rss/app.py-backup

from flask import Flask, render_template, request, redirect
from apscheduler.schedulers.background import BackgroundScheduler
from datetime import datetime
import feedparser
import hashlib
import re
import mysql.connector

app = Flask(__name__)

DB_CONFIG = {
    'host': 'localhost',
    'user': 'x',
    'password': 'x',
    'database': 'noticiasrss'
}

@app.route('/')
def index():
    conn = None
    try:
        conn = mysql.connector.connect(**DB_CONFIG)
        cursor = conn.cursor()
        cursor.execute("SELECT id, nombre, url, categoria_id, activo FROM feeds")
        feeds = cursor.fetchall()
        cursor.execute("SELECT id, nombre FROM categorias_estandar ORDER BY nombre")
        categorias = cursor.fetchall()
    except mysql.connector.Error as db_err:
        app.logger.error(f"[DB ERROR] Al leer feeds o categorías: {db_err}", exc_info=True)
        feeds = []
        categorias = []
    finally:
        if conn:
            conn.close()
    return render_template("index.html", feeds=feeds, categorias=categorias)

@app.route('/add', methods=['POST'])
def add_feed():
    nombre = request.form.get('nombre')
    url = request.form.get('url')
    categoria_id = request.form.get('categoria_id')
    try:
        conn = mysql.connector.connect(**DB_CONFIG)
        cursor = conn.cursor()
        cursor.execute(
            "INSERT INTO feeds (nombre, url, categoria_id) VALUES (%s, %s, %s)",
            (nombre, url, categoria_id)
        )
        conn.commit()
    except mysql.connector.Error as db_err:
        app.logger.error(f"[DB ERROR] Al agregar feed: {db_err}", exc_info=True)
    finally:
        if conn:
            conn.close()
    return redirect('/')

@app.route('/noticias')
def show_noticias():
    conn = None
    noticias = []
    categorias = []
    cat_id = request.args.get('categoria_id')
    try:
        conn = mysql.connector.connect(**DB_CONFIG)
        cursor = conn.cursor()
        cursor.execute("SELECT id, nombre FROM categorias_estandar ORDER BY nombre")
        categorias = cursor.fetchall()

        sql = """
            SELECT n.fecha, n.titulo, n.resumen, n.url, n.imagen_url, c.nombre
              FROM noticias n
              LEFT JOIN categorias_estandar c ON n.categoria_id = c.id
        """
        params = []
        if cat_id:
            sql += " WHERE n.categoria_id = %s"
            params.append(cat_id)
        sql += " ORDER BY n.fecha DESC LIMIT 50"
        cursor.execute(sql, params)
        noticias = cursor.fetchall()
    except mysql.connector.Error as db_err:
        app.logger.error(f"[DB ERROR] Al leer noticias: {db_err}", exc_info=True)
    finally:
        if conn:
            conn.close()
    return render_template('noticias.html', noticias=noticias, categorias=categorias, cat_id=int(cat_id) if cat_id else None)

def fetch_and_store():
    conn = None
    try:
        conn = mysql.connector.connect(**DB_CONFIG)
        cursor = conn.cursor()
        cursor.execute("SELECT url, categoria_id FROM feeds WHERE activo = TRUE")
        feeds = cursor.fetchall()
    except mysql.connector.Error as db_err:
        app.logger.error(f"[DB ERROR] No se pudo conectar o leer feeds: {db_err}", exc_info=True)
        return

    for rss_url, categoria_id in feeds:
        try:
            app.logger.info(f"Procesando feed: {rss_url} [{categoria_id}]")
            parsed = feedparser.parse(rss_url)
        except Exception as e:
            app.logger.error(f"[PARSE ERROR] Al parsear {rss_url}: {e}", exc_info=True)
            continue

        if getattr(parsed, 'bozo', False):
            bozo_exc = getattr(parsed, 'bozo_exception', 'Unknown')
            app.logger.warning(f"[BOZO] Feed mal formado: {rss_url} - {bozo_exc}")
            continue

        for entry in parsed.entries:
            link = entry.get('link') or entry.get('id')
            if not link:
                links_list = entry.get('links', [])
                if isinstance(links_list, list) and links_list:
                    href = next((l.get('href') for l in links_list if l.get('href')), None)
                    link = href
            if not link:
                app.logger.error(f"[ENTRY ERROR] Entrada sin link en feed {rss_url}, salto entrada.")
                continue

            try:
                noticia_id = hashlib.md5(link.encode()).hexdigest()
                titulo = entry.get('title', '')
                resumen = entry.get('summary', '')
                imagen_url = ''
                fecha = None

                if 'media_content' in entry:
                    imagen_url = entry.media_content[0].get('url', '')
                else:
                    img = re.search(r'<img.+?src="(.+?)"', resumen)
                    if img:
                        imagen_url = img.group(1)

                published = entry.get('published_parsed') or entry.get('updated_parsed')
                if published:
                    try:
                        fecha = datetime(*published[:6])
                    except Exception:
                        fecha = None

                cursor.execute(
                    """
                    INSERT IGNORE INTO noticias
                        (id, titulo, resumen, url, fecha, imagen_url, categoria_id)
                    VALUES (%s, %s, %s, %s, %s, %s, %s)
                    """,
                    (noticia_id, titulo, resumen, link, fecha, imagen_url, categoria_id)
                )
            except Exception as entry_err:
                app.logger.error(
                    f"[ENTRY ERROR] Falló procesar entrada {link}: {entry_err}", exc_info=True
                )
                continue

    try:
        conn.commit()
    except Exception as commit_err:
        app.logger.error(f"[DB ERROR] Al confirmar transacción: {commit_err}", exc_info=True)
    finally:
        if conn:
            conn.close()
    app.logger.info(f"[{datetime.now().strftime('%Y-%m-%d %H:%M:%S')}] Feeds procesados.")

if __name__ == '__main__':
    scheduler = BackgroundScheduler()
    scheduler.add_job(fetch_and_store, 'interval', minutes=2, id='rss_job')
    scheduler.start()
    try:
        app.run(host="0.0.0.0", port=5000, debug=True)
    finally:
        scheduler.shutdown()