+    zone = ZoneInfo(profile.timezone)
+    now = timezone.now().astimezone(zone)
+    job.valid_through = datetime(
+        year=now.year,
+        month=now.month,
+        day=now.day,
+        hour=20,
+        tzinfo=zone,
+    )
+    job.save(update_fields=["valid_through"])
+
+    Feedback.objects.create(
+        user=profile.user,
+        profile=profile,
+        job=job,
+        action=Feedback.Action.HIDE,
+    )
+    assert create_closing_reminders(profile, now=now) == []
+
+    Feedback.objects.filter(user=profile.user, profile=profile, job=job).delete()
+    first = create_closing_reminders(profile, now=now)[0]
+    second = create_closing_reminders(profile, now=now)[0]
+    assert first.pk == second.pk
+    assert first.status == ReminderOutbox.Status.PENDING
+
+
+@pytest.mark.django_db
+def test_follow_up_reminder_respects_configured_weekdays(profile, job):
+    zone = ZoneInfo(profile.timezone)
+    now = timezone.now().astimezone(zone)
+    follow_date = now.date()
+    Application.objects.create(
+        user=profile.user,
+        job=job,
+        follow_up_date=follow_date,
+        status=Application.Status.APPLIED,
+        notes="",
+    )
+    profile.follow_up_weekdays = [(follow_date.weekday() + 1) % 7]
+    profile.save(update_fields=["follow_up_weekdays"])
+    assert create_follow_up_reminders(profile, now=now) == []
+
+    profile.follow_up_weekdays = [follow_date.weekday()]
+    profile.save(update_fields=["follow_up_weekdays"])
+    reminders = create_follow_up_reminders(profile, now=now)
+    assert len(reminders) == 1
+    assert reminders[0].reminder_type == ReminderOutbox.ReminderType.FOLLOW_UP
+
+
+@pytest.mark.django_db
+def test_top_match_reminder_requires_threshold_and_profile_setting(profile, job):
+    profile.top_match_reminders_enabled = True
+    profile.top_match_threshold = 90
+    profile.save(update_fields=["top_match_reminders_enabled", "top_match_threshold"])
+    score = score_and_save(job, profile)
+    score.recommendation = ScoreRun.Recommendation.STRONG
+    score.score = 85
+    score.confidence = Decimal("0.9")
+    score.save(update_fields=["recommendation", "score", "confidence"])
+
+    assert create_top_match_reminders(profile) == []
+
+    score.score = 95
+    score.confidence = Decimal("0.90")
+    score.save(update_fields=["score", "confidence"])
+    score.confidence = Decimal("0.84")
+    score.save(update_fields=["confidence"])
+    assert create_top_match_reminders(profile) == []
+
+    score.confidence = Decimal("0.86")
+    score.save(update_fields=["confidence"])
+    reminders = create_top_match_reminders(profile)
+    assert len(reminders) == 1
+    assert reminders[0].reminder_type == ReminderOutbox.ReminderType.TOP_MATCH
+
+    profile.top_match_reminders_enabled = False
+    profile.save(update_fields=["top_match_reminders_enabled"])
+    assert create_top_match_reminders(profile) == []
+
+
+@pytest.mark.django_db
+def test_reminder_obeys_quiet_hours_and_empty_recipient(profile, job):
+    zone = ZoneInfo(profile.timezone)
+    now = timezone.now().astimezone(zone).replace(hour=6, minute=30, second=0, microsecond=0)
+    job.valid_through = datetime(
+        year=now.year,
+        month=now.month,
+        day=now.day,
+        hour=18,
+        tzinfo=zone,
+    )
+    job.save(update_fields=["valid_through"])
+
+    profile.user.email = ""
+    profile.user.save(update_fields=["email"])
+    assert create_closing_reminders(profile, now=now)[0].status == ReminderOutbox.Status.SKIPPED
+
+    profile.user.email = "reminder@example.invalid"
+    profile.user.save(update_fields=["email"])
+    reminder = create_closing_reminders(profile, now=now)[0]
+    assert reminder.status == ReminderOutbox.Status.PENDING
+    assert reminder.scheduled_for.timetz().replace(tzinfo=None) == time(7, 0)
+
+
+@pytest.mark.django_db
+def test_reminder_service_handles_dst_dates(profile, job):
+    zone = ZoneInfo(profile.timezone)
+    now = datetime(2026, 10, 25, 1, 30, tzinfo=zone)
+    job.valid_through = now.replace(hour=20)
+    job.save(update_fields=["valid_through"])
+    reminder = create_closing_reminders(profile, now=now)[0]
+    assert reminder.scheduled_for.tzinfo is not None
+    assert reminder.scheduled_for.tzinfo.key == "Europe/Brussels"

tests/integration/test_source_health.py

@@ -0,0 +1,170 @@
+from datetime import timedelta
+
+from django.utils import timezone
+
+import pytest
+
+from apps.sources.models import RawDocument, Source, SourceRun
+from apps.sources.services.health import collect_source_health
+from apps.sources.tasks import cleanup_raw_documents, enforce_source_health, recover_source_health
+
+
+def _create_run(
+    source: Source,
+    *,
+    status: str,
+    started_at=None,
+    error_category: str = "",
+    extracted_count: int = 0,
+    created_count: int = 0,
+    updated_count: int = 0,
+    duplicate_count: int = 0,
+    http_status: int = 200,
+    metrics: dict[str, object] | None = None,
+):
+    started_at = started_at or timezone.now()
+    finished_at = started_at + timedelta(seconds=1)
+    return SourceRun.objects.create(
+        source=source,
+        started_at=started_at,
+        finished_at=finished_at,
+        status=status,
+        error_category=error_category,
+        http_status=http_status,
+        extracted_count=extracted_count,
+        created_count=created_count,
+        updated_count=updated_count,
+        duplicate_count=duplicate_count,
+        metrics=metrics or {},
+    )
+
+
+@pytest.mark.django_db
+def test_enforce_source_health_quarantines_on_temporary_failures(source):
+    for index in range(6):
+        _create_run(
+            source,
+            status=SourceRun.Status.FAILED,
+            started_at=timezone.now() - timedelta(minutes=index),
+            error_category="timeout",
+        )
+
+    result = enforce_source_health()
+    assert result["quarantined"] == 1
+    source.refresh_from_db()
+    assert source.status == Source.Status.QUARANTINED
+    assert source.policy == Source.Policy.DENY
+
+
+@pytest.mark.django_db
+def test_enforce_source_health_quarantines_on_policy_category(source):
+    for index in range(3):
+        _create_run(
+            source,
+            status=SourceRun.Status.FAILED,
+            started_at=timezone.now() - timedelta(minutes=index),
+            error_category="policy",
+        )
+    result = enforce_source_health()
+    assert result["quarantined"] == 1
+    source.refresh_from_db()
+    assert source.status == Source.Status.QUARANTINED
+    assert "beleid" in source.policy_reason
+
+
+@pytest.mark.django_db
+def test_enforce_source_health_quarantines_on_parser_drift(source):
+    for index in range(3):
+        _create_run(
+            source,
+            status=SourceRun.Status.SUCCESS,
+            started_at=timezone.now() - timedelta(minutes=index),
+            extracted_count=0,
+            created_count=0,
+            updated_count=0,
+            duplicate_count=0,
+            http_status=200,
+            metrics={"parser": "generic-html", "warnings": ["veld ontbreekt", "datum ontbreekt"]},
+        )
+    assert enforce_source_health()["quarantined"] == 1
+    source.refresh_from_db()
+    assert source.status == Source.Status.QUARANTINED
+    assert source.policy == Source.Policy.DENY
+
+
+@pytest.mark.django_db
+def test_collect_source_health_exposes_parser_and_counts(source):
+    _create_run(
+        source,
+        status=SourceRun.Status.SUCCESS,
+        error_category="",
+        extracted_count=2,
+        created_count=1,
+        updated_count=1,
+        duplicate_count=0,
+        metrics={"parser": "jsonld", "warnings": ["warn"]},
+    )
+    row = collect_source_health()[0]
+    assert row.source_id == source.pk
+    assert row.last_parser == "jsonld"
+    assert row.last_parser_warnings == 1
+    assert row.extracted_count == 2
+    assert row.updated_count == 1
+
+
+@pytest.mark.django_db
+def test_recover_source_health_starts_canary_once(source, monkeypatch):
+    source.status = Source.Status.QUARANTINED
+    source.policy = Source.Policy.DENY
+    source.metadata = {
+        "source_health": {
+            "state": "quarantined",
+            "quarantine_reason": "tijdelijke fout",
+            "quarantined_at": timezone.now().isoformat(),
+            "recovery_due_at": timezone.now().isoformat(),
+            "canary_started": False,
+        }
+    }
+    source.save(update_fields=["status", "policy", "metadata"])
+
+    calls: list[int] = []
+
+    def fake_delay(source_id: int, force: bool = False):
+        calls.append(source_id)
+        return source_id
+
+    monkeypatch.setattr("apps.sources.tasks.fetch_source.delay", fake_delay)
+
+    result = recover_source_health()
+    assert result["canary_started"] == 1
+    assert calls == [source.pk]
+
+    source.refresh_from_db()
+    assert source.status == Source.Status.TRIAL
+    assert source.policy == Source.Policy.REVIEW
+
+    # Canary is eenmalig, niet continu herstartbaar.
+    source.refresh_from_db()
+    assert recover_source_health()["canary_started"] == 0
+
+
+@pytest.mark.django_db
+def test_cleanup_retains_quarantined_raw_documents(source):
+    old = RawDocument.objects.create(
+        source=source,
+        kind=RawDocument.Kind.TEXT,
+        content_hash="a" * 64,
+        retain_until=timezone.now() - timedelta(days=1),
+    )
+    keep = RawDocument.objects.create(
+        source=source,
+        kind=RawDocument.Kind.TEXT,
+        content_hash="b" * 64,
+        retain_until=timezone.now() - timedelta(days=1),
+        quarantined=True,
+    )
+
+    result = cleanup_raw_documents()
+    assert result["deleted"] == 1
+    assert RawDocument.objects.filter(pk=old.pk).exists() is False
+    assert RawDocument.objects.filter(pk=keep.pk).exists() is True

tests/integration/test_sources_manual_import.py

@@ -0,0 +1,111 @@
+from __future__ import annotations
+
+from django.core.cache import cache
+from django.test import override_settings
+from django.urls import reverse
+
+import pytest
+
+from apps.sources.services.fetcher import FetchedDocument
+from apps.sources.services.manual_import import ManualImportError
+
+
+def test_manual_import_view_prefills_source_url(client, user):
+    client.force_login(user)
+    response = client.get(
+        reverse("sources:manual_import"), {"source_url": "https://jobs.example.org/vacature"}
+    )
+    assert response.status_code == 200
+    assert 'value="https://jobs.example.org/vacature"' in response.content.decode("utf-8")
+
+
+@pytest.mark.django_db
+def test_manual_import_view_posts_url_and_shows_result(client, user, monkeypatch):
+    def fake_fetch(*_args, **_kwargs):
+        return FetchedDocument(
+            requested_url="https://jobs.example.org/engineering",
+            final_url="https://jobs.example.org/engineering",
+            status_code=200,
+            headers={"content-type": "text/html; charset=utf-8"},
+            content=b"<html><body>vacature</body></html>",
+        )
+
+    def fake_process(_document):
+        return {
+            "extracted": 1,
+            "created": 1,
+            "updated": 0,
+            "duplicates": 0,
+            "parser": "manual-test",
+            "warnings": [],
+        }
+
+    monkeypatch.setattr("apps.sources.services.manual_import.fetch_url", fake_fetch)
+    monkeypatch.setattr("apps.sources.services.manual_import.process_raw_document", fake_process)
+
+    client.force_login(user)
+    response = client.post(
+        reverse("sources:manual_import"),
+        {"source_url": "https://jobs.example.org/engineering"},
+    )
+    html = response.content.decode("utf-8")
+    assert response.status_code == 200
+    assert "Importresultaat" in html
+    assert "Herkenbare items: 1" in html
+
+
+@pytest.mark.django_db
+def test_manual_import_view_paste_prefers_post_flow_and_does_not_fetch(client, user, monkeypatch):
+    def fail_fetch(*_args, **_kwargs):
+        raise AssertionError("Fetcher should not run for paste imports.")
+
+    def fake_process(_document):
+        return {
+            "extracted": 1,
+            "created": 0,
+            "updated": 0,
+            "duplicates": 1,
+            "parser": "manual-test",
+            "warnings": [],
+        }
+
+    monkeypatch.setattr("apps.sources.services.manual_import.fetch_url", fail_fetch)
+    monkeypatch.setattr("apps.sources.services.manual_import.process_raw_document", fake_process)
+
+    client.force_login(user)
+    response = client.post(
+        reverse("sources:manual_import"),
+        {"pasted_text": "Eerste regel\nTweede regel"},
+    )
+    html = response.content.decode("utf-8")
+    assert response.status_code == 200
+    assert "Importresultaat" in html
+    assert "Modus: paste" in html
+
+
+@pytest.mark.django_db
+@override_settings(
+    MANUAL_IMPORT_RATE_LIMIT_MAX_ATTEMPTS=1,
+    MANUAL_IMPORT_RATE_LIMIT_WINDOW_SECONDS=120,
+    MANUAL_IMPORT_RATE_LIMIT_BLOCK_SECONDS=60,
+)
+def test_manual_import_blocks_after_failed_attempts(client, user, monkeypatch):
+    cache.clear()
+
+    def fail_import(*_args, **_kwargs):
+        raise ManualImportError("Handmatige import faalde in test")
+
+    client.force_login(user)
+    monkeypatch.setattr("apps.sources.views.import_manual_source", fail_import)
+
+    response = client.post(
+        reverse("sources:manual_import"),
+        {"source_url": "https://jobs.example.org/engineering"},
+    )
+    assert "Handmatige import faalde in test" in response.content.decode("utf-8")
+
+    response = client.post(
+        reverse("sources:manual_import"),
+        {"source_url": "https://jobs.example.org/engineering"},
+    )
+    assert "Te veel handmatige importverzoeken" in response.content.decode("utf-8")

tests/integration/test_tasks_and_more_views.py

@@ -0,0 +1,217 @@
+from datetime import timedelta
+
+import pytest
+from django.urls import reverse
+from django.utils import timezone
+
+from apps.jobs.models import Feedback, ScoreRun
+from apps.jobs.tasks import rescore_active_jobs, update_job_lifecycle
+from apps.notifications.tasks import build_due_digests
+from apps.sources.models import RawDocument, Source, SourcePolicyReview, SourceLease, SourceRun
+from apps.sources.services.fetcher import RateLimitedError, FetchedDocument
+from apps.sources.tasks import cleanup_raw_documents, fetch_source, schedule_due_sources
+
+
+@pytest.mark.django_db
+def test_task_helpers(profile, job, source, monkeypatch):
+    result = rescore_active_jobs(profile.pk)
+    assert result["scores_created"] == 1
+    assert ScoreRun.objects.filter(job=job, profile=profile).exists()
+    assert update_job_lifecycle() == {"expired": 0, "uncertain": 0, "removed": 0}
+
+    calls = []
+    monkeypatch.setattr(
+        "apps.sources.tasks.fetch_source.delay", lambda source_id: calls.append(source_id)
+    )
+    source.next_run_at = timezone.now() - timedelta(minutes=1)
+    source.save(update_fields=["next_run_at"])
+    assert schedule_due_sources()["scheduled"] == 1
+    assert calls == [source.pk]
+
+    old = RawDocument.objects.create(
+        source=source,
+        kind=RawDocument.Kind.TEXT,
+        content_hash="8" * 64,
+        retain_until=timezone.now() - timedelta(days=1),
+    )
+    assert cleanup_raw_documents()["deleted"] >= 1
+    assert not RawDocument.objects.filter(pk=old.pk).exists()
+
+    source.policy = Source.Policy.DENY
+    source.policy_reason = "test"
+    source.save(update_fields=["policy", "policy_reason"])
+    assert fetch_source(source.pk)["status"] == "skipped"
+    assert SourceRun.objects.filter(source=source, status=SourceRun.Status.SKIPPED).exists()
+
+
+def test_fetch_source_skips_when_lease_blocked(db, source, monkeypatch):
+    SourceLease.objects.create(
+        source=source,
+        token="active-worker",
+        worker_id="active-worker",
+        expires_at=timezone.now() + timedelta(minutes=2),
+    )
+    calls = {"count": 0}
+
+    def fake_fetch_url(*args, **kwargs):
+        calls["count"] += 1
+        return FetchedDocument(
+            requested_url=source.base_url,
+            final_url=source.base_url,
+            status_code=200,
+            headers={},
+            content=b"Vacature",
+        )
+
+    monkeypatch.setattr("apps.sources.tasks.fetch_url", fake_fetch_url)
+    assert fetch_source(source.pk)["status"] == "skipped"
+    assert calls["count"] == 0
+    assert SourceRun.objects.filter(source=source, status=SourceRun.Status.SKIPPED).exists()
+
+
+def test_fetch_source_handles_rate_limit_backoff(db, source, monkeypatch):
+    called = {"count": 0}
+
+    def fake_fetch_url(*args, **kwargs):
+        called["count"] += 1
+        raise RateLimitedError(30)
+
+    monkeypatch.setattr("apps.sources.tasks.fetch_url", fake_fetch_url)
+    result = fetch_source(source.pk)
+    assert result["status"] == "rate_limited"
+    assert called["count"] == 1
+    source.refresh_from_db()
+    assert source.next_run_at is not None
+    assert SourceRun.objects.filter(source=source, status=SourceRun.Status.FAILED).exists()
+
+
+def test_fetch_source_resets_fail_state_on_304(db, source, monkeypatch):
+    source.failure_count = 4
+    source.save(update_fields=["failure_count"])
+
+    def fake_fetch_url(*args, **kwargs):
+        return FetchedDocument(
+            requested_url=source.base_url,
+            final_url=source.base_url,
+            status_code=304,
+            headers={},
+            content=b"",
+        )
+
+    monkeypatch.setattr("apps.sources.tasks.fetch_url", fake_fetch_url)
+    assert fetch_source(source.pk)["status"] == "not_modified"
+    source.refresh_from_db()
+    assert source.failure_count == 0
+
+
+def test_fetch_source_is_blocked_by_conflicting_review(db):
+    source = Source.objects.create(
+        name="Blocked source",
+        source_type=Source.Type.EMPLOYER,
+        base_url="https://jobs.example.org/vacatures/",
+        domain="jobs.example.org",
+        status=Source.Status.ACTIVE,
+        policy=Source.Policy.ALLOW,
+    )
+    SourcePolicyReview.objects.create(
+        source=source,
+        decision=SourcePolicyReview.Decision.DENY,
+        reason="Review blokkeert deze bron",
+    )
+    result = fetch_source(source.pk)
+    assert result["status"] == "skipped"
+    assert SourceRun.objects.filter(source=source, status=SourceRun.Status.SKIPPED).exists()
+
+
+def test_fetch_source_is_blocked_by_expired_review(db):
+    source = Source.objects.create(
+        name="Expired review source",
+        source_type=Source.Type.EMPLOYER,
+        base_url="https://jobs.example.org/vacatures/",
+        domain="jobs.example.org",
+        status=Source.Status.ACTIVE,
+        policy=Source.Policy.ALLOW,
+    )
+    SourcePolicyReview.objects.create(
+        source=source,
+        decision=SourcePolicyReview.Decision.ALLOW,
+        reason="Review verlopen",
+        expires_at=timezone.now() - timedelta(days=1),
+    )
+    result = fetch_source(source.pk)
+    assert result["status"] == "skipped"
+    assert SourceRun.objects.filter(source=source, status=SourceRun.Status.SKIPPED).exists()
+
+    # Outside the digest window this is a safe no-op.
+    assert build_due_digests()["sent"] == 0
+
+
+@pytest.mark.integration
+@pytest.mark.django_db
+def test_job_profile_source_and_application_views(client, user, profile, job, source):
+    client.force_login(user)
+    assert client.get(reverse("jobs:list"), {"q": "Infrastructure"}).status_code == 200
+    assert client.get(reverse("jobs:detail", kwargs={"pk": job.pk})).status_code == 200
+    response = client.post(
+        reverse("jobs:feedback", kwargs={"pk": job.pk}),
+        {"action": Feedback.Action.INTERESTING},
+    )
+    assert response.status_code == 302
+    assert Feedback.objects.filter(user=user, job=job).exists()
+
+    assert client.get(reverse("profiles:list")).status_code == 200
+    assert client.get(reverse("profiles:edit", kwargs={"pk": profile.pk})).status_code == 200
+    assert client.get(reverse("sources:list")).status_code == 200
+    assert client.get(reverse("dashboard:system")).status_code == 200
+    assert client.get(reverse("jobs:applications")).status_code == 200
+
+    source.policy = Source.Policy.DENY
+    source.save(update_fields=["policy"])
+    response = client.post(reverse("sources:retry", kwargs={"pk": source.pk}))
+    assert response.status_code == 302
+
+
+@pytest.mark.integration
+@pytest.mark.django_db
+def test_sources_candidate_bulk_actions_and_evidence_view(client, user):
+    candidate = Source.objects.create(
+        name="Discoverd candidate",
+        source_type=Source.Type.EMPLOYER,
+        base_url="https://jobs.example.org/vacatures/",
+        domain="jobs.example.org",
+        status=Source.Status.CANDIDATE,
+        policy=Source.Policy.REVIEW,
+        metadata={"discovery": [{"url": "https://jobs.example.org/careers", "discovered_from": "html", "reason": "career-link", "confidence": 0.9, "label": "Werken bij"}]},
+    )
+
+    other = Source.objects.create(
+        name="Another source",
+        source_type=Source.Type.EMPLOYER,
+        base_url="https://other.example.org/",
+        domain="other.example.org",
+        status=Source.Status.CANDIDATE,
+        policy=Source.Policy.REVIEW,
+    )
+
+    client.force_login(user)
+    response = client.get(reverse("sources:list"))
+    assert response.status_code == 200
+    content = response.content.decode("utf-8")
+    assert "jobs.example.org/careers" in content
+
+    response = client.post(
+        reverse("sources:bulk"),
+        {"action": "approve_to_trial", "source_ids": [str(candidate.pk)]},
+    )
+    assert response.status_code == 302
+    candidate.refresh_from_db()
+    assert candidate.status == Source.Status.TRIAL
+
+    response = client.post(
+        reverse("sources:bulk"),
+        {"action": "dismiss", "source_ids": [str(other.pk)]},
+    )
+    assert response.status_code == 302
+    other.refresh_from_db()
+    assert other.status == Source.Status.DISABLED
+    assert other.policy == Source.Policy.DENY

tests/integration/test_views.py

@@ -0,0 +1,51 @@
+import pytest
+from django.core.cache import cache
+from django.test import override_settings
+from django.urls import reverse
+
+
+@pytest.mark.integration
+@pytest.mark.django_db
+def test_today_requires_login(client):
+    response = client.get(reverse("dashboard:today"))
+    assert response.status_code == 302
+    assert reverse("login") in response.url
+
+
+@pytest.mark.integration
+@pytest.mark.django_db
+def test_today_loads_for_user(client, user, profile):
+    client.force_login(user)
+    response = client.get(reverse("dashboard:today"))
+    assert response.status_code == 200
+    assert "Vandaag" in response.content.decode()
+
+
+@pytest.mark.integration
+@pytest.mark.django_db
+@override_settings(
+    AUTH_LOGIN_RATE_LIMIT_MAX_ATTEMPTS=2,
+    AUTH_LOGIN_RATE_LIMIT_WINDOW_SECONDS=120,
+    AUTH_LOGIN_RATE_LIMIT_BLOCK_SECONDS=300,
+)
+def test_login_view_rate_limits_failed_attempts(client, user):
+    cache.clear()
+
+    response = client.post(
+        reverse("login"),
+        {"username": user.username, "password": "wrong"},
+    )
+    assert response.status_code == 200
+
+    response = client.post(
+        reverse("login"),
+        {"username": user.username, "password": "wrong"},
+    )
+    body = response.content.decode("utf-8")
+    assert "Te veel inlogpogingen" in body
+
+    response = client.post(
+        reverse("login"),
+        {"username": user.username, "password": "wrong"},
+    )
+    assert "Te veel inlogpogingen" in response.content.decode("utf-8")