Initial deploy setup

2026-07-21 14:00:00 +02:00
commit b8091e59bd
285 changed files with 27854 additions and 0 deletions
@@ -0,0 +1,176 @@
+from __future__ import annotations
+
+from datetime import UTC, datetime
+
+import pytest
+
+from apps.jobs.models import Employer, JobPosting
+from apps.jobs.services.employer_resolution import resolve_direct_employer_match
+from apps.jobs.services.normalization import CanonicalJobDraft
+from apps.sources.models import Source
+
+
+def _draft_for_test(job: JobPosting, *, **overrides) -> CanonicalJobDraft:
+    values = {
+        "source_url": "https://jobs.smartrecruiters.com/example/security-engineer",
+        "canonical_url": "https://jobs.smartrecruiters.com/example/security-engineer",
+        "external_id": "sr-security-123",
+        "title": job.original_title,
+        "normalized_title": job.normalized_title,
+        "job_family": job.job_family,
+        "employer_name": job.employer_name,
+        "employer_domain": "jobs.example.org",
+        "location_text": job.raw_location,
+        "region": job.region,
+        "municipality": job.municipality,
+        "postal_code": job.postal_code,
+        "country": job.country,
+        "workplace_type": job.workplace_type,
+        "employment_types": job.employment_types,
+        "language": "nl",
+        "description_html": "<p>Test</p>",
+        "description_text": "test beschrijving",
+        "date_posted": datetime(2026, 1, 10, tzinfo=UTC),
+        "valid_through": datetime(2026, 6, 1, tzinfo=UTC),
+        "compensation": {},
+        "skills_required": [],
+        "skills_preferred": [],
+        "content_hash": job.content_hash,
+        "canonical_key": "x" * 64,
+        "evidence": [],
+        "raw": {},
+    }
+    values.update(overrides)
+    return CanonicalJobDraft(**values)
+
+
+@pytest.mark.django_db
+def test_resolver_merges_recruiter_alias_to_direct_source():
+    employer = Employer.objects.create(
+        name="Example IT",
+        normalized_name="example it",
+        domain="jobs.example.org",
+    )
+    JobPosting.objects.create(
+        employer=employer,
+        original_title="Security Engineer",
+        normalized_title="security engineer",
+        canonical_url="https://jobs.example.org/vacatures/security-engineer",
+        canonical_key="a" * 64,
+        job_family="security",
+        content_hash="a" * 64,
+        description_text="Security engineer voor beveiliging.",
+        raw_location="Antwerpen",
+        region="Antwerpen",
+        municipality="Antwerpen",
+        workplace_type=JobPosting.Workplace.HYBRID,
+        employment_types=["full_time"],
+        direct_employer=True,
+    )
+    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,
+    )
+    # Use an existing direct employer job as the resolver target.
+    target_job = JobPosting.objects.first()
+    decision = resolve_direct_employer_match(
+        _draft_for_test(target_job),
+        source=source,
+    )
+    assert decision.job == target_job
+    assert decision.reason == "resolved_direct_match"
+    assert decision.canonical_url == "https://jobs.example.org/vacatures/security-engineer"
+
+
+@pytest.mark.django_db
+def test_resolver_flags_conflicting_location_for_review():
+    employer = Employer.objects.create(
+        name="Example IT",
+        normalized_name="example it",
+        domain="jobs.example.org",
+    )
+    JobPosting.objects.create(
+        employer=employer,
+        original_title="Security Engineer",
+        normalized_title="security engineer",
+        canonical_url="https://jobs.example.org/vacatures/security-engineer",
+        canonical_key="b" * 64,
+        job_family="security",
+        content_hash="b" * 64,
+        description_text="Security engineer voor beveiliging.",
+        raw_location="Antwerpen",
+        region="Antwerpen",
+        municipality="Antwerpen",
+        workplace_type=JobPosting.Workplace.HYBRID,
+        employment_types=["full_time"],
+        direct_employer=True,
+    )
+    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,
+    )
+    decision = resolve_direct_employer_match(
+        _draft_for_test(
+            JobPosting.objects.first(),
+            location_text="Gent",
+            municipality="Gent",
+            region="Oost-Vlaanderen",
+        ),
+        source=source,
+    )
+    assert decision.job is None
+    assert decision.reason == "review_direct_conflict"
+
+
+@pytest.mark.django_db
+def test_resolver_keeps_candidate_unmerged_when_similarity_is_conservative():
+    employer = Employer.objects.create(
+        name="Example IT",
+        normalized_name="example it",
+        domain="jobs.example.org",
+    )
+    JobPosting.objects.create(
+        employer=employer,
+        original_title="Security Engineer",
+        normalized_title="security engineer",
+        canonical_url="https://jobs.example.org/vacatures/security-engineer",
+        canonical_key="c" * 64,
+        job_family="security",
+        content_hash="c" * 64,
+        description_text="Security engineer voor beveiliging.",
+        raw_location="Antwerpen",
+        region="Antwerpen",
+        municipality="Antwerpen",
+        workplace_type=JobPosting.Workplace.HYBRID,
+        employment_types=["full_time"],
+        direct_employer=True,
+    )
+    source = Source.objects.create(
+        name="StaffingBoard",
+        source_type=Source.Type.ATS,
+        base_url="https://jobs.staffingboard.com/example",
+        domain="jobs.staffingboard.com",
+        status=Source.Status.ACTIVE,
+        policy=Source.Policy.ALLOW,
+    )
+    decision = resolve_direct_employer_match(
+        _draft_for_test(
+            JobPosting.objects.first(),
+            title="Data Analyst",
+            normalized_title="data analyst",
+            location_text="Brussel",
+            employer_name="Another Employer",
+            employer_domain="jobs.staffingboard.com",
+        ),
+        source=source,
+    )
+    assert decision.job is None
+    assert decision.reason == "no_direct_resolution"