Initial deploy setup

2026-07-21 14:00:00 +02:00
commit b8091e59bd
285 changed files with 27854 additions and 0 deletions
@@ -0,0 +1,87 @@
+from __future__ import annotations
+
+import pytest
+
+from apps.sources.models import Source, SourceRun
+from apps.sources.services.fetcher import FetchedDocument
+from apps.sources.services.manual_import import ManualImportError, import_manual_source
+
+
+def test_manual_import_url_uses_fetch_and_succeeds(monkeypatch, user):
+    def fake_fetch(*_args, **_kwargs):
+        return FetchedDocument(
+            requested_url="https://jobs.example.org/jobs",
+            final_url="https://jobs.example.org/jobs",
+            status_code=200,
+            headers={
+                "content-type": "text/html; charset=utf-8",
+                "etag": "abc",
+                "last-modified": "Mon, 01 Jan 2000 00:00:00 GMT",
+            },
+            content=b"<html><body>vacature</body></html>",
+        )
+
+    def fake_process(document):
+        return {
+            "extracted": 2,
+            "created": 1,
+            "updated": 1,
+            "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)
+
+    summary = import_manual_source(
+        actor=user, source_url="https://jobs.example.org/jobs"
+    )
+    source = Source.objects.get(pk=summary.source_id)
+
+    assert summary.mode == "url"
+    assert summary.extracted_count == 2
+    assert summary.created_count == 1
+    assert source.source_type == Source.Type.MANUAL
+    assert source.domain == "jobs.example.org"
+    assert SourceRun.objects.filter(source=source, status=SourceRun.Status.SUCCESS).exists()
+
+
+def test_manual_import_paste_text_sanitizes_and_skips_fetch(monkeypatch, user):
+    def fail_fetch(*_args, **_kwargs):
+        raise AssertionError("Fetcher should not be called for pasted text.")
+
+    def fake_process(document):
+        return {
+            "extracted": 0,
+            "created": 0,
+            "updated": 0,
+            "duplicates": 0,
+            "parser": "manual-test",
+            "warnings": ["noop"],
+        }
+
+    monkeypatch.setattr("apps.sources.services.manual_import.fetch_url", fail_fetch)
+    monkeypatch.setattr("apps.sources.services.manual_import.process_raw_document", fake_process)
+
+    summary = import_manual_source(
+        actor=user, pasted_text="<script>alert('x')</script>\nVacaturetekst"
+    )
+    source = Source.objects.get(pk=summary.source_id)
+
+    assert summary.mode == "paste"
+    assert source.source_type == Source.Type.MANUAL
+    assert source.base_url.startswith("https://manual-")
+
+
+def test_manual_import_rejects_denied_domain(user):
+    with pytest.raises(ManualImportError):
+        import_manual_source(actor=user, source_url="https://linkedin.com/jobs/123")
+
+
+def test_manual_import_rejects_paste_payload_when_too_large(user, settings):
+    settings.MANUAL_IMPORT_PASTE_MAX_BYTES = 12
+    huge_text = "a" * 20
+
+    with pytest.raises(ManualImportError):
+        import_manual_source(actor=user, pasted_text=huge_text)