assemblyline-v4-service 4.5.1.dev76__py3-none-any.whl → 4.5.1.dev77__py3-none-any.whl

assemblyline_v4_service/VERSION

@@ -1 +1 @@
-4.5.1.dev76
+4.5.1.dev77

assemblyline_v4_service/common/base.py

@@ -19,6 +19,7 @@ from assemblyline.odm.messages.task import Task as ServiceTask
 from assemblyline_v4_service.common import helper
 from assemblyline_v4_service.common.api import PrivilegedServiceAPI, ServiceAPI
 from assemblyline_v4_service.common.ontology_helper import OntologyHelper
+from assemblyline_v4_service.common.ocr import update_ocr_config
 from assemblyline_v4_service.common.request import ServiceRequest
 from assemblyline_v4_service.common.task import Task
 
@@ -84,6 +85,10 @@ class ServiceBase:
         self.rules_hash: str = None
         self.signatures_meta: dict = {}
 
+        # OCR-related
+        if self.config.get('ocr'):
+            update_ocr_config(self.config['ocr'])
+
     @property
     def api_interface(self):
         return self.get_api_interface()

assemblyline_v4_service/common/ocr.py

@@ -1,6 +1,6 @@
 from __future__ import annotations
 
-from typing import Dict, List, TextIO
+from typing import Any, Dict, List, TextIO, Union
 
 from assemblyline_v4_service.common.helper import get_service_manifest
 from assemblyline_v4_service.common.utils import PASSWORD_WORDS
@@ -154,44 +154,47 @@ OCR_INDICATORS_TERMS: dict[str, list[str]] = {
 # The minimum number of indicator hits to avoid FP detections
 OCR_INDICATORS_THRESHOLD: Dict[str, int] = {"ransomware": 2, "macros": 2, "banned": 1, "password": 1}
 
-# Pre-compute the OCR_INDICATOR_* constants on module load so we only load the manifest once rather than per OCR request
-try:
-    # Retrieve service-configured OCR settings on module load
-    ocr_config: Dict = get_service_manifest().get("config", {}).get("ocr", {})
-except Exception:
-    # Service manifest not found
-    ocr_config = {}
-
-indicators = set(list(OCR_INDICATORS_TERMS.keys()) + list(ocr_config.keys()))
-# Iterate over the different indicators and include lines of detection in response
-for indicator in indicators:
-    indicator_config = ocr_config.get(indicator)
-    terms = OCR_INDICATORS_TERMS.get(indicator, [])
-    hit_threshold = OCR_INDICATORS_THRESHOLD.get(indicator, 1)
-    # Backwards compatibility: Check how the OCR configuration is formatted
-    if not indicator_config:
-        # Empty block/no override provided by service
-        pass
-    elif isinstance(indicator_config, list):
-        # Legacy support (before configurable indicator thresholds)
-        terms = indicator_config
-    elif isinstance(indicator_config, dict):
-        # Either you're exclusively overwriting the terms list or you're selectively including/excluding terms
-        if indicator_config.get("terms"):
-            # Overwrite terms list with service configuration
-            terms = indicator_config["terms"]
-        else:
-            included_terms = set(indicator_config.get("include", []))
-            excluded_terms = set(indicator_config.get("exclude", []))
-            # Compute the new terms list for indicator type
-            terms = list(set(terms).union(included_terms) - excluded_terms)
-
-        # Set the indicator hit threshold
-        hit_threshold = indicator_config.get("threshold", 1)
-
-    # Overwrite key-value in respective constants
-    OCR_INDICATORS_TERMS[indicator] = terms
-    OCR_INDICATORS_THRESHOLD[indicator] = hit_threshold
+def update_ocr_config(ocr_config: Dict[str, Union[List[str], Dict[str, Any]]] = None):
+    global OCR_INDICATORS_TERMS
+    global OCR_INDICATORS_THRESHOLD
+    if not ocr_config:
+        try:
+            # Retrieve service-configured OCR settings on module load (primary used in testing)
+            ocr_config: Dict = get_service_manifest().get("config", {}).get("ocr", {})
+        except Exception:
+            # No configuration updates provided
+            return
+
+    indicators = set(list(OCR_INDICATORS_TERMS.keys()) + list(ocr_config.keys()))
+    # Iterate over the different indicators and include lines of detection in response
+    for indicator in indicators:
+        indicator_config = ocr_config.get(indicator)
+        terms = OCR_INDICATORS_TERMS.get(indicator, [])
+        hit_threshold = OCR_INDICATORS_THRESHOLD.get(indicator, 1)
+        # Backwards compatibility: Check how the OCR configuration is formatted
+        if not indicator_config:
+            # Empty block/no override provided by service
+            pass
+        elif isinstance(indicator_config, list):
+            # Legacy support (before configurable indicator thresholds)
+            terms = indicator_config
+        elif isinstance(indicator_config, dict):
+            # Either you're exclusively overwriting the terms list or you're selectively including/excluding terms
+            if indicator_config.get("terms"):
+                # Overwrite terms list with service configuration
+                terms = indicator_config["terms"]
+            else:
+                included_terms = set(indicator_config.get("include", []))
+                excluded_terms = set(indicator_config.get("exclude", []))
+                # Compute the new terms list for indicator type
+                terms = list(set(terms).union(included_terms) - excluded_terms)
+
+            # Set the indicator hit threshold
+            hit_threshold = indicator_config.get("threshold", 1)
+
+        # Overwrite key-value in respective constants
+        OCR_INDICATORS_TERMS[indicator] = terms
+        OCR_INDICATORS_THRESHOLD[indicator] = hit_threshold
 
 
 def ocr_detections(image_path: str, ocr_io: TextIO = None) -> Dict[str, List[str]]:

{assemblyline_v4_service-4.5.1.dev76.dist-info → assemblyline_v4_service-4.5.1.dev77.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: assemblyline-v4-service
-Version: 4.5.1.dev76
+Version: 4.5.1.dev77
 Summary: Assemblyline 4 - Service base
 Home-page: https://github.com/CybercentreCanada/assemblyline-v4-service/
 Author: CCCS Assemblyline development team

{assemblyline_v4_service-4.5.1.dev76.dist-info → assemblyline_v4_service-4.5.1.dev77.dist-info}/RECORD

@@ -1,4 +1,4 @@
-assemblyline_v4_service/VERSION,sha256=iqWsiCUB4qG6Ft4zqwxt5BpBP0Cew2wVsHSog45Go9Y,12
+assemblyline_v4_service/VERSION,sha256=s6mmrhL6AATHlUQnS9l3hT6r003DPEuWyYxn0OsQo8c,12
 assemblyline_v4_service/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 assemblyline_v4_service/healthz.py,sha256=sS1cFkDLw8hUPMpj7tbHXFv8ZmHcazrwZ0l6oQDwwkQ,1575
 assemblyline_v4_service/py.typed,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -6,9 +6,9 @@ assemblyline_v4_service/run_privileged_service.py,sha256=qd4DmHo5G_Tpv8tb0A96qNf
 assemblyline_v4_service/run_service.py,sha256=NiFX52NfsbBQY6E3nrjoB3e2XxIlwFcQpYYY-rADIk4,5996
 assemblyline_v4_service/common/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 assemblyline_v4_service/common/api.py,sha256=Xzp8j4HCCfjPvNSGKiZl5ttH2_Itg47cjlH0NXNtth0,6849
-assemblyline_v4_service/common/base.py,sha256=mKkkzbxVL_wVMy_VieU9mlHYLqZXndga_4dWWbnEnx8,14045
+assemblyline_v4_service/common/base.py,sha256=1wxzg_enESdqxloiAWCL0fUNzcKg0kAVfNNkGAfHX0U,14218
 assemblyline_v4_service/common/helper.py,sha256=xs9quuf-M1JOdKieBqOmWaOece0CtzXFhhe85xQYmuY,3289
-assemblyline_v4_service/common/ocr.py,sha256=dzx5n9fzOHCVX3NnGKONpGAF3Cq16hw_y7M7g_nsJ7A,8588
+assemblyline_v4_service/common/ocr.py,sha256=3fV0PyY3oui_ucAM9dkolP0VRYKACKJuGY4M64DudIE,8841
 assemblyline_v4_service/common/ontology_helper.py,sha256=QpwerYoS5hXjWzpx3Pmwv6j2330PQVYqxYGamjcpW3I,7890
 assemblyline_v4_service/common/request.py,sha256=XXBafAQCV43_OBLXOSHxYoDHmqwERBkNul8fb_X6Ves,11774
 assemblyline_v4_service/common/result.py,sha256=9AqM6qCYiia_Bpyn_fBFhzNQMcqJbtFSiGjp57fXW2E,32713
@@ -32,14 +32,14 @@ test/test_common/__init__.py,sha256=RkOm3vnVp5L947mD1jTo4bdOgLTZJ24_NX-kqfMn5a8,
 test/test_common/test_api.py,sha256=7wlo7wgB12T23zMLbwjJ3GIomLHqE_Qvs3xkibSsR1U,4902
 test/test_common/test_base.py,sha256=fuJSSlPxIDHq6HU1xbvaMFitw2z1spOZNHD2SJ4UUic,13346
 test/test_common/test_helper.py,sha256=sO6YAiBhKTqaxlpLhFYDuy2ZdbuF2cg07Ylzo83ZzQs,2575
-test/test_common/test_ocr.py,sha256=nel1GCkieDRW2F_6kYCbkIwB9Kwj_d2rJBgb8VZWXS8,1685
+test/test_common/test_ocr.py,sha256=mt_PgElgwQKJmNrp2nRVx9NjfMedVk40I6IV317vATI,1753
 test/test_common/test_ontology_helper.py,sha256=KhHEBg_ecJyQbDw79NMT4FzUyA4C1Aak3HEQCwBfM2s,7914
 test/test_common/test_request.py,sha256=PPhHfrwpwMdNZ33P1Z_0h1Zaz9ao9VFiDr_MJrBS3Lg,11492
 test/test_common/test_result.py,sha256=b96bCfyW0ukdTcCsl01jS_l5YhfzXFVYs_VPOwz7IEU,41982
 test/test_common/test_task.py,sha256=jnfF68EgJIu30Pz_4jiJHkncfI-3XpGaut5r79KIXOA,18718
 test/test_common/test_utils.py,sha256=TbnBxqpS_ZC5ptXR9XJX3xtbItD0mTbtiBxxdyP8J5k,5904
-assemblyline_v4_service-4.5.1.dev76.dist-info/LICENCE.md,sha256=NSkYo9EH8h5oOkzg4VhjAHF4339MqPP2cQ8msTPgl-c,1396
-assemblyline_v4_service-4.5.1.dev76.dist-info/METADATA,sha256=WiZF51ibmkKEUtffMRNejEeP4MznC-c1fiujruFi4hU,9498
-assemblyline_v4_service-4.5.1.dev76.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
-assemblyline_v4_service-4.5.1.dev76.dist-info/top_level.txt,sha256=LpTOEaVCatkrvbVq3EZseMSIa2PQZU-2rhuO_FTpZgY,29
-assemblyline_v4_service-4.5.1.dev76.dist-info/RECORD,,
+assemblyline_v4_service-4.5.1.dev77.dist-info/LICENCE.md,sha256=NSkYo9EH8h5oOkzg4VhjAHF4339MqPP2cQ8msTPgl-c,1396
+assemblyline_v4_service-4.5.1.dev77.dist-info/METADATA,sha256=MGwF9WTfmHQfFbCpz0QiOFpdphkhqG7xuOAVdkuRxDY,9498
+assemblyline_v4_service-4.5.1.dev77.dist-info/WHEEL,sha256=GJ7t_kWBFywbagK5eo9IoUwLW6oyOeTKmQ-9iHFVNxQ,92
+assemblyline_v4_service-4.5.1.dev77.dist-info/top_level.txt,sha256=LpTOEaVCatkrvbVq3EZseMSIa2PQZU-2rhuO_FTpZgY,29
+assemblyline_v4_service-4.5.1.dev77.dist-info/RECORD,,

test/test_common/test_ocr.py

@@ -3,10 +3,11 @@ from test.test_common import TESSERACT_LIST
 
 import pytest
 
-from assemblyline_v4_service.common.ocr import ocr_detections, detections
+from assemblyline_v4_service.common.ocr import ocr_detections, detections, update_ocr_config
 
 @pytest.mark.skipif(len(TESSERACT_LIST) < 1, reason="Requires tesseract-ocr apt package")
 def test_ocr_detections():
+    update_ocr_config()
     file_path = os.path.join(os.path.dirname(__file__), "b32969aa664e3905c20f865cdd7b921f922678f5c3850c78e4c803fbc1757a8e")
     assert ocr_detections(file_path) == {
         'ransomware': [
@@ -27,6 +28,8 @@ def test_ocr_detections():
 
 
 def test_detections():
+    update_ocr_config()
+
     # No detection
     assert detections("blah") == {}