feat(error-capture): Add basic exception autocapture (#128)

Author: neilkakkar

CHANGELOG.md

@@ -1,3 +1,7 @@
+## 3.6.0 - 2024-08-28
+
+1. Adds exception autocapture in alpha state. This feature is not yet stable and may change in future versions.
+
 ## 3.5.2 - 2024-08-21
 
 1. Guard for None values in local evaluation

posthog/__init__.py

@@ -19,6 +19,8 @@
 poll_interval = 30  # type: int
 disable_geoip = True  # type: bool
 feature_flags_request_timeout_seconds = 3  # type: int
+# Currently alpha, use at your own risk
+enable_exception_autocapture = False  # type: bool
 
 default_client = None  # type: Optional[Client]
 
@@ -454,6 +456,10 @@ def _proxy(method, *args, **kwargs):
             disabled=disabled,
             disable_geoip=disable_geoip,
             feature_flags_request_timeout_seconds=feature_flags_request_timeout_seconds,
+            # TODO: Currently this monitoring begins only when the Client is initialised (which happens when you do something with the SDK)
+            # This kind of initialisation is very annoying for exception capture. We need to figure out a way around this,
+            # or deprecate this proxy option fully (it's already in the process of deprecation, no new clients should be using this method since like 5-6 months)
+            enable_exception_autocapture=enable_exception_autocapture,
         )
 
     # always set incase user changes it

posthog/client.py

@@ -8,6 +8,7 @@
 from six import string_types
 
 from posthog.consumer import Consumer
+from posthog.exception_capture import ExceptionCapture
 from posthog.feature_flags import InconclusiveMatchError, match_feature_flag_properties
 from posthog.poller import Poller
 from posthog.request import APIError, batch_post, decide, determine_server_host, get
@@ -51,6 +52,7 @@ def __init__(
         disable_geoip=True,
         historical_migration=False,
         feature_flags_request_timeout_seconds=3,
+        enable_exception_autocapture=False,
     ):
         self.queue = queue.Queue(max_queue_size)
 
@@ -77,6 +79,7 @@ def __init__(
         self.disabled = disabled
         self.disable_geoip = disable_geoip
         self.historical_migration = historical_migration
+        self.enable_exception_autocapture = enable_exception_autocapture
 
         # personal_api_key: This should be a generated Personal API Key, private
         self.personal_api_key = personal_api_key
@@ -88,6 +91,9 @@ def __init__(
         else:
             self.log.setLevel(logging.WARNING)
 
+        if self.enable_exception_autocapture:
+            self.exception_capture = ExceptionCapture(self)
+
         if sync_mode:
             self.consumers = None
         else:

posthog/exception_capture.py

@@ -0,0 +1,53 @@
+import logging
+import sys
+import threading
+from typing import TYPE_CHECKING
+
+from posthog.exception_utils import exceptions_from_error_tuple
+
+if TYPE_CHECKING:
+    from posthog.client import Client
+
+
+class ExceptionCapture:
+    # TODO: Add client side rate limiting to prevent spamming the server with exceptions
+
+    log = logging.getLogger("posthog")
+
+    def __init__(self, client: "Client"):
+        self.client = client
+        self.original_excepthook = sys.excepthook
+        sys.excepthook = self.exception_handler
+        threading.excepthook = self.thread_exception_handler
+
+    def exception_handler(self, exc_type, exc_value, exc_traceback):
+        # don't affect default behaviour.
+        self.capture_exception(exc_type, exc_value, exc_traceback)
+        self.original_excepthook(exc_type, exc_value, exc_traceback)
+
+    def thread_exception_handler(self, args):
+        self.capture_exception(args.exc_type, args.exc_value, args.exc_traceback)
+
+    def capture_exception(self, exc_type, exc_value, exc_traceback):
+        try:
+            # if hasattr(sys, "ps1"):
+            #     # Disable the excepthook for interactive Python shells
+            #     return
+
+            # Format stack trace like sentry
+            all_exceptions_with_trace = exceptions_from_error_tuple((exc_type, exc_value, exc_traceback))
+
+            properties = {
+                "$exception_type": all_exceptions_with_trace[0].get("type"),
+                "$exception_message": all_exceptions_with_trace[0].get("value"),
+                "$exception_list": all_exceptions_with_trace,
+                # TODO: Can we somehow get distinct_id from context here? Stateless lib makes this much harder? 😅
+                # '$exception_personURL': f'{self.client.posthog_host}/project/{self.client.token}/person/{self.client.get_distinct_id()}'
+            }
+
+            # TODO: What distinct id should we attach these server-side exceptions to?
+            # Any heuristic seems prone to errors - how can we know if exception occurred in the context of a user that captured some other event?
+
+            self.client.capture("python-exceptions", "$exception", properties=properties)
+        except Exception as e:
+            self.log.exception(f"Failed to capture exception: {e}")