From 5d3a4b47b41fb349648c0ca22d83a1ed127f5ceb Mon Sep 17 00:00:00 2001
From: Rio Liu <rio.liu@r26.me>
Date: Wed, 2 Feb 2022 22:08:16 -0500
Subject: [PATCH] error on 404 and fix artix checker

---
 data.json         |  4 +---
 main.py           |  2 +-
 project.py        | 17 +++++++++--------
 projects/artix.py | 37 ++++++++++++++++++++++++++++++++++++-
 4 files changed, 47 insertions(+), 13 deletions(-)

diff --git a/data.json b/data.json
index cbfa116..31c307a 100644
--- a/data.json
+++ b/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,
diff --git a/main.py b/main.py
index d8ab140..a91cb70 100644
--- a/main.py
+++ b/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:
diff --git a/project.py b/project.py
index e3da361..b438e4a 100644
--- a/project.py
+++ b/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:
diff --git a/projects/artix.py b/projects/artix.py
index c95e6cd..c3ace18 100644
--- a/projects/artix.py
+++ b/projects/artix.py
@@ -2,8 +2,43 @@
 Contains Artix class
 """
 
-from project import Project
+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