LLM-Bridge 1.14.1__tar.gz → 1.15.0a0__tar.gz

{llm_bridge-1.14.1 → llm_bridge-1.15.0a0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: LLM-Bridge
-Version: 1.14.1
+Version: 1.15.0a0
 Summary: A Bridge for LLMs
 Author-email: windsnow1025 <windsnow1025@gmail.com>
 License-Expression: MIT
@@ -37,21 +37,21 @@ PyPI: [https://pypi.org/project/LLM-Bridge/](https://pypi.org/project/LLM-Bridge
     1. **Model Message Converter**: converts general messages to model messages
         1. **Media Processor**: converts general media (Image, Audio, Video, PDF) to model compatible formats.
 3. **Chat Client**: generate stream or non-stream responses
-    - **Model Thoughts**: captures and formats the model's thinking process
-    - **Code Execution**: auto generate and execute Python code
-    - **Web Search + Citations**: extracts and formats citations from search results
+    - **Model Thoughts**: captures the model's thinking process
+    - **Code Execution**: generates and executes Python code
+    - **Web Search**: generates response from search results
     - **Token Counter**: tracks and reports input and output token usage
 
 ### Supported Features for API Types
 
 The features listed represent the maximum capabilities of each API type supported by LLM Bridge.
 
-| API Type | Input Format                   | Capabilities                                                        | Output Format     |
-|----------|--------------------------------|---------------------------------------------------------------------|-------------------|
-| OpenAI   | Text, Image, PDF               | Thinking, Web Search, Code Execution                                | Text              |
-| Gemini   | Text, Image, Video, Audio, PDF | Thinking, Web Search + Citations, Code Execution, Structured Output | Text, Image, File |
-| Claude   | Text, Image, PDF               | Thinking, Web Search, Code Execution                                | Text              |
-| Grok     | Text, Image                    |                                                                     | Text              |
+| API Type | Input Format                   | Capabilities                                            | Output Format     |
+|----------|--------------------------------|---------------------------------------------------------|-------------------|
+| OpenAI   | Text, Image, PDF               | Thinking, Web Search, Code Execution                    | Text              |
+| Gemini   | Text, Image, Video, Audio, PDF | Thinking, Web Search, Code Execution, Structured Output | Text, Image, File |
+| Claude   | Text, Image, PDF               | Thinking, Web Search, Code Execution                    | Text              |
+| Grok     | Text, Image                    |                                                         | Text              |
 
 #### Planned Features
 
@@ -70,7 +70,7 @@ The features listed represent the maximum capabilities of each API type supporte
   uv sync --refresh
   ```
 
-### Pycharm Professional
+### Pycharm
 
 1. Add New Interpreter >> Add Local Interpreter
   - Environment: Select existing

{llm_bridge-1.14.1 → llm_bridge-1.15.0a0}/README.md

@@ -13,21 +13,21 @@ PyPI: [https://pypi.org/project/LLM-Bridge/](https://pypi.org/project/LLM-Bridge
     1. **Model Message Converter**: converts general messages to model messages
         1. **Media Processor**: converts general media (Image, Audio, Video, PDF) to model compatible formats.
 3. **Chat Client**: generate stream or non-stream responses
-    - **Model Thoughts**: captures and formats the model's thinking process
-    - **Code Execution**: auto generate and execute Python code
-    - **Web Search + Citations**: extracts and formats citations from search results
+    - **Model Thoughts**: captures the model's thinking process
+    - **Code Execution**: generates and executes Python code
+    - **Web Search**: generates response from search results
     - **Token Counter**: tracks and reports input and output token usage
 
 ### Supported Features for API Types
 
 The features listed represent the maximum capabilities of each API type supported by LLM Bridge.
 
-| API Type | Input Format                   | Capabilities                                                        | Output Format     |
-|----------|--------------------------------|---------------------------------------------------------------------|-------------------|
-| OpenAI   | Text, Image, PDF               | Thinking, Web Search, Code Execution                                | Text              |
-| Gemini   | Text, Image, Video, Audio, PDF | Thinking, Web Search + Citations, Code Execution, Structured Output | Text, Image, File |
-| Claude   | Text, Image, PDF               | Thinking, Web Search, Code Execution                                | Text              |
-| Grok     | Text, Image                    |                                                                     | Text              |
+| API Type | Input Format                   | Capabilities                                            | Output Format     |
+|----------|--------------------------------|---------------------------------------------------------|-------------------|
+| OpenAI   | Text, Image, PDF               | Thinking, Web Search, Code Execution                    | Text              |
+| Gemini   | Text, Image, Video, Audio, PDF | Thinking, Web Search, Code Execution, Structured Output | Text, Image, File |
+| Claude   | Text, Image, PDF               | Thinking, Web Search, Code Execution                    | Text              |
+| Grok     | Text, Image                    |                                                         | Text              |
 
 #### Planned Features
 
@@ -46,7 +46,7 @@ The features listed represent the maximum capabilities of each API type supporte
   uv sync --refresh
   ```
 
-### Pycharm Professional
+### Pycharm
 
 1. Add New Interpreter >> Add Local Interpreter
   - Environment: Select existing

{llm_bridge-1.14.1 → llm_bridge-1.15.0a0}/llm_bridge/__init__.py

@@ -2,6 +2,6 @@ from .logic.chat_generate.chat_client_factory import create_chat_client
 from .logic.chat_generate.chat_message_converter import *
 from .logic.message_preprocess.message_preprocessor import preprocess_messages
 from .logic.model_prices import ModelPrice, get_model_prices, find_model_prices, calculate_chat_cost
-from .type.chat_response import Citation, ChatResponse
+from .type.chat_response import ChatResponse
 from .type.message import Role, Message, Content, ContentType
 from .type.serializer import serialize

{llm_bridge-1.14.1 → llm_bridge-1.15.0a0}/llm_bridge/client/implementations/gemini/gemini_response_handler.py

@@ -7,7 +7,7 @@ from google.genai.types import Part
 
 from llm_bridge.client.implementations.gemini.gemini_token_counter import count_gemini_tokens
 from llm_bridge.client.implementations.printing_status import PrintingStatus
-from llm_bridge.type.chat_response import Citation, ChatResponse, File
+from llm_bridge.type.chat_response import ChatResponse, File
 
 
 class GeminiResponseHandler:
@@ -26,7 +26,6 @@ class GeminiResponseHandler:
         code_output: str = ""
         files: list[File] = []
         display: Optional[str] = None
-        citations: list[Citation] = extract_citations(response)
         input_tokens, stage_output_tokens = await count_gemini_tokens(response)
 
         parts: list[Part] = []
@@ -89,19 +88,6 @@ class GeminiResponseHandler:
             code_output=code_output,
             files=files,
             display=display,
-            citations=citations,
             input_tokens=input_tokens,
             output_tokens=output_tokens,
         )
-
-
-def extract_citations(response: types.GenerateContentResponse) -> list[Citation]:
-    citations = []
-    if candidates := response.candidates:
-        if grounding_metadata := candidates[0].grounding_metadata:
-            if grounding_supports := grounding_metadata.grounding_supports:
-                for grounding_support in grounding_supports:
-                    citation_indices = [index + 1 for index in grounding_support.grounding_chunk_indices]
-                    citation_text = grounding_support.segment.text
-                    citations.append(Citation(text=citation_text, indices=citation_indices))
-    return citations

{llm_bridge-1.14.1 → llm_bridge-1.15.0a0}/llm_bridge/client/implementations/openai/non_stream_openai_responses_client.py

@@ -1,19 +1,17 @@
 import logging
 import re
-from pprint import pprint
-from typing import Optional
 
 import httpx
 import openai
 from fastapi import HTTPException
 from openai import APIStatusError
-from openai.types.responses import WebSearchToolParam, Response, ResponseOutputItem, ResponseOutputMessage, \
+from openai.types.responses import Response, ResponseOutputItem, ResponseOutputMessage, \
     ResponseOutputText, ResponseReasoningItem
 
 from llm_bridge.client.implementations.openai.openai_token_couter import count_openai_responses_input_tokens, \
     count_openai_output_tokens
 from llm_bridge.client.model_client.openai_client import OpenAIClient
-from llm_bridge.type.chat_response import ChatResponse, Citation, File
+from llm_bridge.type.chat_response import ChatResponse, File
 from llm_bridge.type.serializer import serialize
 
 
@@ -27,7 +25,6 @@ def process_openai_responses_non_stream_response(
     text: str = ""
     thought: str = ""
     files: list[File] = []
-    citations: list[Citation] = []
 
     for output in output_list:
         if output.type == "message":
@@ -36,27 +33,17 @@ def process_openai_responses_non_stream_response(
                 if content.type == "output_text":
                     output_text: ResponseOutputText = content
                     text += output_text.text
-                # Citation is unavailable in OpenAI Responses API
-                # elif annotations := content.annotations:
-                #     for annotation in annotations:
-                #         citations.append(
-                #             Citation(
-                #                 text=content.text[annotation.start_index:annotation.end_index],
-                #                 url=annotation.url
-                #             )
-                #         )
         elif output.type == "reasoning":
             reasoning_item: ResponseReasoningItem = output
             for summary_delta in reasoning_item.summary:
                 thought += summary_delta.text
-        # Image Generation to be tested
-        # if output.type == "image_generation_call":
-        #     file = File(
-        #         name="generated_image.png",
-        #         data=output.result,
-        #         type="image/png",
-        #     )
-        #     files.append(file)
+        if output.type == "image_generation_call":
+            file = File(
+                name="generated_image.png",
+                data=output.result,
+                type="image/png",
+            )
+            files.append(file)
 
     chat_response = ChatResponse(text=text, files=files)
     output_tokens = count_openai_output_tokens(chat_response)
@@ -64,7 +51,6 @@ def process_openai_responses_non_stream_response(
         text=text,
         thought=thought,
         files=files,
-        citations=citations,
         input_tokens=input_tokens,
         output_tokens=output_tokens,
     )

{llm_bridge-1.14.1 → llm_bridge-1.15.0a0}/llm_bridge/client/implementations/openai/steam_openai_responses_client.py

@@ -1,7 +1,6 @@
 import logging
 import re
-from pprint import pprint
-from typing import AsyncGenerator, Optional
+from typing import AsyncGenerator
 
 import httpx
 import openai
@@ -12,7 +11,7 @@ from openai.types.responses import ResponseStreamEvent, ResponseReasoningSummary
 from llm_bridge.client.implementations.openai.openai_token_couter import count_openai_responses_input_tokens, \
     count_openai_output_tokens
 from llm_bridge.client.model_client.openai_client import OpenAIClient
-from llm_bridge.type.chat_response import ChatResponse, Citation, File
+from llm_bridge.type.chat_response import ChatResponse, File
 from llm_bridge.type.serializer import serialize
 
 
@@ -20,7 +19,6 @@ def process_delta(event: ResponseStreamEvent) -> ChatResponse:
     text: str = ""
     thought: str = ""
     files: list[File] = []
-    citations: list[Citation] = []
 
     if event.type == "response.output_text.delta":
         text_delta_event: ResponseTextDeltaEvent = event
@@ -28,23 +26,18 @@ def process_delta(event: ResponseStreamEvent) -> ChatResponse:
     elif event.type == "response.reasoning_summary_text.delta":
         reasoning_summary_text_delta_event: ResponseReasoningSummaryTextDeltaEvent = event
         thought = reasoning_summary_text_delta_event.delta
-    # Citation is unavailable in OpenAI Responses API
-    elif event.type == "response.output_text.annotation.added":
-        pass
-    # Image Generation to be tested
-    # if event.type == "response.image_generation_call.partial_image":
-    #     file = File(
-    #         name="generated_image.png",
-    #         data=event.partial_image_b64,
-    #         type="image/png",
-    #     )
-    #     files.append(file)
+    if event.type == "response.image_generation_call.partial_image":
+        file = File(
+            name="generated_image.png",
+            data=event.partial_image_b64,
+            type="image/png",
+        )
+        files.append(file)
 
     chat_response = ChatResponse(
         text=text,
         thought=thought,
         files=files,
-        citations=citations,
     )
     return chat_response
 
@@ -61,7 +54,6 @@ async def generate_chunk(
                 text=chat_response.text,
                 thought=chat_response.thought,
                 files=chat_response.files,
-                citations=chat_response.citations,
                 input_tokens=input_tokens,
                 output_tokens=output_tokens,
             )

{llm_bridge-1.14.1 → llm_bridge-1.15.0a0}/llm_bridge/type/chat_response.py

@@ -9,20 +9,6 @@ class File:
     type: str
 
 
-@dataclass
-class Citation:
-    text: str
-    indices: list[int]
-
-
-# TODO: adapt to different Citation formats
-# @dataclass
-# class Citation:
-#     text: str
-#     indices: Optional[list[int]] = None
-#     url: Optional[str] = None
-
-
 @dataclass
 class ChatResponse:
     text: Optional[str] = None
@@ -31,7 +17,6 @@ class ChatResponse:
     code_output: Optional[str] = None
     files: Optional[list[File]] = None
     display: Optional[str] = None
-    citations: Optional[list[Citation]] = None
     error: Optional[str] = None
     input_tokens: Optional[int] = 0
     output_tokens: Optional[int] = 0

{llm_bridge-1.14.1 → llm_bridge-1.15.0a0}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "LLM-Bridge"
-version = "1.14.1"
+version = "1.15.0-alpha.0"
 dependencies = [
     "fastapi",
     "httpx",

{llm_bridge-1.14.1 → llm_bridge-1.15.0a0}/usage/main.py

@@ -98,7 +98,7 @@ messages = [
             # Content(type=ContentType.Text, data="Explain the concept of Occam's Razor and provide a simple, everyday example."),
 
             # Web Search
-            # Content(type=ContentType.Text, data="What's the weather in NYC today?"),
+            Content(type=ContentType.Text, data="What's the weather in NYC today?"),
 
             # Image Understanding
             # Content(type=ContentType.File, data="https://www.gstatic.com/webp/gallery3/1.png"),
@@ -111,7 +111,7 @@ messages = [
             # Content(type=ContentType.Text, data="What is in https://www.windsnow1025.com/"),
 
             # Code Execution
-            Content(type=ContentType.Text, data="What is the sum of the first 50 prime numbers? Generate and run code for the calculation, and make sure you get all 50."),
+            # Content(type=ContentType.Text, data="What is the sum of the first 50 prime numbers? Generate and run code for the calculation, and make sure you get all 50."),
 
             # File Output
             # Content(type=ContentType.File, data="https://www.windsnow1025.com/minio/windsnow/uploads/1/1758384216123-script.py"),
@@ -139,23 +139,23 @@ messages = [
 # model = "gpt-5-pro"
 # model = "gpt-5"
 # model = "gpt-4.1"
-# model = "gemini-3-pro-preview"
+model = "gemini-3-pro-preview"
 # model = "gemini-3-pro-image-preview"
-model = "gemini-3-flash-preview"
+# model = "gemini-3-flash-preview"
 # model = "grok-4-1-fast-reasoning"
 # model = "claude-sonnet-4-5"
 # model = "claude-opus-4-5"
-# api_type = "Gemini-Vertex"
+api_type = "Gemini-Vertex"
 # api_type = "Gemini-Free"
-api_type = "Gemini-Paid"
+# api_type = "Gemini-Paid"
 # api_type = "OpenAI"
 # api_type = "OpenAI-Azure"
 # api_type = "OpenAI-GitHub"
 # api_type = "Claude"
 # api_type = "Grok"
 temperature = 0
-stream = True
-# stream = False
+# stream = True
+stream = False
 thought = True
 # thought = False
 code_execution = True