Initial deploy setup

2026-07-21 14:00:00 +02:00
commit b8091e59bd
285 changed files with 27854 additions and 0 deletions
@@ -0,0 +1,120 @@
+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"
+    )