support/scripts/pkg-stats: use aiohttp for upstream URL checking

This commit reworks the code that checks if the upstream URL of each
package (specified by its Config.in file) using the aiohttp
module. This makes the implementation much more elegant, and avoids
the problematic multiprocessing Pool which is causing issues in some
situations.

Suggested-by: Titouan Christophe <titouan.christophe@railnova.eu>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

diff --git a/support/scripts/pkg-stats b/support/scripts/pkg-stats
index 3423c44815545a0eb052d5716df8ef7d9596f4c7..70e7fa7a0c026ff9b8d3e5ae4720be87df40444f 100755 (executable)
--- a/support/scripts/pkg-stats
+++ b/support/scripts/pkg-stats
@@ -25,14 +25,13 @@ import os
 from collections import defaultdict
 import re
 import subprocess
-import requests  # URL checking
+import requests  # NVD database download
 import json
 import ijson
 import distutils.version
 import time
 import gzip
 import sys
-from multiprocessing import Pool
 
 sys.path.append('utils/')
 from getdeveloperlib import parse_developers  # noqa: E402
@@ -499,26 +498,30 @@ def package_init_make_info():
             Package.all_ignored_cves[pkgvar] = value.split()
 
 
-def check_url_status_worker(url, url_status):
-    if url_status[0] == 'ok':
-        try:
-            url_status_code = requests.head(url, timeout=30).status_code
-            if url_status_code >= 400:
-                return ("error", "invalid {}".format(url_status_code))
-        except requests.exceptions.RequestException:
-            return ("error", "invalid (err)")
-        return ("ok", "valid")
-    return url_status
+async def check_url_status(session, pkg, retry=True):
+    try:
+        async with session.get(pkg.url) as resp:
+            if resp.status >= 400:
+                pkg.status['url'] = ("error", "invalid {}".format(resp.status))
+                return
+    except (aiohttp.ClientError, asyncio.TimeoutError):
+        if retry:
+            return await check_url_status(session, pkg, retry=False)
+        else:
+            pkg.status['url'] = ("error", "invalid (err)")
+            return
 
+    pkg.status['url'] = ("ok", "valid")
 
-def check_package_urls(packages):
-    pool = Pool(processes=64)
-    for pkg in packages:
-        pkg.url_worker = pool.apply_async(check_url_status_worker, (pkg.url, pkg.status['url']))
-    for pkg in packages:
-        pkg.status['url'] = pkg.url_worker.get(timeout=3600)
-        del pkg.url_worker
-    pool.terminate()
+
+async def check_package_urls(packages):
+    tasks = []
+    connector = aiohttp.TCPConnector(limit_per_host=5)
+    async with aiohttp.ClientSession(connector=connector, trust_env=True) as sess:
+        packages = [p for p in packages if p.status['url'][0] == 'ok']
+        for pkg in packages:
+            tasks.append(check_url_status(sess, pkg))
+        await asyncio.wait(tasks)
 
 
 def check_package_latest_version_set_status(pkg, status, version, identifier):
@@ -1068,7 +1071,8 @@ def __main__():
         pkg.set_url()
         pkg.set_developers(developers)
     print("Checking URL status")
-    check_package_urls(packages)
+    loop = asyncio.get_event_loop()
+    loop.run_until_complete(check_package_urls(packages))
     print("Getting latest versions ...")
     loop = asyncio.get_event_loop()
     loop.run_until_complete(check_package_latest_version(packages))