refactor(antigravity): 🔨 clarify empty response retry configuration and logic

Renames `EMPTY_RESPONSE_MAX_RETRIES` to `EMPTY_RESPONSE_MAX_ATTEMPTS` and updates the retry logic to be more intuitive and correct.

- Renamed constant to better reflect that it represents total attempts, not retries
- Updated environment variable from `ANTIGRAVITY_EMPTY_RESPONSE_RETRIES` to `ANTIGRAVITY_EMPTY_RESPONSE_ATTEMPTS`
- Changed default from 3 to 4 attempts with minimum enforcement via `max(1, ...)`
- Fixed off-by-one errors in loop conditions: now using `range(EMPTY_RESPONSE_MAX_ATTEMPTS)` instead of `range(EMPTY_RESPONSE_MAX_RETRIES + 1)`
- Corrected retry condition from `attempt < EMPTY_RESPONSE_MAX_RETRIES` to `attempt < EMPTY_RESPONSE_MAX_ATTEMPTS - 1`
- Removed redundant error logging on final attempt failure (caller handles error logging)
- Updated all related comments and log messages to reflect "attempts" terminology
- Applied changes consistently to both streaming and non-streaming handlers

BREAKING CHANGE: The environment variable `ANTIGRAVITY_EMPTY_RESPONSE_RETRIES` has been renamed to `ANTIGRAVITY_EMPTY_RESPONSE_ATTEMPTS`. Users must update their environment configuration to use the new variable name. The semantic meaning has also changed from "number of retries after initial attempt" to "total number of attempts including the first".

Author: Mirrowel

src/rotator_library/providers/antigravity_provider.py

@@ -85,8 +85,8 @@ def _env_int(key: str, default: int) -> int:
 
 # Empty response retry configuration
 # When Antigravity returns an empty response (no content, no tool calls),
-# automatically retry up to this many times before giving up
-EMPTY_RESPONSE_MAX_RETRIES = _env_int("ANTIGRAVITY_EMPTY_RESPONSE_RETRIES", 3)
+# automatically retry up to this many attempts before giving up (minimum 1)
+EMPTY_RESPONSE_MAX_ATTEMPTS = max(1, _env_int("ANTIGRAVITY_EMPTY_RESPONSE_ATTEMPTS", 4))
 EMPTY_RESPONSE_RETRY_DELAY = _env_int("ANTIGRAVITY_EMPTY_RESPONSE_RETRY_DELAY", 2)
 
 # Model alias mappings (internal ↔ public)
@@ -3258,7 +3258,7 @@ async def acompletion(
                         "This may indicate a temporary service issue. Please try again."
                     )
 
-                    for attempt in range(EMPTY_RESPONSE_MAX_RETRIES + 1):
+                    for attempt in range(EMPTY_RESPONSE_MAX_ATTEMPTS):
                         result = await self._handle_non_streaming(
                             client, url, headers, payload, model, file_logger
                         )
@@ -3272,18 +3272,16 @@ async def acompletion(
                         got_response = bool(result_dict.get("choices"))
 
                         if not got_response:
-                            if attempt < EMPTY_RESPONSE_MAX_RETRIES:
+                            if attempt < EMPTY_RESPONSE_MAX_ATTEMPTS - 1:
                                 lib_logger.warning(
                                     f"[Antigravity] Empty response from {model}, "
-                                    f"attempt {attempt + 1}/{EMPTY_RESPONSE_MAX_RETRIES + 1}. Retrying..."
+                                    f"attempt {attempt + 1}/{EMPTY_RESPONSE_MAX_ATTEMPTS}. Retrying..."
                                 )
                                 await asyncio.sleep(EMPTY_RESPONSE_RETRY_DELAY)
                                 continue
                             else:
-                                lib_logger.error(
-                                    f"[Antigravity] Empty response from {model} after "
-                                    f"{EMPTY_RESPONSE_MAX_RETRIES + 1} attempts. Giving up."
-                                )
+                                # Last attempt failed - raise without extra logging
+                                # (caller will log the error)
                                 raise EmptyResponseError(
                                     provider="antigravity",
                                     model=model,
@@ -3493,14 +3491,14 @@ async def _streaming_with_retry(
         Wrapper around _handle_streaming that retries on empty responses.
 
         If the stream yields zero chunks (Antigravity returned nothing),
-        retry up to EMPTY_RESPONSE_MAX_RETRIES times before giving up.
+        retry up to EMPTY_RESPONSE_MAX_ATTEMPTS times before giving up.
         """
         error_msg = (
             "The model returned an empty response after multiple attempts. "
             "This may indicate a temporary service issue. Please try again."
         )
 
-        for attempt in range(EMPTY_RESPONSE_MAX_RETRIES + 1):
+        for attempt in range(EMPTY_RESPONSE_MAX_ATTEMPTS):
             chunk_count = 0
 
             try:
@@ -3514,18 +3512,16 @@ async def _streaming_with_retry(
                     return  # Success - we got data
 
                 # Zero chunks - empty response
-                if attempt < EMPTY_RESPONSE_MAX_RETRIES:
+                if attempt < EMPTY_RESPONSE_MAX_ATTEMPTS - 1:
                     lib_logger.warning(
                         f"[Antigravity] Empty stream from {model}, "
-                        f"attempt {attempt + 1}/{EMPTY_RESPONSE_MAX_RETRIES + 1}. Retrying..."
+                        f"attempt {attempt + 1}/{EMPTY_RESPONSE_MAX_ATTEMPTS}. Retrying..."
                     )
                     await asyncio.sleep(EMPTY_RESPONSE_RETRY_DELAY)
                     continue
                 else:
-                    lib_logger.error(
-                        f"[Antigravity] Empty stream from {model} after "
-                        f"{EMPTY_RESPONSE_MAX_RETRIES + 1} attempts. Giving up."
-                    )
+                    # Last attempt failed - raise without extra logging
+                    # (caller will log the error)
                     raise EmptyResponseError(
                         provider="antigravity",
                         model=model,