make all the things strings

src/supergood/client.py

@@ -13,7 +13,7 @@
 
 from .api import Api
 from .constants import *
-from .helpers import redact_values, safe_decode, safe_parse_json
+from .helpers import decode_headers, redact_values, safe_decode, safe_parse_json
 from .logger import Logger
 from .remote_config import get_vendor_endpoint_from_config, parse_remote_config_json
 from .repeating_thread import RepeatingThread
@@ -158,6 +158,7 @@ def _should_ignore(
     def _cache_request(self, request_id, url, method, body, headers):
         request = {}
         try:
+            url = safe_decode(url)  # we do this first so the urlparse isn't also bytes
             host_domain = urlparse(url).hostname
             request["metadata"] = {}
             # Check that we should cache the request
@@ -178,11 +179,11 @@ def _cache_request(self, request_id, url, method, body, headers):
                 filtered_headers = (
                     {}
                     if (not self.base_config["logRequestHeaders"] or headers is None)
-                    else dict(headers)
+                    else decode_headers(dict(headers))
                 )
                 request["request"] = {
                     "id": request_id,
-                    "method": method,
+                    "method": safe_decode(method),
                     "url": url,
                     "body": filtered_body,
                     "headers": filtered_headers,
@@ -216,13 +217,13 @@ def _cache_response(
                 filtered_headers = (
                     {}
                     if not self.base_config["logResponseHeaders"]
-                    else dict(response_headers)
+                    else decode_headers(dict(response_headers))
                 )
                 response = {
                     "body": filtered_body,
                     "headers": filtered_headers,
                     "status": response_status,
-                    "statusText": response_status_text,
+                    "statusText": safe_decode(response_status_text),
                     "respondedAt": datetime.now().isoformat(),
                 }
                 self._response_cache[request_id] = {
@@ -324,8 +325,17 @@ def flush_cache(self, force=False) -> None:
                 self.log.debug(f"Flushing {len(data)} items")
                 self.api.post_events(data)
         except Exception:
-            payload = self._build_log_payload()
             trace = "".join(traceback.format_exc())
+            try:
+                urls = []
+                for entry in data:
+                    if entry.get("request", None):
+                        urls.append(entry.get("request").get("url"))
+                num_events = len(data)
+                payload = self._build_log_payload(num_events=num_events, urls=urls)
+            except Exception:
+                # something is really messed up, just report out
+                payload = self._build_log_payload()
             self.log.error(ERRORS["POSTING_EVENTS"], trace, payload)
         finally:  # always occurs, even from internal returns
             for response_key in response_keys:

src/supergood/helpers.py

@@ -264,10 +264,21 @@ def safe_decode(input, encoding="utf-8"):
     try:
         if isinstance(input, bytes) and input[:2] == GZIP_START_BYTES:
             return gzip.decompress(input)
-
         if isinstance(input, str):
             return input
-
         return input.decode(encoding)
     except Exception:
         return str(input)
+
+
+def decode_headers(headers, encoding="utf-8"):
+    new_headers = {}
+    for key, value in headers.items():
+        decoded_key = key
+        if isinstance(key, bytes):
+            decoded_key = safe_decode(key, encoding)
+        decoded_value = value
+        if isinstance(value, bytes):
+            decoded_value = safe_decode(value, encoding)
+        new_headers[decoded_key] = decoded_value
+    return new_headers