rag-solution/services/rag/langchain/PLANNING.md

Requirements

Libraries should be installed into the local virtual environment, which is defined in the venv folder. If some libraries are not installed, check online which are best and install them. Use if possible logging, using library loguru, for steps. Use logrotation in file logs/dev.log, also log to stdout

Chosen RAG framework: Langchain Chosen Vector Storage: Qdrant Chosen data folder: relatve ./../../../data - from the current folder

Phase 1 (cli entrypoint)

• Create virtual env in the venv folder in the current directory.
• Create cli.py file, with the usage of click python library. Make default command "ping" which will write output "pong"

Phase 2 (installation of base framework for RAG solution and preparation for data loading)

• Install langchain as base framework for RAG solution
• Analyze the upper data folder (./../../../data), to learn all the possible files extensions of files there. Then, create file in the current directory EXTENSIONS.md with the list of extensions - and loader/loaders for chosen framework (this can be learned online - search for the info), that is needed to load the data in the provided extension. Prioriize libraries that work without external service that require API keys or paid subscriptions. Important: skip stream media files extensions (audio, video). We are not going to load them now.
• Install all needed libraries for loaders, mentioned in the EXTENSIONS.md. If some libraries require API keys for external services, add them to the .env file (create it if it does not exist)

Phase 3 (preparation for storing data in the vector storage + embeddings)

• Install needed library for using Qdrant connection as vector storage. Ensure ports are used (which are needed in the chosen framework): Rest Api: 6333, gRPC Api: 6334. Database available and running on localhost.
• Create file called vector_storage.py, which will contain vector storage initialization, available for import by other modules of initialized. If needed in chosen RAG framework, add embedding model initialization in the same file. Use ollama, model name defined in the .env file: OLLAMA_EMBEDDING_MODEL. Ollama available by the default local port: 11434
• Just in case add possibility to connect via openai embedding, using openrouter api key. Comment this section, so it can be used in the future.

Phase 4 (creating module for loading documents from the folder)

• Create file enrichment.py with the function that will load data with configured data loaders for extensions from the data folder into the chosen vector storage. Remember to specify default embeddings meta properties, such as filename, paragraph, page, section, wherever this is possible (documents can have pages, sections, paragraphs, etc). Use text splitters of the chosen RAG framework accordingly to the documents being loaded. Which chunking/text-splitting strategies framework has, can be learned online.
• Use built-in strategy for marking which documents loaded (if there is such mechanism) and which are not, to avoid re-reading and re-encriching vector storage with the existing data. If there is no built-in mechanism of this type, install sqlite library and use local sqlite database file to store this information.
• Add activation of this function in the cli entrypoint, as a command.

Phase 5 (preparation for the retrieval feature)

• Create file retrieval.py with the configuration for chosen RAG framework, that will retrieve data from the vector storage based on the query. Use retrieving library/plugin, that supports chosen vector storage within the chosen RAG framework. Retrieving configuration should search for the provided text in the query as argument in the function and return found information with the stored meta data, like paragraph, section, page etc. Important: if for chosen RAG framework, there is no need in separation of search, separation of retrieving from the chosen vector storage, this step may be skipped and marked done.

Phase 6 (chat feature, as agent, for usage in the cli)

• Create file agent.py, which will incorporate into itself agent, powered by the chat model. It should use integration with ollama, model specified in .env in property: OLLAMA_CHAT_MODEL
• Integrate this agent with the existing solution for retrieving, with retrieval.py
• Integrate this agent with the cli, as command to start chatting with the agent. If there is a built-in solution for console communication with the agent, initiate this on cli command.

Phase 7 (openai integration for chat model)

• Create openai integration, with using .env variables OPENAI_CHAT_URL, OPENAI_CHAT_KEY. First one for openai compatible URL, second one for Authorization Bearer token.
• Make this integration optional, with using .env variable CHAT_MODEL_STRATEGY. There can be 2 options: "ollama", "openai". Ollama currently already done and working, so we should write code for checking which option is chosen in .env, with ollama being the default.

Phase 8 (http endpoint to retrieve data from the vector storage by query)

• Create file server.py, with web framework fastapi, for example
• Add POST endpoint "/api/test-query" which will use agent, and retrieve response for query, sent in JSON format, field "query"

Phase 9 (simple html web page with chat interface)

• Create html webpage called demo.html, with simple UI for chat interface. It can be taken with predefined data from codepen or something
• Adjust demo.html code, so it would in fact work with the API endpoint, as chat with the agent. API endpoint should be asked beforehand in propmt message.
• After accepting API endpont address, it should be used to send requests and process responses to imitate chat with the agent by the provided API endpoint.
• Show API endpoint in the header of the chat.
• If there is error connecting with the API, imitate bot sending message about error with the connection and suggestion to reload page to provide new API endpoint

Phase 10 (extracting additional metadata from chunks, and filtering where possible with it)

• Create separate function in helpers module (create if does not exist) for retrieving years from the text. It should return found years.
• During enriching vector storage, when loading and splitting documents, extract years from the chunk, and add these years as numbers into metadata field "years" (array of number or best suitable Qdrant type for searching by the year if needed). The helper function for retrieving years from text can be used.
• Updating VectorStoreRetriever._get_relevant_documents: We need to ensure, that when searching for something with the year (user mentiones year in the query, in Russian language), we search vectors with metadata which has these mentioned year in the "years" array of years. The helper function for retrieving years from query can be used to filter out documents with years.
• Create heuristic, regex function in helpers module for extracting name of event, in Russian language. We need to use regex and possible words before, after the event, etc.
• Durint enriching vector storage, try to extract event name from the chunk and save in metadata in field "events", which will contain list of strings, possible evennts. Helper function usage is advised.
• In VectorStoreRetriever._get_relevant_documents add similarity search for the event name, if event name is present in the query. Helper function should be used here to try to extract the event name.