llama-github 0.1.0__tar.gz → 0.1.1__tar.gz

llama_github-0.1.1/CHANGELOG.md

@@ -0,0 +1,34 @@
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [0.1.1] - 2024-08-23
+
+### Added
+- Implemented `answer_with_context` method for direct answer generation (closes #6)
+- Added support for Mistral AI LLM provider
+- Enhanced `retrieve_context` function to include metadata (e.g., URLs) with each context string (closes #2)
+
+### Changed
+- Improved reranking with jina-reranker-v2 for better context retrieval
+- Updated return type of `retrieve_context` to accommodate metadata
+
+### Fixed
+- Resolved warning during context retrieval (closes #3)
+
+### Improved
+- Enhanced overall context retrieval process
+- Expanded LLM support for more versatile use cases
+
+## [0.1.0] - 2024-08-15
+
+### Added
+- Initial release of llama-github
+- Basic functionality for retrieving context from GitHub repositories
+- Integration with LLM for processing and generating responses
+
+[0.1.1]: https://github.com/JetXu-LLM/llama-github/compare/v0.1.0...v0.1.1
+[0.1.0]: https://github.com/JetXu-LLM/llama-github/releases/tag/v0.1.0

llama_github-0.1.1/MANIFEST.in

@@ -0,0 +1,2 @@
+include llama_github/config/config.json
+include CHANGELOG.md

{llama_github-0.1.0 → llama_github-0.1.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: llama-github
-Version: 0.1.0
+Version: 0.1.1
 Summary: Llama-github is an open-source Python library that empowers LLM Chatbots, AI Agents, and Auto-dev Agents to conduct Retrieval from actively selected GitHub public projects. It Augments through LLMs and Generates context for any coding question, in order to streamline the development of sophisticated AI-driven applications.
 Home-page: https://github.com/JetXu-LLM/llama-github
 Author: Jet Xu
@@ -29,9 +29,16 @@ Requires-Dist: flake8; extra == "dev"
 
 # llama-github
 
+[![PyPI version](https://badge.fury.io/py/llama-github.svg)](https://badge.fury.io/py/llama-github)
+[![Downloads](https://static.pepy.tech/badge/Llama-github)](https://pepy.tech/project/Llama-github)
 [![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
 
-Llama-github is an open-source Python library that empowers LLM Chatbots, AI Agents, and Auto-dev Agents to conduct Retrieval from actively selected GitHub public contents (Repository, Issue, Code). It Augments through LLMs and Generates context for any coding question, in order to streamline the development of sophisticated AI-driven applications.
+Llama-github is a powerful tool that helps you retrieve the most relevant code snippets, issues, and repository information from GitHub based on your queries, transforming them into valuable knowledge context. It empowers LLM Chatbots, AI Agents, and Auto-dev Agents to solve complex coding tasks. Whether you're a developer looking for quick solutions or an engineer implementing advanced Auto Dev AI Agents, llama-github makes it easy and efficient.
+
+If you like this project or believe it has potential, please give it a ⭐️. Your support is our greatest motivation!
+
+## Architecture
+![High Level Architecture](./docs/high_level_architecture.drawio.svg)
 
 ## Installation
 ```
@@ -82,6 +89,26 @@ For more advanced usage and examples, please refer to the [documentation](docs/u
 
 - **🛠️ Logging and Error Handling**: We understand the importance of smooth operations and easy troubleshooting. That's why llama-github comes equipped with comprehensive logging and error handling mechanisms. Gain deep insights into the library's behavior, quickly diagnose issues, and maintain a stable and reliable development workflow.
 
+## Vision and Roadmap
+
+### Vision
+
+Our vision is to become a pivotal module in the future of AI-driven development solutions, seamlessly integrating with GitHub to empower LLMs in automatically resolving complex coding tasks.
+
+![Vision Architecture](./docs/vision.drawio.svg)
+
+### Roadmap
+
+For a detailed view of our project roadmap, please visit our [Project Roadmap](https://github.com/users/JetXu-LLM/projects/2).
+
+## Acknowledgments
+
+We would like to express our gratitude to the following open-source projects for their support and contributions:
+
+- **[LangChain](https://github.com/langchain-ai/langchain)**: For providing the foundational framework that empowers the LLM prompting and processing capabilities in llama-github.
+- **[Jina.ai](https://github.com/jina-ai/reader)**: For offering s.jina.ai API and open source reranker and embedding models that enhance the accuracy and relevance of the generated contexts in llama-github.
+
+Their contributions have been instrumental in the development of llama-github, and we highly recommend checking out their projects for more innovative solutions.
 
 ## Contributing
 
@@ -93,8 +120,43 @@ This project is licensed under the terms of the Apache 2.0 license. See the [LIC
 
 ## Contact
 
-If you have any questions, suggestions, or feedback, please feel free to reach out to us at [Voldemort.xu@foxmail.com](mailto:Voldemort.xu@foxmail.com).
+If you have any questions, suggestions, or feedback, please feel free to reach out to us at [Jet Xu's email](mailto:Voldemort.xu@foxmail.com).
 
 ---
 
 Thank you for choosing llama-github! We hope this library enhances your AI development experience and helps you build powerful applications with ease.
+
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [0.1.1] - 2024-08-23
+
+### Added
+- Implemented `answer_with_context` method for direct answer generation (closes #6)
+- Added support for Mistral AI LLM provider
+- Enhanced `retrieve_context` function to include metadata (e.g., URLs) with each context string (closes #2)
+
+### Changed
+- Improved reranking with jina-reranker-v2 for better context retrieval
+- Updated return type of `retrieve_context` to accommodate metadata
+
+### Fixed
+- Resolved warning during context retrieval (closes #3)
+
+### Improved
+- Enhanced overall context retrieval process
+- Expanded LLM support for more versatile use cases
+
+## [0.1.0] - 2024-08-15
+
+### Added
+- Initial release of llama-github
+- Basic functionality for retrieving context from GitHub repositories
+- Integration with LLM for processing and generating responses
+
+[0.1.1]: https://github.com/JetXu-LLM/llama-github/compare/v0.1.0...v0.1.1
+[0.1.0]: https://github.com/JetXu-LLM/llama-github/releases/tag/v0.1.0

llama_github-0.1.0/llama_github.egg-info/PKG-INFO → llama_github-0.1.1/README.md

@@ -1,37 +1,15 @@
-Metadata-Version: 2.1
-Name: llama-github
-Version: 0.1.0
-Summary: Llama-github is an open-source Python library that empowers LLM Chatbots, AI Agents, and Auto-dev Agents to conduct Retrieval from actively selected GitHub public projects. It Augments through LLMs and Generates context for any coding question, in order to streamline the development of sophisticated AI-driven applications.
-Home-page: https://github.com/JetXu-LLM/llama-github
-Author: Jet Xu
-Author-email: Voldemort.xu@foxmail.com
-Classifier: Programming Language :: Python :: 3
-Classifier: License :: OSI Approved :: Apache Software License
-Classifier: Operating System :: OS Independent
-Requires-Python: >=3.6
-Description-Content-Type: text/markdown
-License-File: LICENSE
-Requires-Dist: aiohttp
-Requires-Dist: asyncio
-Requires-Dist: dataclasses
-Requires-Dist: PyGithub
-Requires-Dist: langchain
-Requires-Dist: numpy
-Requires-Dist: pydantic
-Requires-Dist: requests
-Requires-Dist: torch
-Requires-Dist: transformers
-Requires-Dist: urllib3
-Provides-Extra: dev
-Requires-Dist: pytest; extra == "dev"
-Requires-Dist: black; extra == "dev"
-Requires-Dist: flake8; extra == "dev"
-
 # llama-github
 
+[![PyPI version](https://badge.fury.io/py/llama-github.svg)](https://badge.fury.io/py/llama-github)
+[![Downloads](https://static.pepy.tech/badge/Llama-github)](https://pepy.tech/project/Llama-github)
 [![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
 
-Llama-github is an open-source Python library that empowers LLM Chatbots, AI Agents, and Auto-dev Agents to conduct Retrieval from actively selected GitHub public contents (Repository, Issue, Code). It Augments through LLMs and Generates context for any coding question, in order to streamline the development of sophisticated AI-driven applications.
+Llama-github is a powerful tool that helps you retrieve the most relevant code snippets, issues, and repository information from GitHub based on your queries, transforming them into valuable knowledge context. It empowers LLM Chatbots, AI Agents, and Auto-dev Agents to solve complex coding tasks. Whether you're a developer looking for quick solutions or an engineer implementing advanced Auto Dev AI Agents, llama-github makes it easy and efficient.
+
+If you like this project or believe it has potential, please give it a ⭐️. Your support is our greatest motivation!
+
+## Architecture
+![High Level Architecture](./docs/high_level_architecture.drawio.svg)
 
 ## Installation
 ```
@@ -82,6 +60,26 @@ For more advanced usage and examples, please refer to the [documentation](docs/u
 
 - **🛠️ Logging and Error Handling**: We understand the importance of smooth operations and easy troubleshooting. That's why llama-github comes equipped with comprehensive logging and error handling mechanisms. Gain deep insights into the library's behavior, quickly diagnose issues, and maintain a stable and reliable development workflow.
 
+## Vision and Roadmap
+
+### Vision
+
+Our vision is to become a pivotal module in the future of AI-driven development solutions, seamlessly integrating with GitHub to empower LLMs in automatically resolving complex coding tasks.
+
+![Vision Architecture](./docs/vision.drawio.svg)
+
+### Roadmap
+
+For a detailed view of our project roadmap, please visit our [Project Roadmap](https://github.com/users/JetXu-LLM/projects/2).
+
+## Acknowledgments
+
+We would like to express our gratitude to the following open-source projects for their support and contributions:
+
+- **[LangChain](https://github.com/langchain-ai/langchain)**: For providing the foundational framework that empowers the LLM prompting and processing capabilities in llama-github.
+- **[Jina.ai](https://github.com/jina-ai/reader)**: For offering s.jina.ai API and open source reranker and embedding models that enhance the accuracy and relevance of the generated contexts in llama-github.
+
+Their contributions have been instrumental in the development of llama-github, and we highly recommend checking out their projects for more innovative solutions.
 
 ## Contributing
 
@@ -93,7 +91,7 @@ This project is licensed under the terms of the Apache 2.0 license. See the [LIC
 
 ## Contact
 
-If you have any questions, suggestions, or feedback, please feel free to reach out to us at [Voldemort.xu@foxmail.com](mailto:Voldemort.xu@foxmail.com).
+If you have any questions, suggestions, or feedback, please feel free to reach out to us at [Jet Xu's email](mailto:Voldemort.xu@foxmail.com).
 
 ---
 

{llama_github-0.1.0 → llama_github-0.1.1}/llama_github/config/config.json

@@ -1,12 +1,12 @@
 {
-    "general_prompt": "You are a highly intelligent assistant with expertise in GitHub repositories and coding practices. Your task is to analyze questions related to GitHub projects, coding issues, or programming concepts. Using your extensive knowledge base, you will provide detailed, accurate, and contextually relevant answers. You have the ability to understand complex coding queries, retrieve pertinent information from GitHub repositories, and augment this data with your advanced reasoning capabilities. Your responses should guide developers towards solving their problems, understanding new concepts, or finding the information they seek related to GitHub projects and software development.",
+    "general_prompt": "You are a highly intelligent assistant with expertise in GitHub repositories and coding practices. Your primary task is to provide comprehensive and accurate answers to questions related to GitHub projects, coding issues, or programming concepts. When analyzing queries, focus on delivering a complete response that directly addresses the original question. While you may be provided with additional context, use this information judiciously to enhance your answer without deviating from the main point. Your extensive knowledge base, combined with your ability to understand complex coding queries and retrieve pertinent information, should be the foundation of your responses. When referencing provided context, integrate it seamlessly into your answer without explicitly evaluating or critiquing it. Your goal is to guide developers towards solutions, explain concepts clearly, or provide the information they seek about GitHub projects and software development, always ensuring that your final response is a cohesive and complete answer to the original question.",
     "always_answer_prompt": "**Instructions:**\nAs an advanced AI assistant with deep expertise in GitHub repositories, coding practices, and programming concepts, your primary goal is to provide concise, accurate, and contextually relevant answers to complex coding queries. When presented with a question, your first step is to analyze the query and generate a succinct abstraction that captures its core essence by using only one sentence, especially if the original question is lengthy or convoluted.\n\nNext, leverage your extensive knowledge base and reasoning capabilities to craft a coherent and informative response. If possible, enhance your answer with sample code snippets that demonstrate the practical application of the concepts discussed. Remember, your responses should guide developers towards solving their problems, understanding new concepts, or finding the information they seek related to GitHub projects and software development. Please keep your responses concise and to the point, focusing on the most essential information needed to address the query. Avoid generating long articles or overly detailed explanations.\n\nIn addition to the answer itself, provide a brief analysis of how you would approach searching for relevant code and issues within GitHub repositories. This analysis should outline your thought process and the key factors you would consider when conducting these searches. However, keep this analysis concise and focused on the high-level logic rather than delving into specific search criteria or keywords.\n\nThroughout your responses, prioritize clarity and brevity. Focus on delivering the most essential information needed to address the query effectively. Even if certain details are unknown, ensure that your answers are plausible, useful, and serve as a foundation for further exploration and context generation.\n\nRemember, your ultimate aim is to empower developers with the knowledge and guidance they need to overcome challenges, expand their understanding, and navigate the vast landscape of GitHub repositories and software development practices.",
     "code_search_criteria_prompt": "**Instructions:**\n- **Expertise-Driven Github Code Search Criteria Generation:** Generate GitHub code search criteria strings based on the provided question and its draft answer. Analyze both the question and answer to identify key concepts, technologies, and coding practices that can help locate relevant code snippets on GitHub. Always include the `language:` qualifier to focus your search on language-related content.\n\n**Output Format:** Present each search criteria string on a new line, formatted for immediate use in GitHub's code search, without additional explanations or commentary.\n\n**Optimization Considerations:**\n- **Keyword Relevance:** Extract keywords and phrases tightly related to the question from the question and answer that are likely to appear in relevant code and code comments. Prioritize terms that reflect specific coding concepts, libraries, or techniques. Avoid generic terms like \"example\" or \"integration\" that may not be present in actual code.\n- **Contextual Understanding:** Use the provided answer as additional context to inform your keyword selection. Identify key insights, technologies, or approaches mentioned in the answer tightly related to the question that can help refine the search criteria.\n- **Language and Platform Specificity:** If the question is specific to a certain programming language or platform, ensure to include relevant language or platform-specific keywords, libraries, or frameworks in the search criteria. This helps filter out irrelevant results from other languages or platforms.\n- **Simplicity and Effectiveness:** Craft search criteria with simple and limited keywords which could lead to precise search results to relevant code snippets tightly related to original question. Strike a balance between specificity and breadth to ensure the criteria capture the essential aspects of the question and answer. The search criteria should be neither too narrow that no results are returned, nor too broad that many irrelevant results are included.\n- **Multiple Perspectives:** Generate multiple search criteria strings that approach the question from different angles or emphasize different aspects mentioned in the question and answer. This increases the chances of finding relevant code snippets.",
     "issue_search_criteria_prompt": "**Instructions:**\n- **Question-Driven GitHub Issue Search Criteria Generation:** Generate GitHub issue search criteria strings based on the provided question. Analyze the question to identify key concepts, technologies, and problem-solving approaches that can help locate relevant issues on GitHub. Consider using relevant `label:` or `is:` qualifiers when applicable.\n\n**Output Format:** Present each search criteria string on a new line, formatted for immediate use in GitHub's issue search, without additional explanations or commentary.\n\n**Optimization Considerations:**\n- **Keyword Relevance:** Extract keywords and phrases tightly related to the question that are likely to appear in issue titles, descriptions, and discussions. Prioritize terms that reflect specific problems, error messages, or technologies. Avoid generic terms like \"help\" or \"problem\" that may not effectively narrow down the search results.\n- **Contextual Understanding:** Use the question's draft answer to inform your keyword selection. Identify key aspects, technologies, or potential troubleshooting areas tightly related to the question but not only specific aspects of answers that can help refine the search criteria.\n- **Simplicity and Effectiveness:** Craft search criteria with simple and limited keywords which could lead to precise search results relevant to the original question. Strike a balance between specificity and breadth to ensure the criteria capture the essential aspects of the question without being overly restrictive.\n- **Multiple Perspectives:** Generate multiple search criteria strings that approach the question from different angles or emphasize different aspects mentioned in the question. This increases the chances of finding relevant issues that discuss similar problems or solutions.\n- **Leveraging Labels:** When appropriate, include relevant `label:` qualifiers in the search criteria to narrow down the results to issues with specific labels, such as \"bug,\" \"enhancement,\" or \"documentation.\" This can help focus the search on issues that align with the nature of the question.\n- **Considering Issue Discussions:** Keep in mind that issue discussions often contain valuable information, experiences, and workarounds shared by other developers. Craft search criteria that not only match the issue title and description but also consider the likelihood of the keywords appearing in the issue's comments and discussions.",
     "repo_search_criteria_prompt": "**Instructions:**\n- **Expertise-Driven Github Repository Search Criteria Generation:** Generate GitHub repo search criteria strings based on the provided question. Analyze the question leverage your expertise for related key concepts, technologies, and problem-solving approaches that can help locate relevant repositories on GitHub. Focus on practical keywords and phrases likely to be present in repository names, descriptions, and topics. Use the `language:` qualifier to direct your search toward repositories written in a specific language, keeping the criteria simple and effective.\n- **Necessity Score Determination:** Evaluate the necessity of conducting a GitHub repository search based on the difficulty of question. Determine if repository-level information is essential to comprehensively address the question. Assign a necessity score indicating the importance of performing a repository search.\n\n**Output Format:**\n- **Necessity Score:** Begin your output with a necessity score (0-100) indicating the importance of performing a separate GitHub repository search. Use the following scale:\n  - 0-59: Low necessity - Only code and issue search results is sufficient.\n  - 60-79: Medium necessity - One repository search may offer additional insights and context.\n  - 80-100: High necessity - Two repository searches are crucial to gather comprehensive information, such as project structure, documentation, or community engagement, to thoroughly address the question.\n\n- **Search Criteria:** Present each search criteria string on a new line, formatted for immediate use in GitHub's repository search, without additional explanations or commentary.\n**Optimization Considerations:**\n- **Keyword Relevance:** Generate search criteria keywords and phrases from the question that are uniquely relevant to repository names, descriptions, and topics. Prioritize terms that reflect the broader context, expertise, and strategic thinking required to address the question effectively. Avoid generic terms that may lead to irrelevant search results.\n- **Simplicity and Effectiveness:** Craft search criteria that are simple yet effective in narrowing down the repository search results to the most relevant and informative ones. Strike a balance between specificity and breadth, ensuring that the criteria capture the essential aspects of the question without being overly restrictive. Aim for criteria that yield a manageable number of high-quality repository results.\n- **Language and Platform Specificity:** If the question pertains to a specific programming language or platform, incorporate relevant language or platform-specific keywords in the search criteria. Use the `language:` qualifier to filter repositories based on the language of interest. This helps focus the search on repositories that are more likely to contain relevant code, documentation, and community expertise.\n- **Multiple Criteria Flexibility:** Generate multiple search criteria strings that approach the question from different angles or emphasize different aspects mentioned in the question. This flexibility allows for a more comprehensive repository search, increasing the chances of discovering relevant repositories that may offer valuable insights, code samples, or best practices related to the question at hand.",
     "scoring_context_prompt": "You are an expert in evaluating the relevance of coding-related contexts to given questions. Your primary function is to analyze the provided context and question, and output a single integer score between 0 and 100, indicating how well the context supports answering the question.\n\nScoring criteria:\n0-20: The context is completely irrelevant to the question and provides no useful information to answer it.\n21-40: The context is slightly relevant to the question but lacks crucial information to provide a complete answer.\n41-60: The context is somewhat relevant to the question and provides some useful information, but it may not be sufficient to fully answer the question.\n61-80: The context is highly relevant to the question and provides most of the necessary information to answer it, but some minor details may be missing.\n81-100: The context is extremely relevant to the question and provides all the necessary information to comprehensively answer it.\n\nRemember, your output should consist of only a single integer score without any additional text or explanation. Analyze the context and question carefully, and provide a score that accurately reflects the relevance of the context in answering the question.",
     "default_embedding": "jinaai/jina-embeddings-v2-base-code",
-    "default_reranker": "jinaai/jina-reranker-v1-turbo-en",
+    "default_reranker": "jinaai/jina-reranker-v2-base-multilingual",
     "min_stars_to_keep_result": 20,
     "max_workers": 8,
     "code_search_max_hits": 30,
@@ -16,5 +16,5 @@
     "issue_chunk_size": 7000,
     "repo_chunk_size": 7000,
     "google_chunk_size": 7000,
-    "top_n_contexts": 5
+    "top_n_contexts": 4
 }

{llama_github-0.1.0 → llama_github-0.1.1}/llama_github/github_rag.py

@@ -1,11 +1,10 @@
 from llama_github.logger import logger
 from llama_github.config.config import config
-from typing import List, Optional, Any
+from typing import List, Optional, Any, Dict
 from dataclasses import dataclass
 from pprint import pformat
 
 from llama_github.llm_integration.initial_load import LLMManager
-from llama_github.llm_integration.llm_handler import LLMHandler
 from llama_github.rag_processing.rag_processor import RAGProcessor
 
 from llama_github.github_integration.github_auth_manager import GitHubAuthManager
@@ -32,6 +31,7 @@ class GithubRAG:
     def __init__(self, github_access_token: Optional[str] = None,
                  github_app_credentials: Optional[GitHubAppCredentials] = None,
                  openai_api_key: Optional[str] = None,
+                 mistral_api_key: Optional[str] = None,
                  huggingface_token: Optional[str] = None,
                  jina_api_key: Optional[str] = None,
                  open_source_models_hg_dir: Optional[str] = None,
@@ -92,7 +92,7 @@ class GithubRAG:
             logger.debug(
                 "Initializing llm manager, embedding model & reranker model...")
             self.llm_manager = LLMManager(
-                openai_api_key, huggingface_token, open_source_models_hg_dir, embedding_model, rerank_model, llm)
+                openai_api_key, mistral_api_key, huggingface_token, open_source_models_hg_dir, embedding_model, rerank_model, llm)
             logger.debug(
                 "LLM Manager, Embedding model & Reranker model Initialized.")
 
@@ -112,8 +112,8 @@ class GithubRAG:
         try:
             logger.info("Retrieving context...")
             if simple_mode:
-                #In simple mode, only a Google search will be conducted based on the user's question.
-                #This model is not suitable for long questions (e.g., questions with more than 20 words).
+                # In simple mode, only a Google search will be conducted based on the user's question.
+                # This model is not suitable for long questions (e.g., questions with more than 20 words).
                 task_google_search = asyncio.create_task(
                     self.google_search_retrieval(query=query))
                 await asyncio.gather(task_google_search)
@@ -121,7 +121,7 @@ class GithubRAG:
                     f"Google search: {str(len(task_google_search.result()))}")
                 context_list = self.rag_processor.arrange_context(
                     google_search_result=task_google_search.result())
-                if len(context_list)>0:
+                if len(context_list) > 0:
                     topn_contexts = await self.rag_processor.retrieve_topn_contexts(
                         context_list=context_list, query=query, top_n=config.get("top_n_contexts"))
             else:
@@ -166,7 +166,7 @@ class GithubRAG:
                     repo_search_result=task_repo_search.result(),
                     google_search_result=task_google_search.result())
 
-                if len(context_list)>0:
+                if len(context_list) > 0:
                     topn_contexts = await self.rag_processor.retrieve_topn_contexts(
                         context_list=context_list, query=query, answer=analyzed_strategy[1], top_n=config.get("top_n_contexts"))
 
@@ -237,9 +237,13 @@ class GithubRAG:
             seen = set()
             unique_list = []
             for d in result:
-                value = d["url"]
-                if value not in seen:
-                    seen.add(value)
+                api_url = d["url"]
+                if api_url not in seen:
+                    seen.add(api_url)
+                    # Transform the API URL to the official GitHub issue webpage URL
+                    html_url = api_url.replace(
+                        'api.github.com/repos', 'github.com').replace('issues/', 'issues/')
+                    d["url"] = html_url
                     unique_list.append(d)
             result = unique_list
 
@@ -330,6 +334,7 @@ class GithubRAG:
                             results_with_index.append({
                                 'index': index,
                                 'full_name': repository.full_name,
+                                'url': repository.html_url,
                                 'content': repo_readme,
                             })
                         # if repo_simple_structure:
@@ -345,4 +350,57 @@ class GithubRAG:
             logger.info("Repo search retrieved successfully.")
         except Exception as e:
             logger.error(f"Error retrieving repos search: {e}")
-        return results_with_index
+        return results_with_index
+
+    def answer_with_context(self, query: str, contexts: Optional[List[Dict[str, Any]]] = None, simple_mode=False) -> str:
+        """
+        Generate an answer based on the given query and optional contexts.
+
+        This method can be called in different environments (Jupyter notebook, 
+        synchronous Python program, asynchronous Python program) and will 
+        handle the async call appropriately.
+
+        Args:
+            query (str): The user's query.
+            contexts (Optional[List[Dict[str, Any]]]): Optional list of context dictionaries.
+                Each dictionary should contain 'content' and 'url' keys.
+
+        Returns:
+            str: The generated answer.
+        """
+        self.loop = asyncio.get_event_loop()
+        ipython = get_ipython()
+        if ipython and ipython.has_trait('kernel'):
+            logger.debug("Running in Jupyter notebook, nest_asyncio applied.")
+            import nest_asyncio
+            nest_asyncio.apply()
+            return asyncio.run(self.async_answer_with_context(query, contexts, simple_mode))
+        if self.loop.is_running():
+            return asyncio.ensure_future(self.async_answer_with_context(query, contexts, simple_mode))
+        return self.loop.run_until_complete(self.async_answer_with_context(query, contexts, simple_mode))
+
+    async def async_answer_with_context(self, query: str, contexts: Optional[List[Dict[str, Any]]] = None, simple_mode=False) -> str:
+        """
+        Asynchronously generate an answer based on the given query and optional contexts.
+
+        Args:
+            query (str): The user's query.
+            contexts (Optional[List[Dict[str, Any]]]): Optional list of context dictionaries.
+                Each dictionary should contain 'content' and 'url' keys.
+
+        Returns:
+            str: The generated answer.
+        """
+        if contexts is None:
+            contexts = await self.async_retrieve_context(query, simple_mode)
+            logger.debug(f"Retrieved contexts: {contexts}")
+        context_contents = [context['context'] for context in contexts]
+        context_urls = [context['url'] for context in contexts]
+
+        answer = await self.rag_processor.llm_handler.ainvoke(
+            human_question=query,
+            context=context_contents,
+            # context_urls=context_urls
+        )
+
+        return answer

{llama_github-0.1.0 → llama_github-0.1.1}/llama_github/llm_integration/initial_load.py

@@ -5,6 +5,7 @@ from threading import Lock
 from langchain_openai import ChatOpenAI
 from langchain_community.embeddings import HuggingFaceEmbeddings
 from langchain_community.cross_encoders import HuggingFaceCrossEncoder
+from langchain_mistralai.chat_models import ChatMistralAI
 
 from llama_github.config.config import config
 from llama_github.logger import logger
@@ -33,6 +34,7 @@ class LLMManager:
 
     def __init__(self,
                  openai_api_key: Optional[str] = None,
+                 mistral_api_key: Optional[str] = None,
                  huggingface_token: Optional[str] = None,
                  open_source_models_hg_dir: Optional[str] = None,
                  embedding_model: Optional[str] = config.get(
@@ -48,11 +50,16 @@ class LLMManager:
         if llm is not None:
             self.llm = llm
             self.model_type = "Custom_langchain_llm"
+        elif mistral_api_key is not None and mistral_api_key != "" and self.llm is None:
+            logger.info("Initializing Mistral API...")
+            self.llm = ChatMistralAI(mistral_api_key=mistral_api_key, model="mistral-large-latest")
+            self.llm_simple = ChatMistralAI(mistral_api_key=mistral_api_key, model="open-mistral-nemo")
+            self.model_type = "OpenAI"
         elif openai_api_key is not None and openai_api_key != "" and self.llm is None:
             logger.info("Initializing OpenAI API...")
             self.llm = ChatOpenAI(api_key=openai_api_key, model="gpt-4-turbo")
             self.llm_simple = ChatOpenAI(
-                api_key=openai_api_key, model="gpt-3.5-turbo")
+                api_key=openai_api_key, model="gpt-4o-mini")
             self.model_type = "OpenAI"
         # Initialize for Open Source Models
         elif open_source_models_hg_dir is not None and open_source_models_hg_dir != "" and self.llm is None:

{llama_github-0.1.0 → llama_github-0.1.1}/llama_github/llm_integration/llm_handler.py

@@ -8,11 +8,10 @@ from langchain_core.output_parsers import StrOutputParser
 from llama_github.config.config import config
 from llama_github.logger import logger
 from langchain_core.messages import AIMessage, HumanMessage, SystemMessage
-from langchain_core.pydantic_v1 import BaseModel
+from pydantic import BaseModel
 from typing import Optional
 from langchain_openai import output_parsers
 
-
 class LLMHandler:
     def __init__(self, llm_manager: Optional[LLMManager] = None):
         """
@@ -80,7 +79,6 @@ class LLMHandler:
 
                 # Format the prompt with the provided parameters.
                 formatted_prompt = chat_prompt.format_prompt(**prompt_params)
-                logger.debug(f"Formatted prompt: {formatted_prompt}")
                 # Determine the processing chain based on the presence of an output structure.
                 if output_structure is not None:
                     chain = llm.with_structured_output(output_structure)
@@ -92,7 +90,7 @@ class LLMHandler:
                     response = await chain.ainvoke(formatted_prompt.to_messages())
                 except Exception as e:
                     logger.exception(
-                        f"Call llm with #{human_question}# generated an exception:{e}")
+                        f"Call {'simple ' if simple_llm else ''}llm with #{human_question}# generated an exception:{e}")
                     if output_structure is not None:
                         response = await chain.ainvoke(formatted_prompt.to_messages())
                 return response

{llama_github-0.1.0 → llama_github-0.1.1}/llama_github/rag_processing/rag_processor.py

@@ -209,20 +209,21 @@ class RAGProcessor:
             logger.error(f"Error in get_issue_search_criteria: {e}")
             return []
 
-    def _arrange_code_search_result(self, code_search_result: List[Dict]) -> List[Dict]:
+    def _arrange_code_search_result(self, code_search_result: List[Dict]) -> List[Dict[str, str]]:
         """
-        Arrange the result of code search into chunks suitable for LLM RAG contexts.
+        Arrange the result of Code search with metadata.
 
         Args:
-            code_search_result (list): the result of code search.
+            _arrange_code_search_result (dict): The result of Code search.
 
         Returns:
-            list: the arranged result of code search.
+            List[Dict[str, str]]: The arranged result of Code search with metadata.
         """
         arranged_results = []
 
         for result in code_search_result:
             content = result['content']
+            url = result.get('url', '')
 
             # Split content into chunks
             chunks = self._split_content_into_chunks(
@@ -249,7 +250,7 @@ class RAGProcessor:
                     f"programming language is: {language}\n\n"
                     f"{chunk}"
                 )
-                arranged_results.append(chunk_text)
+                arranged_results.append({'context': chunk_text, 'url': url})
 
         return arranged_results
 
@@ -290,173 +291,182 @@ class RAGProcessor:
         chunks = splitter.split_text(content)
         return chunks
 
-    def _arrange_issue_search_result(self, issue_search_result: dict) -> dict:
+    def _arrange_issue_search_result(self, issue_search_result: dict) -> List[Dict[str, str]]:
         """
-        arrange the result of issue search
+        Arrange the result of Issue search with metadata.
 
         Args:
-            issue_search_result (dict): the result of issue search.
+            _arrange_issue_search_result (dict): The result of Issue search.
 
         Returns:
-            dict: the arranged result of issue search.
+            List[Dict[str, str]]: The arranged result of Issue search with metadata.
         """
         arranged_results = []
 
         for result in issue_search_result:
             content = result['issue_content']
+            url = result.get('url', '')
 
             # Split content into chunks
             chunks = self._split_content_into_chunks(
                 content, max_tokens=config.get('issue_chunk_size'))
 
             for chunk in chunks:
-                arranged_results.append(chunk)
+                arranged_results.append({'context': chunk, 'url': url})
 
         return arranged_results
 
-    def _arrange_repo_search_result(self, repo_search_result: dict) -> dict:
+    def _arrange_repo_search_result(self, repo_search_result: dict) -> List[Dict[str, str]]:
         """
-        arrange the result of repo search
+        Arrange the result of Repo search with metadata.
 
         Args:
-            repo_search_result (dict): the result of repo search.
+            _arrange_repo_search_result (dict): The result of Repo search.
 
         Returns:
-            dict: the arranged result of repo search.
+            List[Dict[str, str]]: The arranged result of Repo search with metadata.
         """
         arranged_results = []
 
         for result in repo_search_result:
             content = result['content']
+            url = result.get('url', '')
 
             # Split content into chunks
             chunks = self._split_content_into_chunks(
                 content, max_tokens=config.get('repo_chunk_size'))
 
             for chunk in chunks:
-                arranged_results.append(chunk)
+                arranged_results.append({'context': chunk, 'url': url})
 
         return arranged_results
 
-    def _arrange_google_search_result(self, google_search_result: dict) -> dict:
+    def _arrange_google_search_result(self, google_search_result: dict) -> List[Dict[str, str]]:
         """
-        arrange the result of google search
+        Arrange the result of Google search with metadata.
 
         Args:
-            google_search_result (dict): the result of google search.
+            google_search_result (dict): The result of Google search.
 
         Returns:
-            dict: the arranged result of google search.
+            List[Dict[str, str]]: The arranged result of Google search with metadata.
         """
         arranged_results = []
 
         for result in google_search_result:
             content = result['content']
+            url = result.get('url', '')  # Extract the URL if available
 
             # Split content into chunks
-            chunks = self._split_content_into_chunks(
-                content, max_tokens=config.get('google_chunk_size'))
+            chunks = self._split_content_into_chunks(content, max_tokens=config.get('google_chunk_size'))
 
             for chunk in chunks:
-                arranged_results.append(chunk)
+                arranged_results.append({'context': chunk, 'url': url})
 
         return arranged_results
 
-    def arrange_context(self, code_search_result: Optional[dict] = None, issue_search_result: Optional[dict] = None, repo_search_result: Optional[dict] = None, google_search_result: Optional[dict] = None) -> dict:
+    def arrange_context(self, code_search_result: Optional[dict] = None, issue_search_result: Optional[dict] = None,
+                        repo_search_result: Optional[dict] = None, google_search_result: Optional[dict] = None) -> List[
+        Dict[str, str]]:
         """
-        arrange the context before RAG
+        Arrange the context before RAG with metadata.
 
         Args:
-            code_search_result (dict, optional): the result of code search. Defaults to None.
-            issue_search_result (dict, optional): the result of issue search. Defaults to None.
-            repo_search_result (dict, optional): the result of repo search. Defaults to None.
-            google_search_result (dict, optional): the result of google search. Defaults to None.
+            code_search_result (dict, optional): The result of code search. Defaults to None.
+            issue_search_result (dict, optional): The result of issue search. Defaults to None.
+            repo_search_result (dict, optional): The result of repo search. Defaults to None.
+            google_search_result (dict, optional): The result of Google search. Defaults to None.
 
         Returns:
-            dict: the arranged context.
+            List[Dict[str, str]]: The arranged context with metadata.
         """
         context = []
         if code_search_result:
-            context.extend(
-                self._arrange_code_search_result(code_search_result))
+            context.extend(self._arrange_code_search_result(code_search_result))
         if issue_search_result:
-            context.extend(
-                self._arrange_issue_search_result(issue_search_result))
+            context.extend(self._arrange_issue_search_result(issue_search_result))
         if repo_search_result:
-            context.extend(
-                self._arrange_repo_search_result(repo_search_result))
+            context.extend(self._arrange_repo_search_result(repo_search_result))
         if google_search_result:
-            context.extend(
-                self._arrange_google_search_result(google_search_result))
+            context.extend(self._arrange_google_search_result(google_search_result))
         return context
 
-    async def retrieve_topn_contexts(self, context_list: List[str], query: str, answer: Optional[str] = None, top_n: Optional[int] = 5) -> List[str]:
+    async def retrieve_topn_contexts(self, context_list: List[Dict[str, str]], query: str, answer: Optional[str] = None,
+                                     top_n: Optional[int] = 5) -> List[Dict[str, str]]:
         """
-        Retrieve top n context strings from the context list.
+        Retrieve top n context dictionaries from the context list.
 
         Args:
-            context_list (List[str]): List of context strings to retrieve top n from.
+            context_list (List[Dict[str, str]]): List of context dictionaries to retrieve top n from.
+                Each dictionary should have at least 'context' and 'url' keys.
             query (str): The query string.
             answer (Optional[str]): The answer string (optional).
             top_n (Optional[int]): Number of top context strings to retrieve (default: 5).
 
         Returns:
-            List[str]: A list of top n context strings.
+            List[Dict[str, str]]: A list of top n context dictionaries.
         """
         top_contexts = []
         try:
             reranker = self.llm_manager.get_rerank_model()
 
-            sentence_pairs = [[query, doc] for doc in context_list]
+            # Extract contexts from the dictionaries
+            contexts = [context_item['context'] for context_item in context_list]
+
+            # Create sentence pairs for reranking
+            sentence_pairs = [[query, doc] for doc in contexts]
             rerank_scores = reranker.compute_score(sentence_pairs)
 
-            # Zip scores with contexts
+            # Zip scores with context dictionaries
             scored_contexts = list(zip(rerank_scores, context_list))
             sorted_scored_contexts = sorted(
                 scored_contexts, key=lambda x: x[0], reverse=True)
 
-            # Extract top 3*top_n contexts after rerank
-            selected_context = [context for score, context in sorted_scored_contexts[:min(
-                top_n*3, len(sorted_scored_contexts))]]
+            # Extract top 3*top_n context dictionaries after rerank
+            selected_contexts = [context for score, context in
+                                 sorted_scored_contexts[:min(top_n * 3, len(sorted_scored_contexts))]]
 
             # If there are too few contexts, skip embedding comparison step
-            if len(selected_context) < top_n*2:
-                return selected_context[:min(top_n, len(selected_context))]
+            if len(selected_contexts) < top_n * 2:
+                return selected_contexts[:min(top_n, len(selected_contexts))]
 
-            # Calculate embedding to select top 2*top_n
+            # Calculate embeddings to select top 2*top_n
             logger.debug("Embedding start...")
             embedding_model = self.llm_manager.get_embedding_model()
-            query_embedding = embedding_model.encode(
-                query + "\n" + answer if answer is not None else "")
-            context_embeddings = [embedding_model.encode(
-                context) for context in selected_context]
+            query_embedding = embedding_model.encode(query + "\n" + answer if answer is not None else "")
+            context_embeddings = [embedding_model.encode(context_item['context']) for context_item in selected_contexts]
 
-            cos_similarities = [(query_embedding @ context_embedding.T) / (norm(query_embedding) * norm(context_embedding))
-                                for context_embedding in context_embeddings]
+            # Calculate cosine similarities
+            cos_similarities = [
+                (query_embedding @ context_embedding.T) / (norm(query_embedding) * norm(context_embedding))
+                for context_embedding in context_embeddings]
 
-            top_indices = np.argsort(cos_similarities)[-(top_n*2):][::-1]
-            top_contexts = [selected_context[i] for i in top_indices]
+            # Get top indices based on cosine similarities
+            top_indices = np.argsort(cos_similarities)[-(top_n * 2):][::-1]
+            top_contexts = [selected_contexts[i] for i in top_indices]
             top_cos_similarities = [cos_similarities[i] for i in top_indices]
-            top_rerank_scores = [rerank_scores[context_list.index(
-                context)] for context in top_contexts]
+            top_rerank_scores = [rerank_scores[contexts.index(context_item['context'])] for context_item in
+                                 top_contexts]
 
-            # Use simple LLM to calculate context score
-            llm_scores = await asyncio.gather(*[self.get_context_relevance_score(query, context) for context in top_contexts])
+            # Use simple LLM to calculate context relevance scores
+            llm_scores = await asyncio.gather(
+                *[self.get_context_relevance_score(query, context_item['context']) for context_item in top_contexts])
             logger.debug(f"Simple LLM scores: {llm_scores}")
 
             # Combine scores for final ranking
             combined_scores = [llm_score * cos_sim * rerank_score
-                               for llm_score, cos_sim, rerank_score in zip(llm_scores, top_cos_similarities, top_rerank_scores)]
+                               for llm_score, cos_sim, rerank_score in
+                               zip(llm_scores, top_cos_similarities, top_rerank_scores)]
 
             combined_context_scores = list(zip(top_contexts, combined_scores))
             sorted_combined_context_scores = sorted(
                 combined_context_scores, key=lambda x: x[1], reverse=True)
-            logger.debug(
-                f"Combined sorted context scores: {sorted_combined_context_scores}")
+            logger.debug(f"Combined sorted context scores: {sorted_combined_context_scores}")
 
-            top_contexts = [context for context,
-                            _ in sorted_combined_context_scores[:top_n]]
+            # Extract top n context dictionaries
+            top_contexts = [context for context, _ in sorted_combined_context_scores[:top_n]]
             logger.debug(f"Final top contexts: {top_contexts}")
+
         except Exception as e:
             logger.error(f"Error retrieving top n context: {e}")
 

llama_github-0.1.1/llama_github/version.py

@@ -0,0 +1 @@
+__version__ = '0.1.1'

llama_github-0.1.0/README.md → llama_github-0.1.1/llama_github.egg-info/PKG-INFO

@@ -1,8 +1,44 @@
+Metadata-Version: 2.1
+Name: llama-github
+Version: 0.1.1
+Summary: Llama-github is an open-source Python library that empowers LLM Chatbots, AI Agents, and Auto-dev Agents to conduct Retrieval from actively selected GitHub public projects. It Augments through LLMs and Generates context for any coding question, in order to streamline the development of sophisticated AI-driven applications.
+Home-page: https://github.com/JetXu-LLM/llama-github
+Author: Jet Xu
+Author-email: Voldemort.xu@foxmail.com
+Classifier: Programming Language :: Python :: 3
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Operating System :: OS Independent
+Requires-Python: >=3.6
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: aiohttp
+Requires-Dist: asyncio
+Requires-Dist: dataclasses
+Requires-Dist: PyGithub
+Requires-Dist: langchain
+Requires-Dist: numpy
+Requires-Dist: pydantic
+Requires-Dist: requests
+Requires-Dist: torch
+Requires-Dist: transformers
+Requires-Dist: urllib3
+Provides-Extra: dev
+Requires-Dist: pytest; extra == "dev"
+Requires-Dist: black; extra == "dev"
+Requires-Dist: flake8; extra == "dev"
+
 # llama-github
 
+[![PyPI version](https://badge.fury.io/py/llama-github.svg)](https://badge.fury.io/py/llama-github)
+[![Downloads](https://static.pepy.tech/badge/Llama-github)](https://pepy.tech/project/Llama-github)
 [![License](https://img.shields.io/badge/License-Apache%202.0-blue.svg)](https://opensource.org/licenses/Apache-2.0)
 
-Llama-github is an open-source Python library that empowers LLM Chatbots, AI Agents, and Auto-dev Agents to conduct Retrieval from actively selected GitHub public contents (Repository, Issue, Code). It Augments through LLMs and Generates context for any coding question, in order to streamline the development of sophisticated AI-driven applications.
+Llama-github is a powerful tool that helps you retrieve the most relevant code snippets, issues, and repository information from GitHub based on your queries, transforming them into valuable knowledge context. It empowers LLM Chatbots, AI Agents, and Auto-dev Agents to solve complex coding tasks. Whether you're a developer looking for quick solutions or an engineer implementing advanced Auto Dev AI Agents, llama-github makes it easy and efficient.
+
+If you like this project or believe it has potential, please give it a ⭐️. Your support is our greatest motivation!
+
+## Architecture
+![High Level Architecture](./docs/high_level_architecture.drawio.svg)
 
 ## Installation
 ```
@@ -53,6 +89,26 @@ For more advanced usage and examples, please refer to the [documentation](docs/u
 
 - **🛠️ Logging and Error Handling**: We understand the importance of smooth operations and easy troubleshooting. That's why llama-github comes equipped with comprehensive logging and error handling mechanisms. Gain deep insights into the library's behavior, quickly diagnose issues, and maintain a stable and reliable development workflow.
 
+## Vision and Roadmap
+
+### Vision
+
+Our vision is to become a pivotal module in the future of AI-driven development solutions, seamlessly integrating with GitHub to empower LLMs in automatically resolving complex coding tasks.
+
+![Vision Architecture](./docs/vision.drawio.svg)
+
+### Roadmap
+
+For a detailed view of our project roadmap, please visit our [Project Roadmap](https://github.com/users/JetXu-LLM/projects/2).
+
+## Acknowledgments
+
+We would like to express our gratitude to the following open-source projects for their support and contributions:
+
+- **[LangChain](https://github.com/langchain-ai/langchain)**: For providing the foundational framework that empowers the LLM prompting and processing capabilities in llama-github.
+- **[Jina.ai](https://github.com/jina-ai/reader)**: For offering s.jina.ai API and open source reranker and embedding models that enhance the accuracy and relevance of the generated contexts in llama-github.
+
+Their contributions have been instrumental in the development of llama-github, and we highly recommend checking out their projects for more innovative solutions.
 
 ## Contributing
 
@@ -64,8 +120,43 @@ This project is licensed under the terms of the Apache 2.0 license. See the [LIC
 
 ## Contact
 
-If you have any questions, suggestions, or feedback, please feel free to reach out to us at [Voldemort.xu@foxmail.com](mailto:Voldemort.xu@foxmail.com).
+If you have any questions, suggestions, or feedback, please feel free to reach out to us at [Jet Xu's email](mailto:Voldemort.xu@foxmail.com).
 
 ---
 
-Thank you for choosing llama-github! We hope this library enhances your AI development experience and helps you build powerful applications with ease.
+Thank you for choosing llama-github! We hope this library enhances your AI development experience and helps you build powerful applications with ease.
+
+# Changelog
+
+All notable changes to this project will be documented in this file.
+
+The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
+and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
+
+## [0.1.1] - 2024-08-23
+
+### Added
+- Implemented `answer_with_context` method for direct answer generation (closes #6)
+- Added support for Mistral AI LLM provider
+- Enhanced `retrieve_context` function to include metadata (e.g., URLs) with each context string (closes #2)
+
+### Changed
+- Improved reranking with jina-reranker-v2 for better context retrieval
+- Updated return type of `retrieve_context` to accommodate metadata
+
+### Fixed
+- Resolved warning during context retrieval (closes #3)
+
+### Improved
+- Enhanced overall context retrieval process
+- Expanded LLM support for more versatile use cases
+
+## [0.1.0] - 2024-08-15
+
+### Added
+- Initial release of llama-github
+- Basic functionality for retrieving context from GitHub repositories
+- Integration with LLM for processing and generating responses
+
+[0.1.1]: https://github.com/JetXu-LLM/llama-github/compare/v0.1.0...v0.1.1
+[0.1.0]: https://github.com/JetXu-LLM/llama-github/releases/tag/v0.1.0

{llama_github-0.1.0 → llama_github-0.1.1}/llama_github.egg-info/SOURCES.txt

@@ -1,3 +1,4 @@
+CHANGELOG.md
 LICENSE
 MANIFEST.in
 README.md

{llama_github-0.1.0 → llama_github-0.1.1}/setup.cfg

@@ -1,10 +1,10 @@
 [metadata]
 name = llama-github
-version = 0.1.0
+version = 0.1.1
 author = Jet Xu
 author_email = Voldemort.xu@foxmail.com
 description = Llama-github is an open-source Python library that empowers LLM Chatbots, AI Agents, and Auto-dev Agents to conduct Retrieval from actively selected GitHub public projects. It Augments through LLMs and Generates context for any coding question, in order to streamline the development of sophisticated AI-driven applications.
-long_description = file: README.md
+long_description = file: README.md, CHANGELOG.md
 long_description_content_type = text/markdown
 url = https://github.com/JetXu-LLM/llama-github
 classifiers = 

{llama_github-0.1.0 → llama_github-0.1.1}/setup.py

@@ -1,9 +1,16 @@
 from setuptools import setup, find_packages
+from configparser import ConfigParser
 
 # Read the requirements from requirements.txt
 with open("requirements.txt", "r", encoding="utf-8") as fh:
     requirements = fh.readlines()
 
+# Read version from setup.cfg
+config = ConfigParser()
+config.read('setup.cfg')
+version = config['metadata']['version']
+
 setup(
+    version=version,
     install_requires=[req.strip() for req in requirements],
 )

llama_github-0.1.0/MANIFEST.in

@@ -1 +0,0 @@
-include llama_github/config/config.json

llama_github-0.1.0/llama_github/version.py

@@ -1 +0,0 @@
-__version__ = '0.1.0'