Add actual scraping

arkose_prometheus/__init__.py

@@ -0,0 +1,34 @@
+import logging
+import argparse
+import asyncio
+from prometheus_client import start_http_server
+from . import arkose_gauge
+
+logger = logging.getLogger(__name__)
+
+
+def main():
+    parser = argparse.ArgumentParser("arkose_prometheus")
+    parser.add_argument("-p", "--port", type=int, default=9154)
+    parser.add_argument(
+        "-i",
+        "--scrape-interval",
+        type=int,
+        default=180,
+        help="Scrape interval, in seconds.",
+    )
+
+    args = parser.parse_args()
+    logging.basicConfig(level=logging.INFO)
+
+    start_http_server(args.port)
+    logger.info("Listening on port %d...", args.port)
+
+    try:
+        asyncio.run(arkose_gauge.scrape_metric_forever(interval=args.scrape_interval))
+    except KeyboardInterrupt:
+        logger.info("Closing server (keyboard interrupt)")
+
+
+if __name__ == "__main__":
+    main()

arkose_prometheus/arkose_gauge.py

@@ -0,0 +1,59 @@
+import asyncio
+import aiohttp
+import datetime
+import re
+import logging
+from prometheus_client import Gauge
+
+logger = logging.getLogger(__name__)
+
+ARKOSES = {
+    "genevois": "https://genevois.arkose.com/",
+    "massy": "https://massy.arkose.com/",
+    "montreuil": "https://montreuil.arkose.com/",
+    "nation": "https://nation.arkose.com/",
+}
+
+BLOCKPARK_OCCUPATION = Gauge(
+    "blockpark_occupation", "Blockpark occupation", ["blockpark"]
+)
+
+GAUGE_RE = re.compile(r'<div class="jauge-bar"><div style="width: ([0-9]*)%"></div>')
+
+
+class FailedScrape(Exception):
+    pass
+
+
+async def _scrape_arkose(arkose, session):
+    async with session.get(ARKOSES[arkose]) as resp:
+        if resp.status != 200:
+            raise FailedScrape("Non-200 error")
+        content = await resp.text()
+
+    with open("/tmp/arkose_{}.html".format(arkose), "w") as handle:
+        handle.write(content)
+
+    match = GAUGE_RE.search(content)
+    if not match:
+        raise FailedScrape("Could not extract gauge")
+    return int(match[1])
+
+
+async def scrape_metric_forever(interval: int):
+    async with aiohttp.ClientSession() as session:
+        while True:
+            start_time = datetime.datetime.now()
+            occupation_data = await asyncio.gather(
+                *[_scrape_arkose(arkose, session) for arkose in ARKOSES],
+                return_exceptions=True
+            )
+
+            for arkose, occup in zip(ARKOSES, occupation_data):
+                if isinstance(occup, Exception):
+                    logger.warning("%s failed: %s", arkose, occup)
+                    continue
+                BLOCKPARK_OCCUPATION.labels(blockpark=arkose).set(occup)
+
+            delta_time = datetime.datetime.now() - start_time
+            await asyncio.sleep(interval - delta_time.total_seconds())