import json, os, sqlite3, threading, unittest from unittest.mock import patch from http.client import HTTPConnection from tempfile import TemporaryDirectory from app.main import ORGANIZATION_ID, connect, create_server from app.sources import ApprovedDirectorySource, CsvSource, ManualSource, available_adapters class SourceAdapterTests(unittest.TestCase): def test_adapter_catalog_exposes_ready_and_gated_sources(self): catalog = {item['source_code']: item for item in available_adapters()} for code in ('manual', 'csv', 'public_website', 'ct_logs', 'dns', 'rdap'): self.assertTrue(catalog[code]['available'], code) for code in ('bing_local', 'permitted_social'): self.assertFalse(catalog[code]['available'], code) self.assertTrue(catalog[code]['optional'], code) for code in ('openstreetmap','wikidata','common_crawl'): self.assertTrue(catalog[code]['available'], code) self.assertFalse(catalog[code]['optional'], code) self.assertTrue(catalog['google_places']['available']) self.assertTrue(catalog['google_places']['optional']) def test_csv_adapter_is_deterministic_and_normalizes(self): src = CsvSource() a = src.discover({'csv': 'Name,Website,Email\n Acme ,https://acme.test,a@acme.test\n'}) b = src.discover({'csv': 'Name,Website,Email\n Acme ,https://acme.test,a@acme.test\n'}) self.assertEqual(a.records, b.records) self.assertEqual(a.records[0]['name'], 'Acme') self.assertEqual(a.records[0]['email'], 'a@acme.test') def test_manual_validation_rejects_secret_fields(self): result = ManualSource().validate({'rows': [{'name': 'x', 'api_key': 'secret'}]}) self.assertFalse(result.valid) self.assertIn('secret', result.errors[0].lower()) def test_openstreetmap_uses_provider_side_service_tag_query(self): class Response: def read(self, _): return b'{"elements":[{"tags":{"name":"Cape Plumber","craft":"plumber"}}]}' def __enter__(self): return self def __exit__(self, *_): return False config={'provider':'openstreetmap','approved':True,'public_access':True,'terms_accepted':True,'rate_limit':1} criteria={'keywords':['plumbers'],'city':'Cape Town'} with patch('app.sources.urlopen',return_value=Response()) as request: result=ApprovedDirectorySource().discover(config,criteria=criteria,limits={'max_records':10}) query=request.call_args.args[0].data.decode() self.assertIn('"craft"~"plumber|plumbers",i]',query) self.assertIn('"shop"~"plumber|plumbers",i]',query) self.assertEqual(result.records[0]['name'],'Cape Plumber') def test_legacy_source_kind_constraint_is_migrated(self): with TemporaryDirectory() as tmp: path=os.path.join(tmp,'legacy.db') legacy=sqlite3.connect(path) legacy.execute("CREATE TABLE sources (id INTEGER PRIMARY KEY AUTOINCREMENT, organization_id TEXT NOT NULL, name TEXT NOT NULL, kind TEXT NOT NULL CHECK(kind IN ('csv','manual')), enabled INTEGER NOT NULL DEFAULT 0, config_json TEXT NOT NULL DEFAULT '{}', health_status TEXT NOT NULL DEFAULT 'unknown', consecutive_failures INTEGER NOT NULL DEFAULT 0, circuit_open INTEGER NOT NULL DEFAULT 0, last_success_at TEXT, last_failure_at TEXT, last_error TEXT, created_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP, updated_at TEXT NOT NULL DEFAULT CURRENT_TIMESTAMP, source_code TEXT NOT NULL DEFAULT '', display_name TEXT NOT NULL DEFAULT '', approved INTEGER NOT NULL DEFAULT 0, policy_json TEXT NOT NULL DEFAULT '{}', quota_json TEXT NOT NULL DEFAULT '{}', UNIQUE(organization_id,name))") legacy.execute("INSERT INTO sources(organization_id,name,kind) VALUES(?, 'Existing manual','manual')",(ORGANIZATION_ID,)); legacy.commit(); legacy.close() db=connect(path) self.assertEqual(db.execute("SELECT kind FROM sources WHERE name='Existing manual'").fetchone()[0],'manual') db.execute("INSERT INTO sources(organization_id,name,kind,source_code) VALUES(?,?,?,?)",(ORGANIZATION_ID,'OpenStreetMap / Overpass · plumbers','approved_directory','openstreetmap')) db.execute("INSERT INTO sources(organization_id,name,kind,source_code) VALUES(?,?,?,?)",(ORGANIZATION_ID,'Experimental Google','google_browser_search','google_browser_search')) db.commit(); self.assertEqual(db.execute("SELECT kind FROM sources WHERE source_code='openstreetmap'").fetchone()[0],'approved_directory'); self.assertEqual(db.execute("SELECT kind FROM sources WHERE source_code='google_browser_search'").fetchone()[0],'google_browser_search'); db.close() class SourceApiTests(unittest.TestCase): def setUp(self): self.tmp=TemporaryDirectory(); os.environ['BOOTSTRAP_ADMIN_EMAIL']='owner@example.test'; os.environ['BOOTSTRAP_ADMIN_PASSWORD']='development-password' self.server=create_server('127.0.0.1',0,self.tmp.name+'/x.db'); self.thread=threading.Thread(target=self.server.serve_forever,daemon=True); self.thread.start(); self.c=HTTPConnection('127.0.0.1',self.server.server_port); self.cookie=None self.req('POST','/api/v1/auth/login',{'email':'owner@example.test','password':'development-password'}) def tearDown(self): self.server.shutdown(); self.server.server_close(); self.thread.join(2); self.tmp.cleanup() def req(self,m,p,x=None,cookie=True): body=json.dumps(x).encode() if x is not None else None; h={'Content-Type':'application/json'} if body else {}; if cookie and self.cookie:h['Cookie']=self.cookie self.c.request(m,p,body,h); r=self.c.getresponse(); sc=r.getheader('Set-Cookie'); if sc:self.cookie=sc.split(';',1)[0] raw=r.read(); return r.status,json.loads(raw or b'{}') def test_source_lifecycle_ingestion_idempotency_health_and_disabled(self): s, source=self.req('POST','/api/v1/sources',{'name':'Import','kind':'manual','config':{}}); self.assertEqual(s,201) sid=source['id']; self.assertFalse(source['enabled']) self.assertEqual(self.req('PATCH',f'/api/v1/sources/{sid}',{'enabled':True})[0],200) payload={'rows':[{'name':'Acme','website':'https://acme.test'}], 'source_url':'file://import.csv','query_context':{'q':'test'}} self.assertEqual(self.req('POST',f'/api/v1/sources/{sid}/ingest',payload)[0],201) self.assertEqual(self.req('POST',f'/api/v1/sources/{sid}/ingest',payload)[0],200) self.assertEqual(len(self.req('GET','/api/v1/source-records')[1]['items']),1) self.assertEqual(self.req('POST',f'/api/v1/sources/{sid}/test')[0],200) self.assertEqual(self.req('PATCH',f'/api/v1/sources/{sid}',{'enabled':False})[0],200) self.assertEqual(self.req('POST',f'/api/v1/sources/{sid}/ingest',payload)[0],409) db=sqlite3.connect(self.tmp.name+'/x.db'); self.assertTrue(db.execute("select 1 from audit_log where action='source.disabled'").fetchone()); db.close() def test_queries_enqueue_and_records_are_tenant_scoped(self): _,source=self.req('POST','/api/v1/sources',{'name':'CSV','kind':'csv','enabled':True,'config':{'csv':'name\nA'}}) _,q=self.req('POST','/api/v1/discovery-queries',{'source_id':source['id'],'name':'q','query':{}}) status,job=self.req('POST',f"/api/v1/discovery-queries/{q['id']}/run",{}) self.assertEqual(status,202); self.assertEqual(job['type'],'source_discovery') for _ in range(100): _, current=self.req('GET',f"/api/v1/jobs/{job['id']}") if current['status'] in ('succeeded','failed'): break threading.Event().wait(.01) self.assertEqual(current['status'],'succeeded') self.assertEqual(len(self.req('GET','/api/v1/businesses')[1]['items']),1) self.assertEqual(self.req('GET','/api/v1/source-records')[1]['items'][0]['discovery_query_id'],q['id']) self.assertEqual(self.req('GET','/api/v1/source-records?page_size=101')[0],400) def test_sources_require_auth(self): self.cookie=None; self.assertEqual(self.req('GET','/api/v1/sources',cookie=False)[0],401) def test_unavailable_or_unconfigured_sources_cannot_be_enabled(self): status, gated = self.req('POST', '/api/v1/sources', {'name': 'Google', 'kind': 'google_places', 'config': {}}) self.assertEqual(status, 201) self.assertEqual(self.req('PATCH', f"/api/v1/sources/{gated['id']}", {'enabled': True})[0], 409) status, website = self.req('POST', '/api/v1/sources', {'name': 'Web', 'kind': 'public_website', 'config': {}}) self.assertEqual(status, 201) self.assertEqual(self.req('PATCH', f"/api/v1/sources/{website['id']}", {'enabled': True})[0], 409) def test_duplicate_source_registration_is_idempotent(self): payload={'name':'OpenStreetMap / Overpass · plumbers','kind':'openstreetmap','config':{'provider':'openstreetmap','approved':True,'public_access':True,'terms_accepted':True,'rate_limit':1}} status, created=self.req('POST','/api/v1/sources',payload); self.assertEqual(status,201); self.assertTrue(created['created']) status, reused=self.req('POST','/api/v1/sources',payload); self.assertEqual(status,200); self.assertFalse(reused['created']); self.assertEqual(reused['id'],created['id']) second={**payload,'name':'OpenStreetMap / Overpass · plumbers · Durban'} status, other=self.req('POST','/api/v1/sources',second); self.assertEqual(status,201); self.assertTrue(other['created']); self.assertNotEqual(other['id'],created['id']) def test_source_registry_exposes_first_class_governance_fields(self): payload = { 'name': 'Governed manual', 'kind': 'manual', 'config': {'rows': []}, 'policy': {'owner': 'Research Ops', 'terms_url': 'https://example.test/terms', 'terms_status': 'approved', 'rate_limit': '60/hour'}, 'quota': {'daily_quota': 200}, } status, created = self.req('POST', '/api/v1/sources', payload) self.assertEqual(status, 201) status, items = self.req('GET', '/api/v1/sources') self.assertEqual(status, 200) source = next(item for item in items['items'] if item['id'] == created['id']) self.assertEqual(source['owner'], 'Research Ops') self.assertEqual(source['terms_url'], 'https://example.test/terms') self.assertEqual(source['terms_status'], 'approved') self.assertEqual(source['rate_limit'], '60/hour') self.assertEqual(source['daily_quota'], 200) self.assertFalse(source['credentials_configured']) def test_source_configuration_rejects_discovery_criteria(self): forbidden = {'query': 'plumbers', 'category': 'trades', 'city': 'Cape Town', 'location': 'Western Cape'} for field, value in forbidden.items(): with self.subTest(field=field): status, body = self.req('POST', '/api/v1/sources', { 'name': 'Manual ' + field, 'kind': 'manual', 'config': {'rows': [], field: value}, }) self.assertEqual(status, 400) self.assertEqual(body['error'], 'source_configuration_contains_criteria') def test_source_worker_passes_query_criteria_and_effective_limits_to_connector(self): status, source = self.req('POST', '/api/v1/sources', { 'name': 'Criteria manual', 'kind': 'manual', 'enabled': True, 'config': {'rows': [{'name': 'Criteria Acme'}]}, }) self.assertEqual(status, 201) status, query = self.req('POST', '/api/v1/discovery-queries', { 'source_id': source['id'], 'name': 'Cape solar', 'query': {'keywords': ['solar'], 'city': 'Cape Town'}, 'max_records': 7, 'daily_limit': 9, }) self.assertEqual(status, 201) observed = [] from app.sources import ManualSource original = ManualSource.discover def spy(adapter, config, cursor=None, criteria=None, limits=None): observed.append((criteria, limits)) return original(adapter, config, cursor, criteria=criteria, limits=limits) with patch('app.sources.ManualSource.discover', new=spy): status, job = self.req('POST', f"/api/v1/discovery-queries/{query['id']}/run", {}) self.assertEqual(status, 202) for _ in range(100): _, current = self.req('GET', f"/api/v1/jobs/{job['id']}") if current['status'] in ('succeeded', 'failed'): break threading.Event().wait(.01) self.assertEqual(current['status'], 'succeeded') self.assertEqual(observed, [({'keywords': ['solar'], 'city': 'Cape Town'}, {'max_records': 7, 'daily_limit': 9, 'per_run_limit': 7})]) def test_source_configuration_can_be_saved_before_enablement(self): status, source = self.req('POST', '/api/v1/sources', {'name': 'DNS', 'kind': 'dns', 'config': {}}) self.assertEqual(status, 201) status, configured = self.req('PATCH', f"/api/v1/sources/{source['id']}", {'config': {'domains': ['example.co.za']}}) self.assertEqual(status, 200) self.assertEqual(self.req('PATCH', f"/api/v1/sources/{source['id']}", {'enabled': True})[0], 200) def test_fresh_schema_accepts_optional_source_kind_fail_closed(self): status, source = self.req('POST', '/api/v1/sources', {'name': 'RDAP', 'kind': 'rdap', 'config': {}}) self.assertEqual(status, 201) self.assertEqual(source['kind'], 'rdap') self.assertEqual(self.req('POST', f"/api/v1/sources/{source['id']}/test", {})[0], 200) status, health = self.req('GET', f"/api/v1/sources/{source['id']}/health") self.assertEqual(status, 200) self.assertFalse(health['configured']) def test_disabled_selected_source_fails_instead_of_succeeding_with_zero_records(self): status, source=self.req('POST','/api/v1/sources',{'name':'Disabled manual source','kind':'manual','config':{'rows':[]}}); self.assertEqual(status,201) status, query=self.req('POST','/api/v1/discovery-queries',{'source_id':source['id'],'name':'disabled-source-run','query':{},'selected_adapters':['manual']}); self.assertEqual(status,201) status, job=self.req('POST',f"/api/v1/discovery-queries/{query['id']}/run",{}); self.assertEqual(status,202) for _ in range(100): _, current=self.req('GET',f"/api/v1/jobs/{job['id']}") if current['status'] in ('succeeded','failed'): break threading.Event().wait(.01) self.assertEqual(current['status'],'failed'); self.assertEqual(current['error_code'],'SOURCE_DISABLED') def test_google_browser_source_cannot_enable_without_runtime_feature_flag(self): with patch.dict(os.environ, {'GOOGLE_BROWSER_SEARCH_ENABLED': 'false'}): status, source = self.req('POST', '/api/v1/sources', { 'name': 'Disabled Experimental Google', 'kind': 'google_browser_search', 'config': {'approved': True, 'public_access': True, 'terms_accepted': True, 'rate_limit': 6}, }) self.assertEqual(status, 201) status, body = self.req('PATCH', f"/api/v1/sources/{source['id']}", {'enabled': True}) self.assertEqual(status, 409) self.assertEqual(body['error'], 'source_feature_disabled') with patch.dict(os.environ, {'GOOGLE_BROWSER_SEARCH_ENABLED': 'false'}): status, body = self.req('POST', '/api/v1/sources', { 'name': 'Still Disabled Experimental Google', 'kind': 'google_browser_search', 'enabled': True, 'config': {'approved': True, 'public_access': True, 'terms_accepted': True, 'rate_limit': 6}, }) self.assertEqual(status, 409) self.assertEqual(body['error'], 'source_feature_disabled') def test_google_browser_block_is_structured_and_stops_source_job(self): class Response: def read(self, _limit): return b"Our systems have detected unusual traffic from your computer network." def __enter__(self): return self def __exit__(self, *_): return False with patch.dict(os.environ, {'GOOGLE_BROWSER_SEARCH_ENABLED': 'true'}): status, source = self.req('POST', '/api/v1/sources', { 'name': 'Experimental Google', 'kind': 'google_browser_search', 'enabled': True, 'config': {'approved': True, 'public_access': True, 'terms_accepted': True, 'rate_limit': 6}, }) self.assertEqual(status, 201) status, job = self.req('POST', '/api/v1/discovery', { 'criteria': {'keywords': ['solar'], 'city': 'Cape Town'}, 'selected_adapters': ['google_browser_search'], 'idempotency_key': 'google-blocked', 'max_records': 2, }) self.assertEqual(status, 202) with patch('app.sources.urlopen', return_value=Response()) as network: for _ in range(100): _, current = self.req('GET', f"/api/v1/jobs/{job['id']}") if current['status'] in ('succeeded', 'failed'): break threading.Event().wait(.01) self.assertEqual(current['status'], 'failed') self.assertEqual(current['error_code'], 'GOOGLE_BROWSER_BLOCKED') self.assertEqual(network.call_count, 1) events = self.req('GET', f"/api/v1/jobs/{job['id']}/events")[1]['items'] self.assertIn('GOOGLE_BROWSER_BLOCKED', [event.get('error_code') for event in events]) def test_source_discovery_persists_pipeline_and_is_idempotent(self): status, source = self.req('POST', '/api/v1/sources', { 'name': 'Manual leads', 'kind': 'manual', 'enabled': True, 'config': {'rows': [{'name': 'Acme Solar', 'website': 'https://acme.test', 'email': 'hello@acme.test', 'phone': '011 555 0100', 'description': 'solar installers', 'location': 'Cape Town'}]}}) self.assertEqual(status, 201) payload = {'criteria': {'keywords': ['solar']}, 'selected_adapters': ['manual'], 'idempotency_key': 'source-run-1', 'max_records': 10} status, job = self.req('POST', '/api/v1/discovery', payload) self.assertEqual(status, 202) for _ in range(100): _, current = self.req('GET', f"/api/v1/jobs/{job['id']}") if current['status'] in ('succeeded', 'failed'): break threading.Event().wait(.01) self.assertEqual(current['status'], 'succeeded') events = self.req('GET', f"/api/v1/jobs/{job['id']}/events")[1]['items'] event_types = [event['event_type'] for event in events] for stage in ('source.started', 'source.raw_persisted', 'source.normalized', 'business.created', 'enrichment.queued', 'review.queued', 'discovery.completed'): self.assertIn(stage, event_types) businesses = self.req('GET', '/api/v1/businesses')[1]['items'] self.assertEqual(len(businesses), 1) detail = self.req('GET', f"/api/v1/businesses/{businesses[0]['id']}")[1] self.assertTrue(detail['domains']); self.assertTrue(detail['websites']); self.assertTrue(detail['evidence']) self.assertTrue(detail['contacts']); self.assertEqual(detail['review_status'], 'pending') self.assertEqual(detail['score_version'], 'opportunity-v1') self.assertTrue(all({'code', 'name', 'points', 'version'} <= set(item) for item in detail['score_factors'])) db = sqlite3.connect(self.tmp.name + '/x.db') self.assertEqual(db.execute('SELECT processing_status FROM source_records').fetchone()[0], 'processed') self.assertEqual(db.execute('SELECT status FROM enrichment_queue').fetchone()[0], 'completed') db.close() status, second = self.req('POST', '/api/v1/discovery', {**payload, 'idempotency_key': 'source-run-2'}) self.assertEqual(status, 202) for _ in range(100): _, current = self.req('GET', f"/api/v1/jobs/{second['id']}") if current['status'] in ('succeeded', 'failed'): break threading.Event().wait(.01) self.assertEqual(current['status'], 'succeeded') self.assertEqual(len(self.req('GET', '/api/v1/businesses')[1]['items']), 1) self.assertEqual(self.req('GET', '/api/v1/source-records')[1]['items'].__len__(), 1) def test_source_limits_and_run_lifecycle_are_enforced(self): status, source = self.req('POST', '/api/v1/sources', {'name': 'Limited', 'kind': 'manual', 'enabled': True, 'config': {'rows': [{'name': 'A'}, {'name': 'B'}]}, 'quota': {'daily_limit': 1, 'per_run_limit': 1}}) self.assertEqual(status, 201) status, job = self.req('POST', '/api/v1/discovery', {'criteria': {}, 'selected_adapters': ['manual'], 'idempotency_key': 'limited-1', 'max_records': 10, 'daily_limit': 1}) self.assertEqual(status, 202) runs = self.req('GET', '/api/v1/discovery-runs')[1]['items']; rid = runs[0]['id'] self.assertEqual(self.req('POST', f'/api/v1/discovery-runs/{rid}/pause', {})[0], 200) self.assertEqual(self.req('POST', f'/api/v1/discovery-runs/{rid}/resume', {})[0], 200) self.assertEqual(self.req('POST', f'/api/v1/discovery-runs/{rid}/cancel', {})[0], 200) db = sqlite3.connect(self.tmp.name + '/x.db') self.assertEqual(db.execute("SELECT lifecycle FROM discovery_runs WHERE id=?", (rid,)).fetchone()[0], 'cancelled') self.assertIn(db.execute("SELECT status FROM jobs WHERE id=?", (job['id'],)).fetchone()[0], ('cancelled', 'succeeded')) db.close() if __name__=='__main__': unittest.main()