Merge remote-tracking branch 'origin/main' into feat-fileparsing

Author: mengqian

examples/basic_usage.py

@@ -6,9 +6,9 @@ def completion():
 
 	# Create chat completion
 	response = client.chat.completions.create(
-		model='glm-4',
+		model='glm-4.6',
 		messages=[{'role': 'user', 'content': 'Hello, Z.ai!'}],
-		temperature=0.7,
+		temperature=1.0,
 	)
 	print(response.choices[0].message.content)
 
@@ -19,7 +19,7 @@ def completion_with_stream():
 
 	# Create chat completion
 	response = client.chat.completions.create(
-		model='glm-4',
+		model='glm-4.6',
 		messages=[
 			{'role': 'system', 'content': 'You are a helpful assistant.'},
 			{'role': 'user', 'content': 'Tell me a story about AI.'},
@@ -38,7 +38,7 @@ def completion_with_websearch():
 
 	# Create chat completion
 	response = client.chat.completions.create(
-		model='glm-4',
+		model='glm-4.6',
 		messages=[
 			{'role': 'system', 'content': 'You are a helpful assistant.'},
 			{'role': 'user', 'content': 'What is artificial intelligence?'},
@@ -52,7 +52,7 @@ def completion_with_websearch():
 				},
 			}
 		],
-		temperature=0.5,
+		temperature=1.0,
 		max_tokens=2000,
 	)
 
@@ -234,10 +234,10 @@ def ofZhipu():
 	print(response.choices[0].message.content)
 
 if __name__ == '__main__':
-    # completion()
-    # completion_with_stream()
+    completion()
+    completion_with_stream()
     # completion_with_websearch()
-    multi_modal_chat()
+    # multi_modal_chat()
     # role_play()
     # assistant_conversation()
     # video_generation()

examples/stream_tools.py

@@ -0,0 +1,81 @@
+from zai import ZhipuAiClient
+
+def main():
+    client = ZhipuAiClient()
+    # create chat completion with tool calls and streaming
+    response = client.chat.completions.create(
+        model="glm-4.6",
+        messages=[
+            {"role": "user", "content": "How is the weather in Beijing and Shanghai? Please provide the answer in Celsius."},
+        ],
+        tools=[
+            {
+                "type": "function",
+                "function": {
+                    "name": "get_weather",
+                    "description": "Get the weather information for a specific location",
+                    "parameters": {
+                        "type": "object",
+                        "properties": {
+                            "location": {"type": "string", "description": "City, eg: Beijing, Shanghai"},
+                            "unit": {"type": "string", "enum": ["celsius", "fahrenheit"]}
+                        },
+                        "required": ["location"]
+                    }
+                }
+            }
+        ],
+        stream=True,  # enable streaming
+        tool_stream=True  # enable tool call streaming
+    )
+
+    # init variables to collect streaming data
+    reasoning_content = ""  # reasoning content
+    content = ""  # response content
+    final_tool_calls = {}  # tool call data
+    reasoning_started = False  # is reasoning started
+    content_started = False  # is content started
+
+    # process streaming response
+    for chunk in response:
+        if not chunk.choices:
+            continue
+
+        delta = chunk.choices[0].delta
+
+        # process streaming reasoning output
+        if hasattr(delta, 'reasoning_content') and delta.reasoning_content:
+            if not reasoning_started and delta.reasoning_content.strip():
+                print("\n🧠 Thinking: ")
+                reasoning_started = True
+            reasoning_content += delta.reasoning_content
+            print(delta.reasoning_content, end="", flush=True)
+
+        # process streaming answer content output
+        if hasattr(delta, 'content') and delta.content:
+            if not content_started and delta.content.strip():
+                print("\n\n💬 Answer: ")
+                content_started = True
+            content += delta.content
+            print(delta.content, end="", flush=True)
+
+        # process streaming tool call info
+        if delta.tool_calls:
+            for tool_call in delta.tool_calls:
+                index = tool_call.index
+                if index not in final_tool_calls:
+                    # add new tool call
+                    final_tool_calls[index] = tool_call
+                    final_tool_calls[index].function.arguments = tool_call.function.arguments
+                else:
+                    # append tool call params by streaming index
+                    final_tool_calls[index].function.arguments += tool_call.function.arguments
+
+    # output the final construct tool call info
+    if final_tool_calls:
+        print("\n📋 Function Calls Triggered:")
+        for index, tool_call in final_tool_calls.items():
+            print(f"  {index}: Function Name: {tool_call.function.name}, Params: {tool_call.function.arguments}")
+
+if __name__ == "__main__":
+    main()

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "zai-sdk"
-version = "0.0.3.6"
+version = "0.0.4"
 description = "A SDK library for accessing big model apis from Z.ai"
 authors = ["Z.ai"]
 readme = "README.md"

src/zai/_version.py

@@ -1,2 +1,2 @@
 __title__ = 'Z.ai'
-__version__ = '0.0.3.6'
+__version__ = '0.0.4'

src/zai/api_resource/audio/audio.py

@@ -22,6 +22,8 @@
 from zai.types.sensitive_word_check import SensitiveWordCheckRequest
 
 from .transcriptions import Transcriptions
+from zai.core._streaming import StreamResponse
+from zai.types.audio import AudioSpeechChunk
 
 if TYPE_CHECKING:
 	from zai._client import ZaiClient
@@ -55,9 +57,11 @@ def speech(
 		extra_headers: Headers | None = None,
 		extra_body: Body | None = None,
 		timeout: float | httpx.Timeout | None | NotGiven = NOT_GIVEN,
+		encode_format: str = None,
 		speed: float | None = 1.0,
 		volume: float | None = 1.0,
-	) -> HttpxBinaryResponseContent:
+		stream: bool | None = False
+	) -> HttpxBinaryResponseContent | StreamResponse[AudioSpeechChunk]:
 		"""
 		Generate speech audio from text input
 
@@ -79,18 +83,21 @@ def speech(
 				'input': input,
 				'voice': voice,
 				'response_format': response_format,
-				'sensitive_word_check': sensitive_word_check,
+				'encode_format': encode_format,
 				'request_id': request_id,
 				'user_id': user_id,
 				'speed': speed,
 				'volume': volume,
+				'stream': stream
 			}
 		)
 		return self._post(
 			'/audio/speech',
 			body=maybe_transform(body, AudioSpeechParams),
 			options=make_request_options(extra_headers=extra_headers, extra_body=extra_body, timeout=timeout),
 			cast_type=HttpxBinaryResponseContent,
+			stream=stream or False,
+			stream_cls=StreamResponse[AudioSpeechChunk]
 		)
 
 	def customization(

src/zai/api_resource/chat/async_completions.py

@@ -88,25 +88,6 @@ def create(
 			watermark_enabled (Optional[bool]): Whether to enable watermark on generated audio
 		"""
 		_cast_type = AsyncTaskStatus
-		logger.debug(f'temperature:{temperature}, top_p:{top_p}')
-		if temperature is not None and temperature != NOT_GIVEN:
-			if temperature <= 0:
-				do_sample = False
-				temperature = 0.01
-				# logger.warning("temperature: value range is (0.0, 1.0) open interval,"
-				# "do_sample rewritten as false (parameters top_p temperature do not take effect)")
-			if temperature >= 1:
-				temperature = 0.99
-				# logger.warning("temperature: value range is (0.0, 1.0) open interval")
-		if top_p is not None and top_p != NOT_GIVEN:
-			if top_p >= 1:
-				top_p = 0.99
-				# logger.warning("top_p: value range is (0.0, 1.0) open interval, cannot equal 0 or 1")
-			if top_p <= 0:
-				top_p = 0.01
-				# logger.warning("top_p: value range is (0.0, 1.0) open interval, cannot equal 0 or 1")
-
-		logger.debug(f'temperature:{temperature}, top_p:{top_p}')
 		if isinstance(messages, List):
 			for item in messages:
 				if item.get('content'):

src/zai/api_resource/chat/completions.py

@@ -66,6 +66,7 @@ def create(
 		response_format: object | None = None,
 		thinking: object | None = None,
 		watermark_enabled: Optional[bool] | NotGiven = NOT_GIVEN,
+        tool_stream: bool | NotGiven = NOT_GIVEN,
 	) -> Completion | StreamResponse[ChatCompletionChunk]:
 		"""
 		Create a chat completion
@@ -93,6 +94,7 @@ def create(
 			response_format (object): Response format specification
 			thinking (Optional[object]): Configuration parameters for model reasoning
 			watermark_enabled (Optional[bool]): Whether to enable watermark on generated audio
+			tool_stream (Optional[bool]): Whether to enable tool streaming
 		"""
 		logger.debug(f'temperature:{temperature}, top_p:{top_p}')
 		if temperature is not None and temperature != NOT_GIVEN:
@@ -141,6 +143,7 @@ def create(
 				'response_format': response_format,
 				'thinking': thinking,
 				'watermark_enabled': watermark_enabled,
+                'tool_stream': tool_stream,
 			}
 		)
 		return self._post(

src/zai/types/audio/__init__.py

@@ -1,5 +1,6 @@
 from .audio_customization_param import AudioCustomizationParam
+from .audio_speech_chunk import AudioSpeechChunk
 from .audio_speech_params import AudioSpeechParams
 from .transcriptions_create_param import TranscriptionsParam
 
-__all__ = ['AudioSpeechParams', 'AudioCustomizationParam', 'TranscriptionsParam']
+__all__ = ['AudioSpeechParams', 'AudioCustomizationParam', 'TranscriptionsParam', 'AudioSpeechChunk']

src/zai/types/audio/audio_speech_chunk.py

@@ -0,0 +1,32 @@
+from typing import List, Optional, Dict, Any
+
+from ...core import BaseModel
+
+__all__ = [
+    "AudioSpeechChunk",
+    "AudioError",
+    "AudioSpeechChoice",
+    "AudioSpeechDelta"
+]
+
+
+class AudioSpeechDelta(BaseModel):
+    content: Optional[str] = None
+    role: Optional[str] = None
+
+
+class AudioSpeechChoice(BaseModel):
+    delta: AudioSpeechDelta
+    finish_reason: Optional[str] = None
+    index: int
+
+class AudioError(BaseModel):
+    code: Optional[str] = None
+    message: Optional[str] = None
+
+
+class AudioSpeechChunk(BaseModel):
+    choices: List[AudioSpeechChoice]
+    request_id: Optional[str] = None
+    created: Optional[int] = None
+    error: Optional[AudioError] = None

src/zai/types/audio/audio_speech_params.py

@@ -29,3 +29,7 @@ class AudioSpeechParams(TypedDict, total=False):
 	sensitive_word_check: Optional[SensitiveWordCheckRequest]
 	request_id: str
 	user_id: str
+	encode_format: str
+	speed: float
+	volume: float
+	stream: bool