Merge pull request 'Fix Artix checker and make wrong http code raise error' (#3) from artix-checker into master

Reviewed-on: public/mirror-checker#3

data.json

@@ -31,9 +31,7 @@
 		"out_of_sync_since": null,
 		"out_of_sync_interval": 86400,
 		"csc": "artixlinux/",
-		"upstream": "https://mirror1.artixlinux.org/repos/",
-		"file": "lastupdate",
-		"exclude": true
+		"upstream": "https://mirror1.artixlinux.org/repos/"
 	},
 	"CentOS": {
 		"out_of_sync_since": null,

main.py

@@ -36,7 +36,7 @@ if __name__ == "__main__":
             if data[project].get('exclude', False):
                 continue
             checker_result = project_class.check(data, project, current_time)
-            if project in ["CPAN", "ubuntu_releases", "manjaro", "mxlinux", "cran", "ctan", "gentooportage"]:
+            if project in ["CPAN", "ubuntu_releases", "manjaro", "mxlinux", "cran", "ctan", "gentooportage", "Artix"]:
                 if checker_result:
                     safe_print(f"Success: {project} up-to-date")
                 else:

project.py

@@ -18,17 +18,18 @@ class Project(ABC):
 		"""Check if project packages are up-to-date"""
 		csc_url = CSC_MIRROR + data[project]["csc"] + data[project]["file"]
 		upstream_url = data[project]["upstream"] + data[project]["file"]
-		CSC = requests.get(csc_url).text
-		upstream = requests.get(upstream_url).text
+
+		req = requests.get(csc_url)
+		req.raise_for_status()
+		CSC = req.text
+
+		req = requests.get(upstream_url)
+		req.raise_for_status()
+		upstream = req.text
+
 		if upstream == CSC:
 			return True
 
-		# Parse number differences
-		bad_re = '[a-zA-Z \-\n]+'
-		if re.search(bad_re, CSC):
-			# print(re.search(bad_re, CSC).group().strip())
-			CSC = re.sub(bad_re, '', CSC)
-			upstream = re.sub(bad_re, '', upstream)
 		try:
 			return int(upstream) - int(CSC) < data[project]["out_of_sync_interval"]
 		except ValueError:

projects/artix.py

@@ -2,8 +2,43 @@
 Contains Artix class
 """
 
+import re
+from datetime import datetime
+import requests
+
 from project import Project
+from shared import CSC_MIRROR
 
+def _get_date(url):
+	req = requests.head(url)
+	req.raise_for_status()
+	return datetime.strptime(req.headers['Last-Modified'], '%a, %d %b %Y %H:%M:%S %Z')
 
 class Artix(Project):
 	"""Artix class"""
+	@classmethod
+	def check(cls, data, project, current_time):
+		csc_url = CSC_MIRROR + data[project]['csc']
+		upstream_url = data[project]['upstream']
+
+		req = requests.get(upstream_url)
+		req.raise_for_status()
+		index = req.text
+
+		repos = re.findall(r'href="(\w+)/"', index);
+
+		outdated_since = None
+		for repo in repos:
+			# Good enough for now, we can change it if Artix gets more arch in the future
+			db_path = repo + '/os/x86_64/' + repo + '.db'
+			upstream_date = _get_date(data[project]['upstream'] + db_path)
+			csc_date = _get_date(CSC_MIRROR + data[project]['csc'] + db_path)
+			if csc_date < upstream_date:
+				if outdated_since is None or upstream_date < outdated_since:
+					outdated_since = upstream_date
+
+		if outdated_since is not None:
+			data[project]['out_of_sync_since'] = int(outdated_since.timestamp())
+			return (datetime.now() - outdated_since).total_seconds() < data[project]['out_of_sync_interval']
+
+		return True