fix: increase retry limit and provide override

src/gpt_review/_ask.py

@@ -164,7 +164,7 @@ def _call_gpt(
         )
         return completion.choices[0].message.content  # type: ignore
     except RateLimitError as error:
-        if retry < 5:
+        if retry < C.MAX_RETRIES:
             logging.warning("Call to GPT failed due to rate limit, retry attempt: %s", retry)
             time.sleep(retry * 10)
             return _call_gpt(prompt, temperature, max_tokens, top_p, frequency_penalty, presence_penalty, retry + 1)

src/gpt_review/_llama_index.py

@@ -9,6 +9,8 @@
 from llama_index import GPTVectorStoreIndex, LLMPredictor, LangchainEmbedding, ServiceContext, SimpleDirectoryReader
 from llama_index.indices.base import BaseGPTIndex
 
+import gpt_review.constants as C
+
 
 def _ask_doc(question: str, files: List[str], fast: bool = False, large: bool = False) -> str:
     """
@@ -54,7 +56,7 @@ def _document_indexer(
             "api_type": "azure",
             "api_version": "2023-03-15-preview",
         },
-        max_retries=10,
+        max_retries=MAX_RETRIES,
     )
 
     llm_predictor = LLMPredictor(llm=llm)

src/gpt_review/constants.py

@@ -1,4 +1,5 @@
 """Contains constants for minimum and maximum values of various parameters used in GPT Review."""
+import os
 import sys
 
 MAX_TOKENS_DEFAULT = 100
@@ -20,3 +21,5 @@
 PRESENCE_PENALTY_DEFAULT = 0
 PRESENCE_PENALTY_MIN = 0
 PRESENCE_PENALTY_MAX = 2
+
+MAX_RETRIES = os.getenv("MAX_RETRIES", 10)