import { BLOCKED_DOMAINS, BLOCKED_EXTENSIONS, STOP_WORDS } from "./constants";

const NOISE_PATH_PATTERNS = [
    /\/(login|signup|register|cart|checkout|account|profile)\b/i,
    /\/(wp-admin|wp-login|xmlrpc|wp-content|wp-includes)\//i,
    /\/(cgi-bin|admin|administrator)\//i,
];

export function isNoiseUrl(url: string, query: string): boolean {
    try {
        const parsed = new URL(url);
        const host = parsed.hostname.toLowerCase().replace(/^www\./, "");
        if (BLOCKED_DOMAINS.has(host)) return true;
        if (BLOCKED_EXTENSIONS.some(ext => parsed.pathname.toLowerCase().endsWith(ext))) return true;
        if (NOISE_PATH_PATTERNS.some(p => p.test(parsed.pathname))) return true;
        return false;
    } catch {
        return true;
    }
}

export const MIN_RELEVANCE_SCORE = 8;

function buildWordRegex(word: string): RegExp {
    return new RegExp(`\\b${word}\\b`, "i");
}

/**
 * BM25-inspired relevance scoring with word-boundary matching and frequency awareness.
 */
export function pageRelevanceScore(query: string, title: string, content: string): number {
    const queryWords = query.toLowerCase().split(/\s+/).filter(w => w.length > 1 && !STOP_WORDS.has(w));
    if (queryWords.length === 0) return 100;

    const firstWordRegex = buildWordRegex(queryWords[0]);
    let score = 0;

    // First query word scoring: title match = 50, content match = 20
    if (firstWordRegex.test(title)) {
        score += 50;
    } else if (firstWordRegex.test(content)) {
        score += 20;
    }

    // Subsequent words with diminishing returns
    let extraMatched = 0;
    for (let i = 1; i < queryWords.length; i++) {
        const wordRegex = buildWordRegex(queryWords[i]);
        if (!wordRegex.test(title) && !wordRegex.test(content)) continue;

        let points = 0;
        if (wordRegex.test(title)) {
            // Title position bonus: earlier in title = more weight
            const titleMatch = title.match(wordRegex);
            if (titleMatch?.index !== undefined) {
                const posBonus = Math.max(0, 1 - titleMatch.index / 100);
                points = 15 + Math.round(posBonus * 10); // 15-25 range
            } else {
                points = 15;
            }
        } else if (wordRegex.test(content)) {
            const freq = (content.match(wordRegex) || []).length;
            points = 8 + Math.min(freq, 3); // 8-11 range based on frequency
        }

        score += points;
        extraMatched++;
        if (extraMatched >= 4) break; // cap at 4 additional words
    }

    return Math.min(Math.round(score), 100);
}

import { BLOCKED_DOMAINS, BLOCKED_EXTENSIONS, STOP_WORDS } from "./constants";

const NOISE_PATH_PATTERNS = [
    /\/(login|signup|register|cart|checkout|account|profile)\b/i,
    /\/(wp-admin|wp-login|xmlrpc|wp-content|wp-includes)\//i,
    /\/(cgi-bin|admin|administrator)\//i,
];

export function isNoiseUrl(url: string, query: string): boolean {
    try {
        const parsed = new URL(url);
        const host = parsed.hostname.toLowerCase().replace(/^www\./, "");
        if (BLOCKED_DOMAINS.has(host)) return true;
        if (BLOCKED_EXTENSIONS.some(ext => parsed.pathname.toLowerCase().endsWith(ext))) return true;
        if (NOISE_PATH_PATTERNS.some(p => p.test(parsed.pathname))) return true;
        return false;
    } catch {
        return true;
    }
}

export const MIN_RELEVANCE_SCORE = 8;

function buildWordRegex(word: string): RegExp {
    return new RegExp(`\\b${word}\\b`, "i");
}

/**
 * BM25-inspired relevance scoring with word-boundary matching and frequency awareness.
 */
export function pageRelevanceScore(query: string, title: string, content: string): number {
    const queryWords = query.toLowerCase().split(/\s+/).filter(w => w.length > 1 && !STOP_WORDS.has(w));
    if (queryWords.length === 0) return 100;

    const firstWordRegex = buildWordRegex(queryWords[0]);
    let score = 0;

    // First query word scoring: title match = 50, content match = 20
    if (firstWordRegex.test(title)) {
        score += 50;
    } else if (firstWordRegex.test(content)) {
        score += 20;
    }

    // Subsequent words with diminishing returns
    let extraMatched = 0;
    for (let i = 1; i < queryWords.length; i++) {
        const wordRegex = buildWordRegex(queryWords[i]);
        if (!wordRegex.test(title) && !wordRegex.test(content)) continue;

        let points = 0;
        if (wordRegex.test(title)) {
            // Title position bonus: earlier in title = more weight
            const titleMatch = title.match(wordRegex);
            if (titleMatch?.index !== undefined) {
                const posBonus = Math.max(0, 1 - titleMatch.index / 100);
                points = 15 + Math.round(posBonus * 10); // 15-25 range
            } else {
                points = 15;
            }
        } else if (wordRegex.test(content)) {
            const freq = (content.match(wordRegex) || []).length;
            points = 8 + Math.min(freq, 3); // 8-11 range based on frequency
        }

        score += points;
        extraMatched++;
        if (extraMatched >= 4) break; // cap at 4 additional words
    }

    return Math.min(Math.round(score), 100);
}