Enable S rule

.ruff.toml

@@ -1,4 +1,4 @@
-select = ["A", "ASYNC", "I", "E", "F", "B", "C4", "T10", "T20"]
+select = ["A", "ASYNC", "I", "E", "F", "B", "C4", "T10", "T20", "S"]
 ignore = ["E501"]
 
 # Allow autofix for all enabled rules (when `--fix`) is provided.
@@ -16,7 +16,8 @@ dummy-variable-rgx = "^(_+|(_+[a-zA-Z0-9_]*[a-zA-Z0-9]+?))$"
 target-version = "py310"
 
 [per-file-ignores]
-"tests/*" = ["B017"]
+"tests/*" = ["B017", "S101", "S"]
+"connectors/*" = ["S608"]
 
 [isort]
 known-first-party=["connectors", "tests"]

connectors/cli.py

@@ -25,7 +25,6 @@
 from connectors.preflight_check import PreflightCheck
 from connectors.services import get_services
 from connectors.source import get_source_klass, get_source_klasses
-from connectors.utils import get_event_loop
 
 __all__ = ["main"]
 
@@ -128,6 +127,28 @@ async def _start_service(actions, config, loop):
         return await multiservice.run()
 
 
+def get_event_loop(uvloop=False):
+    if uvloop:
+        # activate uvloop if lib is present
+        try:
+            import uvloop
+
+            asyncio.set_event_loop_policy(uvloop.EventLoopPolicy())
+        except Exception:
+            logger.warning(
+                "Unable to enable uvloop: {e}. Running with default event loop"
+            )
+            pass
+    try:
+        loop = asyncio.get_running_loop()
+    except RuntimeError:
+        loop = asyncio.get_event_loop_policy().get_event_loop()
+        if loop is None:
+            loop = asyncio.new_event_loop()
+            asyncio.set_event_loop(loop)
+    return loop
+
+
 def run(args):
     """Loads the config file, sets the logger and executes an action.
 

connectors/sources/directory.py

@@ -7,15 +7,14 @@
 Demo of a standalone source
 """
 import functools
-import hashlib
 import os
 from datetime import datetime, timezone
 from pathlib import Path
 
 import aiofiles
 
 from connectors.source import BaseDataSource
-from connectors.utils import TIKA_SUPPORTED_FILETYPES, get_base64_value
+from connectors.utils import TIKA_SUPPORTED_FILETYPES, get_base64_value, hash_id
 
 DEFAULT_DIR = os.environ.get("SYSTEM_DIR", os.path.dirname(__file__))
 
@@ -57,7 +56,7 @@ async def changed(self):
         return True
 
     def get_id(self, path):
-        return hashlib.md5(str(path).encode("utf8")).hexdigest()
+        return hash_id(str(path))
 
     async def _download(self, path, timestamp=None, doit=None):
         if not (doit and os.path.splitext(path)[-1] in TIKA_SUPPORTED_FILETYPES):

connectors/sources/dropbox.py

@@ -64,7 +64,7 @@
 
 
 class EndpointName(Enum):
-    ACCESS_TOKEN = "access_token"
+    ACCESS_TOKEN = "access_token"  # noqa S105
     PING = "ping"
     CHECK_PATH = "check_path"
     FILES_FOLDERS = "files_folders"

connectors/sources/microsoft_teams.py

@@ -60,7 +60,7 @@
     GRAPH_ACQUIRE_TOKEN_URL = override_url
 else:
     GRAPH_API_AUTH_URL = "https://login.microsoftonline.com"
-    GRAPH_ACQUIRE_TOKEN_URL = "https://graph.microsoft.com/.default"
+    GRAPH_ACQUIRE_TOKEN_URL = "https://graph.microsoft.com/.default"  # noqa S105
     BASE_URL = "https://graph.microsoft.com/v1.0"
 
 SCOPE = [

connectors/sources/outlook.py

@@ -617,7 +617,7 @@ async def get_mails(self, account):
                     folder_object = (
                         account.root / "Top of Information Store" / "Archive"
                     )
-                except Exception:
+                except Exception:  # noqa S112
                     continue
             else:
                 folder_object = getattr(account, mail_type["folder"])

connectors/sources/s3.py

@@ -8,7 +8,6 @@
 import os
 from contextlib import AsyncExitStack
 from functools import partial
-from hashlib import md5
 
 import aioboto3
 from aiobotocore.config import AioConfig
@@ -17,6 +16,7 @@
 
 from connectors.logger import logger, set_extra_logger
 from connectors.source import BaseDataSource
+from connectors.utils import hash_id
 
 DEFAULT_PAGE_SIZE = 100
 DEFAULT_MAX_RETRY_ATTEMPTS = 5
@@ -190,7 +190,7 @@ async def format_document(self, bucket_name, bucket_object):
             document: Modified document.
         """
 
-        doc_id = md5(f"{bucket_name}/{bucket_object.key}".encode("utf8")).hexdigest()
+        doc_id = hash_id(f"{bucket_name}/{bucket_object.key}")
         owner = await bucket_object.owner
         document = {
             "_id": doc_id,

connectors/sources/salesforce.py

@@ -29,7 +29,7 @@
 
 BASE_URL = "https://<domain>.my.salesforce.com"
 API_VERSION = "v58.0"
-TOKEN_ENDPOINT = "/services/oauth2/token"
+TOKEN_ENDPOINT = "/services/oauth2/token"  # noqa S105
 QUERY_ENDPOINT = f"/services/data/{API_VERSION}/query"
 DESCRIBE_ENDPOINT = f"/services/data/{API_VERSION}/sobjects"
 DESCRIBE_SOBJECT_ENDPOINT = f"/services/data/{API_VERSION}/sobjects/<sobject>/describe"

connectors/utils.py

@@ -210,14 +210,15 @@ def convert_to_b64(source, target=None, overwrite=False):
             version = int(platform.mac_ver()[0].split(".")[0])
             # MacOS 13 has changed base64 util
             if version >= 13:
-                cmd = f"{_BASE64} -i {source} -o {temp_target}"
+                cmd = [_BASE64, "-i", source, "-o", temp_target]
             else:
-                cmd = f"{_BASE64} {source} > {temp_target}"
+                cmd = [_BASE64, source, ">", temp_target]
         else:
             # In Linuces, avoid line wrapping
-            cmd = f"{_BASE64} -w 0 {source} > {temp_target}"
+            cmd = [_BASE64, "-w", "0", source, ">", temp_target]
         logger.debug(f"Calling {cmd}")
-        subprocess.check_call(cmd, shell=True)
+        # TODO: make it more robust
+        subprocess.check_call(cmd)  # noqa S603
     else:
         # Pure Python version
         with open(source, "rb") as sf, open(temp_target, "wb") as tf:
@@ -408,25 +409,6 @@ def cancel(self):
             task.cancel()
 
 
-def get_event_loop(uvloop=False):
-    if uvloop:
-        # activate uvloop if lib is present
-        try:
-            import uvloop
-
-            asyncio.set_event_loop_policy(uvloop.EventLoopPolicy())
-        except Exception:
-            pass
-    try:
-        loop = asyncio.get_running_loop()
-    except RuntimeError:
-        loop = asyncio.get_event_loop_policy().get_event_loop()
-        if loop is None:
-            loop = asyncio.new_event_loop()
-            asyncio.set_event_loop(loop)
-    return loop
-
-
 class RetryStrategy(Enum):
     CONSTANT = 0
     LINEAR_BACKOFF = 1
@@ -638,7 +620,9 @@ def get_pem_format(key, postfix="-----END CERTIFICATE-----"):
 
 def hash_id(_id):
     # Collision probability: 1.47*10^-29
-    return hashlib.md5(_id.encode("utf8")).hexdigest()
+    # S105 rule considers this code unsafe, but we're not using it for security-related
+    # things, only to generate pseudo-ids for documents
+    return hashlib.md5(_id.encode("utf8")).hexdigest()  # noqa S105
 
 
 def truncate_id(_id):