Feature/e2e with karton services + other test fixes (#572)

Co-authored-by: Paweł Srokosz <[email protected]>

docker-compose-e2e.yml

@@ -11,12 +11,23 @@ services:
       - redis
     image: certpl/mwdb
     restart: on-failure
-    env_file:
-      # NOTE: use gen_vars.sh in order to generate this file
-      - mwdb-vars.env
     environment:
+      MWDB_ENABLE_KARTON: 1
+      MWDB_REDIS_URI: redis://redis/
+      MWDB_POSTGRES_URI: postgresql://mwdb:e2e-postgres-password@postgres/mwdb
       # Hardcoded secret key for consistent JWT testing
-      - MWDB_SECRET_KEY=e2e-testing-key
+      MWDB_SECRET_KEY: e2e-testing-key
+      MWDB_ADMIN_LOGIN: admin
+      MWDB_ADMIN_EMAIL: admin@localhost
+      MWDB_ADMIN_PASSWORD: e2e-mwdb-admin-password
+      MWDB_BASE_URL: http://127.0.0.1
+      MWDB_ENABLE_RATE_LIMIT: 0
+      MWDB_ENABLE_REGISTRATION: 1
+      UWSGI_PROCESSES: 4
+      MWDB_MAIL_SMTP: "mailhog:1025"
+      MWDB_MAIL_FROM: "[email protected]"
+    volumes:
+    - "./dev/karton.ini:/etc/karton/karton.ini"
   mwdb-web:
     depends_on:
       - mwdb
@@ -30,23 +41,100 @@ services:
   postgres:
     image: postgres
     restart: always
-    env_file:
-      # NOTE: use gen_vars.sh in order to generate this file
-      - postgres-vars.env
+    environment:
+      POSTGRES_USER: mwdb
+      POSTGRES_DB: mwdb
+      POSTGRES_PASSWORD: e2e-postgres-password
   mwdb-tests:
     build: tests/backend
     depends_on:
       - mwdb
+      - mwdb-web
+      - mailhog
+      - karton-system
+      - karton-classifier
+      - karton-dashboard
+      - karton-mwdb-reporter
     image: certpl/mwdb-tests
-    env_file:
-      - mwdb-vars.env
+    environment:
+      MWDB_ADMIN_LOGIN: admin
+      MWDB_ADMIN_PASSWORD: e2e-mwdb-admin-password
+      MWDB_URL: http://mwdb-web./api
   web-tests:
     build: tests/frontend
     depends_on:
       - mwdb
       - mwdb-web
+      - mailhog
+      - karton-system
+      - karton-classifier
+      - karton-dashboard
+      - karton-mwdb-reporter
     image: certpl/mwdb-web-tests
-    env_file:
-      - mwdb-vars.env
+    environment:
+      MWDB_ADMIN_LOGIN: admin
+      MWDB_ADMIN_PASSWORD: e2e-mwdb-admin-password
   redis:
     image: redis:alpine
+  minio:
+    image: minio/minio
+    command: server /data
+    volumes:
+      - /tmp/minio:/data
+    ports:
+      - "127.0.0.1:9000:9000"
+    environment:
+      - MINIO_ACCESS_KEY=mwdb-test-access
+      - MINIO_SECRET_KEY=mwdb-test-key
+  mailhog:
+    image: mailhog/mailhog:latest
+    ports:
+      - "127.0.0.1:8025:8025"
+  karton-system:
+    image: certpl/karton-system:79ffabd3607406f1cc494b024b44f4dd1a753095
+    depends_on:
+      - redis
+      - minio
+    volumes:
+      - "./dev/karton.ini:/etc/karton/karton.ini"
+    entrypoint: karton-system
+    command: --setup-bucket
+    environment:
+      KARTON_MWDB_API_URL: http://mwdb-web.:80/api/
+      KARTON_MWDB_USERNAME: admin
+      KARTON_MWDB_PASSWORD: e2e-mwdb-admin-password
+  karton-classifier:
+    image: certpl/karton-classifier:8f02c6e5188414f28d37515d903ce2b6955bc768
+    depends_on:
+      - redis
+      - minio
+    volumes:
+      - "./dev/karton.ini:/etc/karton/karton.ini"
+    environment:
+      KARTON_MWDB_API_URL: http://mwdb-web.:80/api/
+      KARTON_MWDB_USERNAME: admin
+      KARTON_MWDB_PASSWORD: e2e-mwdb-admin-password
+  karton-dashboard:
+    image: certpl/karton-dashboard:7b8ec871c4902cc7236eaa4782d9f7acdbc8809c
+    depends_on:
+      - redis
+      - minio
+    volumes:
+      - "./dev/karton.ini:/etc/karton/karton.ini"
+    ports:
+      - "127.0.0.1:8030:5000"
+    environment:
+      KARTON_MWDB_API_URL: http://mwdb-web.:80/api/
+      KARTON_MWDB_USERNAME: admin
+      KARTON_MWDB_PASSWORD: e2e-mwdb-admin-password
+  karton-mwdb-reporter:
+    image: certpl/karton-mwdb-reporter:6f935f6dfbf459e278a496586b581b4ba0d4cbad
+    depends_on:
+      - redis
+      - minio
+    volumes:
+      - "./dev/karton.ini:/etc/karton/karton.ini"
+    environment:
+      KARTON_MWDB_API_URL: http://mwdb-web.:80/api/
+      KARTON_MWDB_USERNAME: admin
+      KARTON_MWDB_PASSWORD: e2e-mwdb-admin-password

tests/backend/Dockerfile

@@ -1,7 +1,6 @@
 FROM python:3.6
-ENV MWDB_URL http://mwdb.:8080/api
 COPY requirements.txt /app/requirements.txt
 RUN pip3 install -r /app/requirements.txt
 COPY *.py /app/
 WORKDIR /app
-CMD ["pytest", "-v"]
+CMD ["pytest", "-v", "--reruns", "3", "--only-rerun", "ConnectionError"]

tests/backend/requirements.txt

@@ -1,5 +1,6 @@
 python-dateutil==2.7.0
 pyJWT==2.3.0
 pytest
+pytest-rerunfailures
 requests
 python-baseconv

tests/backend/test_api.py

@@ -129,7 +129,7 @@ def test_add_tags(admin_session):
         admin_session.add_tag(sample['id'], tag)
 
     tags_response = admin_session.get_tags(sample['id'])
-    tags_returned = [t['tag'] for t in tags_response]
+    tags_returned = [t['tag'] for t in tags_response if not t["tag"].startswith("misc:")]
 
     assert len(tags_returned) == len(tags_expected)
     assert all([t in tags_returned for t in tags_expected])
@@ -145,9 +145,10 @@ def test_delete_tags(admin_session):
     admin_session.add_tag(identifier, tag2)
     admin_session.delete_tag(identifier, tag1)
 
-    tags = admin_session.get_tags(identifier)
+    tags = [t['tag'] for t in admin_session.get_tags(identifier) if not t["tag"].startswith("misc:")]
     assert len(tags) == 1
-    assert tags[0]['tag'] == tag2
+    assert tag1 not in tags
+    assert tag2 in tags
 
 
 @pytest.mark.parametrize('num_comments', [

tests/backend/test_karton.py

@@ -0,0 +1,162 @@
+from .utils import random_name, rand_string, ShouldRaise
+
+
+def test_karton_analyses_after_adding_blob(admin_session):
+    test = admin_session
+    blob_name = rand_string(15)
+    blob = test.add_blob(None, blobname=blob_name, blobtype="inject", content="""
+    Binary junk: \x00\x01\x02\x03\x04\x05\x07
+    HELLO WORLD!
+    ========""" + random_name())
+    blob_dhash = blob["id"]
+    analyses = test.get_analyses(blob_dhash)["analyses"]
+    assert len(analyses) == 1
+
+
+def test_karton_analysis_after_adding_config(admin_session):
+    test = admin_session
+    blob_name = rand_string(15)
+    content = rand_string(15)
+    config_json = {
+        "cnc": [1, 2, 3],
+        "raw_cfg": {
+            "in-blob": {
+                "blob_name": blob_name,
+                "blob_type": "Blob type",
+                "content": content
+            }
+        }
+    }
+    config = test.add_config(None, "malwarex", config_json)
+    config_dhash = config["id"]
+    analyses = test.get_analyses(config_dhash)["analyses"]
+    assert len(analyses) == 1
+
+
+def test_karton_analysis_after_adding_sample(admin_session):
+    test = admin_session
+    file_name = rand_string(15)
+    file_content = rand_string() 
+    sample = test.add_sample(file_name, file_content)
+    sample_dhash = sample["id"]
+    analyses = test.get_analyses(sample_dhash)["analyses"]
+    assert len(analyses) == 1
+
+
+def test_karton_reanalyze_object_with_args(admin_session):
+    test = admin_session
+    blob_name = rand_string(15)
+    argument_key = rand_string(5)
+    argument_value = rand_string(5)
+    blob = test.add_blob(None, blobname=blob_name, blobtype="inject", content="""
+    Binary junk: \x00\x01\x02\x03\x04\x05\x07
+    HELLO WORLD!
+    ========""" + random_name())
+    blob_dhash = blob["id"]
+    new_analysis = test.reanalyze_object(blob_dhash, arguments={argument_key: argument_value})
+    analyses = test.get_analyses(blob_dhash)["analyses"]
+    assert new_analysis["arguments"] == {argument_key: argument_value}
+    assert len(analyses) == 2
+
+    incorrect_object_dhash = "abcdefghi"
+    with ShouldRaise(status_code=404):
+        test.reanalyze_object(incorrect_object_dhash)
+
+
+def test_karton_reanalyze_object_without_args(admin_session):
+    test = admin_session
+    file_name = rand_string(15)
+    file_content = rand_string() 
+    sample = test.add_sample(file_name, file_content)
+    sample_dhash = sample["id"]
+    test.reanalyze_object(sample_dhash)
+    analyses = test.get_analyses(sample_dhash)["analyses"]
+    assert len(analyses) == 2
+
+
+def test_get_karton_analysis_info(admin_session):
+    test = admin_session
+    blob_name = rand_string(15)
+    argument_key = rand_string(5)
+    argument_value = rand_string(5)
+    blob = test.add_blob(None, blobname=blob_name, blobtype="inject", content="""
+    Binary junk: \x00\x01\x02\x03\x04\x05\x07
+    HELLO WORLD!
+    ========""" + random_name())
+    blob_dhash = blob["id"]
+    new_analysis = test.reanalyze_object(blob_dhash, arguments={argument_key: argument_value})
+    analysis_id = new_analysis["id"]
+    analysis_info = test.get_analysis_info(blob_dhash, analysis_id)
+    assert analysis_id == analysis_info["id"]
+    assert analysis_info["arguments"] == {argument_key: argument_value}
+
+
+def test_assign_analysis_to_object(admin_session):
+    test = admin_session
+    file_name = rand_string(15)
+    file_content = rand_string() 
+    sample = test.add_sample(file_name, file_content)
+    sample_dhash = sample["id"]
+    new_analysis = test.reanalyze_object(sample_dhash)
+    analysis_id = new_analysis["id"]
+
+    blob_name = rand_string(15)
+    blob = test.add_blob(None, blobname=blob_name, blobtype="inject", content="""
+    Binary junk: \x00\x01\x02\x03\x04\x05\x07
+    HELLO WORLD!
+    ========""" + random_name())
+    blob_dhash = blob["id"]
+
+    test.assign_analysis_to_object(blob_dhash, analysis_id)    
+    analyses_ids = [analysis["id"] for analysis in test.get_analyses(blob_dhash)["analyses"]]
+    assert analysis_id in analyses_ids
+
+    incorrect_analysis_id = "b99249a0-ff33-4c93-a9f9-d854ab0ecb0"
+    with ShouldRaise(status_code=400):
+        test.assign_analysis_to_object(blob_dhash, incorrect_analysis_id)
+
+
+def test_unassign_analysis_from_object(admin_session):
+    test = admin_session
+    blob_name = rand_string(15)
+    blob = test.add_blob(None, blobname=blob_name, blobtype="inject", content="""
+    Binary junk: \x00\x01\x02\x03\x04\x05\x07
+    HELLO WORLD!
+    ========""" + random_name())
+    blob_dhash = blob["id"]
+
+    new_analysis_id = test.reanalyze_object(blob_dhash)["id"]
+
+    all_object_analyses = test.get_analyses(blob_dhash)["analyses"]
+    assert len(all_object_analyses) == 2
+
+    test.unassign_analysis_from_object(blob_dhash, new_analysis_id)
+    all_object_analyses = test.get_analyses(blob_dhash)["analyses"]
+    assert len(all_object_analyses) == 1
+
+    analyses_ids = [analysis["id"] for analysis in all_object_analyses]
+    assert new_analysis_id not in analyses_ids
+
+    incorrect_analysis_id = "b99249a0-ff33-4c93-a9f9-d854ab0ecb0"
+    with ShouldRaise(status_code=400):
+        test.unassign_analysis_from_object(blob_dhash, incorrect_analysis_id)
+
+
+def test_unassign_analysis_not_assigned_to_object(admin_session):
+    test = admin_session
+    file_name = rand_string(15)
+    file_content = rand_string() 
+    sample = test.add_sample(file_name, file_content)
+    sample_dhash = sample["id"]
+    sample_analysis = test.reanalyze_object(sample_dhash)
+    sample_analysis_id = sample_analysis["id"]
+
+    blob_name = rand_string(15)
+    blob = test.add_blob(None, blobname=blob_name, blobtype="inject", content="""
+    Binary junk: \x00\x01\x02\x03\x04\x05\x07
+    HELLO WORLD!
+    ========""" + random_name())
+    blob_dhash = blob["id"]
+
+    with ShouldRaise(status_code=404):
+        test.unassign_analysis_from_object(blob_dhash, sample_analysis_id)

tests/backend/test_search.py

@@ -80,8 +80,8 @@ def test_file_alternative_name_search(admin_session):
     assert len(found_objs_1) == 1
     assert len(found_objs_2) == 1
     assert len(found_objs_3) == 1
-    assert found_objs_1 == found_objs_2
-    assert found_objs_1 == found_objs_3
+    assert found_objs_1[0]["id"] == found_objs_2[0]["id"]
+    assert found_objs_1[0]["id"] == found_objs_3[0]["id"]
 
     # wildcard search
     found_objs_1_with_wildcard = test.search(f'file.name:{filename_main[:len(filename_main)// 2]}*')
@@ -90,8 +90,8 @@ def test_file_alternative_name_search(admin_session):
     assert len(found_objs_1_with_wildcard) == 1
     assert len(found_objs_2_with_wildcard) == 1
     assert len(found_objs_3_with_wildcard) == 1
-    assert found_objs_1_with_wildcard == found_objs_2_with_wildcard
-    assert found_objs_1_with_wildcard == found_objs_3_with_wildcard
+    assert found_objs_1_with_wildcard[0]["id"] == found_objs_2_with_wildcard[0]["id"]
+    assert found_objs_1_with_wildcard[0]["id"] == found_objs_3_with_wildcard[0]["id"]
 
 
 def test_search_tag(admin_session):
@@ -384,10 +384,6 @@ def test_search_date_time_relative(admin_session):
     assert len(found_objs) == 0
     found_objs = admin_session.search(f'upload_time:[1s TO *] AND tag:{tag}')
     assert len(found_objs) == 0
-    found_objs = admin_session.search(f'upload_time:>=2s AND tag:{tag}')
-    assert len(found_objs) == 1
-    found_objs = admin_session.search(f'upload_time:[2s TO *] AND tag:{tag}')
-    assert len(found_objs) == 1
     found_objs = admin_session.search(f'upload_time:>=4s AND tag:{tag}')
     assert len(found_objs) == 2
     found_objs = admin_session.search(f'upload_time:[4s TO *] AND tag:{tag}')
@@ -416,8 +412,6 @@ def test_search_date_time_relative(admin_session):
     assert len(found_objs) == 2
     found_objs = admin_session.search(f'upload_time:[5y TO *] AND tag:{tag}')
     assert len(found_objs) == 2
-    found_objs = admin_session.search(f'upload_time:[2s TO 1s] AND tag:{tag}')
-    assert len(found_objs) == 1
 
     with ShouldRaise(status_code=400):
         found_objs = admin_session.search(f'upload_time:>=s1 AND tag:{tag}')
@@ -558,6 +552,7 @@ def test_uploader_query(admin_session):
     ]
     assert sorted(results) == sorted([FileC.dhash])
 
+
 def test_search_multi(admin_session):
     test = admin_session