vectorvein 0.1.23__py3-none-any.whl → 0.1.25__py3-none-any.whl

vectorvein/chat_clients/utils.py

@@ -3,6 +3,7 @@
 import re
 import json
 from math import ceil
+from typing import Iterable
 import httpx
 import tiktoken
 from anthropic import Anthropic
@@ -13,6 +14,11 @@ from ..settings import settings
 from ..utilities.retry import Retry
 from ..types.enums import BackendType
 from ..utilities.media_processing import ImageProcessor
+from ..types.llm_parameters import (
+    NotGiven,
+    NOT_GIVEN,
+    ToolParam,
+)
 
 
 chatgpt_encoding = tiktoken.encoding_for_model("gpt-3.5-turbo")
@@ -127,8 +133,10 @@ def get_token_counts(text: str | dict, model: str = "") -> int:
             .sleep_time(10)
             .run()
         )
-        response = response.json()
-        return response["segments_num"]
+        if response is None:
+            return 1000
+        result = response.json()
+        return result["segments_num"]
     elif model in ("moonshot-v1-8k", "moonshot-v1-32k", "moonshot-v1-128k"):
         model_setting = settings.moonshot.models[model]
         if len(model_setting.endpoints) == 0:
@@ -150,8 +158,10 @@ def get_token_counts(text: str | dict, model: str = "") -> int:
             .sleep_time(10)
             .run()
         )
-        response = response.json()
-        return response["data"]["total_tokens"]
+        if response is None:
+            return 1000
+        result = response.json()
+        return result["data"]["total_tokens"]
     elif model.startswith("gemini"):
         model_setting = settings.gemini.models[model]
         if len(model_setting.endpoints) == 0:
@@ -175,6 +185,8 @@ def get_token_counts(text: str | dict, model: str = "") -> int:
             .sleep_time(10)
             .run()
         )
+        if response is None:
+            return 1000
         result = response.json()
         return result["totalTokens"]
     elif model.startswith("claude"):
@@ -207,7 +219,11 @@ def calculate_image_tokens(width: int, height: int, model: str = "gpt-4o"):
     return total_tokens
 
 
-def get_message_token_counts(messages: list, tools: dict | None = None, model: str = "gpt-4o") -> int:
+def get_message_token_counts(
+    messages: list,
+    tools: list | Iterable[ToolParam] | NotGiven = NOT_GIVEN,
+    model: str = "gpt-4o",
+) -> int:
     tokens = 0
     formatted_messages = format_messages(messages, backend=BackendType.OpenAI, native_multimodal=True)
     for message in formatted_messages:
@@ -221,7 +237,7 @@ def get_message_token_counts(messages: list, tools: dict | None = None, model: s
                 elif isinstance(item, dict) and item["type"].startswith("image"):
                     # TODO: Get real image size
                     tokens += calculate_image_tokens(2048, 2048, model)
-    if tools is not None:
+    if tools:
         tokens += get_token_counts(str(tools), model)
 
     return tokens
@@ -297,15 +313,20 @@ def cutoff_messages(
         if index == 0:
             # 一条消息就超过长度则将该消息内容进行截断，保留该消息最后的一部分
             if backend == BackendType.Gemini:
-                message[content_key] = [{"text": message[content_key][-max_count:]}]
+                return system_message + [
+                    {
+                        "role": message["role"],
+                        content_key: [{"text": message[content_key][-max_count:]}],
+                    }
+                ]
             else:
                 content = message[content_key][max_count - messages_length :]
-            return system_message + [
-                {
-                    "role": message["role"],
-                    content_key: content,
-                }
-            ]
+                return system_message + [
+                    {
+                        "role": message["role"],
+                        content_key: content,
+                    }
+                ]
         return system_message + messages[-index:]
     return system_message + messages
 
@@ -340,7 +361,7 @@ def format_image_message(image: str, backend: BackendType = BackendType.OpenAI)
         }
 
 
-def format_workflow_messages(message, content, backend):
+def format_workflow_messages(message: dict, content: str, backend: BackendType):
     formatted_messages = []
 
     # 工具调用消息
@@ -505,7 +526,6 @@ def format_messages(
     def is_vectorvein_message(message):
         return "content_type" in message
 
-    backend = backend.lower()
     formatted_messages = []
 
     for message in messages:
@@ -528,7 +548,13 @@ def format_messages(
     return formatted_messages
 
 
-def format_text_message(content, role, attachments, backend, native_multimodal):
+def format_text_message(
+    content: str,
+    role: str,
+    attachments: list,
+    backend: BackendType,
+    native_multimodal: bool,
+):
     images_extensions = ("jpg", "jpeg", "png", "bmp")
     has_images = any(attachment.lower().endswith(images_extensions) for attachment in attachments)
 
@@ -564,7 +590,7 @@ def format_text_message(content, role, attachments, backend, native_multimodal):
             return {"role": role, "content": content}
 
 
-def generate_tool_use_system_prompt(tools: list, format_type: str = "json") -> str:
+def generate_tool_use_system_prompt(tools: list | str, format_type: str = "json") -> str:
     if format_type == "json":
         return (
             "You have access to the following tools. Use them if required and wait for the tool call result. Stop output after calling a tool.\n\n"
@@ -585,3 +611,5 @@ def generate_tool_use_system_prompt(tools: list, format_type: str = "json") -> s
             "## Output format\n<|▶|><invoke><tool_name>[function name:str]</tool_name><parameters><parameter_1_name>[parameter_1_value]</parameter_1_name><parameter_2_name>[parameter_2_value]</parameter_2_name>...</parameters></invoke><|◀|>\n\n"
             "## Example output\n<|▶|><invoke><tool_name>calculator</tool_name><parameters><first_operand>1984135</first_operand><second_operand>9343116</second_operand><operator>*</operator></parameters></invoke><|◀|>"
         )
+    else:
+        return ""

vectorvein/types/defaults.py

@@ -150,7 +150,7 @@ GROQ_MODELS = {
 }
 
 # Qwen models
-QWEN_DEFAULT_MODEL = "qwen2-72b-instruct"
+QWEN_DEFAULT_MODEL = "qwen2.5-72b-instruct"
 QWEN_MODELS = {
     "qwen1.5-1.8b-chat": {
         "id": "qwen1.5-1.8b-chat",
@@ -208,6 +208,62 @@ QWEN_MODELS = {
         "function_call_available": False,
         "response_format_available": True,
     },
+    "qwen2.5-7b-instruct": {
+        "id": "qwen2.5-7b-instruct",
+        "context_length": 131072,
+        "max_output_tokens": 8192,
+        "function_call_available": False,
+        "response_format_available": True,
+    },
+    "qwen2.5-14b-instruct": {
+        "id": "qwen2.5-14b-instruct",
+        "context_length": 131072,
+        "max_output_tokens": 8192,
+        "function_call_available": False,
+        "response_format_available": True,
+    },
+    "qwen2.5-32b-instruct": {
+        "id": "qwen2.5-32b-instruct",
+        "context_length": 131072,
+        "max_output_tokens": 8192,
+        "function_call_available": False,
+        "response_format_available": True,
+    },
+    "qwen2.5-72b-instruct": {
+        "id": "qwen2.5-72b-instruct",
+        "context_length": 131072,
+        "max_output_tokens": 8192,
+        "function_call_available": False,
+        "response_format_available": True,
+    },
+    "qwen-max": {
+        "id": "qwen-max",
+        "context_length": 8096,
+        "max_output_tokens": 2048,
+        "function_call_available": False,
+        "response_format_available": True,
+    },
+    "qwen-max-longcontext": {
+        "id": "qwen-max-longcontext",
+        "context_length": 30000,
+        "max_output_tokens": 2048,
+        "function_call_available": False,
+        "response_format_available": True,
+    },
+    "qwen-plus": {
+        "id": "qwen-plus",
+        "context_length": 131072,
+        "max_output_tokens": 8096,
+        "function_call_available": False,
+        "response_format_available": True,
+    },
+    "qwen-turbo": {
+        "id": "qwen-turbo",
+        "context_length": 8096,
+        "max_output_tokens": 1500,
+        "function_call_available": False,
+        "response_format_available": True,
+    },
 }
 
 # Yi models

vectorvein/types/llm_parameters.py

@@ -1,10 +1,19 @@
 # @Author: Bi Ying
 # @Date:   2024-07-26 23:48:04
-from typing import List, Dict, Optional
+from typing import List, Dict, Optional, Union, Iterable
 
 from pydantic import BaseModel, Field
-from openai.types.chat.chat_completion_message import ChatCompletionMessageToolCall
+
+from anthropic.types import ToolParam as AnthropicToolParam
+from anthropic._types import NotGiven as AnthropicNotGiven
+from anthropic.types.message_create_params import ToolChoice as AnthropicToolChoice
+
+from openai._types import NotGiven as OpenAINotGiven
+from openai._types import NOT_GIVEN as OpenAINOT_GIVEN
 from openai.types.chat.chat_completion_chunk import ChoiceDeltaToolCall
+from openai.types.chat.chat_completion_tool_param import ChatCompletionToolParam
+from openai.types.chat.chat_completion_message_tool_call import ChatCompletionMessageToolCall
+from openai.types.chat.chat_completion_tool_choice_option_param import ChatCompletionToolChoiceOptionParam
 
 from . import defaults as defs
 
@@ -26,7 +35,7 @@ class EndpointSetting(BaseModel):
 
 
 class ModelSetting(BaseModel):
-    id: Optional[str] = Field(None, description="The id of the model.")
+    id: str = Field(..., description="The id of the model.")
     endpoints: List[str] = Field(default_factory=list, description="Available endpoints for the model.")
     function_call_available: bool = Field(False, description="Indicates if function call is available.")
     response_format_available: bool = Field(False, description="Indicates if response format is available.")
@@ -80,3 +89,15 @@ class ChatCompletionDeltaMessage(BaseModel):
     function_call_arguments: Optional[dict] = None
 
     usage: Optional[Usage] = None
+
+
+NotGiven = Union[AnthropicNotGiven, OpenAINotGiven]
+
+NOT_GIVEN = OpenAINOT_GIVEN
+
+OpenAIToolParam = ChatCompletionToolParam
+ToolParam = OpenAIToolParam
+
+Tools = Iterable[ToolParam]
+
+ToolChoice = ChatCompletionToolChoiceOptionParam

{vectorvein-0.1.23.dist-info → vectorvein-0.1.25.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vectorvein
-Version: 0.1.23
+Version: 0.1.25
 Summary: Default template for PDM package
 Author-Email: Anderson <andersonby@163.com>
 License: MIT

{vectorvein-0.1.23.dist-info → vectorvein-0.1.25.dist-info}/RECORD

@@ -1,28 +1,28 @@
-vectorvein-0.1.23.dist-info/METADATA,sha256=WlWH7IJNvTkNwtDxV2IAtUEgfGvjhhfkLlGK7vZluPM,502
-vectorvein-0.1.23.dist-info/WHEEL,sha256=Yaoh9rlmnhBUmykFXwmFW_cmVAW3ZIWH8QkQVngJtyw,90
-vectorvein-0.1.23.dist-info/entry_points.txt,sha256=6OYgBcLyFCUgeqLgnvMyOJxPCWzgy7se4rLPKtNonMs,34
+vectorvein-0.1.25.dist-info/METADATA,sha256=DtT3VCr5gTtj-j25i_7Kf28DmLVG8Ui-QPc0soqt9w8,502
+vectorvein-0.1.25.dist-info/WHEEL,sha256=Vza3XR51HW1KmFP0iIMUVYIvz0uQuKJpIXKYOBGQyFQ,90
+vectorvein-0.1.25.dist-info/entry_points.txt,sha256=6OYgBcLyFCUgeqLgnvMyOJxPCWzgy7se4rLPKtNonMs,34
 vectorvein/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 vectorvein/chat_clients/__init__.py,sha256=lOGrIEBGN-EoxJ-dF5uMsO6viNCIFIeNL8whDwE6x3g,4657
-vectorvein/chat_clients/anthropic_client.py,sha256=VJQi7cKLbbLkJtmEFN9M41JUaGjwEuJaXIuQLK-3rdw,20332
+vectorvein/chat_clients/anthropic_client.py,sha256=h82GxBi7h22B7leBuPofwBstxH_c12tEgGjpnKg6UDc,25007
 vectorvein/chat_clients/baichuan_client.py,sha256=CVMvpgjdrZGv0BWnTOBD-f2ufZ3wq3496wqukumsAr4,526
-vectorvein/chat_clients/base_client.py,sha256=nGml8rGyKc71Wl2nMbqt_5L9gbX-MCRU2jvHJvxF4AY,5108
+vectorvein/chat_clients/base_client.py,sha256=jSUSZNowUBg1Fl0Z2c9soPwx6glomoPyi7NFZIPLVBQ,7402
 vectorvein/chat_clients/deepseek_client.py,sha256=3qWu01NlJAP2N-Ff62d5-CZXZitlizE1fzb20LNetig,526
-vectorvein/chat_clients/gemini_client.py,sha256=N08OPbCDK-_jN_CzEG8rIKlrMMNqW6cq9ucjREqF6dY,15295
+vectorvein/chat_clients/gemini_client.py,sha256=2SrREa_wJqTXOAvv-2LSo3DVR2vONsMzmBV9WFkNQuA,17640
 vectorvein/chat_clients/groq_client.py,sha256=Uow4pgdmFi93ZQSoOol2-0PhhqkW-S0XuSldvppz5U4,498
 vectorvein/chat_clients/local_client.py,sha256=55nOsxzqUf79q3Y14MKROA71zxhsT7p7FsDZ89rts2M,422
-vectorvein/chat_clients/minimax_client.py,sha256=xykptVg0qzOrJKuswHLG3M7r8H37jEMOb0gAnYAUdKQ,13947
+vectorvein/chat_clients/minimax_client.py,sha256=dNMhCP74gRCnReR_xNosUkGc0_NP3IfNhp48WvBpU-4,16189
 vectorvein/chat_clients/mistral_client.py,sha256=1aKSylzBDaLYcFnaBIL4-sXSzWmXfBeON9Q0rq-ziWw,534
 vectorvein/chat_clients/moonshot_client.py,sha256=gbu-6nGxx8uM_U2WlI4Wus881rFRotzHtMSoYOcruGU,526
 vectorvein/chat_clients/openai_client.py,sha256=Nz6tV45pWcsOupxjnsRsGTicbQNJWIZyxuJoJ5DGMpg,527
-vectorvein/chat_clients/openai_compatible_client.py,sha256=NR88EgyLtXt1XiOKibZFtgMPb3_haTlKmsxoPvcYnb0,14130
+vectorvein/chat_clients/openai_compatible_client.py,sha256=Ojsxs7j6s4Hne7vMDbR0u-CtDx0WZQZ1NIsQIFm5BcA,17758
 vectorvein/chat_clients/qwen_client.py,sha256=-ryh-m9PgsO0fc4ulcCmPTy1155J8YUy15uPoJQOHA0,513
-vectorvein/chat_clients/utils.py,sha256=EbSodMwLCR8wnzyt2J6iOsL8WvSzlwdu71xZikzNdlg,22483
+vectorvein/chat_clients/utils.py,sha256=8Md6XOF_io0ACKRQ7ruqP1eJu7g9uo80eUiGbWrLx7k,23041
 vectorvein/chat_clients/yi_client.py,sha256=RNf4CRuPJfixrwLZ3-DEc3t25QDe1mvZeb9sku2f8Bc,484
 vectorvein/chat_clients/zhipuai_client.py,sha256=Ys5DSeLCuedaDXr3PfG1EW2zKXopt-awO2IylWSwY0s,519
 vectorvein/settings/__init__.py,sha256=jVHbhHn1BuMcyfZGXrxWKiI4NdY9wzvYyGMvKYmUtqg,3378
-vectorvein/types/defaults.py,sha256=GaloEG4rNCcHz8DGRAF9XV0wHmIOiIqwI6sIzJHSY8s,15994
+vectorvein/types/defaults.py,sha256=EouXmZvjbvDQhYJ-5FIz6Ee6Xyc7Ud1wlOPMIDnaAfY,17811
 vectorvein/types/enums.py,sha256=PNK_pTIyjJFy-yAG2PHaMIO1ey3W6fReMCkH8M8VRW4,1595
-vectorvein/types/llm_parameters.py,sha256=mmJjJZz4bPRi0nHzYNUNdWsQLHa9lbf3-MNVnU78vaY,3608
+vectorvein/types/llm_parameters.py,sha256=N6RQ8tqO1RCywMFRWPooffeAEPd9x3JW6Bl4UgQtF5I,4379
 vectorvein/utilities/media_processing.py,sha256=BujciRmw1GMmc3ELRvafL8STcy6r5b2rVnh27-uA7so,2256
 vectorvein/utilities/retry.py,sha256=9ePuJdeUUGx-qMWfaFxmlOvG_lQPwCQ4UB1z3Edlo34,993
-vectorvein-0.1.23.dist-info/RECORD,,
+vectorvein-0.1.25.dist-info/RECORD,,

{vectorvein-0.1.23.dist-info → vectorvein-0.1.25.dist-info}/WHEEL

@@ -1,4 +1,4 @@
 Wheel-Version: 1.0
-Generator: pdm-backend (2.4.0)
+Generator: pdm-backend (2.4.1)
 Root-Is-Purelib: true
 Tag: py3-none-any