Implementation:PacktPublishing LLM Engineers Handbook SelfQuery Generate

Field	Value
Type	API Doc
Workflow	RAG_Inference
Repository	PacktPublishing/LLM-Engineers-Handbook
Source	self_query.py:L15-37
Implements	Principle:PacktPublishing_LLM_Engineers_Handbook_Self_Query_Metadata_Extraction

API Signature

SelfQuery.generate(self, query: Query) -> Query

Import

from llm_engineering.application.rag.self_query import SelfQuery

Key Code

class SelfQuery(RAGStep):
    @opik.track(name="SelfQuery.generate")
    def generate(self, query: Query) -> Query:
        prompt = SelfQueryTemplate().create_template()
        chain = prompt | self._llm
        response = chain.invoke({"question": query.content})
        # Extracts author_full_name from LLM response
        # Looks up author_id from UserDocument
        query.author_full_name = author_full_name
        query.author_id = author_id
        return query

Parameters

Parameter	Type	Description
query	Query	The raw user query object containing the natural language question

Inputs and Outputs

Inputs:

• query: Query - The raw user query with the content field containing the natural language question

Outputs:

• Query - The same query object enriched with extracted metadata:
  • author_full_name - The full name of the author extracted from the query
  • author_id - The corresponding author ID looked up from UserDocument

How It Works

1. A SelfQueryTemplate generates a prompt that instructs the LLM to identify and extract the author name from the user's query
2. The prompt is chained with the LLM using LangChain's pipe operator (prompt | self._llm)
3. The LLM response is parsed to extract the author_full_name
4. The author name is looked up against UserDocument records to obtain the corresponding author_id
5. Both fields are set on the query object, which is returned for use in downstream filtering

External Dependencies

• langchain_openai (ChatOpenAI) - LLM used for metadata extraction
• opik - Observability and tracing decorator for monitoring
• loguru - Structured logging

Source File

• llm_engineering/application/rag/self_query.py (lines 15-37)

See Also

• Principle:PacktPublishing_LLM_Engineers_Handbook_Self_Query_Metadata_Extraction
• Environment:PacktPublishing_LLM_Engineers_Handbook_API_Credentials
• Heuristic:PacktPublishing_LLM_Engineers_Handbook_RAG_Retrieval_Parameters
• Heuristic:PacktPublishing_LLM_Engineers_Handbook_Temperature_Selection_By_Task