portales-hispanohablantes/inmuebles/elempleo_co/backfill.py

#!/usr/bin/env python3
"""
Backfill scraper ElEmpleo.com Colombia
Checkpoint: /opt/scrapers/elempleo_co/backfill_state.json
  {"page": N, "total_new": M}   -> reanudar desde pagina N
Para reiniciar desde cero: borrar el archivo de checkpoint.
"""
import os, re, time, logging, json
import requests
import psycopg2
from psycopg2.extras import execute_values
from datetime import datetime

logging.basicConfig(
    level=logging.INFO,
    format='%(asctime)s %(levelname)s %(message)s',
    datefmt='%Y-%m-%d %H:%M:%S',
)
log = logging.getLogger('elempleo_backfill')

PROXY      = 'socks5h://127.0.0.1:1090'
PROXIES    = {'http': PROXY, 'https': PROXY}
DB_HOST    = '100.75.240.87'
DB_NAME    = 'elempleo_co'
DB_USER    = 'pgadmin'
DB_PASS    = 'J5BVlq65JvedWxZVrcY96OQX'
UA         = 'Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 Chrome/124 Safari/537.36'
BASE       = 'https://www.elempleo.com'
LIST_URL   = BASE + '/co/ofertas-empleo/?pagina={}'
DELAY      = 0.8
MAX_PAGES  = int(os.environ.get('BF_MAX_PAGES', '500'))
MAX_STREAK = int(os.environ.get('BF_STREAK', '8'))

STATE_FILE = os.path.join(os.path.dirname(os.path.abspath(__file__)), 'backfill_state.json')


def load_state():
    if os.path.exists(STATE_FILE):
        with open(STATE_FILE) as f:
            st = json.load(f)
        log.info('Checkpoint cargado: pagina=%d total_new=%d', st['page'], st.get('total_new', 0))
        return st
    return {'page': 1, 'total_new': 0}


def save_state(page, total_new):
    with open(STATE_FILE, 'w') as f:
        json.dump({'page': page, 'total_new': total_new, 'ts': datetime.utcnow().isoformat()}, f)


def get_session():
    s = requests.Session()
    s.proxies.update(PROXIES)
    s.headers.update({'User-Agent': UA, 'Accept': 'text/html,application/xhtml+xml,*/*;q=0.8',
                      'Accept-Language': 'es-CO,es;q=0.9'})
    return s


def get_conn():
    return psycopg2.connect(host=DB_HOST, port=5432, dbname=DB_NAME,
                            user=DB_USER, password=DB_PASS, connect_timeout=10)


def create_table(conn):
    cur = conn.cursor()
    cur.execute("""
        CREATE TABLE IF NOT EXISTS jobs (
            job_id     TEXT PRIMARY KEY,
            titulo     TEXT,
            empresa    TEXT,
            ubicacion  TEXT,
            url        TEXT UNIQUE,
            fecha_pub  TEXT,
            categoria  TEXT,
            scraped_at TIMESTAMP DEFAULT NOW()
        )
    """)
    conn.commit()
    cur.close()


def upsert(conn, rows):
    if not rows:
        return 0
    deduped = list({r['job_id']: r for r in rows}.values())
    cur = conn.cursor()
    execute_values(cur,
        """INSERT INTO jobs (job_id, titulo, empresa, ubicacion, url, scraped_at)
           VALUES %s
           ON CONFLICT (job_id) DO UPDATE SET
             titulo=EXCLUDED.titulo, scraped_at=EXCLUDED.scraped_at""",
        [(r['job_id'], r['titulo'], r['empresa'], r['ubicacion'],
          r['url'], datetime.utcnow()) for r in deduped]
    )
    n = cur.rowcount
    conn.commit()
    cur.close()
    return n


def fetch_page(sess, page):
    url = LIST_URL.format(page)
    for attempt in range(3):
        try:
            resp = sess.get(url, timeout=30)
            if resp.status_code == 404:
                return None
            if resp.status_code in (403, 429):
                log.warning('HTTP %d en %s', resp.status_code, url)
                time.sleep(60)
                continue
            resp.raise_for_status()
            break
        except Exception as e:
            if attempt == 2:
                log.error('Error %s: %s', url, e)
                return []
            time.sleep([2, 5, 15][attempt])

    html = resp.text
    links = list(dict.fromkeys(re.findall(r'href="(/co/ofertas-trabajo/[^"]{10,150})"', html)))
    if not links:
        return []

    jobs = []
    for href in links:
        m = re.search(r'-(\d{7,})$', href)
        if not m:
            continue
        job_id = m.group(1)
        slug = href.split('/')[-1]
        titulo_raw = slug[:-(len(job_id) + 1)].replace('-', ' ').strip()
        titulo = titulo_raw.title() if titulo_raw else 'Sin titulo'
        jobs.append({
            'job_id':    job_id,
            'titulo':    titulo[:300],
            'empresa':   '',
            'ubicacion': 'Colombia',
            'url':       BASE + href,
        })

    log.info('Pagina %d: %d ofertas', page, len(jobs))
    return jobs


def main():
    state     = load_state()
    start_page = state['page']
    total_new  = state.get('total_new', 0)
    log.info('=== Backfill ElEmpleo CO — desde pagina %d ===', start_page)

    sess = get_session()
    conn = get_conn()
    create_table(conn)

    streak     = 0
    last_page  = start_page

    for page in range(start_page, MAX_PAGES + 1):
        last_page = page
        jobs = fetch_page(sess, page)

        if jobs is None:
            log.info('Pagina %d: 404 — fin de paginacion', page)
            save_state(page + 1, total_new)
            break

        if not jobs:
            streak += 1
            log.info('Pagina %d: sin resultados (racha %d/%d)', page, streak, MAX_STREAK)
            save_state(page + 1, total_new)
            if streak >= MAX_STREAK:
                log.info('Deteniendo: %d paginas consecutivas sin resultados', streak)
                break
            time.sleep(DELAY)
            continue

        n = upsert(conn, jobs)
        streak = 0 if n > 0 else streak + 1
        total_new += n
        save_state(page + 1, total_new)
        log.info('Pagina %d: %d nuevos (acumulado %d)', page, n, total_new)

        if streak >= MAX_STREAK:
            log.info('Deteniendo: %d paginas sin nuevos', streak)
            break

        time.sleep(DELAY)

    conn.close()
    log.info('=== Backfill finalizado: %d nuevos (ultima pagina: %d) ===', total_new, last_page)
    print(f'Finalizado: {total_new} nuevos')


if __name__ == '__main__':
    main()