Merge pull request middlewarehq#548 from VipinDevelops/fix/middleware…

…hq#546

[FIX]: Handle Error response from AI API

backend/analytics_server/mhq/service/ai/ai_analytics_service.py

@@ -2,7 +2,7 @@
 import requests
 from http import HTTPStatus
 from enum import Enum
-from typing import Dict, List
+from typing import Dict, List, Union
 
 
 class AIProvider(Enum):
@@ -44,7 +44,30 @@ def __init__(self, llm: LLM, access_token: str):
     def _get_message(self, message: str, role: str = "user"):
         return {"role": role, "content": message}
 
+    def _handle_api_response(self, response) -> Dict[str, Union[str, int]]:
+        """
+        Handles the API response, returning a success or error structure that the frontend can use.
+        """
+        if response.status_code == HTTPStatus.OK:
+            return {
+                "status": "success",
+                "data": response.json()["choices"][0]["message"]["content"],
+            }
+        elif response.status_code == HTTPStatus.UNAUTHORIZED:
+            return {
+                "status": "error",
+                "message": "Unauthorized Access: Your access token is either missing, expired, or invalid. Please ensure that you are providing a valid token. ",
+            }
+        else:
+            return {
+                "status": "error",
+                "message": f"Unexpected error: {response.text}",
+            }
+
     def _open_ai_fetch_completion_open_ai(self, messages: List[Dict[str, str]]):
+        """
+        Handles the request to OpenAI API for fetching completions.
+        """
         payload = {
             "model": self.LLM_NAME_TO_MODEL_MAP[self._llm],
             "temperature": 0.6,
@@ -53,13 +76,12 @@ def _open_ai_fetch_completion_open_ai(self, messages: List[Dict[str, str]]):
         api_url = "https://api.openai.com/v1/chat/completions"
         response = requests.post(api_url, headers=self._headers, json=payload)
 
-        print(payload, api_url, response)
-        if response.status_code != HTTPStatus.OK:
-            raise Exception(response.json())
-
-        return response.json()
+        return self._handle_api_response(response)
 
     def _fireworks_ai_fetch_completions(self, messages: List[Dict[str, str]]):
+        """
+        Handles the request to Fireworks AI API for fetching completions.
+        """
         payload = {
             "model": self.LLM_NAME_TO_MODEL_MAP[self._llm],
             "temperature": 0.6,
@@ -73,28 +95,28 @@ def _fireworks_ai_fetch_completions(self, messages: List[Dict[str, str]]):
         api_url = "https://api.fireworks.ai/inference/v1/chat/completions"
         response = requests.post(api_url, headers=self._headers, json=payload)
 
-        if response.status_code != HTTPStatus.OK:
-            raise Exception(response.json())
-
-        return response.json()
-
-    def _fetch_completion(self, messages: List[Dict[str, str]]):
+        return self._handle_api_response(response)
 
+    def _fetch_completion(
+        self, messages: List[Dict[str, str]]
+    ) -> Dict[str, Union[str, int]]:
+        """
+        Fetches the completion using the appropriate AI provider based on the LLM.
+        """
         if self._ai_provider == AIProvider.FIREWORKS_AI:
-            return self._fireworks_ai_fetch_completions(messages)["choices"][0][
-                "message"
-            ]["content"]
+            return self._fireworks_ai_fetch_completions(messages)
 
         if self._ai_provider == AIProvider.OPEN_AI:
-            return self._open_ai_fetch_completion_open_ai(messages)["choices"][0][
-                "message"
-            ]["content"]
+            return self._open_ai_fetch_completion_open_ai(messages)
 
-        raise Exception(f"Invalid AI provider {self._ai_provider}")
+        return {
+            "status": "error",
+            "message": f"Invalid AI provider {self._ai_provider}",
+        }
 
     def get_dora_metrics_score(
         self, four_keys_data: Dict[str, float]
-    ) -> Dict[str, str]:
+    ) -> Dict[str, Union[str, int]]:
         """
         Calculate the DORA metrics score using input data and an LLM (Language Learning Model).
 

web-server/pages/api/internal/ai/dora_metrics.ts

@@ -64,50 +64,78 @@ const postSchema = yup.object().shape({
 });
 
 const endpoint = new Endpoint(nullSchema);
-
 endpoint.handle.POST(postSchema, async (req, res) => {
   const { data, model, access_token } = req.payload;
+  const dora_data = data as TeamDoraMetricsApiResponseType;
+
+  try {
+    const [
+      doraMetricsScore,
+      leadTimeSummary,
+      CFRSummary,
+      MTTRSummary,
+      deploymentFrequencySummary,
+      doraTrendSummary
+    ] = await Promise.all(
+      [
+        getDoraMetricsScore,
+        getLeadTimeSummary,
+        getCFRSummary,
+        getMTTRSummary,
+        getDeploymentFrequencySummary,
+        getDoraTrendsCorrelationSummary
+      ].map((fn) => fn(dora_data, model, access_token))
+    );
+
+    const aggregatedData = {
+      ...doraMetricsScore,
+      ...leadTimeSummary,
+      ...CFRSummary,
+      ...MTTRSummary,
+      ...deploymentFrequencySummary,
+      ...doraTrendSummary
+    };
 
-  const dora_data = data as unknown as TeamDoraMetricsApiResponseType;
-
-  const [
-    dora_metrics_score,
-    lead_time_trends_summary,
-    change_failure_rate_trends_summary,
-    mean_time_to_recovery_trends_summary,
-    deployment_frequency_trends_summary,
-    dora_trend_summary
-  ] = await Promise.all(
-    [
-      getDoraMetricsScore,
-      getLeadTimeSummary,
-      getCFRSummary,
-      getMTTRSummary,
-      getDeploymentFrequencySummary,
-      getDoraTrendsCorrelationSummary
-    ].map((f) => f(dora_data, model, access_token))
-  );
+    const compiledSummary = await getDORACompiledSummary(
+      aggregatedData,
+      model,
+      access_token
+    );
 
-  const aggregated_dora_data = {
-    ...dora_metrics_score,
-    ...lead_time_trends_summary,
-    ...change_failure_rate_trends_summary,
-    ...mean_time_to_recovery_trends_summary,
-    ...deployment_frequency_trends_summary,
-    ...dora_trend_summary
-  } as AggregatedDORAData;
-
-  const dora_compiled_summary = await getDORACompiledSummary(
-    aggregated_dora_data,
-    model,
-    access_token
-  );
+    const responses = {
+      ...aggregatedData,
+      ...compiledSummary
+    };
 
-  res.send({
-    ...aggregated_dora_data,
-    ...dora_compiled_summary
-  });
+    const { status, message } = checkForErrors(responses);
+
+    if (status === 'error') {
+      return res.status(400).send({ message });
+    }
+
+    const simplifiedData = Object.fromEntries(
+      Object.entries(responses).map(([key, value]) => [key, value.data])
+    );
+
+    return res.status(200).send(simplifiedData);
+  } catch (error) {
+    return res.status(500).send({
+      message: 'Internal Server Error',
+      error: error.message
+    });
+  }
 });
+const checkForErrors = (
+  responses: Record<string, { status: string; message: string }>
+): { status: string; message: string } => {
+  const errorResponse = Object.values(responses).find(
+    (value) => value.status === 'error'
+  );
+
+  return errorResponse
+    ? { status: 'error', message: errorResponse.message }
+    : { status: 'success', message: '' };
+};
 
 const getDoraMetricsScore = (
   dora_data: TeamDoraMetricsApiResponseType,