backstory/src/rag_cli.py

# cli.py - Command-line interface
import sys

class RagCLI:
    """Command-line interface for the RAG application"""
    
    def __init__(self, core):
        self.core = core
        
    def show_help(self):
        """Display available commands"""
        print("""help>
docs       Show RAG docs
full       Show last full top match
headlines  Show the RAG headlines
prompt     Show the last prompt
response   Show the last response
scores     Show last RAG scores
why|think  Show last response's <think>
context|match  Show RAG match info to last prompt
exit|quit  Exit the application
""")

    def print_document_summary(self, documents):
        """Print brief summary of all documents"""
        if not documents:
            print("No documents available.")
            return
        
        for i, doc in enumerate(documents):
            print(f"Document {i+1}:")
            print(f"  Title: {doc['title']}")
            print(f"  Text: {doc['text'][:100]}...")
            print()

    def print_headlines(self, headlines):
        """Print headlines"""
        if not headlines:
            print("No headlines available.")
            return
        
        for source, title in headlines:
            print(f"{source}: {title}")

    def print_match_info(self, match_info):
        """Print detailed information about a matching document"""
        if not match_info:
            print("No matching document found.")
            return
        
        print("="*50)
        print("MATCHING DOCUMENT")
        print("="*50)
        print(f"Chunk ID: {match_info['id']}")
        print(f"Similarity Score: {match_info['distance']:.4f}")
        
        print("\nCHUNK METADATA:")
        for key, value in match_info['metadata'].items():
            print(f"  {key}: {value}")
        
        print("\nMATCHING CHUNK CONTENT:")
        print(match_info['chunk'][:500].strip() + ("..." if len(match_info['chunk']) > 500 else ""))

    def print_full_document(self, doc, distance=None):
        """Print full document details"""
        if not doc:
            print("No document found.")
            return
            
        print(f"""Context:
Source: {doc["source"]}
Title: {doc["title"]}
Link: {doc["link"]}
{"Distance: " + str(distance) if distance is not None else ""}
Full text:
{doc["text"]}""")

    def run(self):
        """Run the command-line interface"""
        # Load documents
        doc_count = self.core.load_documents()
        print(f"Loaded {doc_count} documents.")
        
        while True:
            try:
                query = input("> ").strip()
            except KeyboardInterrupt:
                print("\nExiting.")
                break

            if query in ["exit", "quit"]:
                print("\nExiting.")
                break
                
            elif query == "docs":
                self.print_document_summary(self.core.get_documents())
                
            elif query == "prompt":
                last_prompt = self.core.get_last_prompt()
                last_system = self.core.get_last_system()
                if last_prompt:
                    print(f"""last prompt>
{"="*10}system{"="*10}
{last_system}
{"="*10}prompt{"="*10}
{last_prompt}""")
                else:
                    print("No prompts yet")
                    
            elif query == "response":
                last_response = self.core.get_last_response()
                if last_response:
                    print(f"""last response>
{"="*10}response{"="*10}
{last_response}""")
                else:
                    print("No responses yet")
                    
            elif query in ["", "help"]:
                self.show_help()
                
            elif query == "headlines":
                self.print_headlines(self.core.get_headlines())
                
            elif query in ["match", "context"]:
                match_info = self.core.get_match_info()
                self.print_match_info(match_info)
                    
            elif query in ["why", "think"]:
                last_thinking = self.core.get_last_thinking()
                if last_thinking:
                    print(f"""
why>
{last_thinking}
""")
                else:
                    print("No processed prompts")
                    
            elif query == "scores":
                match_count = self.core.get_match_count()
                if match_count > 0:
                    for i in range(match_count):
                        match_info = self.core.get_match_info(index=i)
                        self.print_match_info(match_info)
                else:
                    print("No matches to display")
                    
            elif query == "full":
                doc = self.core.get_top_match()
                distance = None
                if self.core.last_results and self.core.last_results.get("distances"):
                    distance = self.core.last_results["distances"][0][0]
                self.print_full_document(doc, distance)
                    
            else:
                # Process the user query
                response, debug_info = self.core.process_query(query)
                print(f"system len: {debug_info['system_len']}")            
                print(f"prompt len: {debug_info['prompt_len']}")
                print(f"Response>\n{response}")