const axios = require('axios');
const fs = require('fs');
const cheerio = require('cheerio');
const stringSimilarity = require('string-similarity');
const { Client, Databases, Storage, ID, Query } = require('node-appwrite');
require('dotenv').config({ path: '.env.local' });

let InputFile;
try { InputFile = require('node-appwrite/file').InputFile; } catch (e) { InputFile = null; }

const client = new Client()
    .setEndpoint(process.env.NEXT_PUBLIC_APPWRITE_ENDPOINT)
    .setProject(process.env.NEXT_PUBLIC_APPWRITE_PROJECT_ID)
    .setKey(process.env.APPWRITE_API_KEY);

const databases = new Databases(client);
const storage = new Storage(client);
const DB_ID = process.env.APPWRITE_DATABASE_ID;
const BUCKET_ID = process.env.NEXT_PUBLIC_APPWRITE_MEDIA_BUCKET_ID;

const excludedNodes = ['962', '963', '895', '894', '893', '964'];

async function deleteOldScrapes(postsCollectionId) {
    console.log('Deleting all poorly scraped multivision posts...');
    let cursor = null;
    let deletedCount = 0;
    while (true) {
        const queries = [Query.limit(100)];
        if (cursor) queries.push(Query.cursorAfter(cursor));
        const res = await databases.listDocuments(DB_ID, postsCollectionId, queries);
        const toDelete = res.documents.filter(p => p.slug && (p.slug.startsWith('multivision-amj-') || p.slug.startsWith('multivision-final-') || p.slug.startsWith('multivision-marzo-')));
        for (const d of toDelete) {
            await databases.deleteDocument(DB_ID, postsCollectionId, d.$id);
            deletedCount++;
        }
        if (res.documents.length < 100) break;
        cursor = res.documents[res.documents.length - 1].$id;
    }
    console.log(`Deleted ${deletedCount} old posts.`);
}

async function scrapeNode(url) {
    try {
        const res = await axios.get(url);
        const $ = cheerio.load(res.data);

        const title = $('h1').first().text().replace(/\|.*/, '').trim() || $('title').text().replace(/\|.*/, '').trim();

        // Extract date
        let dateObj = null;
        const timeText = $('span').filter((i, el) => $(el).text().match(/\d{2}\/\d{2}\/\d{4} - \d{2}:\d{2}/)).first().text();
        if (timeText) {
            const match = timeText.match(/(\d{2})\/(\d{2})\/(\d{4}) - (\d{2}:\d{2})/);
            if (match) {
                dateObj = new Date(`${match[3]}-${match[2]}-${match[1]}T${match[4]}:00.000Z`);
            }
        }

        if (!dateObj) return null;

        // We only want March (2), April (3), May (4), June (5) 2026
        const validMonths = [2, 3, 4, 5];
        if (!validMonths.includes(dateObj.getMonth()) || dateObj.getFullYear() !== 2026) {
            return { skipped: true, reason: 'Not in Mar, Apr, May, or Jun 2026' };
        }

        // Extract image
        let image = null;
        $('img').each((i, el) => {
            const src = $(el).attr('src');
            // Changed filter to match Multivision's actual image paths
            if (src && src.includes('sites/default/files') && !src.includes('logo')) {
                image = src.startsWith('http') ? src : 'https://www.multivisiondecolombia.com' + src;
                return false; 
            }
        });

        // Extract content
        let contentHtml = '';
        let maxLen = 0;
        
        // Find the block with the most text
        $('.field--name-body, .field--name-field-resumen-de-la-noticia').each((i, el) => {
            const text = $(el).text().trim();
            if (text.length > maxLen) {
                maxLen = text.length;
                contentHtml = $(el).html();
            }
        });
        
        if (!contentHtml || maxLen < 100) {
            // Fallback to finding the longest div if both fail
            maxLen = 0;
            $('div').each((i, el) => {
                const text = $(el).text().trim();
                if (text.length > maxLen && $(el).children().length < 5) {
                    maxLen = text.length;
                    contentHtml = $(el).html();
                }
            });
        }

        // Clean content just a bit
        let cleanContent = (contentHtml || '<p></p>').trim();
        
        // Auto-generate fields
        const plainText = cheerio.load(cleanContent).text().replace(/\s+/g, ' ').trim();
        let resumen = plainText.substring(0, 150);
        if (plainText.length > 150) resumen += '...';
        
        // Keywords
        const words = plainText.split(' ').filter(w => w.length > 5).slice(0, 8);
        const keywords = words.join(', ').replace(/[^a-zA-ZáéíóúÁÉÍÓÚñÑ, ]/g, '');

        return { 
            title, 
            image, 
            date: dateObj.toISOString(), 
            content: cleanContent,
            resumen,
            keywords
        };
    } catch (e) {
        console.error('Error scraping node', url, e.message);
        return null;
    }
}

async function run() {
    try {
        const collections = await databases.listCollections(DB_ID);
        const map = {};
        collections.collections.forEach(c => { map[c.name.toLowerCase()] = c.$id; });
        const postsCollectionId = map['posts'] || 'posts';
        const mediaCollectionId = map['media'] || 'media';
        const pubTypesCollectionId = map['publication types'] || map['publication_types'] || 'publication_types';

        const typeRes = await databases.listDocuments(DB_ID, pubTypesCollectionId, [Query.equal('codigo', 'NOTICIA')]);
        const typeId = typeRes.documents.length > 0 ? typeRes.documents[0].$id : null;

        await deleteOldScrapes(postsCollectionId);

        console.log('Fetching existing titles from Appwrite to prevent duplicates...');
        let allExistingTitles = [];
        let cursor = null;
        while (true) {
            const queries = [Query.limit(100), Query.select(['titulo', 'slug'])];
            if (cursor) queries.push(Query.cursorAfter(cursor));
            const existingRes = await databases.listDocuments(DB_ID, postsCollectionId, queries);
            
            existingRes.documents.forEach(d => {
                // Don't add multivision ones since we just deleted them or are replacing them
                if (!d.slug || !d.slug.startsWith('multivision-')) {
                    allExistingTitles.push(d.titulo);
                }
            });
            
            if (existingRes.documents.length < 100) break;
            cursor = existingRes.documents[existingRes.documents.length - 1].$id;
        }
        console.log(`Found ${allExistingTitles.length} existing native posts.`);

        console.log('Fetching news list from Multivision (Pages 0-25)...');
        const allNodes = new Set();
        
        for (let page = 0; page <= 25; page++) {
            try {
                const listUrl = `https://www.multivisiondecolombia.com/noticias?page=${page}`;
                const res = await axios.get(listUrl);
                const $ = cheerio.load(res.data);
                
                $('a').each((i, el) => {
                    const href = $(el).attr('href');
                    if (href && href.startsWith('/node/')) {
                        const nodeId = href.split('/')[2];
                        if (nodeId && !excludedNodes.includes(nodeId)) {
                            allNodes.add(`https://www.multivisiondecolombia.com/node/${nodeId}`);
                        }
                    }
                });
            } catch(e) {
                console.log(`Failed to fetch page ${page}`);
            }
        }

        console.log(`Found ${allNodes.size} valid nodes to process.`);
        
        let count = 0;
        for (const url of allNodes) {
            console.log(`\nProcessing: ${url}`);
            const data = await scrapeNode(url);
            
            if (!data) {
                console.log('No data extracted.');
                continue;
            }
            if (data.skipped) {
                console.log('Skipped:', data.reason);
                continue;
            }

            // Check similarity
            let highestSimilarity = 0;
            let mostSimilarTitle = '';

            for (const exTitle of allExistingTitles) {
                if (!exTitle) continue;
                const sim = stringSimilarity.compareTwoStrings(data.title.toLowerCase(), exTitle.toLowerCase());
                if (sim > highestSimilarity) {
                    highestSimilarity = sim;
                    mostSimilarTitle = exTitle;
                }
            }

            if (highestSimilarity > 0.60) {
                console.log(`Skipped due to similarity (${(highestSimilarity*100).toFixed(1)}%) with: "${mostSimilarTitle}"`);
                continue;
            }

            console.log('Extracted Title:', data.title);
            console.log('Extracted Date:', data.date);

            let mediaId = null;
            if (data.image) {
                try {
                    const imgRes = await axios.get(data.image, { responseType: 'arraybuffer' });
                    const buffer = Buffer.from(imgRes.data, 'binary');
                    const filePath = `/tmp/multivision_${count}.jpg`;
                    fs.writeFileSync(filePath, buffer);

                    let fileParam;
                    if (InputFile && InputFile.fromPath) fileParam = InputFile.fromPath(filePath, `image${count}.jpg`);
                    else fileParam = fs.createReadStream(filePath);

                    const file = await storage.createFile(BUCKET_ID, ID.unique(), fileParam);
                    const mediaDoc = await databases.createDocument(DB_ID, mediaCollectionId, ID.unique(), {
                        file_id: file.$id, bucket_id: BUCKET_ID,
                        titulo: data.title, tipo: 'imagen'
                    });
                    mediaId = mediaDoc.$id;
                } catch(err) {
                    console.error('Error uploading image:', err.message);
                }
            }

            const doc = {
                titulo: data.title.substring(0, 100),
                slug: `multivision-final-${Date.now()}-${count}`,
                tipo_publicacion_id: typeId,
                estado: 'Publicado',
                fecha_publicacion: data.date,
                contenido: data.content,
                resumen_redes: data.resumen,
                seo_descripcion: data.resumen,
                seo_keywords: data.keywords,
                imagen_destacada_id: mediaId
            };

            const res = await databases.createDocument(DB_ID, postsCollectionId, ID.unique(), doc);
            console.log('Published! ID:', res.$id);
            // Add to existing to prevent duplicates within this run
            allExistingTitles.push(data.title);
            count++;
        }
        console.log(`\nAll done! Successfully published ${count} perfect news articles.`);
    } catch(e) {
        console.error(e);
    }
}
run();