Nye-TeeOff/backend/scrape_status.py

618 lines
28 KiB
Python
Raw Permalink Normal View History

2026-02-28 09:20:56 +01:00
import asyncio
import os
import asyncpg
import smtplib
2026-03-02 19:39:40 +01:00
import re
2026-03-05 05:18:03 +01:00
import argparse
2026-02-28 09:20:56 +01:00
from datetime import datetime
from email.mime.text import MIMEText
from email.mime.multipart import MIMEMultipart
from playwright.async_api import async_playwright
try:
from playwright_stealth import stealth_async as apply_stealth
except ImportError:
from playwright_stealth import stealth as apply_stealth
2026-03-05 05:18:03 +01:00
from google import genai
2026-02-28 09:20:56 +01:00
from dotenv import load_dotenv
2026-04-27 08:56:58 +02:00
from course_status_history import ensure_course_status_history_table, log_course_status_change
from env_config import get_database_url
2026-05-04 15:30:29 +02:00
from scrape_utils import (
ProgressCallback,
emit_progress,
exclude_discontinued_facilities_clause,
make_progress_event,
)
2026-02-28 09:20:56 +01:00
load_dotenv()
DB_URL = get_database_url()
2026-05-04 15:30:29 +02:00
DISABLED_STATUS_METHODS = {"", "disabled", "manual"}
2026-02-28 09:20:56 +01:00
2026-03-05 05:18:03 +01:00
# ==========================================
# KONFIGURERER GEMINI AI (NY SDK)
# ==========================================
client = genai.Client()
2026-03-05 09:25:15 +01:00
async def ask_llm_status(text, course_name, is_single_course, ai_instruction=None):
2026-03-05 05:18:03 +01:00
if is_single_course:
bane_instruks = "Finn den generelle banestatusen for dette golfanlegget. Se bort fra spesifikke banenavn, da anlegget kun har én bane."
else:
bane_instruks = f'Finn banestatusen SPESIFIKT for banen som heter/omtales som: "{course_name}".'
2026-03-05 15:11:04 +01:00
ekstra_tekst = f"\n!!! VIKTIG EKSTRA-INSTRUKS FRA ADMIN (DENNE OVERSTYRER ALLE ANDRE REGLER) !!!:\n{ai_instruction}\n" if ai_instruction else ""
2026-03-05 09:25:15 +01:00
2026-03-05 05:18:03 +01:00
prompt = f"""
Du er en ekspert å lese norske golfklubbers nettsider for å finne banestatus.
{bane_instruks}
2026-03-05 09:25:15 +01:00
{ekstra_tekst}
2026-03-05 05:18:03 +01:00
Svar KUN med nøyaktig ETT av disse ordene:
- aapen (hvis banen er åpen/sommergreener)
- stengt (hvis banen er lukket/stengt/frost/snø)
- aapen_med_vintergreener (hvis det spilles vintergreener)
- aapner_snart (hvis den åpner om kort tid)
- stenger_snart (hvis den stenger for sesongen om kort tid)
- under_utvikling (hvis den er under utvikling)
- nedlagt (hvis den er nedlagt)
- ukjent (hvis du ikke finner noe info om banen i teksten)
Tekst fra nettsiden:
{text[:15000]}
"""
2026-03-05 15:11:04 +01:00
print("\n" + "="*60)
print(f"🤖 SENDER PROMPT TIL GEMINI FOR: '{course_name}'")
print(f"👉 STANDARD-INSTRUKS: {bane_instruks}")
if ai_instruction:
print(f"👉 ADMIN-HVISKER: {ai_instruction}")
clean_text_sample = " ".join(text.split())[:250]
print(f"👉 TEKST FRA NETTSIDEN (utdrag): '{clean_text_sample}...'")
print("="*60 + "\n")
2026-03-05 05:18:03 +01:00
try:
response = await client.aio.models.generate_content(
model='gemini-2.5-flash',
contents=prompt
)
svar = response.text.strip().lower()
2026-03-05 15:11:04 +01:00
print(f" 🧠 GEMINI RÅ-SVAR: '{svar}'")
# --- NYTT: SORTERT SIKKERHETSFILTER ---
2026-03-05 05:18:03 +01:00
gyldige_svar = [
"aapen_med_vintergreener",
"aapner_snart",
"stenger_snart",
"under_utvikling",
"nedlagt",
2026-03-05 15:11:04 +01:00
"stengt",
"aapen",
2026-03-05 05:18:03 +01:00
"ukjent"
]
for gyldig in gyldige_svar:
if gyldig in svar:
return gyldig
return "ukjent"
except Exception as e:
print(f"❌ Gemini Feil: {e}")
return "ukjent"
# ==========================================
# EKSISTERENDE LOGIKK FOR MANUELL SCRAPING
# ==========================================
2026-02-28 09:20:56 +01:00
def clean_text(text):
return re.sub(r'[^a-zA-Z0-9æøåÆØÅ]', '', text).lower()
def interpret_status(text, keyword=None):
t_raw = text.lower()
if keyword:
k_clean = clean_text(keyword)
if k_clean not in clean_text(t_raw):
return "NOT_FOUND"
parts = re.split(re.escape(keyword), t_raw, flags=re.IGNORECASE)
if len(parts) > 1:
t_raw = parts[1][:150]
else:
2026-03-02 19:39:40 +01:00
t_raw = t_raw[-200:]
2026-02-28 09:20:56 +01:00
if any(word in t_raw for word in ["stengt", "lukket", "frost", "snø", "is", "closed", "stenger"]):
return "stengt"
if any(word in t_raw for word in ["vintergreen", "vintergrønn", "vinter"]):
return "aapen_med_vintergreener"
if any(word in t_raw for word in ["snart", "åpner kl"]):
return "aapner_snart"
if any(word in t_raw for word in ["åpen", "åpent", "aapen", "open"]):
return "aapen"
return "ukjent"
2026-03-02 19:39:40 +01:00
def send_report(changes, warnings, successes):
if not changes and not warnings and not successes: return
2026-02-28 09:20:56 +01:00
subject = f"TeeOff Banestatus Rapport - {datetime.now().strftime('%d.%m.%Y')}"
2026-03-05 05:18:03 +01:00
2026-02-28 09:20:56 +01:00
body = "BANESTATUS RAPPORT\n" + "="*30 + "\n\n"
2026-03-02 19:39:40 +01:00
2026-02-28 09:20:56 +01:00
if changes: body += "✅ OPPDATERINGER:\n" + "\n".join(changes) + "\n\n"
2026-03-02 19:39:40 +01:00
if warnings: body += "⚠️ MERKNADER / ADVARSLER:\n" + "\n".join(warnings) + "\n\n"
if successes: body += "🆗 VELLYKKEDE SJEKKER (INGEN ENDRING):\n" + "\n".join(successes) + "\n"
2026-02-28 09:20:56 +01:00
2026-03-02 19:39:40 +01:00
msg = MIMEMultipart()
msg['From'] = os.getenv("SMTP_USER")
msg['To'] = os.getenv("EMAIL_TO")
msg['Subject'] = subject
2026-02-28 09:20:56 +01:00
msg.attach(MIMEText(body, 'plain'))
try:
with smtplib.SMTP_SSL(os.getenv("SMTP_SERVER"), int(os.getenv("SMTP_PORT"))) as server:
server.login(os.getenv("SMTP_USER"), os.getenv("SMTP_PASS"))
server.send_message(msg)
print("✅ Rapport sendt på e-post.")
2026-03-02 19:39:40 +01:00
except Exception as e:
print(f"❌ E-post feil: {e}")
2026-02-28 09:20:56 +01:00
2026-03-05 05:18:03 +01:00
# ==========================================
# HOVEDMOTOR
# ==========================================
2026-04-12 10:11:23 +02:00
async def run_daily_scraping(facility_ids=None, progress_callback: ProgressCallback | None = None):
2026-02-28 09:20:56 +01:00
print(f"🚀 Starter sjekk {datetime.now().strftime('%H:%M:%S')}...")
conn = await asyncpg.connect(DB_URL)
2026-04-27 08:56:58 +02:00
await ensure_course_status_history_table(conn)
2026-05-04 15:30:29 +02:00
facility_filter = exclude_discontinued_facilities_clause("facilities")
2026-02-28 09:20:56 +01:00
2026-03-05 05:18:03 +01:00
if facility_ids:
print(f"📌 Kjører skraping KUN for anlegg-ID(er): {facility_ids}")
facilities = await conn.fetch(
2026-05-04 15:30:29 +02:00
f"SELECT id, name, scrape_status_url, scrape_status_selector, scrape_method, ai_instruction FROM facilities WHERE scrape_status_url IS NOT NULL AND id = ANY($1::int[]){facility_filter}",
2026-03-05 05:18:03 +01:00
facility_ids
)
else:
print("🌍 Kjører skraping for ALLE anlegg med scrape_status_url...")
facilities = await conn.fetch(
2026-05-04 15:30:29 +02:00
f"SELECT id, name, scrape_status_url, scrape_status_selector, scrape_method, ai_instruction FROM facilities WHERE scrape_status_url IS NOT NULL{facility_filter}"
2026-03-05 05:18:03 +01:00
)
if not facilities:
print("⚠️ Fant ingen anlegg å skrape.")
await conn.close()
2026-04-10 18:37:33 +02:00
return {
"processed_facilities": 0,
"updated_courses": 0,
"warnings": 0,
"successes": 0,
2026-04-12 10:11:23 +02:00
"failed_facilities": 0,
"skipped_facilities": 0,
2026-04-10 18:37:33 +02:00
}
2026-03-05 05:18:03 +01:00
2026-03-02 19:39:40 +01:00
changes, warnings, successes = [], [], []
2026-04-12 10:11:23 +02:00
total_facilities = len(facilities)
ok_facilities = 0
failed_facilities = 0
skipped_facilities = 0
await emit_progress(
progress_callback,
progress_total=total_facilities,
progress_completed=0,
progress_ok=0,
progress_failed=0,
progress_skipped=0,
event=make_progress_event(
facility_id=None,
facility_name="Banestatus",
outcome="info",
message=f"Starter banestatusskraping for {total_facilities} anlegg.",
processed=0,
total=total_facilities,
),
)
2026-02-28 09:20:56 +01:00
async with async_playwright() as p:
browser = await p.chromium.launch(headless=True)
context = await browser.new_context()
2026-04-12 10:11:23 +02:00
for index, f in enumerate(facilities, start=1):
2026-05-04 15:30:29 +02:00
raw_method = (f.get('scrape_method') or "").strip()
method = raw_method or 'css_selector'
method_label = "disabled" if raw_method in {"", "disabled"} else method
2026-04-12 10:11:23 +02:00
facility_id = f['id']
facility_name = f['name']
await emit_progress(
progress_callback,
current_facility_id=facility_id,
current_facility_name=facility_name,
event=make_progress_event(
facility_id=facility_id,
facility_name=facility_name,
outcome="info",
2026-05-04 15:30:29 +02:00
message=f"Starter sjekk med metode {method_label}.",
2026-04-12 10:11:23 +02:00
processed=index - 1,
total=total_facilities,
),
)
2026-03-05 09:25:15 +01:00
2026-05-04 15:30:29 +02:00
if raw_method in DISABLED_STATUS_METHODS:
skip_reason = "Manuell overstyring" if raw_method == "manual" else "Skraping avslått"
progress_message = (
"Hoppet over fordi anlegget er satt til manuell overstyring."
if raw_method == "manual"
else "Hoppet over fordi banestatusskraping er avslått for anlegget."
)
successes.append(f"⏸️ {f['name']}: Hoppet over ({skip_reason})")
print(f" ⏸️ Hopper over skraping av {f['name']} ({skip_reason})")
2026-04-12 10:11:23 +02:00
skipped_facilities += 1
await emit_progress(
progress_callback,
progress_completed=index,
progress_ok=ok_facilities,
progress_failed=failed_facilities,
progress_skipped=skipped_facilities,
current_facility_id=facility_id,
current_facility_name=facility_name,
event=make_progress_event(
facility_id=facility_id,
facility_name=facility_name,
outcome="warning",
2026-05-04 15:30:29 +02:00
message=progress_message,
2026-04-12 10:11:23 +02:00
processed=index,
total=total_facilities,
),
)
2026-03-05 09:25:15 +01:00
continue
2026-02-28 09:20:56 +01:00
page = await context.new_page()
try: await apply_stealth(page)
except: pass
try:
2026-03-05 09:25:15 +01:00
print(f"🔍 Besøker {f['name']} (Metode: {method})...")
2026-03-02 19:39:40 +01:00
await page.goto(f['scrape_status_url'], timeout=60000, wait_until="domcontentloaded")
2026-03-05 17:44:18 +01:00
await page.wait_for_timeout(3000)
2026-02-28 09:20:56 +01:00
2026-03-02 19:39:40 +01:00
full_text = ""
if method == 'css_selector':
element = page.locator(f['scrape_status_selector']).first
if await element.count() == 0:
warnings.append(f"{f['name']}: Fant ikke CSS-elementet '{f['scrape_status_selector']}'")
2026-04-12 10:11:23 +02:00
failed_facilities += 1
await emit_progress(
progress_callback,
progress_completed=index,
progress_ok=ok_facilities,
progress_failed=failed_facilities,
progress_skipped=skipped_facilities,
current_facility_id=facility_id,
current_facility_name=facility_name,
event=make_progress_event(
facility_id=facility_id,
facility_name=facility_name,
outcome="error",
message=f"Fant ikke CSS-elementet {f['scrape_status_selector']}.",
processed=index,
total=total_facilities,
),
)
2026-03-02 19:39:40 +01:00
continue
full_text = await element.inner_text()
elif method == 'iframe_golfbox':
frame = page.frame_locator('iframe[src*="golfbox"]')
element = frame.locator(f['scrape_status_selector']).first
if await element.count() == 0:
warnings.append(f"{f['name']}: Fant ikke elementet '{f['scrape_status_selector']}' i iframen")
2026-04-12 10:11:23 +02:00
failed_facilities += 1
await emit_progress(
progress_callback,
progress_completed=index,
progress_ok=ok_facilities,
progress_failed=failed_facilities,
progress_skipped=skipped_facilities,
current_facility_id=facility_id,
current_facility_name=facility_name,
event=make_progress_event(
facility_id=facility_id,
facility_name=facility_name,
outcome="error",
message=f"Fant ikke elementet {f['scrape_status_selector']} i Golfbox-iframe.",
processed=index,
total=total_facilities,
),
)
2026-03-02 19:39:40 +01:00
continue
full_text = await element.inner_text()
2026-03-04 13:17:10 +01:00
elif method == 'click_then_css':
parts = f['scrape_status_selector'].split('||')
if len(parts) != 2:
warnings.append(f"{f['name']}: Ugyldig selector for click_then_css (mangler ||)")
2026-04-12 10:11:23 +02:00
failed_facilities += 1
await emit_progress(
progress_callback,
progress_completed=index,
progress_ok=ok_facilities,
progress_failed=failed_facilities,
progress_skipped=skipped_facilities,
current_facility_id=facility_id,
current_facility_name=facility_name,
event=make_progress_event(
facility_id=facility_id,
facility_name=facility_name,
outcome="error",
message="Ugyldig click_then_css-selector i konfigurasjonen.",
processed=index,
total=total_facilities,
),
)
2026-03-04 13:17:10 +01:00
continue
btn_selector, text_selector = parts
btn = page.locator(btn_selector).first
if await btn.count() == 0:
warnings.append(f"{f['name']}: Fant ikke knappen å klikke på: '{btn_selector}'")
2026-04-12 10:11:23 +02:00
failed_facilities += 1
await emit_progress(
progress_callback,
progress_completed=index,
progress_ok=ok_facilities,
progress_failed=failed_facilities,
progress_skipped=skipped_facilities,
current_facility_id=facility_id,
current_facility_name=facility_name,
event=make_progress_event(
facility_id=facility_id,
facility_name=facility_name,
outcome="error",
message=f"Fant ikke knappen {btn_selector} som skulle klikkes.",
processed=index,
total=total_facilities,
),
)
2026-03-04 13:17:10 +01:00
continue
2026-03-05 17:44:18 +01:00
await btn.click(force=True)
await page.wait_for_timeout(2000)
2026-03-04 13:17:10 +01:00
element = page.locator(text_selector).first
if await element.count() == 0:
warnings.append(f"{f['name']}: Fant ikke tekstboksen '{text_selector}' etter klikk")
2026-04-12 10:11:23 +02:00
failed_facilities += 1
await emit_progress(
progress_callback,
progress_completed=index,
progress_ok=ok_facilities,
progress_failed=failed_facilities,
progress_skipped=skipped_facilities,
current_facility_id=facility_id,
current_facility_name=facility_name,
event=make_progress_event(
facility_id=facility_id,
facility_name=facility_name,
outcome="error",
message=f"Fant ikke tekstboksen {text_selector} etter klikk.",
processed=index,
total=total_facilities,
),
)
2026-03-04 13:17:10 +01:00
continue
full_text = await element.inner_text()
2026-02-28 09:20:56 +01:00
2026-03-05 05:18:03 +01:00
elif method == 'llm_parse':
2026-03-05 15:11:04 +01:00
print(" 🖱️ Leter etter knapper å klikke på for å avdekke skjult tekst...")
2026-03-05 17:44:18 +01:00
knapper = await page.get_by_text(re.compile(r"banestatus|dagens status|se status|se dagens status|baneinfo|\bstatus\b", re.IGNORECASE)).all()
2026-03-05 05:18:03 +01:00
2026-03-05 17:44:18 +01:00
klikk_count = 0
2026-03-05 05:18:03 +01:00
for knapp in knapper:
try:
if await knapp.is_visible():
2026-03-05 17:44:18 +01:00
await knapp.click(timeout=2000, force=True)
klikk_count += 1
await page.wait_for_timeout(2000)
2026-03-05 05:18:03 +01:00
except Exception:
pass
2026-03-05 17:44:18 +01:00
if klikk_count > 0:
print(f" 🎯 Tvangsklikket på {klikk_count} status-knapp(er)! Venter ekstra på at innholdet laster...")
await page.wait_for_timeout(2000)
else:
print(" ⚠️ Fant ingen knapper å klikke på.")
# --- NYTT: HENTER OGSÅ SKJULT TEKST (For Scangolf megamenyer) ---
2026-03-05 05:18:03 +01:00
element = page.locator("body").first
if await element.count() == 0:
warnings.append(f"{f['name']}: Klarte ikke å lese siden for AI-tolkning")
2026-04-12 10:11:23 +02:00
failed_facilities += 1
await emit_progress(
progress_callback,
progress_completed=index,
progress_ok=ok_facilities,
progress_failed=failed_facilities,
progress_skipped=skipped_facilities,
current_facility_id=facility_id,
current_facility_name=facility_name,
event=make_progress_event(
facility_id=facility_id,
facility_name=facility_name,
outcome="error",
message="Klarte ikke å lese siden for AI-tolkning.",
processed=index,
total=total_facilities,
),
)
2026-03-05 05:18:03 +01:00
continue
2026-03-05 17:44:18 +01:00
synlig_tekst = await element.inner_text() or ""
skjult_tekst = await element.text_content() or ""
# Slår sammen all tekst slik at Gemini får med seg menyer som er gjemt med CSS
råtekst = synlig_tekst + " " + skjult_tekst
2026-03-05 05:18:03 +01:00
full_text = " ".join(råtekst.split())
2026-03-05 17:44:18 +01:00
# ----------------------------------------------------------------
2026-03-05 05:18:03 +01:00
2026-03-02 19:39:40 +01:00
else:
warnings.append(f"⚠️ {f['name']}: Ukjent skrapemetode i databasen: '{method}'")
2026-04-12 10:11:23 +02:00
failed_facilities += 1
await emit_progress(
progress_callback,
progress_completed=index,
progress_ok=ok_facilities,
progress_failed=failed_facilities,
progress_skipped=skipped_facilities,
current_facility_id=facility_id,
current_facility_name=facility_name,
event=make_progress_event(
facility_id=facility_id,
facility_name=facility_name,
outcome="error",
message=f"Ukjent skrapemetode: {method}.",
processed=index,
total=total_facilities,
),
)
2026-03-02 19:39:40 +01:00
continue
2026-02-28 09:20:56 +01:00
await conn.execute("UPDATE facilities SET status_updated_at = CURRENT_DATE WHERE id = $1", f['id'])
courses = await conn.fetch("SELECT id, name, status, scrape_keyword FROM courses WHERE facility_id = $1", f['id'])
2026-03-05 05:18:03 +01:00
is_single_course = len(courses) == 1
2026-04-12 10:11:23 +02:00
facility_changed = 0
facility_confirmed = 0
facility_unresolved = 0
2026-03-05 05:18:03 +01:00
2026-02-28 09:20:56 +01:00
for c in courses:
2026-03-05 17:44:18 +01:00
old_status = c['status'] or "ukjent"
2026-03-05 05:18:03 +01:00
if method == 'llm_parse':
print(f" 🤖 Spør Gemini om status for '{c['name']}' (Singelbane: {is_single_course})...")
2026-03-05 09:25:15 +01:00
new_status = await ask_llm_status(full_text, c['name'], is_single_course, f.get('ai_instruction'))
2026-03-05 15:11:04 +01:00
print(" ⏳ Tar 5 sekunders pause for å spare Gemini-kvoten...")
await asyncio.sleep(5)
2026-03-05 05:18:03 +01:00
else:
new_status = interpret_status(full_text, c['scrape_keyword'])
2026-02-28 09:20:56 +01:00
if new_status == "NOT_FOUND":
2026-03-05 05:18:03 +01:00
warnings.append(f"{f['name']} ({c['name']}): Fant ikke søkeordet '{c['scrape_keyword']}' i teksten.")
2026-04-12 10:11:23 +02:00
facility_unresolved += 1
2026-02-28 09:20:56 +01:00
continue
2026-03-05 17:44:18 +01:00
# --- OPPDATERT LOGIKK (Fikser logg-buggen) ---
if new_status == "ukjent":
# Sikkerhetsnettet slår inn: Vi beholder gammel status!
warnings.append(f"⚠️ {f['name']} ({c['name']}): Fant ikke status. Beholder '{old_status.upper()}'.")
print(f" 🟡 KONKLUSJON: Fant ikke status i teksten (Sikkerhetsnett). Beholder gammel status ({old_status.upper()}).")
2026-04-12 10:11:23 +02:00
facility_unresolved += 1
2026-03-05 17:44:18 +01:00
elif new_status != old_status:
2026-04-27 08:56:58 +02:00
await log_course_status_change(
conn,
course_id=int(c["id"]),
facility_id=int(f["id"]),
old_status=old_status,
new_status=new_status,
change_source="scraper",
changed_by="scraper",
)
2026-02-28 09:20:56 +01:00
await conn.execute("UPDATE courses SET status = $1 WHERE id = $2", new_status, c['id'])
changes.append(f"🔹 {f['name']} ({c['name']}): {old_status.upper()}{new_status.upper()}")
2026-03-05 15:11:04 +01:00
print(f" 🟢 KONKLUSJON: Status endret fra {old_status.upper()} til {new_status.upper()}")
2026-04-12 10:11:23 +02:00
facility_changed += 1
2026-02-28 09:20:56 +01:00
else:
2026-03-02 19:39:40 +01:00
successes.append(f"{f['name']} ({c['name']}): {new_status.upper()}")
2026-03-05 17:44:18 +01:00
print(f" ⚪ KONKLUSJON: Ingen endring. Banen er fortsatt {old_status.upper()}")
2026-04-12 10:11:23 +02:00
facility_confirmed += 1
2026-03-05 17:44:18 +01:00
# ---------------------------------------------
2026-02-28 09:20:56 +01:00
2026-04-12 10:11:23 +02:00
ok_facilities += 1
if facility_changed > 0:
facility_outcome = "success"
facility_message = (
f"{facility_changed} baner oppdatert, {facility_confirmed} bekreftet"
+ (f", {facility_unresolved} beholdt som før" if facility_unresolved > 0 else "")
+ "."
)
elif facility_unresolved > 0:
facility_outcome = "warning"
facility_message = (
f"Ingen statusendring. {facility_confirmed} baner bekreftet og "
f"{facility_unresolved} beholdt som før."
)
else:
facility_outcome = "success"
facility_message = f"Ingen endring. {facility_confirmed} baner bekreftet."
await emit_progress(
progress_callback,
progress_completed=index,
progress_ok=ok_facilities,
progress_failed=failed_facilities,
progress_skipped=skipped_facilities,
current_facility_id=facility_id,
current_facility_name=facility_name,
event=make_progress_event(
facility_id=facility_id,
facility_name=facility_name,
outcome=facility_outcome,
message=facility_message,
processed=index,
total=total_facilities,
),
)
2026-02-28 09:20:56 +01:00
except Exception as e:
2026-03-02 19:39:40 +01:00
err_msg = str(e).split('\n')[0]
warnings.append(f"🔥 {f['name']}: Feil under skraping: {err_msg}")
2026-04-12 10:11:23 +02:00
failed_facilities += 1
await emit_progress(
progress_callback,
progress_completed=index,
progress_ok=ok_facilities,
progress_failed=failed_facilities,
progress_skipped=skipped_facilities,
current_facility_id=facility_id,
current_facility_name=facility_name,
event=make_progress_event(
facility_id=facility_id,
facility_name=facility_name,
outcome="error",
message=f"Feil under skraping: {err_msg}",
processed=index,
total=total_facilities,
),
)
2026-02-28 09:20:56 +01:00
finally:
await page.close()
2026-03-05 05:18:03 +01:00
2026-02-28 09:20:56 +01:00
await browser.close()
await conn.close()
2026-03-02 19:39:40 +01:00
send_report(changes, warnings, successes)
2026-02-28 09:20:56 +01:00
print("🏁 Ferdig.")
2026-04-10 18:37:33 +02:00
return {
"processed_facilities": len(facilities),
"updated_courses": len(changes),
"warnings": len(warnings),
"successes": len(successes),
2026-04-12 10:11:23 +02:00
"failed_facilities": failed_facilities,
"skipped_facilities": skipped_facilities,
2026-04-10 18:37:33 +02:00
}
2026-02-28 09:20:56 +01:00
if __name__ == "__main__":
2026-03-05 05:18:03 +01:00
parser = argparse.ArgumentParser(description="TeeOff Status Scraper")
parser.add_argument("--ids", type=str, help="Kommaseparert liste med anleggs-IDer", default=None)
args = parser.parse_args()
facility_ids_list = None
if args.ids:
try:
facility_ids_list = [int(id_str.strip()) for id_str in args.ids.split(",") if id_str.strip()]
except ValueError:
print("❌ Feil format på --ids. Må være kommaseparerte tall, f.eks: 1,4,12")
exit(1)
2026-04-10 18:37:33 +02:00
asyncio.run(run_daily_scraping(facility_ids_list))