Update api.py

Hi,

When the generate_text() is called by the API, there is no error catch in case of failure. 
This change allows to see the error message on the answer cell of the web app instead of a blank.

Useful when openai account is exceeding the quota or other types of error.

MA

api.py

@@ -110,16 +110,19 @@ def load_recommender(path, start_page=1):
 
 def generate_text(openAI_key, prompt, engine="text-davinci-003"):
     openai.api_key = openAI_key
-    completions = openai.Completion.create(
-        engine=engine,
-        prompt=prompt,
-        max_tokens=512,
-        n=1,
-        stop=None,
-        temperature=0.7,
-    )
-    message = completions.choices[0].text
-    return message
+    try:
+        completions = openai.Completion.create(
+            engine=engine,
+            prompt=prompt,
+            max_tokens=512,
+            n=1,
+            stop=None,
+            temperature=0.7,
+        )
+        message = completions.choices[0].text
+    except Exception as e:
+        message = f'API Error: {str(e)}'
+    return message 
 
 
 def generate_answer(question, openAI_key):