aient 1.0.44__py3-none-any.whl → 1.0.46__py3-none-any.whl

aient/core/models.py

@@ -86,6 +86,9 @@ class Thinking(BaseModel):
     budget_tokens: Optional[int] = None
     type: Optional[Literal["enabled", "disabled"]] = None
 
+class StreamOptions(BaseModel):
+    include_usage: Optional[bool] = None
+
 class RequestModel(BaseRequest):
     model: str
     messages: List[Message]
@@ -105,6 +108,7 @@ class RequestModel(BaseRequest):
     tools: Optional[List[Tool]] = None
     response_format: Optional[ResponseFormat] = None
     thinking: Optional[Thinking] = None
+    stream_options: Optional[StreamOptions] = None
 
     def get_last_text_message(self) -> Optional[str]:
         for message in reversed(self.messages):

aient/core/request.py

@@ -96,7 +96,8 @@ async def get_gemini_payload(request, engine, provider, api_key=None):
             content[0]["text"] = re.sub(r"_+", "_", content[0]["text"])
             systemInstruction = {"parts": content}
 
-    if "gemini-2.0-flash-exp" in original_model or "gemini-1.5" in original_model:
+    off_models = ["gemini-2.0-flash-exp", "gemini-1.5", "gemini-2.5-pro"]
+    if any(off_model in original_model for off_model in off_models):
         safety_settings = "OFF"
     else:
         safety_settings = "BLOCK_NONE"
@@ -119,7 +120,11 @@ async def get_gemini_payload(request, engine, provider, api_key=None):
             {
                 "category": "HARM_CATEGORY_DANGEROUS_CONTENT",
                 "threshold": safety_settings
-            }
+            },
+            {
+                "category": "HARM_CATEGORY_CIVIC_INTEGRITY",
+                "threshold": "BLOCK_NONE"
+            },
         ]
     }
 
@@ -144,7 +149,8 @@ async def get_gemini_payload(request, engine, provider, api_key=None):
         'include_usage',
         'logprobs',
         'top_logprobs',
-        'response_format'
+        'response_format',
+        'stream_options',
     ]
     generation_config = {}
 
@@ -190,9 +196,12 @@ async def get_gemini_payload(request, engine, provider, api_key=None):
             else:
                 payload[field] = value
 
-    if generation_config:
-        payload["generationConfig"] = generation_config
-        if "maxOutputTokens" not in generation_config:
+    max_token_65k_models = ["gemini-2.5-pro", "gemini-2.0-pro", "gemini-2.0-flash-thinking"]
+    payload["generationConfig"] = generation_config
+    if "maxOutputTokens" not in generation_config:
+        if any(pro_model in original_model for pro_model in max_token_65k_models):
+            payload["generationConfig"]["maxOutputTokens"] = 65536
+        else:
             payload["generationConfig"]["maxOutputTokens"] = 8192
 
     if request.model.endswith("-search"):
@@ -277,7 +286,8 @@ async def get_vertex_gemini_payload(request, engine, provider, api_key=None):
     original_model = model_dict[request.model]
     search_tool = None
 
-    if "gemini-2.0" in original_model or "gemini-exp" in original_model:
+    pro_models = ["gemini-2.5-pro", "gemini-2.0-pro", "gemini-exp"]
+    if any(pro_model in original_model for pro_model in pro_models):
         location = gemini2
         search_tool = {"googleSearch": {}}
     else:
@@ -384,7 +394,8 @@ async def get_vertex_gemini_payload(request, engine, provider, api_key=None):
         'user',
         'include_usage',
         'logprobs',
-        'top_logprobs'
+        'top_logprobs',
+        'stream_options',
     ]
     generation_config = {}
 
@@ -549,6 +560,7 @@ async def get_vertex_claude_payload(request, engine, provider, api_key=None):
         'n',
         'user',
         'include_usage',
+        'stream_options',
     ]
 
     for field, value in request.model_dump(exclude_unset=True).items():
@@ -845,6 +857,7 @@ async def get_openrouter_payload(request, engine, provider, api_key=None):
         'n',
         'user',
         'include_usage',
+        'stream_options',
     ]
 
     for field, value in request.model_dump(exclude_unset=True).items():
@@ -912,7 +925,8 @@ async def get_cohere_payload(request, engine, provider, api_key=None):
         'user',
         'include_usage',
         'logprobs',
-        'top_logprobs'
+        'top_logprobs',
+        'stream_options',
     ]
 
     for field, value in request.model_dump(exclude_unset=True).items():
@@ -959,7 +973,8 @@ async def get_cloudflare_payload(request, engine, provider, api_key=None):
         'user',
         'include_usage',
         'logprobs',
-        'top_logprobs'
+        'top_logprobs',
+        'stream_options',
     ]
 
     for field, value in request.model_dump(exclude_unset=True).items():
@@ -1141,6 +1156,7 @@ async def get_claude_payload(request, engine, provider, api_key=None):
         'n',
         'user',
         'include_usage',
+        'stream_options',
     ]
 
     for field, value in request.model_dump(exclude_unset=True).items():

aient/core/response.py

@@ -50,7 +50,6 @@ async def fetch_gemini_response_stream(client, url, headers, payload, model):
                     try:
                         json_data = json.loads( "{" + line + "}")
                         content = json_data.get('text', '')
-                        content = "\n".join(content.split("\\n"))
                         # content = content.replace("\n", "\n\n")
                         # if last_text_line == 0 and is_thinking:
                         #     content = "> " + content.lstrip()
@@ -108,7 +107,6 @@ async def fetch_vertex_claude_response_stream(client, url, headers, payload, mod
                     try:
                         json_data = json.loads( "{" + line + "}")
                         content = json_data.get('text', '')
-                        content = "\n".join(content.split("\\n"))
                         sse_string = await generate_sse_response(timestamp, model, content=content)
                         yield sse_string
                     except json.JSONDecodeError:

aient/core/utils.py

@@ -155,7 +155,8 @@ def update_initial_model(provider):
         proxy = safe_get(provider, "preferences", "proxy", default=None)
         client_config = get_proxy(proxy)
         if engine == "gemini":
-            url = "https://generativelanguage.googleapis.com/v1beta/models"
+            before_v1 = api_url.split("/v1beta")[0]
+            url = before_v1 + "/v1beta/models"
             params = {"key": api}
             with httpx.Client(**client_config) as client:
                 response = client.get(url, params=params)
@@ -288,7 +289,7 @@ class ThreadSafeCircularList:
             # self.requests[item] = []
             logger.warning(f"API key {item} 已进入冷却状态，冷却时间 {cooling_time} 秒")
 
-    async def is_rate_limited(self, item, model: str = None) -> bool:
+    async def is_rate_limited(self, item, model: str = None, is_check: bool = False) -> bool:
         now = time()
         # 检查是否在冷却中
         if now < self.cooling_until[item]:
@@ -321,7 +322,8 @@ class ThreadSafeCircularList:
             # 使用特定模型的请求记录进行计算
             recent_requests = sum(1 for req in self.requests[item][model_key] if req > now - limit_period)
             if recent_requests >= limit_count:
-                logger.warning(f"API key {item} 对模型 {model_key} 已达到速率限制 ({limit_count}/{limit_period}秒)")
+                if not is_check:
+                    logger.warning(f"API key {item}: model: {model_key} has been rate limited ({limit_count}/{limit_period} seconds)")
                 return True
 
         # 清理太旧的请求记录
@@ -329,7 +331,9 @@ class ThreadSafeCircularList:
         self.requests[item][model_key] = [req for req in self.requests[item][model_key] if req > now - max_period]
 
         # 记录新的请求
-        self.requests[item][model_key].append(now)
+        if not is_check:
+            self.requests[item][model_key].append(now)
+
         return False
 
     async def next(self, model: str = None):
@@ -349,6 +353,30 @@ class ThreadSafeCircularList:
                     logger.warning(f"All API keys are rate limited!")
                     raise HTTPException(status_code=429, detail="Too many requests")
 
+    async def is_all_rate_limited(self, model: str = None) -> bool:
+        """检查是否所有的items都被速率限制
+
+        与next方法不同，此方法不会改变任何内部状态（如self.index），
+        仅返回一个布尔值表示是否所有的key都被限制。
+
+        Args:
+            model: 要检查的模型名称，默认为None
+
+        Returns:
+            bool: 如果所有items都被速率限制返回True，否则返回False
+        """
+        if len(self.items) == 0:
+            return False
+
+        async with self.lock:
+            for item in self.items:
+                if not await self.is_rate_limited(item, model, is_check=True):
+                    return False
+
+            # 如果遍历完所有items都被限制，返回True
+            # logger.debug(f"Check result: all items are rate limited!")
+            return True
+
     async def after_next_current(self):
         # 返回当前取出的 API，因为已经调用了 next，所以当前API应该是上一个
         if len(self.items) == 0:

aient/models/chatgpt.py

@@ -159,7 +159,9 @@ class chatgpt(BaseLLM):
 
         conversation_len = len(self.conversation[convo_id]) - 1
         message_index = 0
-        # print(json.dumps(self.conversation[convo_id], indent=4, ensure_ascii=False))
+        # if self.print_log:
+        #     replaced_text = json.loads(re.sub(r';base64,([A-Za-z0-9+/=]+)', ';base64,***', json.dumps(self.conversation[convo_id])))
+        #     print(json.dumps(replaced_text, indent=4, ensure_ascii=False))
         while message_index < conversation_len:
             if self.conversation[convo_id][message_index]["role"] == self.conversation[convo_id][message_index + 1]["role"]:
                 if self.conversation[convo_id][message_index].get("content") and self.conversation[convo_id][message_index + 1].get("content"):
@@ -180,6 +182,9 @@ class chatgpt(BaseLLM):
                     and type(self.conversation[convo_id][message_index + 1]["content"]) == dict:
                         self.conversation[convo_id][message_index]["content"] = [self.conversation[convo_id][message_index]["content"]]
                         self.conversation[convo_id][message_index + 1]["content"] = [self.conversation[convo_id][message_index + 1]["content"]]
+                    if type(self.conversation[convo_id][message_index]["content"]) == list \
+                    and type(self.conversation[convo_id][message_index + 1]["content"]) == dict:
+                        self.conversation[convo_id][message_index + 1]["content"] = [self.conversation[convo_id][message_index + 1]["content"]]
                     self.conversation[convo_id][message_index]["content"] += self.conversation[convo_id][message_index + 1]["content"]
                 self.conversation[convo_id].pop(message_index + 1)
                 conversation_len = conversation_len - 1

{aient-1.0.44.dist-info → aient-1.0.46.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: aient
-Version: 1.0.44
+Version: 1.0.46
 Summary: Aient: The Awakening of Agent.
 Description-Content-Type: text/markdown
 License-File: LICENSE

{aient-1.0.44.dist-info → aient-1.0.46.dist-info}/RECORD

@@ -2,17 +2,17 @@ aient/__init__.py,sha256=SRfF7oDVlOOAi6nGKiJIUK6B_arqYLO9iSMp-2IZZps,21
 aient/core/.git,sha256=lrAcW1SxzRBUcUiuKL5tS9ykDmmTXxyLP3YYU-Y-Q-I,45
 aient/core/__init__.py,sha256=NxjebTlku35S4Dzr16rdSqSTWUvvwEeACe8KvHJnjPg,34
 aient/core/log_config.py,sha256=kz2_yJv1p-o3lUQOwA3qh-LSc3wMHv13iCQclw44W9c,274
-aient/core/models.py,sha256=8MsuiYHBHVR5UMQ_cNLkvntoxalS7NpVwaNwHA0iZmk,7379
-aient/core/request.py,sha256=-nyFwGM86LB8Zn6ScRJvAbkJ9LPCHjgg51tO_edAIZ4,48422
-aient/core/response.py,sha256=7s1Jil0E5nnbL9xQldcjHIqSp0MFeWQo9mNX_iAuvSk,25954
-aient/core/utils.py,sha256=i9ZwyywBLIhRM0fNmFSD3jF3dBL5QqVMOtSlG_ddv-I,24101
+aient/core/models.py,sha256=H3_XuWA7aS25MWZPK1c-5RBiiuxWJbTfE3RAk0Pkc9A,7504
+aient/core/request.py,sha256=OlMkjGMcFAH-ItA1PgPuf2HT-RbI-Ca4JXncWApc3gM,49088
+aient/core/response.py,sha256=7RVSFfGHisejv2SlsHvp0t-N_8OpTS4edQU_NOi5BGU,25822
+aient/core/utils.py,sha256=I0u3WLWaMd4j1ShqKg_tz67m-1wr_uXlWgxGeUjIIiE,25098
 aient/core/test/test_base_api.py,sha256=CjfFzMG26r8C4xCPoVkKb3Ac6pp9gy5NUCbZJHoSSsM,393
 aient/core/test/test_image.py,sha256=_T4peNGdXKBHHxyQNx12u-NTyFE8TlYI6NvvagsG2LE,319
 aient/core/test/test_payload.py,sha256=8jBiJY1uidm1jzL-EiK0s6UGmW9XkdsuuKFGrwFhFkw,2755
 aient/models/__init__.py,sha256=ouNDNvoBBpIFrLsk09Q_sq23HR0GbLAKfGLIFmfEuXE,219
 aient/models/audio.py,sha256=kRd-8-WXzv4vwvsTGwnstK-WR8--vr9CdfCZzu8y9LA,1934
 aient/models/base.py,sha256=Loyt2F2WrDMBbK-sdmTtgkLVtdUXxK5tg4qoI6nc0Xo,7527
-aient/models/chatgpt.py,sha256=rF95RmO4C3h4PKRqE3Qk6fKoR0yIf-3zp8t7KBF_kjA,41685
+aient/models/chatgpt.py,sha256=QGMx2szrYlK-uqe18Vbem3ou37nrQFhS7vonpLxHrUo,42173
 aient/models/claude.py,sha256=thK9P8qkaaoUN3OOJ9Shw4KDs-pAGKPoX4FOPGFXva8,28597
 aient/models/duckduckgo.py,sha256=1l7vYCs9SG5SWPCbcl7q6pCcB5AUF_r-a4l9frz3Ogo,8115
 aient/models/gemini.py,sha256=chGLc-8G_DAOxr10HPoOhvVFW1RvMgHd6mt--VyAW98,14730
@@ -29,8 +29,8 @@ aient/plugins/websearch.py,sha256=yiBzqXK5X220ibR-zko3VDsn4QOnLu1k6E2YOygCeTQ,15
 aient/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 aient/utils/prompt.py,sha256=UcSzKkFE4-h_1b6NofI6xgk3GoleqALRKY8VBaXLjmI,11311
 aient/utils/scripts.py,sha256=obrf5oxzFQPCu1A5MYDDiZv_LM6l9C1QSkgWIqcu28k,25690
-aient-1.0.44.dist-info/licenses/LICENSE,sha256=XNdbcWldt0yaNXXWB_Bakoqnxb3OVhUft4MgMA_71ds,1051
-aient-1.0.44.dist-info/METADATA,sha256=KQvQXJbHv_lAv8h22O-Ez1Rc2jF_5VRHdtpwhHagMqk,4986
-aient-1.0.44.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
-aient-1.0.44.dist-info/top_level.txt,sha256=3oXzrP5sAVvyyqabpeq8A2_vfMtY554r4bVE-OHBrZk,6
-aient-1.0.44.dist-info/RECORD,,
+aient-1.0.46.dist-info/licenses/LICENSE,sha256=XNdbcWldt0yaNXXWB_Bakoqnxb3OVhUft4MgMA_71ds,1051
+aient-1.0.46.dist-info/METADATA,sha256=nYfiefitlFshZCNddR3PTfypDm1mrCtJhjboAJmoNOQ,4986
+aient-1.0.46.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
+aient-1.0.46.dist-info/top_level.txt,sha256=3oXzrP5sAVvyyqabpeq8A2_vfMtY554r4bVE-OHBrZk,6
+aient-1.0.46.dist-info/RECORD,,