Compare commits
No commits in common. "master" and "master" have entirely different histories.
23
data.json
23
data.json
|
@ -48,8 +48,7 @@
|
||||||
"file": "timestamp"
|
"file": "timestamp"
|
||||||
},
|
},
|
||||||
"CPAN": {
|
"CPAN": {
|
||||||
"out_of_sync_interval": 172800,
|
"out_of_sync_interval": 172800
|
||||||
"out_of_sync_since": null
|
|
||||||
},
|
},
|
||||||
"cran": {
|
"cran": {
|
||||||
"out_of_sync_since": null,
|
"out_of_sync_since": null,
|
||||||
|
@ -87,7 +86,7 @@
|
||||||
"file": "debian-cd/project/trace/cdimage.debian.org"
|
"file": "debian-cd/project/trace/cdimage.debian.org"
|
||||||
},
|
},
|
||||||
"DebianMultimedia": {
|
"DebianMultimedia": {
|
||||||
"out_of_sync_since": 1659116719,
|
"out_of_sync_since": null,
|
||||||
"out_of_sync_interval": 86400,
|
"out_of_sync_interval": 86400,
|
||||||
"csc": "debian-multimedia/",
|
"csc": "debian-multimedia/",
|
||||||
"upstream": "http://debian-mirrors.sdinet.de/deb-multimedia/",
|
"upstream": "http://debian-mirrors.sdinet.de/deb-multimedia/",
|
||||||
|
@ -117,7 +116,7 @@
|
||||||
},
|
},
|
||||||
"Fedora": {
|
"Fedora": {
|
||||||
"out_of_sync_since": null,
|
"out_of_sync_since": null,
|
||||||
"out_of_sync_interval": 259200,
|
"out_of_sync_interval": 86400,
|
||||||
"csc": "fedora/",
|
"csc": "fedora/",
|
||||||
"upstream": "http://fedora.mirror.iweb.com/",
|
"upstream": "http://fedora.mirror.iweb.com/",
|
||||||
"file": "linux/development/rawhide/COMPOSE_ID"
|
"file": "linux/development/rawhide/COMPOSE_ID"
|
||||||
|
@ -174,7 +173,7 @@
|
||||||
"out_of_sync_interval": 172800
|
"out_of_sync_interval": 172800
|
||||||
},
|
},
|
||||||
"KDE": {
|
"KDE": {
|
||||||
"out_of_sync_since": 1659116720,
|
"out_of_sync_since": null,
|
||||||
"out_of_sync_interval": 86400,
|
"out_of_sync_interval": 86400,
|
||||||
"csc": "kde/",
|
"csc": "kde/",
|
||||||
"upstream": "https://kde.c3sl.ufpr.br/",
|
"upstream": "https://kde.c3sl.ufpr.br/",
|
||||||
|
@ -209,7 +208,7 @@
|
||||||
"file": "dists/"
|
"file": "dists/"
|
||||||
},
|
},
|
||||||
"macports": {
|
"macports": {
|
||||||
"out_of_sync_since": 1642827723,
|
"out_of_sync_since": null,
|
||||||
"out_of_sync_interval": 86400,
|
"out_of_sync_interval": 86400,
|
||||||
"csc": "MacPorts/mpdistfiles/",
|
"csc": "MacPorts/mpdistfiles/",
|
||||||
"upstream": "https://distfiles.macports.org/",
|
"upstream": "https://distfiles.macports.org/",
|
||||||
|
@ -270,7 +269,7 @@
|
||||||
"exclude": true
|
"exclude": true
|
||||||
},
|
},
|
||||||
"opensuse": {
|
"opensuse": {
|
||||||
"out_of_sync_since": 1648699331,
|
"out_of_sync_since": null,
|
||||||
"out_of_sync_interval": 86400,
|
"out_of_sync_interval": 86400,
|
||||||
"csc": "opensuse/update/",
|
"csc": "opensuse/update/",
|
||||||
"upstream": "http://opensuse-mirror-gce-us.opensu.se/update/",
|
"upstream": "http://opensuse-mirror-gce-us.opensu.se/update/",
|
||||||
|
@ -321,7 +320,7 @@
|
||||||
"file": "dists/"
|
"file": "dists/"
|
||||||
},
|
},
|
||||||
"raspbian": {
|
"raspbian": {
|
||||||
"out_of_sync_since": 1659116721,
|
"out_of_sync_since": null,
|
||||||
"out_of_sync_interval": 86400,
|
"out_of_sync_interval": 86400,
|
||||||
"csc": "raspbian/",
|
"csc": "raspbian/",
|
||||||
"upstream": "http://archive.raspbian.org/",
|
"upstream": "http://archive.raspbian.org/",
|
||||||
|
@ -342,7 +341,7 @@
|
||||||
"file": ""
|
"file": ""
|
||||||
},
|
},
|
||||||
"slackware": {
|
"slackware": {
|
||||||
"out_of_sync_since": 1642827723,
|
"out_of_sync_since": null,
|
||||||
"out_of_sync_interval": 86400,
|
"out_of_sync_interval": 86400,
|
||||||
"csc": "slackware/",
|
"csc": "slackware/",
|
||||||
"upstream": "https://mirrors.slackware.com/slackware/",
|
"upstream": "https://mirrors.slackware.com/slackware/",
|
||||||
|
@ -374,7 +373,7 @@
|
||||||
"file": ""
|
"file": ""
|
||||||
},
|
},
|
||||||
"ubuntu_ports": {
|
"ubuntu_ports": {
|
||||||
"out_of_sync_since": 1651550528,
|
"out_of_sync_since": null,
|
||||||
"out_of_sync_interval": 86400,
|
"out_of_sync_interval": 86400,
|
||||||
"csc": "ubuntu-ports/",
|
"csc": "ubuntu-ports/",
|
||||||
"upstream": "http://ports.ubuntu.com/ubuntu-ports/",
|
"upstream": "http://ports.ubuntu.com/ubuntu-ports/",
|
||||||
|
@ -391,7 +390,7 @@
|
||||||
"out_of_sync_since": null,
|
"out_of_sync_since": null,
|
||||||
"out_of_sync_interval": 172800,
|
"out_of_sync_interval": 172800,
|
||||||
"csc": "",
|
"csc": "",
|
||||||
"upstream": "https://launchpad.net/ubuntu/+mirror/mirror.csclub.uwaterloo.ca-archive",
|
"upstream": "https://launchpad.net/ubuntu/+mirror/mirror.csclub.uwaterloo.ca-release",
|
||||||
"file": ""
|
"file": ""
|
||||||
},
|
},
|
||||||
"vlc": {
|
"vlc": {
|
||||||
|
@ -402,7 +401,7 @@
|
||||||
"file": "trace"
|
"file": "trace"
|
||||||
},
|
},
|
||||||
"x_org": {
|
"x_org": {
|
||||||
"out_of_sync_since": 1657512131,
|
"out_of_sync_since": null,
|
||||||
"out_of_sync_interval": 86400,
|
"out_of_sync_interval": 86400,
|
||||||
"csc": "x.org/individual/",
|
"csc": "x.org/individual/",
|
||||||
"upstream": "https://www.x.org/releases/individual/",
|
"upstream": "https://www.x.org/releases/individual/",
|
||||||
|
|
91
main.py
91
main.py
|
@ -7,88 +7,59 @@ This mirror status checker determines whether CSC mirror is up-to-date with upst
|
||||||
import time
|
import time
|
||||||
import sys
|
import sys
|
||||||
import requests
|
import requests
|
||||||
from multiprocessing import Pool, Manager
|
|
||||||
from typing import Optional
|
|
||||||
from time import sleep, localtime, strftime
|
|
||||||
|
|
||||||
from projects import *
|
from projects import *
|
||||||
import json
|
import json
|
||||||
|
|
||||||
NUM_THREAD = 16
|
|
||||||
MAX_RETRY = 3
|
|
||||||
RETRY_TIMEOUT = 30 # In seconds
|
|
||||||
|
|
||||||
def safe_print(*args, **kwargs):
|
def safe_print(*args, **kwargs):
|
||||||
# When run with 'chronic' and 'timeout', stdout gets suppressed
|
# When run with 'chronic' and 'timeout', stdout gets suppressed
|
||||||
# due to buffering. Make sure to always flush the output.
|
# due to buffering. Make sure to always flush the output.
|
||||||
print(*args, **kwargs, flush=True)
|
print(*args, **kwargs, flush=True)
|
||||||
|
|
||||||
# Return None if no error occurs and a string for error message otherwise
|
|
||||||
def check_project(args) -> Optional[str]:
|
if __name__ == "__main__":
|
||||||
|
|
||||||
|
exit_code = 0
|
||||||
|
|
||||||
|
data_file = 'data.json'
|
||||||
|
if len(sys.argv) > 1:
|
||||||
|
data_file = sys.argv[1]
|
||||||
|
data = json.load(open(data_file))
|
||||||
|
|
||||||
current_time = int(time.time())
|
current_time = int(time.time())
|
||||||
project, data = args
|
for project in data:
|
||||||
try:
|
try:
|
||||||
project_class = getattr(sys.modules[__name__], project)
|
project_class = getattr(sys.modules[__name__], project)
|
||||||
|
|
||||||
# Skip projects we no longer mirror
|
# Skip projects we no longer mirror
|
||||||
if data[project].get('exclude', False):
|
if data[project].get('exclude', False):
|
||||||
return None
|
continue
|
||||||
|
|
||||||
checker_result = project_class.check(data, project, current_time)
|
checker_result = project_class.check(data, project, current_time)
|
||||||
|
if project in ["CPAN", "ubuntu_releases", "manjaro", "mxlinux", "cran", "ctan", "gentooportage", "Artix"]:
|
||||||
|
if checker_result:
|
||||||
|
safe_print(f"Success: {project} up-to-date")
|
||||||
|
else:
|
||||||
|
safe_print(f"Failure: {project} out-of-sync")
|
||||||
|
|
||||||
|
# Exit with non-zero status if any of the projects are not up-to-date
|
||||||
|
exit_code = 1
|
||||||
|
continue
|
||||||
if checker_result:
|
if checker_result:
|
||||||
data[project]["out_of_sync_since"] = None
|
data[project]["out_of_sync_since"] = None
|
||||||
return None
|
elif data[project]["out_of_sync_since"] is None:
|
||||||
|
|
||||||
elif (data[project]["out_of_sync_since"] is not None
|
|
||||||
and current_time - data[project]["out_of_sync_since"] > data[project]["out_of_sync_interval"]):
|
|
||||||
now_str = strftime("%d %b %Y %H:%M:%S (local time)", localtime())
|
|
||||||
duration = current_time - data[project]["out_of_sync_since"]
|
|
||||||
return f"{project} out-of-sync at {now_str} for {duration}s"
|
|
||||||
|
|
||||||
else:
|
|
||||||
data[project]["out_of_sync_since"] = current_time
|
data[project]["out_of_sync_since"] = current_time
|
||||||
return None
|
elif current_time - data[project]["out_of_sync_since"] \
|
||||||
|
> data[project]["out_of_sync_interval"]:
|
||||||
|
safe_print(f"Failure: {project} out-of-sync")
|
||||||
|
|
||||||
except requests.exceptions.RequestException as err:
|
# Exit with non-zero status if any of the projects are not up-to-date
|
||||||
return f"{project}\n{err}"
|
exit_code = 1
|
||||||
|
continue
|
||||||
def check_project_with_retry(args) -> bool:
|
|
||||||
project, _ = args
|
|
||||||
errs = []
|
|
||||||
for _ in range(MAX_RETRY):
|
|
||||||
res = check_project(args)
|
|
||||||
if res == None:
|
|
||||||
safe_print(f"Success: {project} up-to-date")
|
safe_print(f"Success: {project} up-to-date")
|
||||||
return True
|
except requests.exceptions.RequestException as err:
|
||||||
else:
|
safe_print(f"Error: {project}\n{err}")
|
||||||
errs.append(res)
|
|
||||||
# Do nothing, try again later
|
|
||||||
sleep(RETRY_TIMEOUT)
|
|
||||||
|
|
||||||
# Max try reached, print errors
|
|
||||||
safe_print(f"Error: {project}")
|
|
||||||
for reason in errs:
|
|
||||||
safe_print(f" {reason}")
|
|
||||||
|
|
||||||
return False
|
|
||||||
|
|
||||||
def main():
|
|
||||||
data_file = 'data.json'
|
|
||||||
if len(sys.argv) > 1:
|
|
||||||
data_file = sys.argv[1]
|
|
||||||
|
|
||||||
manager = Manager()
|
|
||||||
data = json.load(open(data_file))
|
|
||||||
sync_data = manager.dict({k: manager.dict(v) for k, v in data.items()})
|
|
||||||
|
|
||||||
with Pool(NUM_THREAD) as pool:
|
|
||||||
all_pass = all(pool.imap(check_project_with_retry, ((k, sync_data) for k in data.keys())))
|
|
||||||
|
|
||||||
with open(data_file, "w", encoding="utf-8") as file:
|
with open(data_file, "w", encoding="utf-8") as file:
|
||||||
json.dump({k: dict(v) for k, v in sync_data.items()}, file, indent=' ')
|
json.dump(data, file, indent='\t')
|
||||||
|
|
||||||
sys.exit(0 if all_pass else 1)
|
sys.exit(exit_code)
|
||||||
|
|
||||||
if __name__ == "__main__":
|
|
||||||
main()
|
|
||||||
|
|
|
@ -3,35 +3,7 @@ Contains Apache class
|
||||||
"""
|
"""
|
||||||
|
|
||||||
from project import Project
|
from project import Project
|
||||||
from shared import CSC_MIRROR
|
|
||||||
import requests
|
|
||||||
|
|
||||||
class Apache(Project):
|
class Apache(Project):
|
||||||
"""Apache class"""
|
"""Apache class"""
|
||||||
|
|
||||||
# Apache's time file has two segments, so we need a special function
|
|
||||||
# Example: 1648323001 rsync-he-fi
|
|
||||||
def check(data, project, current_time):
|
|
||||||
"""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"]
|
|
||||||
|
|
||||||
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
|
|
||||||
try:
|
|
||||||
return get_timestamp_from_apache(upstream) - get_timestamp_from_apache(CSC) < data[project]["out_of_sync_interval"]
|
|
||||||
except ValueError:
|
|
||||||
print("failed to parse apache")
|
|
||||||
return False
|
|
||||||
|
|
||||||
def get_timestamp_from_apache(s: str) -> int:
|
|
||||||
real_time = s.split(" ")[0]
|
|
||||||
return int(real_time)
|
|
||||||
|
|
|
@ -39,6 +39,6 @@ class Artix(Project):
|
||||||
|
|
||||||
if outdated_since is not None:
|
if outdated_since is not None:
|
||||||
data[project]['out_of_sync_since'] = int(outdated_since.timestamp())
|
data[project]['out_of_sync_since'] = int(outdated_since.timestamp())
|
||||||
return current_time - data[project]['out_of_sync_since'] < data[project]['out_of_sync_interval']
|
return (datetime.now() - outdated_since).total_seconds() < data[project]['out_of_sync_interval']
|
||||||
|
|
||||||
return True
|
return True
|
||||||
|
|
|
@ -15,8 +15,6 @@ class CPAN(Project):
|
||||||
def check(data, project, current_time):
|
def check(data, project, current_time):
|
||||||
res_json = requests.get("http://mirrors.cpan.org/cpan-json.txt").json()
|
res_json = requests.get("http://mirrors.cpan.org/cpan-json.txt").json()
|
||||||
for mirror in res_json:
|
for mirror in res_json:
|
||||||
if mirror["url"] == f"{CSC_MIRROR}CPAN/" and mirror["last_status"] == "ok":
|
if mirror["url"] == f"{CSC_MIRROR}CPAN/":
|
||||||
# This is an improvised method: report we're good if CPAN think we are good
|
return current_time - int(mirror["age"]) <= data[project]["out_of_sync_interval"]
|
||||||
# Change this to a more precise method if you find a better way to do it
|
|
||||||
return True
|
|
||||||
return False
|
return False
|
||||||
|
|
|
@ -8,6 +8,7 @@ from shared import CSC_MIRROR
|
||||||
import requests
|
import requests
|
||||||
import datefinder # another date finding library
|
import datefinder # another date finding library
|
||||||
from datetime import timedelta
|
from datetime import timedelta
|
||||||
|
from datetime import datetime
|
||||||
import re
|
import re
|
||||||
import pandas as pd
|
import pandas as pd
|
||||||
|
|
||||||
|
@ -18,9 +19,8 @@ class cran(Project):
|
||||||
page = requests.get(data[project]["upstream"]).text
|
page = requests.get(data[project]["upstream"]).text
|
||||||
indexOfFile = page.find("mirror.csclub.uwaterloo.ca")
|
indexOfFile = page.find("mirror.csclub.uwaterloo.ca")
|
||||||
|
|
||||||
m = re.search(r'(\d+ minutes?)|(\d+ hours?)|(\d+(\.)?\d+ days?)', page[indexOfFile:])
|
m = re.search(r'(\d+ hour)|(\d+ hours)|(\d+(\.)?\d+ days)', page[indexOfFile:]) # solution from: https://stackoverflow.com/questions/21074100/how-to-convert-standard-timedelta-string-to-timedelta-object/21074460
|
||||||
|
|
||||||
duration = pd.to_timedelta(m.group(0))
|
duration = pd.to_timedelta(m.group(0))
|
||||||
data[project]["out_of_sync_since"] = current_time - duration.total_seconds()
|
|
||||||
|
|
||||||
return duration <= pd.to_timedelta(data[project]["out_of_sync_interval"], unit='s')
|
return duration <= pd.to_timedelta(data[project]["out_of_sync_interval"], unit='s')
|
|
@ -7,7 +7,8 @@ from project import Project
|
||||||
from shared import CSC_MIRROR
|
from shared import CSC_MIRROR
|
||||||
import requests
|
import requests
|
||||||
import datefinder # another date finding library
|
import datefinder # another date finding library
|
||||||
from datetime import datetime, timedelta
|
from datetime import timedelta
|
||||||
|
from datetime import datetime
|
||||||
import re
|
import re
|
||||||
import pandas as pd
|
import pandas as pd
|
||||||
|
|
||||||
|
@ -18,10 +19,8 @@ class ctan(Project):
|
||||||
page = requests.get(data[project]["upstream"]).text
|
page = requests.get(data[project]["upstream"]).text
|
||||||
indexOfFile = page.find("mirror.csclub.uwaterloo.ca")
|
indexOfFile = page.find("mirror.csclub.uwaterloo.ca")
|
||||||
|
|
||||||
m = re.search(r'(\d+ minutes?)|(\d+ hours?)|(\d+(\.)?\d+ days?)', page[indexOfFile:])
|
m = re.search(r'(\d+ hour)|(\d+ hours)|(\d+(\.)?\d+ days)', page[indexOfFile:]) # solution from: https://stackoverflow.com/questions/21074100/how-to-convert-standard-timedelta-string-to-timedelta-object/21074460
|
||||||
|
|
||||||
|
|
||||||
duration = pd.to_timedelta(m.group(0))
|
duration = pd.to_timedelta(m.group(0))
|
||||||
data[project]["out_of_sync_since"] = datetime.now() - duration
|
|
||||||
|
|
||||||
return duration <= pd.to_timedelta(data[project]["out_of_sync_interval"], unit='s')
|
return duration <= pd.to_timedelta(data[project]["out_of_sync_interval"], unit='s')
|
||||||
|
|
|
@ -3,37 +3,7 @@ Contains Debian class
|
||||||
"""
|
"""
|
||||||
|
|
||||||
from project import Project
|
from project import Project
|
||||||
from shared import CSC_MIRROR
|
|
||||||
import requests
|
|
||||||
from datetime import datetime
|
|
||||||
import time
|
|
||||||
|
|
||||||
class Debian(Project):
|
class Debian(Project):
|
||||||
"""Debian class"""
|
"""Debian class"""
|
||||||
|
|
||||||
@staticmethod
|
|
||||||
def check(data, project, current_time):
|
|
||||||
csc_url = CSC_MIRROR + data[project]["csc"] + data[project]["file"]
|
|
||||||
upstream_url = data[project]["upstream"] + data[project]["file"]
|
|
||||||
|
|
||||||
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
|
|
||||||
|
|
||||||
CSC_date = datetime.strptime(CSC.partition('\n')[0], "%a %b %d %H:%M:%S UTC %Y")
|
|
||||||
CSC_utc_time = time.mktime(CSC_date.timetuple())
|
|
||||||
|
|
||||||
upstream_date = datetime.strptime(upstream.partition('\n')[0], "%a %b %d %H:%M:%S UTC %Y")
|
|
||||||
upstream_utc_time = time.mktime(upstream_date.timetuple())
|
|
||||||
|
|
||||||
try:
|
|
||||||
return int(upstream_utc_time) - int(CSC_utc_time) < data[project]["out_of_sync_interval"]
|
|
||||||
except ValueError:
|
|
||||||
return False
|
|
||||||
|
|
|
@ -3,40 +3,7 @@ Contains DebianCD class
|
||||||
"""
|
"""
|
||||||
|
|
||||||
from project import Project
|
from project import Project
|
||||||
from shared import CSC_MIRROR
|
|
||||||
import requests
|
|
||||||
from datetime import datetime
|
|
||||||
import time
|
|
||||||
|
|
||||||
class DebianCD(Project):
|
class DebianCD(Project):
|
||||||
"""DebianCD class"""
|
"""DebianCD class"""
|
||||||
|
|
||||||
@staticmethod
|
|
||||||
def check(data, project, current_time):
|
|
||||||
csc_url = CSC_MIRROR + data[project]["csc"] + data[project]["file"]
|
|
||||||
upstream_url = data[project]["upstream"] + data[project]["file"]
|
|
||||||
|
|
||||||
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
|
|
||||||
|
|
||||||
# Date Format Example: Sun 27 Mar 00:20:12 UTC 2022
|
|
||||||
date_format = "%a %d %b %H:%M:%S UTC %Y\n"
|
|
||||||
|
|
||||||
CSC_date = datetime.strptime(CSC, date_format)
|
|
||||||
CSC_utc_time = time.mktime(CSC_date.timetuple())
|
|
||||||
|
|
||||||
upstream_date = datetime.strptime(upstream, date_format)
|
|
||||||
upstream_utc_time = time.mktime(upstream_date.timetuple())
|
|
||||||
|
|
||||||
try:
|
|
||||||
return int(upstream_utc_time) - int(CSC_utc_time) < data[project]["out_of_sync_interval"]
|
|
||||||
except ValueError:
|
|
||||||
return False
|
|
||||||
|
|
|
@ -3,40 +3,7 @@ Contains Fedora class
|
||||||
"""
|
"""
|
||||||
|
|
||||||
from project import Project
|
from project import Project
|
||||||
from shared import CSC_MIRROR
|
|
||||||
import requests
|
|
||||||
from datetime import datetime
|
|
||||||
import time
|
|
||||||
|
|
||||||
class Fedora(Project):
|
class Fedora(Project):
|
||||||
"""Fedora class"""
|
"""Fedora class"""
|
||||||
|
|
||||||
@staticmethod
|
|
||||||
def check(data, project, current_time):
|
|
||||||
csc_url = CSC_MIRROR + data[project]["csc"] + data[project]["file"]
|
|
||||||
upstream_url = data[project]["upstream"] + data[project]["file"]
|
|
||||||
|
|
||||||
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
|
|
||||||
|
|
||||||
# Date example: Fedora-Rawhide-20220725.n.1
|
|
||||||
date_format = "%Y%m%d"
|
|
||||||
|
|
||||||
CSC_date = datetime.strptime(CSC[15:23], "%Y%m%d")
|
|
||||||
CSC_utc_time = time.mktime(CSC_date.timetuple())
|
|
||||||
|
|
||||||
upstream_date = datetime.strptime(upstream[15:23], "%Y%m%d")
|
|
||||||
upstream_utc_time = time.mktime(upstream_date.timetuple())
|
|
||||||
|
|
||||||
try:
|
|
||||||
return upstream_utc_time - CSC_utc_time < data[project]["out_of_sync_interval"]
|
|
||||||
except ValueError:
|
|
||||||
return False
|
|
||||||
|
|
|
@ -3,35 +3,7 @@ Contains FreeBSD class
|
||||||
"""
|
"""
|
||||||
|
|
||||||
from project import Project
|
from project import Project
|
||||||
from shared import CSC_MIRROR
|
|
||||||
import requests
|
|
||||||
|
|
||||||
class FreeBSD(Project):
|
class FreeBSD(Project):
|
||||||
"""FreeBSD class"""
|
"""FreeBSD class"""
|
||||||
|
|
||||||
# FreeBSD's time file has two segments, so we need a special function
|
|
||||||
# Example TIMESTAMP file: 1648308600 2022-03-26T15:30:00 UTC
|
|
||||||
def check(data, project, current_time):
|
|
||||||
"""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"]
|
|
||||||
|
|
||||||
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
|
|
||||||
try:
|
|
||||||
return get_timestamp_for_freebsd(upstream) - get_timestamp_for_freebsd(CSC) < data[project]["out_of_sync_interval"]
|
|
||||||
except ValueError:
|
|
||||||
print("failed to parse apache")
|
|
||||||
return False
|
|
||||||
|
|
||||||
def get_timestamp_for_freebsd(s: str) -> int:
|
|
||||||
real_time = s.split(" ")[0]
|
|
||||||
return int(real_time)
|
|
||||||
|
|
|
@ -3,35 +3,7 @@ Contains GentooDistfiles class
|
||||||
"""
|
"""
|
||||||
|
|
||||||
from project import Project
|
from project import Project
|
||||||
from shared import CSC_MIRROR
|
|
||||||
import requests
|
|
||||||
from datetime import datetime
|
|
||||||
import time
|
|
||||||
|
|
||||||
class GentooDistfiles(Project):
|
class GentooDistfiles(Project):
|
||||||
"""GentooDistfiles class"""
|
"""GentooDistfiles class"""
|
||||||
|
|
||||||
@staticmethod
|
|
||||||
def check(data, project, current_time):
|
|
||||||
csc_url = CSC_MIRROR + data[project]["csc"] + data[project]["file"]
|
|
||||||
upstream_url = data[project]["upstream"] + data[project]["file"]
|
|
||||||
|
|
||||||
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
|
|
||||||
|
|
||||||
CSC_utc_time = CSC[0:11]
|
|
||||||
|
|
||||||
upstream_utc_time = upstream[0:11]
|
|
||||||
|
|
||||||
try:
|
|
||||||
return int(upstream_utc_time) - int(CSC_utc_time) < data[project]["out_of_sync_interval"]
|
|
||||||
except ValueError:
|
|
||||||
return False
|
|
||||||
|
|
|
@ -9,6 +9,7 @@ from project import Project
|
||||||
import requests
|
import requests
|
||||||
import datefinder # another date finding library
|
import datefinder # another date finding library
|
||||||
from datetime import timedelta
|
from datetime import timedelta
|
||||||
|
from datetime import datetime
|
||||||
import re
|
import re
|
||||||
import pandas as pd
|
import pandas as pd
|
||||||
|
|
||||||
|
@ -41,9 +42,8 @@ class GentooPortage(Project):
|
||||||
page = requests.get(data[project]["upstream"]).text
|
page = requests.get(data[project]["upstream"]).text
|
||||||
indexOfFile = page.find("rsync4.ca.gentoo.org")
|
indexOfFile = page.find("rsync4.ca.gentoo.org")
|
||||||
|
|
||||||
m = re.search(r'(\d+ minutes?)|(\d+ hours?)|(\d+(\.)?\d+ days?)', page[indexOfFile:])
|
m = re.search(r'(\d+ hour)|(\d+ hours)|(\d+(\.)?\d+ days)', page[indexOfFile:])
|
||||||
|
|
||||||
duration = pd.to_timedelta(m.group(0))
|
duration = pd.to_timedelta(m.group(0))
|
||||||
data[project]["out_of_sync_since"] = current_time - duration.total_seconds()
|
|
||||||
|
|
||||||
return duration <= pd.to_timedelta(data[project]["out_of_sync_interval"], unit='s')
|
return duration <= pd.to_timedelta(data[project]["out_of_sync_interval"], unit='s')
|
||||||
|
|
|
@ -20,7 +20,6 @@ class manjaro(Project):
|
||||||
|
|
||||||
m = re.search(r'(?P<hours>\d+):(?P<minutes>\d+)', page[indexOfFile:]) # solution from: https://stackoverflow.com/questions/21074100/how-to-convert-standard-timedelta-string-to-timedelta-object/21074460
|
m = re.search(r'(?P<hours>\d+):(?P<minutes>\d+)', page[indexOfFile:]) # solution from: https://stackoverflow.com/questions/21074100/how-to-convert-standard-timedelta-string-to-timedelta-object/21074460
|
||||||
duration = timedelta(**{key: float(val) for key, val in m.groupdict().items()})
|
duration = timedelta(**{key: float(val) for key, val in m.groupdict().items()})
|
||||||
data[project]["out_of_sync_since"] = current_time - duration.total_seconds()
|
|
||||||
|
|
||||||
return duration <= pd.to_timedelta(data[project]["out_of_sync_interval"], unit='s')
|
return duration <= pd.to_timedelta(data[project]["out_of_sync_interval"], unit='s')
|
||||||
|
|
||||||
|
|
|
@ -8,6 +8,7 @@ from shared import CSC_MIRROR
|
||||||
import requests
|
import requests
|
||||||
import datefinder # another date finding library
|
import datefinder # another date finding library
|
||||||
from datetime import timedelta
|
from datetime import timedelta
|
||||||
|
from datetime import datetime
|
||||||
import re
|
import re
|
||||||
import pandas as pd
|
import pandas as pd
|
||||||
|
|
||||||
|
@ -18,9 +19,8 @@ class mxlinux(Project):
|
||||||
page = requests.get(data[project]["upstream"]).text
|
page = requests.get(data[project]["upstream"]).text
|
||||||
indexOfFile = page.find("mirror.csclub.uwaterloo.ca")
|
indexOfFile = page.find("mirror.csclub.uwaterloo.ca")
|
||||||
|
|
||||||
m = re.search(r'(\d+ minutes?)|(\d+ hours?)|(\d+(\.)?\d+ days?)', page[indexOfFile:])
|
m = re.search(r'(\d+ hour)|(\d+ hours)|(\d+(\.)?\d+ days)', page[indexOfFile:]) # solution from: https://stackoverflow.com/questions/21074100/how-to-convert-standard-timedelta-string-to-timedelta-object/21074460
|
||||||
|
|
||||||
duration = pd.to_timedelta(m.group(0))
|
duration = pd.to_timedelta(m.group(0))
|
||||||
data[project]["out_of_sync_since"] = current_time - duration.total_seconds()
|
|
||||||
|
|
||||||
return duration <= pd.to_timedelta(data[project]["out_of_sync_interval"], unit='s')
|
return duration <= pd.to_timedelta(data[project]["out_of_sync_interval"], unit='s')
|
|
@ -8,6 +8,7 @@ from shared import CSC_MIRROR
|
||||||
import requests
|
import requests
|
||||||
import datefinder # another date finding library
|
import datefinder # another date finding library
|
||||||
from datetime import timedelta
|
from datetime import timedelta
|
||||||
|
from datetime import datetime
|
||||||
import re
|
import re
|
||||||
import pandas as pd
|
import pandas as pd
|
||||||
|
|
||||||
|
@ -19,7 +20,7 @@ class mxlinux_iso(Project):
|
||||||
page = requests.get(data[project]["upstream"]).text
|
page = requests.get(data[project]["upstream"]).text
|
||||||
indexOfFile = page.find("mirror.csclub.uwaterloo.ca")
|
indexOfFile = page.find("mirror.csclub.uwaterloo.ca")
|
||||||
|
|
||||||
m = re.search(r'(\d+ minutes?)|(\d+ hours?)|(\d+(\.)?\d+ days?)', page[indexOfFile:])
|
m = re.search(r'(\d+ hour)|(\d+ hours)|(\d+(\.)?\d+ days)', page[indexOfFile:]) # solution from: https://stackoverflow.com/questions/21074100/how-to-convert-standard-timedelta-string-to-timedelta-object/21074460
|
||||||
|
|
||||||
duration = pd.to_timedelta(m.group(0))
|
duration = pd.to_timedelta(m.group(0))
|
||||||
|
|
||||||
|
|
|
@ -3,31 +3,6 @@ Contains nongnu class
|
||||||
"""
|
"""
|
||||||
|
|
||||||
from project import Project
|
from project import Project
|
||||||
from shared import CSC_MIRROR
|
|
||||||
import requests
|
|
||||||
from datetime import datetime
|
|
||||||
import time
|
|
||||||
|
|
||||||
class nongnu(Project):
|
class nongnu(Project):
|
||||||
"""nongnu class"""
|
"""nongnu class"""
|
||||||
|
|
||||||
@staticmethod
|
|
||||||
def check(data, project, current_time):
|
|
||||||
csc_url = CSC_MIRROR + data[project]["csc"] + data[project]["file"]
|
|
||||||
upstream_url = data[project]["upstream"] + data[project]["file"]
|
|
||||||
|
|
||||||
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
|
|
||||||
|
|
||||||
try:
|
|
||||||
return int(upstream.partition('\n')[0]) - int(CSC.partition('\n')[0]) < data[project]["out_of_sync_interval"]
|
|
||||||
except ValueError:
|
|
||||||
return False
|
|
||||||
|
|
|
@ -23,8 +23,7 @@ class raspberrypi(Project):
|
||||||
for i in s.find_all("a"): # for a href directories
|
for i in s.find_all("a"): # for a href directories
|
||||||
href = i.attrs['href']
|
href = i.attrs['href']
|
||||||
|
|
||||||
# The raspberry pi server doesn't use a relative path to the parent directory
|
if href.endswith("/") and href != "../" and href != "/":
|
||||||
if href.endswith("/") and href != "../" and href != "/" and href != "/debian/":
|
|
||||||
site_next = site+href+"Release"
|
site_next = site+href+"Release"
|
||||||
|
|
||||||
if site_next not in urls:
|
if site_next not in urls:
|
||||||
|
@ -45,7 +44,6 @@ class raspberrypi(Project):
|
||||||
cls.scrape(urls1, csc_url)
|
cls.scrape(urls1, csc_url)
|
||||||
cls.scrape(urls2, upstream_url)
|
cls.scrape(urls2, upstream_url)
|
||||||
|
|
||||||
|
|
||||||
if (len(urls1) != len(urls2)):
|
if (len(urls1) != len(urls2)):
|
||||||
return False
|
return False
|
||||||
urls1.sort()
|
urls1.sort()
|
||||||
|
|
|
@ -34,7 +34,7 @@ class slackware(Project):
|
||||||
hrefs2 = [i.attrs['href'] for i in s2.find_all("a")]
|
hrefs2 = [i.attrs['href'] for i in s2.find_all("a")]
|
||||||
|
|
||||||
for href in hrefs1: # for a href directories
|
for href in hrefs1: # for a href directories
|
||||||
if href.endswith("/") and href != "../" and href != "/" and not href.startswith("/") and not re.match(r'slackware-([1-7]|8\.0).*', href) and href != "slackware-iso/" and href != "slackware-current/" and href != "slackware-pre-1.0-beta/" and href != "unsupported/" and not href.startswith("http"):
|
if href.endswith("/") and href != "../" and href != "/" and not href.startswith("/") and not re.match(r'slackware-([1-7]|8\.0).*', href) and href != "slackware-iso/" and href != "slackware-current/" and href != "slackware-pre-1.0-beta/" and href != "unsupported/":
|
||||||
# print(href)
|
# print(href)
|
||||||
if href not in hrefs2:
|
if href not in hrefs2:
|
||||||
return False
|
return False
|
||||||
|
|
|
@ -5,7 +5,6 @@ Contains ubuntu class
|
||||||
import os
|
import os
|
||||||
from project import Project
|
from project import Project
|
||||||
from shared import CSC_MIRROR
|
from shared import CSC_MIRROR
|
||||||
from shared import NUM_UBUNTU_RELEASES
|
|
||||||
import requests
|
import requests
|
||||||
import datefinder # another date finding library
|
import datefinder # another date finding library
|
||||||
from datetime import timedelta
|
from datetime import timedelta
|
||||||
|
@ -18,4 +17,4 @@ class ubuntu(Project):
|
||||||
@staticmethod
|
@staticmethod
|
||||||
def check(data, project, current_time):
|
def check(data, project, current_time):
|
||||||
page = requests.get(data[project]["upstream"]).text
|
page = requests.get(data[project]["upstream"]).text
|
||||||
return page.count("Up to date") == NUM_UBUNTU_RELEASES
|
return page.count("Up to date") == 21
|
||||||
|
|
|
@ -5,7 +5,6 @@ Contains ubuntu_releases class
|
||||||
import os
|
import os
|
||||||
from project import Project
|
from project import Project
|
||||||
from shared import CSC_MIRROR
|
from shared import CSC_MIRROR
|
||||||
from shared import NUM_UBUNTU_RELEASES
|
|
||||||
import requests
|
import requests
|
||||||
import datefinder # another date finding library
|
import datefinder # another date finding library
|
||||||
from datetime import timedelta
|
from datetime import timedelta
|
||||||
|
@ -18,4 +17,9 @@ class ubuntu_releases(Project):
|
||||||
@staticmethod
|
@staticmethod
|
||||||
def check(data, project, current_time):
|
def check(data, project, current_time):
|
||||||
page = requests.get(data[project]["upstream"]).text
|
page = requests.get(data[project]["upstream"]).text
|
||||||
return page.count("Up to date") == NUM_UBUNTU_RELEASES
|
indexOfFile = page.find("last verified")
|
||||||
|
matches = list(datefinder.find_dates(page[indexOfFile:]))
|
||||||
|
date = matches[0] # date is of type datetime.datetime
|
||||||
|
return(pd.to_datetime(current_time, unit='s') - date.replace(tzinfo=None) <= pd.to_timedelta(data[project]["out_of_sync_interval"], unit='s'))
|
||||||
|
|
||||||
|
# https://launchpad.net/ubuntu/+mirror/mirror.csclub.uwaterloo.ca-release
|
Loading…
Reference in New Issue