source_reputation.py

import json
import os
import re

class ReputationEngine:
    def __init__(self, db_path=None):
        if not db_path:
            db_path = os.path.join(os.path.dirname(__file__), 'data', 'source_reputation.json')
        
        self.db_path = db_path
        self.reputation_data = {}
        self.load_db()

    def load_db(self):
        if os.path.exists(self.db_path):
            with open(self.db_path, 'r') as f:
                self.reputation_data = json.load(f)
        else:
            print(f"Warning: Reputation database not found at {self.db_path}")

    def get_source_stats(self, identifier):
        """
        identifier can be a speaker name (e.g. 'john-mccain') 
        or a domain (e.g. 'reuters.com').
        """
        # Clean identifier
        slug = identifier.lower().strip().replace(' ', '-')
        
        # Try direct match
        if slug in self.reputation_data:
            return self.reputation_data[slug]
        
        # Try regex search for domains in URLs/metadata
        for key in self.reputation_data:
            if key in slug or slug in key:
                return self.reputation_data[key]
        
        return None

    def analyze_source(self, text, speaker=None, context=None):
        """
        Returns a dictionary with trust_score, bias, and label.
        """
        target = speaker or "unknown"
        stats = self.get_source_stats(target)
        
        # Fallback: check context for domain-like strings
        if not stats and context:
            domains = re.findall(r'([a-z0-9]+(?:-[a-z0-9]+)*\.[a-z]{2,})', context.lower())
            for d in domains:
                stats = self.get_source_stats(d)
                if stats: break

        if stats:
            score = stats.get('trust_score', 0)
            party = stats.get('metadata', {}).get('party', 'unknown').lower()
            
            # Simple bias mapping
            bias = "Center"
            if 'republican' in party: bias = "Right-Leaning"
            elif 'democrat' in party: bias = "Left-Leaning"
            elif 'libertarian' in party: bias = "Libertarian"
            
            # Labeling
            label = "Unknown"
            if score >= 90: label = "EXCEPTIONAL"
            elif score >= 75: label = "HIGHLY RELIABLE"
            elif score >= 50: label = "MIXED RECORD"
            elif score >= 30: label = "QUESTIONABLE"
            else: label = "LOW CREIBILITY"

            return {
                "source_name": stats.get('name', target),
                "trust_score": score,
                "bias": bias,
                "veracity_label": label,
                "total_history": stats.get('total_claims', 0),
                "counts": stats.get('counts', {})
            }
        
        return {
            "source_name": target,
            "trust_score": 0,
            "bias": "Undetermined",
            "veracity_label": "NO DATA",
            "total_history": 0,
            "counts": {}
        }