This commit is contained in:
2026-01-14 15:40:16 +01:00
parent eb84dbc387
commit 9da7594052
+102 -35
View File
@@ -12,22 +12,31 @@ data:
import requests
from bs4 import BeautifulSoup
from fastapi import FastAPI, Query
from fastapi import FastAPI, Query, Response
from fastapi.responses import JSONResponse
from prometheus_client import Gauge, Counter, Histogram, generate_latest, CONTENT_TYPE_LATEST
app = FastAPI()
IDOKEP_BASE = "https://www.idokep.hu"
DEFAULT_PLACE = os.getenv("IDOKEP_PLACE", "Budapest VII. ker")
DEFAULT_PLACE = os.getenv("IDOKEP_PLACE", "Budapest VIII. ker")
USER_AGENT = os.getenv(
"IDOKEP_UA",
"Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120 Safari/537.36",
)
# simple in-memory cache: {place: (expires_epoch, payload)}
CACHE_TTL_SEC = int(os.getenv("CACHE_TTL_SEC", "600")) # 10 min
CACHE_TTL_SEC = int(os.getenv("CACHE_TTL_SEC", "900")) # 15 minutes
_cache: Dict[str, Tuple[float, Dict[str, Any]]] = {}
# --- Prometheus metrics (low-cardinality, place as label) ---
SCRAPES_TOTAL = Counter("idokep_scrapes_total", "Total Időkép scrapes", ["place", "status"])
SCRAPE_SECONDS = Histogram("idokep_scrape_seconds", "Időkép scrape duration in seconds", ["place"])
CURRENT_TEMP_C = Gauge("idokep_current_temp_c", "Current temperature in Celsius", ["place"])
DAILY_TMIN_C = Gauge("idokep_daily_tmin_c", "Daily minimum temperature in Celsius", ["place", "dow"])
DAILY_TMAX_C = Gauge("idokep_daily_tmax_c", "Daily maximum temperature in Celsius", ["place", "dow"])
DAILY_PREC_MM = Gauge("idokep_daily_precip_mm", "Daily precipitation in mm", ["place", "dow"])
def _num(s: str) -> Optional[float]:
if s is None:
return None
@@ -49,11 +58,9 @@ data:
return el.get_text(" ", strip=True)
def _fetch_place_html(place: str) -> str:
# Időkép uses the place slug in path; requests will percent-encode automatically if we build it carefully.
# We'll do a conservative encoding by replacing spaces with %20, keep UTF-8.
place_path = requests.utils.requote_uri(place)
url = f"{IDOKEP_BASE}/idojaras/{place_path}"
r = requests.get(url, headers={"User-Agent": USER_AGENT}, timeout=15)
r = requests.get(url, headers={"User-Agent": USER_AGENT}, timeout=20)
r.raise_for_status()
return r.text
@@ -70,21 +77,17 @@ data:
cond_hu_el = soup.select_one(".weather-short-desc")
condition_hu = _pick_text(cond_hu_el)
# HOURLY (take first 6)
# HOURLY (first 6)
hourly_cards = soup.select(".new-hourly-forecast-card")
hourly: List[Dict[str, Any]] = []
for card in hourly_cards[:6]:
hour_el = card.select_one(".new-hourly-forecast-hour")
hour_txt = _pick_text(hour_el)
hour_txt = _pick_text(card.select_one(".new-hourly-forecast-hour"))
htemp_el = card.select_one(".tempValue .hover-over")
htemp_c = _num(_pick_text(htemp_el) or "")
htemp_c = _num(_pick_text(card.select_one(".tempValue .hover-over")) or "")
hicon_el = card.select_one(".forecast-icon")
hicon_url = _abs_url(hicon_el.get("src") if hicon_el else None)
hicon_url = _abs_url((card.select_one(".forecast-icon") or {}).get("src")) if card.select_one(".forecast-icon") else None
hprec_el = card.select_one(".hourly-rain-chance a")
hprec_pct = _num(_pick_text(hprec_el) or "")
hprec_pct = _num(_pick_text(card.select_one(".hourly-rain-chance a")) or "")
hourly.append(
{
@@ -95,33 +98,26 @@ data:
}
)
# DAILY (take next 5 columns; Időkép layout usually has an extra leading column, HA used nth-child(2) as day1)
# DAILY (next 5, skip first like your HA template did)
daily_cols = soup.select(".dailyForecastCol")
daily: List[Dict[str, Any]] = []
# Skip first column if it looks like a header-ish column; keep behavior close to your HA selectors.
cols = daily_cols[1:] if len(daily_cols) >= 2 else daily_cols
daily_raw: List[Dict[str, Any]] = []
for col in cols[:5]:
dow_el = col.select_one(".dfDay")
dow = _pick_text(dow_el)
dow = _pick_text(col.select_one(".dfDay"))
daynum = _pick_text(col.select_one(".dfDayNum"))
daynum_el = col.select_one(".dfDayNum")
daynum = _pick_text(daynum_el)
dicon_url = _abs_url((col.select_one(".forecast") or {}).get("src")) if col.select_one(".forecast") else None
dicon_el = col.select_one(".forecast")
dicon_url = _abs_url(dicon_el.get("src") if dicon_el else None)
# max/min sometimes show in different wrappers; try a few options
# various layouts: try a few
max_el = col.select_one(".max a") or col.select_one(".min-max-close a:nth-child(1)")
min_el = col.select_one(".min a") or col.select_one(".min-max-close a:nth-child(2)")
tmax_c = _num(_pick_text(max_el) or "")
tmin_c = _num(_pick_text(min_el) or "")
mm_el = col.select_one(".mm")
prec_mm = _num(_pick_text(mm_el) or "")
prec_mm = _num(_pick_text(col.select_one(".mm")) or "")
daily.append(
daily_raw.append(
{
"dow": dow, # e.g. "Sze"
"daynum": daynum, # e.g. "14"
@@ -132,6 +128,28 @@ data:
}
)
# Compute weekly min/max for HA-like bars (left/width)
mins = [d["tmin_c"] for d in daily_raw if d.get("tmin_c") is not None]
maxs = [d["tmax_c"] for d in daily_raw if d.get("tmax_c") is not None]
week_min = min(mins) if mins else None
week_max = max(maxs) if maxs else None
denom = (week_max - week_min) if (week_min is not None and week_max is not None and week_max != week_min) else None
daily: List[Dict[str, Any]] = []
for d in daily_raw:
left = None
width = None
if denom is not None and d.get("tmin_c") is not None and d.get("tmax_c") is not None:
left = ((d["tmin_c"] - week_min) / denom) * 100.0
width = ((d["tmax_c"] - d["tmin_c"]) / denom) * 100.0
# clamp for safety
left = max(0.0, min(100.0, left))
width = max(1.0, min(100.0, width))
d2 = dict(d)
d2["bar_left_pct"] = left
d2["bar_width_pct"] = width
daily.append(d2)
return {
"source": {
"name": "Időkép",
@@ -145,21 +163,60 @@ data:
},
"hourly": hourly,
"daily": daily,
"weekly": {
"tmin_c": week_min,
"tmax_c": week_max,
},
"fetched_at_unix": int(time.time()),
}
@app.get("/api/idokep")
def api_idokep(place: str = Query(default=DEFAULT_PLACE, description="Időkép place name as used in the /idojaras/<place> URL")):
def api_idokep(place: str = Query(default=DEFAULT_PLACE, description="Időkép place name as in /idojaras/<place>")):
now = time.time()
cached = _cache.get(place)
if cached and cached[0] > now:
return JSONResponse(cached[1])
with SCRAPE_SECONDS.labels(place=place).time():
try:
html = _fetch_place_html(place)
payload = _parse_idokep(html, place)
_cache[place] = (now + CACHE_TTL_SEC, payload)
# update metrics (best-effort)
t = payload.get("current", {}).get("temp_c")
if t is not None:
CURRENT_TEMP_C.labels(place=place).set(float(t))
for d in payload.get("daily", []):
dow = d.get("dow") or "?"
if d.get("tmin_c") is not None:
DAILY_TMIN_C.labels(place=place, dow=dow).set(float(d["tmin_c"]))
if d.get("tmax_c") is not None:
DAILY_TMAX_C.labels(place=place, dow=dow).set(float(d["tmax_c"]))
if d.get("prec_mm") is not None:
DAILY_PREC_MM.labels(place=place, dow=dow).set(float(d["prec_mm"]))
SCRAPES_TOTAL.labels(place=place, status="ok").inc()
return JSONResponse(payload)
except Exception:
SCRAPES_TOTAL.labels(place=place, status="error").inc()
# return a structured error Glance can show
return JSONResponse(
{
"place": place,
"error": "Failed to scrape Időkép. Check the place string or Időkép page layout changes.",
"fetched_at_unix": int(time.time()),
},
status_code=502,
)
@app.get("/metrics")
def metrics():
return Response(generate_latest(), media_type=CONTENT_TYPE_LATEST)
---
apiVersion: apps/v1
kind: Deployment
@@ -175,6 +232,10 @@ spec:
metadata:
labels:
app: idokep-proxy
annotations:
prometheus.io/scrape: "true"
prometheus.io/port: "8000"
prometheus.io/path: "/metrics"
spec:
containers:
- name: idokep-proxy
@@ -183,9 +244,15 @@ spec:
- containerPort: 8000
env:
- name: IDOKEP_PLACE
value: "Budapest VII. ker"
value: "Budapest VIII. ker"
- name: CACHE_TTL_SEC
value: "900" # 15 minutes, matches your HA scan_interval philosophy
value: "900"
resources:
requests:
cpu: 25m
memory: 128Mi
limits:
memory: 256Mi
volumeMounts:
- name: app
mountPath: /app
@@ -193,7 +260,7 @@ spec:
command: ["/bin/sh","-lc"]
args:
- |
pip install --no-cache-dir fastapi uvicorn requests beautifulsoup4 lxml &&
pip install --no-cache-dir fastapi uvicorn requests beautifulsoup4 lxml prometheus_client &&
uvicorn app:app --host 0.0.0.0 --port 8000
volumes:
- name: app