Refactor mirrors (#4274)

This commit is contained in:
Daniel Girtler 2026-03-21 17:57:33 +11:00 committed by GitHub
parent 8d6c56ca2b
commit 1764b4971d
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
8 changed files with 176 additions and 172 deletions

View File

@ -11,7 +11,8 @@ from archinstall.lib.interactions.general_conf import add_number_of_parallel_dow
from archinstall.lib.interactions.system_conf import select_kernel, select_swap
from archinstall.lib.locale.locale_menu import LocaleMenu
from archinstall.lib.menu.abstract_menu import AbstractMenu, SpecialMenuKey
from archinstall.lib.mirrors import MirrorListHandler, MirrorMenu
from archinstall.lib.mirror.mirror_handler import MirrorListHandler
from archinstall.lib.mirror.mirror_menu import MirrorMenu
from archinstall.lib.models.application import ApplicationConfiguration, ZramConfiguration
from archinstall.lib.models.authentication import AuthenticationConfiguration
from archinstall.lib.models.bootloader import Bootloader, BootloaderConfiguration

View File

@ -30,7 +30,7 @@ from archinstall.lib.exceptions import DiskError, HardwareIncompatibilityError,
from archinstall.lib.hardware import SysInfo
from archinstall.lib.locale.utils import verify_keyboard_layout, verify_x11_keyboard_layout
from archinstall.lib.luks import Luks2, unlock_luks2_dev
from archinstall.lib.mirrors import MirrorListHandler
from archinstall.lib.mirror.mirror_handler import MirrorListHandler
from archinstall.lib.models.application import ZramAlgorithm
from archinstall.lib.models.bootloader import Bootloader
from archinstall.lib.models.device import (

View File

View File

@ -0,0 +1,163 @@
import time
import urllib
from pathlib import Path
from archinstall.lib.models import MirrorRegion
from archinstall.lib.models.mirrors import MirrorStatusEntryV3, MirrorStatusListV3
from archinstall.lib.networking import fetch_data_from_url
from archinstall.lib.output import debug, info
class MirrorListHandler:
def __init__(
self,
local_mirrorlist: Path = Path('/etc/pacman.d/mirrorlist'),
offline: bool = False,
verbose: bool = False,
) -> None:
self._local_mirrorlist = local_mirrorlist
self._status_mappings: dict[str, list[MirrorStatusEntryV3]] | None = None
self._fetched_remote: bool = False
self.offline = offline
self.verbose = verbose
def _mappings(self) -> dict[str, list[MirrorStatusEntryV3]]:
if self._status_mappings is None:
self.load_mirrors()
assert self._status_mappings is not None
return self._status_mappings
def get_mirror_regions(self) -> list[MirrorRegion]:
available_mirrors = []
mappings = self._mappings()
for region_name, status_entry in mappings.items():
urls = [entry.server_url for entry in status_entry]
region = MirrorRegion(region_name, urls)
available_mirrors.append(region)
return available_mirrors
def load_mirrors(self) -> None:
if self.offline:
self._fetched_remote = False
self.load_local_mirrors()
else:
self._fetched_remote = self.load_remote_mirrors()
debug(f'load mirrors: {self._fetched_remote}')
if not self._fetched_remote:
self.load_local_mirrors()
def load_remote_mirrors(self) -> bool:
url = 'https://archlinux.org/mirrors/status/json/'
attempts = 3
for attempt_nr in range(attempts):
try:
mirrorlist = fetch_data_from_url(url)
self._status_mappings = self._parse_remote_mirror_list(mirrorlist)
return True
except Exception as e:
debug(f'Error while fetching mirror list: {e}')
time.sleep(attempt_nr + 1)
debug('Unable to fetch mirror list remotely, falling back to local mirror list')
return False
def load_local_mirrors(self) -> None:
with self._local_mirrorlist.open('r') as fp:
mirrorlist = fp.read()
self._status_mappings = self._parse_local_mirrors(mirrorlist)
def get_status_by_region(self, region: str, speed_sort: bool) -> list[MirrorStatusEntryV3]:
mappings = self._mappings()
region_list = mappings[region]
# Only sort if we have remote mirror data with score/speed info
# Local mirrors lack this data and can be modified manually before-hand
# Or reflector potentially ran already
if self._fetched_remote and speed_sort:
info('Sorting your selected mirror list based on the speed between you and the individual mirrors (this might take a while)')
# Sort by speed descending (higher is better in bitrate form core.db download)
return sorted(region_list, key=lambda mirror: -mirror.speed)
# just return as-is without sorting?
return region_list
def _parse_remote_mirror_list(self, mirrorlist: str) -> dict[str, list[MirrorStatusEntryV3]]:
context = {'verbose': self.verbose}
mirror_status = MirrorStatusListV3.model_validate_json(mirrorlist, context=context)
sorting_placeholder: dict[str, list[MirrorStatusEntryV3]] = {}
for mirror in mirror_status.urls:
# We filter out mirrors that have bad criteria values
if any(
[
mirror.active is False, # Disabled by mirror-list admins
mirror.last_sync is None, # Has not synced recently
# mirror.score (error rate) over time reported from backend:
# https://github.com/archlinux/archweb/blob/31333d3516c91db9a2f2d12260bd61656c011fd1/mirrors/utils.py#L111C22-L111C66
(mirror.score is None or mirror.score >= 100),
]
):
continue
if mirror.country == '':
# TODO: This should be removed once RFC!29 is merged and completed
# Until then, there are mirrors which lacks data in the backend
# and there is no way of knowing where they're located.
# So we have to assume world-wide
mirror.country = 'Worldwide'
if mirror.url.startswith('http'):
sorting_placeholder.setdefault(mirror.country, []).append(mirror)
sorted_by_regions: dict[str, list[MirrorStatusEntryV3]] = dict(
{region: unsorted_mirrors for region, unsorted_mirrors in sorted(sorting_placeholder.items(), key=lambda item: item[0])}
)
return sorted_by_regions
def _parse_local_mirrors(self, mirrorlist: str) -> dict[str, list[MirrorStatusEntryV3]]:
lines = mirrorlist.splitlines()
# remove empty lines
# lines = [line for line in lines if line]
mirror_list: dict[str, list[MirrorStatusEntryV3]] = {}
current_region = ''
for line in lines:
line = line.strip()
if line.startswith('## '):
current_region = line.replace('## ', '').strip()
mirror_list.setdefault(current_region, [])
if line.startswith('Server = '):
if not current_region:
current_region = 'Local'
mirror_list.setdefault(current_region, [])
url = line.removeprefix('Server = ')
mirror_entry = MirrorStatusEntryV3(
url=url.removesuffix('$repo/os/$arch'),
protocol=urllib.parse.urlparse(url).scheme,
active=True,
country=current_region or 'Worldwide',
# The following values are normally populated by
# archlinux.org mirror-list endpoint, and can't be known
# from just the local mirror-list file.
country_code='WW',
isos=True,
ipv4=True,
ipv6=True,
details='Locally defined mirror',
)
mirror_list[current_region].append(mirror_entry)
return mirror_list

View File

@ -1,24 +1,19 @@
import time
import urllib.parse
from pathlib import Path
from typing import override
from archinstall.lib.menu.abstract_menu import AbstractSubMenu
from archinstall.lib.menu.helpers import Input, Loading, Selection
from archinstall.lib.menu.list_manager import ListManager
from archinstall.lib.mirror.mirror_handler import MirrorListHandler
from archinstall.lib.models.mirrors import (
CustomRepository,
CustomServer,
MirrorConfiguration,
MirrorRegion,
MirrorStatusEntryV3,
MirrorStatusListV3,
SignCheck,
SignOption,
)
from archinstall.lib.models.packages import Repository
from archinstall.lib.networking import fetch_data_from_url
from archinstall.lib.output import FormattedOutput, debug, info
from archinstall.lib.output import FormattedOutput
from archinstall.lib.translationhandler import tr
from archinstall.tui.ui.menu_item import MenuItem, MenuItemGroup
from archinstall.tui.ui.result import ResultType
@ -206,161 +201,6 @@ class CustomMirrorServersList(ListManager[CustomServer]):
return None
class MirrorListHandler:
def __init__(
self,
local_mirrorlist: Path = Path('/etc/pacman.d/mirrorlist'),
offline: bool = False,
verbose: bool = False,
) -> None:
self._local_mirrorlist = local_mirrorlist
self._status_mappings: dict[str, list[MirrorStatusEntryV3]] | None = None
self._fetched_remote: bool = False
self.offline = offline
self.verbose = verbose
def _mappings(self) -> dict[str, list[MirrorStatusEntryV3]]:
if self._status_mappings is None:
self.load_mirrors()
assert self._status_mappings is not None
return self._status_mappings
def get_mirror_regions(self) -> list[MirrorRegion]:
available_mirrors = []
mappings = self._mappings()
for region_name, status_entry in mappings.items():
urls = [entry.server_url for entry in status_entry]
region = MirrorRegion(region_name, urls)
available_mirrors.append(region)
return available_mirrors
def load_mirrors(self) -> None:
if self.offline:
self._fetched_remote = False
self.load_local_mirrors()
else:
self._fetched_remote = self.load_remote_mirrors()
debug(f'load mirrors: {self._fetched_remote}')
if not self._fetched_remote:
self.load_local_mirrors()
def load_remote_mirrors(self) -> bool:
url = 'https://archlinux.org/mirrors/status/json/'
attempts = 3
for attempt_nr in range(attempts):
try:
mirrorlist = fetch_data_from_url(url)
self._status_mappings = self._parse_remote_mirror_list(mirrorlist)
return True
except Exception as e:
debug(f'Error while fetching mirror list: {e}')
time.sleep(attempt_nr + 1)
debug('Unable to fetch mirror list remotely, falling back to local mirror list')
return False
def load_local_mirrors(self) -> None:
with self._local_mirrorlist.open('r') as fp:
mirrorlist = fp.read()
self._status_mappings = self._parse_local_mirrors(mirrorlist)
def get_status_by_region(self, region: str, speed_sort: bool) -> list[MirrorStatusEntryV3]:
mappings = self._mappings()
region_list = mappings[region]
# Only sort if we have remote mirror data with score/speed info
# Local mirrors lack this data and can be modified manually before-hand
# Or reflector potentially ran already
if self._fetched_remote and speed_sort:
info('Sorting your selected mirror list based on the speed between you and the individual mirrors (this might take a while)')
# Sort by speed descending (higher is better in bitrate form core.db download)
return sorted(region_list, key=lambda mirror: -mirror.speed)
# just return as-is without sorting?
return region_list
def _parse_remote_mirror_list(self, mirrorlist: str) -> dict[str, list[MirrorStatusEntryV3]]:
context = {'verbose': self.verbose}
mirror_status = MirrorStatusListV3.model_validate_json(mirrorlist, context=context)
sorting_placeholder: dict[str, list[MirrorStatusEntryV3]] = {}
for mirror in mirror_status.urls:
# We filter out mirrors that have bad criteria values
if any(
[
mirror.active is False, # Disabled by mirror-list admins
mirror.last_sync is None, # Has not synced recently
# mirror.score (error rate) over time reported from backend:
# https://github.com/archlinux/archweb/blob/31333d3516c91db9a2f2d12260bd61656c011fd1/mirrors/utils.py#L111C22-L111C66
(mirror.score is None or mirror.score >= 100),
]
):
continue
if mirror.country == '':
# TODO: This should be removed once RFC!29 is merged and completed
# Until then, there are mirrors which lacks data in the backend
# and there is no way of knowing where they're located.
# So we have to assume world-wide
mirror.country = 'Worldwide'
if mirror.url.startswith('http'):
sorting_placeholder.setdefault(mirror.country, []).append(mirror)
sorted_by_regions: dict[str, list[MirrorStatusEntryV3]] = dict(
{region: unsorted_mirrors for region, unsorted_mirrors in sorted(sorting_placeholder.items(), key=lambda item: item[0])}
)
return sorted_by_regions
def _parse_local_mirrors(self, mirrorlist: str) -> dict[str, list[MirrorStatusEntryV3]]:
lines = mirrorlist.splitlines()
# remove empty lines
# lines = [line for line in lines if line]
mirror_list: dict[str, list[MirrorStatusEntryV3]] = {}
current_region = ''
for line in lines:
line = line.strip()
if line.startswith('## '):
current_region = line.replace('## ', '').strip()
mirror_list.setdefault(current_region, [])
if line.startswith('Server = '):
if not current_region:
current_region = 'Local'
mirror_list.setdefault(current_region, [])
url = line.removeprefix('Server = ')
mirror_entry = MirrorStatusEntryV3(
url=url.removesuffix('$repo/os/$arch'),
protocol=urllib.parse.urlparse(url).scheme,
active=True,
country=current_region or 'Worldwide',
# The following values are normally populated by
# archlinux.org mirror-list endpoint, and can't be known
# from just the local mirror-list file.
country_code='WW',
isos=True,
ipv4=True,
ipv6=True,
details='Locally defined mirror',
)
mirror_list[current_region].append(mirror_entry)
return mirror_list
class MirrorMenu(AbstractSubMenu[MirrorConfiguration]):
def __init__(
self,

View File

@ -16,7 +16,7 @@ from archinstall.lib.networking import DownloadTimer, ping
from archinstall.lib.output import debug
if TYPE_CHECKING:
from archinstall.lib.mirrors import MirrorListHandler
from archinstall.lib.mirror.mirror_handler import MirrorListHandler
class MirrorStatusEntryV3(BaseModel):
@ -64,17 +64,17 @@ class MirrorStatusEntryV3(BaseModel):
assert timer.time is not None
self._speed = size / timer.time
debug(f' speed: {self._speed} ({int(self._speed / 1024 / 1024 * 100) / 100}MiB/s)')
debug(f' speed: {self._speed} ({int(self._speed / 1024 / 1024 * 100) / 100}MiB/s)')
# Do not retry error
except urllib.error.URLError as error:
debug(f' speed: <undetermined> ({error}), skip')
debug(f' speed: <undetermined> ({error}), skip')
self._speed = 0
# Do retry error
except (http.client.IncompleteRead, ConnectionResetError) as error:
debug(f' speed: <undetermined> ({error}), retry')
debug(f' speed: <undetermined> ({error}), retry')
# Catch all
except Exception as error:
debug(f' speed: <undetermined> ({error}), skip')
debug(f' speed: <undetermined> ({error}), skip')
self._speed = 0
retry += 1
@ -104,7 +104,7 @@ class MirrorStatusEntryV3(BaseModel):
def validate_score(cls, value: float) -> int | None:
if value is not None:
value = round(value)
debug(f' score: {value}')
debug(f' score: {value}')
return value

View File

@ -12,7 +12,7 @@ from archinstall.lib.global_menu import GlobalMenu
from archinstall.lib.installer import Installer, accessibility_tools_in_use, run_custom_user_commands
from archinstall.lib.interactions.general_conf import PostInstallationAction, select_post_installation
from archinstall.lib.menu.util import delayed_warning
from archinstall.lib.mirrors import MirrorListHandler
from archinstall.lib.mirror.mirror_handler import MirrorListHandler
from archinstall.lib.models import Bootloader
from archinstall.lib.models.device import DiskLayoutType, EncryptionType
from archinstall.lib.models.users import User

View File

@ -1,6 +1,6 @@
from pathlib import Path
from archinstall.lib.mirrors import MirrorListHandler
from archinstall.lib.mirror.mirror_handler import MirrorListHandler
def test_mirrorlist_no_country(mirrorlist_no_country_fixture: Path) -> None: