Merge pull request #33 from Theodo-UK/feat/console-log-fusion

Fuse the data from the request and the response console logs

sdk-python/omnilogger/logger.py

@@ -18,48 +18,58 @@ def start_listener(url: str):
     sys.stderr = StreamToLogger(logger, logging.ERROR)
 
 
-def handle_openai_response(data: str, url: str):
-    output = data.split("body='")[1].split("' headers=")[0]
-    output = bytes(output, "utf-8").decode("unicode_escape")
-    output = json.loads(output)
-
-    log = {
-        "input": "This is a new dummy input!",
-        "datetime_utc": datetime.utcfromtimestamp(output["created"]),
-        "output": output["choices"][0]["text"],
-        "total_tokens": output["usage"]["total_tokens"],
-    }
-    send_to_db(url, log)
-
-
 class OpenAIFilter(logging.Filter):
     url: str
+    prompt: str
 
     def __init__(self, url):
         self.url = url
+        self.prompt = ""
 
     def filter(self, record) -> bool:
         msg = record.getMessage()
 
-        if "OpenAI" in msg or "openai" in msg:
+        if "OpenAI" in msg or "openai" in msg or "text-davinci-003" in msg:
             if "message='Request to OpenAI API'" in msg:
                 return False
 
             if "message='Post details'" in msg:
+                try:
+                    self.extract_prompt_from_request(msg)
+                except IndexError:
+                    record.msg = "Prompt not found in OpenAI request: " + msg
+                    return True
                 return False
 
             if "message='OpenAI API response'" in msg:
                 return False
 
             if "message='API response body'" in msg:
                 try:
-                    handle_openai_response(msg, self.url)
+                    self.handle_openai_response(msg)
                 except json.decoder.JSONDecodeError:
                     record.msg = "JSONDecodeError for OpenAI response: " + msg
                     return True
                 return False
         return True
 
+    def extract_output_from_response(self, data: str):
+        output = data.split("body='")[1].split("' headers=")[0]
+        output = bytes(output, "utf-8").decode("unicode_escape")
+        output = json.loads(output)
+
+        log = {
+            "input": self.prompt,
+            "datetime_utc": datetime.utcfromtimestamp(output["created"]),
+            "output": output["choices"][0]["text"],
+            "total_tokens": output["usage"]["total_tokens"],
+        }
+        send_to_db(self.url, log)
+
+    def extract_prompt_from_request(self, data: str):
+        self.prompt = data.split('"prompt": "')[1]
+        self.prompt = self.prompt.split('", "temperature"')[0]
+
 
 class StreamToLogger(object):
     """

sdk-python/pyproject.toml

@@ -20,12 +20,13 @@ classifiers = [
 ]
 
 [tool.poetry.dependencies]
-python = "^3.7.1"
+python = "^3.8"
 psycopg2-binary = "^2.9.7"
 openai = "^0.27.8"
 
 [tool.poetry.group.dev.dependencies]
 pytest = "^7.4.0"
+python-dotenv = "^1.0.0"
 
 [project.urls]
 "Homepage" = "https://github.com/Theodo-UK/OmniLog"

sdk-python/test/openai_filter_test.py

@@ -56,7 +56,7 @@ def test_filter_post_details(self):
         When it is processed by the filter,
         then the record is removed from the stream.
         """
-        self.input = "message='Post details' path=https://api.openai.com"
+        self.input = 'message=\'Post details\' path=https://api.openai.com body={"prompt": "This is a prompt"}'
         self.compute_result("test_filter_post_details")
         self.assertEqual(self.result, "")
 

sdk-python/test/playground.py

@@ -0,0 +1,18 @@
+import os
+from dotenv import load_dotenv
+import openai
+
+from omnilogger import start_listener
+
+load_dotenv()
+DATABASE_URL = os.getenv("DATABASE_URL")
+OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
+
+start_listener(DATABASE_URL)
+
+openai.api_key = OPENAI_API_KEY
+openai.Completion.create(
+    model="text-davinci-003",
+    prompt="How are rubber ducks used in IT",
+    temperature=0.6,
+)