adding some stuff for repomirror

This commit is contained in:
brent s. 2020-06-13 03:46:49 -04:00
parent 19fddfaacf
commit 5526111743
Signed by: bts
GPG Key ID: 8C004C2F93481F6B
6 changed files with 387 additions and 0 deletions

18
repomirror/__init__.py Normal file
View File

@ -0,0 +1,18 @@
from . import logger
##
import logging
##
from . import config
from . import constants


_logger = logging.getLogger()


class Sync(object):
def __init__(self, cfg = None, dummy = False, distro = None, logdir = None, *args, **kwargs):
_args = dict(locals())
del(_args['self'])
_logger.debug('Sync class instantiated with args: {0}'.format(_args))
self.cfg = config.Config(cfg)

174
repomirror/config.py Normal file
View File

@ -0,0 +1,174 @@
import copy
import datetime
import os
import logging
import re
import shutil
##
import requests
import requests.auth
from lxml import etree


_logger = logging.getLogger()


def create_default_cfg():
# Create a stripped sample config.
ws_re = re.compile(r'^\s*$')
cur_dir = os.path.dirname(os.path.abspath(os.path.expanduser(__file__)))
samplexml = os.path.abspath(os.path.join(cur_dir, '..', 'example.config.xml'))
with open(samplexml, 'rb') as fh:
xml = etree.fromstring(fh.read())
# Create a stripped sample config.
# First we strip comments (and fix the ensuing whitespace).
# etree has a .canonicalize(), but it chokes on a default namespace.
# https://bugs.launchpad.net/lxml/+bug/1869455
# So everything we do is kind of a hack.
# for c in xml.xpath("//comment()"):
# parent = c.getparent()
# parent.remove(c)
xmlstr = etree.tostring(xml, with_comments = False, method = 'c14n', pretty_print = True).decode('utf-8')
newstr = []
for line in xmlstr.splitlines():
r = ws_re.search(line)
if not r:
newstr.append(line.strip())
xml = etree.fromstring(''.join(newstr).encode('utf-8'))
# Remove text and attr text.
xpathq = "descendant-or-self::*[namespace-uri()!='']"
for e in xml.xpath(xpathq):
if e.tag == '{{{0}}}mirror'.format(xml.nsmap[None]):
continue
if e.text is not None and e.text.strip() != '':
e.text = ''
for k, v in e.attrib.items():
if v is not None:
e.attrib[k] = ''
# Remove multiple children of same type to simplify.
for e in xml.xpath(xpathq):
if e.tag == '{{{0}}}mirror'.format(xml.nsmap[None]):
continue
parent = e.getparent()
try:
for idx, child in enumerate(parent.findall(e.tag)):
if idx == 0:
continue
parent.remove(child)
except AttributeError:
pass
# And add a comment pointing them to the fully commented config.
xml.insert(0, etree.Comment(('\n Please reference the fully commented example.config.xml found either '
'at:\n '
' * {0}\n * https://git.square-r00t.net/RepoMirror/tree/'
'example.config.xml\n and then configure this according to those '
'instructions.\n ').format(samplexml)))
return(etree.tostring(xml,
pretty_print = True,
with_comments = True,
with_tail = True,
encoding = 'UTF-8',
xml_declaration = True))


class Config(object):
default_xsd = 'http://schema.xml.r00t2.io/projects/repomirror.xsd'
default_xml_path = '~/.config/repomirror.xml'

def __init__(self, xml_path, *args, **kwargs):
if not xml_path:
xml_path = self.default_xml_path
self.xml_path = os.path.abspath(os.path.expanduser(xml_path))
if not os.path.isfile(self.xml_path):
with open(self.xml_path, 'wb') as fh:
fh.write(create_default_cfg())
_logger.error(('{0} does not exist so a sample configuration file has been created in its place. '
'Be sure to configure it appropriately.').format(self.default_xml_path))
raise ValueError('Config does not exist')
else:
with open(xml_path, 'rb') as fh:
self.raw = fh.read()
self.xml = None
self.xsd = None
self.ns_xml = None
self.tree = None
self.ns_tree = None
self.defaults_parser = None
self.parse_xml()
_logger.info('Instantiated {0}.'.format(type(self).__name__))

def get_xsd(self):
raw_xsd = None
base_url = None
xsi = self.xml.nsmap.get('xsi', 'http://www.w3.org/2001/XMLSchema-instance')
schemaLocation = '{{{0}}}schemaLocation'.format(xsi)
schemaURL = self.xml.attrib.get(schemaLocation, self.default_xsd)
split_url = schemaURL.split()
if len(split_url) == 2: # a properly defined schemaLocation
schemaURL = split_url[1]
else:
schemaURL = split_url[0] # a LAZY schemaLocation
if schemaURL.startswith('file://'):
schemaURL = re.sub(r'^file://', r'', schemaURL)
with open(schemaURL, 'rb') as fh:
raw_xsd = fh.read()
base_url = os.path.dirname(schemaURL) + '/'
else:
req = requests.get(schemaURL)
if not req.ok:
raise RuntimeError('Could not download XSD')
raw_xsd = req.content
base_url = os.path.split(req.url)[0] + '/' # This makes me feel dirty.
self.xsd = etree.XMLSchema(etree.XML(raw_xsd, base_url = base_url))
return(None)

def parse_xml(self):
self.parse_raw()
self.get_xsd()
self.populate_defaults()
self.validate()
return(None)

def parse_raw(self, parser = None):
self.xml = etree.fromstring(self.raw, parser = parser)
self.ns_xml = etree.fromstring(self.raw, parser = parser)
self.tree = self.xml.getroottree()
self.ns_tree = self.ns_xml.getroottree()
self.tree.xinclude()
self.ns_tree.xinclude()
self.strip_ns()
return(None)

def populate_defaults(self):
if not self.xsd:
self.get_xsd()
if not self.defaults_parser:
self.defaults_parser = etree.XMLParser(schema = self.xsd, attribute_defaults = True)
self.parse_raw(parser = self.defaults_parser)
return(None)

def remove_defaults(self):
self.parse_raw()
return(None)

def strip_ns(self, obj = None):
# https://stackoverflow.com/questions/30232031/how-can-i-strip-namespaces-out-of-an-lxml-tree/30233635#30233635
xpathq = "descendant-or-self::*[namespace-uri()!='']"
if not obj:
for x in (self.tree, self.xml):
for e in x.xpath(xpathq):
e.tag = etree.QName(e).localname
elif isinstance(obj, (etree._Element, etree._ElementTree)):
obj = copy.deepcopy(obj)
for e in obj.xpath(xpathq):
e.tag = etree.QName(e).localname
return(obj)
else:
raise ValueError('Did not know how to parse obj parameter')
return(None)

def validate(self):
if not self.xsd:
self.get_xsd()
self.xsd.assertValid(self.ns_tree)
return(None)

2
repomirror/constants.py Normal file
View File

@ -0,0 +1,2 @@
PROTO_DEF_PORTS = {'ftp': 21,
'rsync': 873}

57
repomirror/logger.py Normal file
View File

@ -0,0 +1,57 @@
import logging
import logging.handlers
import os
try:
# https://www.freedesktop.org/software/systemd/python-systemd/journal.html#journalhandler-class
from systemd import journal
_has_journald = True
except ImportError:
_has_journald = False


def preplog(logfile = None):
if not logfile:
if os.geteuid() == 0:
logfile = '/var/log/repo/main.log'
else:
logfile = '~/.cache/repo.log'
# Prep the log file.
logfile = os.path.abspath(os.path.expanduser(logfile))
os.makedirs(os.path.dirname(logfile), exist_ok = True, mode = 0o0700)
if not os.path.isfile(logfile):
with open(logfile, 'w') as fh:
fh.write('')
os.chmod(logfile, 0o0600)
return(logfile)


# And set up logging.
_cfg_args = {'handlers': [],
'level': logging.DEBUG}
if _has_journald:
# There were some weird changes somewhere along the line.
try:
# But it's *probably* this one.
h = journal.JournalHandler()
except AttributeError:
h = journal.JournaldLogHandler()
# Systemd includes times, so we don't need to.
h.setFormatter(logging.Formatter(style = '{',
fmt = ('{name}:{levelname}:{name}:{filename}:'
'{funcName}:{lineno}: {message}')))
_cfg_args['handlers'].append(h)

filehandler = logging.handlers.RotatingFileHandler(preplog(),
encoding = 'utf8',
# Disable rotating for now.
# maxBytes = 50000000000,
# backupCount = 30
)
filehandler.setFormatter(logging.Formatter(style = '{',
fmt = ('{asctime}:'
'{levelname}:{name}:{filename}:'
'{funcName}:{lineno}: {message}')))
_cfg_args['handlers'].append(filehandler)
logging.basicConfig(**_cfg_args)
logger = logging.getLogger('Repo Mirror')
logger.info('Logging initialized.')

55
reposync.py Normal file
View File

@ -0,0 +1,55 @@
#!/usr/bin/env python3

import argparse
import os
import pathlib
import sys
##
##
_cur_file = pathlib.Path(os.path.abspath(os.path.expanduser(__file__))).resolve()
_cur_path = os.path.dirname(_cur_file)
sys.path.insert(1, _cur_path)
import repomirror


if os.geteuid() == 0:
_def_logdir = '/var/log/repo'
else:
_def_logdir = '~/.cache/logs'


def parseArgs():
args = argparse.ArgumentParser(description = 'Sync repositories for various distributions to local paths')
args.add_argument('-c', '--config',
default = '~/.config/repomirror.xml',
dest = 'cfg',
help = ('The path to the config file. If it does not exist, a bare version will be created. '
'Default: ~/.config/repomirror.xmlost'))
args.add_argument('-n', '--dry-run',
action = 'store_true',
dest = 'dummy',
help = ('If specified, do not actually sync anything (other than timestamp files if '
'applicable to determine logic); do not actually sync any repositories'))
args.add_argument('-d', '--distro',
dest = 'distro',
action = 'append',
help = ('If specified, only sync the specified distro in the config file (otherwise sync all '
'specified). May be given multiple times'))
args.add_argument('-l', '--logdir',
default = _def_logdir,
dest = 'logdir',
help = ('The path to the directory where logs should be written. The actual log files will be '
'named after their respective distro names in the config file. '
'Defailt: {0}'.format(_def_logdir)))
return(args)


def main():
args = parseArgs().parse_args()
r = repomirror.Sync()

return(None)


if __name__ == '__main__':
main()

81
sample.config.xml Normal file
View File

@ -0,0 +1,81 @@
<?xml version="1.0" encoding="UTF-8" ?>
<!--
This is by default placed in ~/.config/repomirror.xml
-->
<mirror xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xmlns="https://git.square-r00t.net/RepoMirror/"
xsi:schemaLocation="https://git.square-r00t.net/RepoMirror/ http://schema.xml.r00t2.io/projects/repomirror.xsd">
<distro name="arch">
<!--
The local path to where the hierarchy/files should be synced to.
-->
<dest>/srv/repos/arch/.</dest>
<!--
The local file to update with a timestamp with the last time we checked for updates.
-->
<lastLocalCheck>/srv/http/arch.lastcheck</lastLocalCheck>
<!--
The file to update with a timestamp with the last time we synced from our upstream.
-->
<lastLocalSync>/srv/http/arch.lastsync</lastLocalSync>
<!--
The path to a file on the upstream(s) that gives a time when it last updated.
-->
<lastRemoteUpdate>/lastupdate</lastRemoteUpdate>
<!--
The path to a file on the upstream(s) that gives a time when it last synced from its upstream.
-->
<lastRemoteSync>/lastsync</lastRemoteSync>
<!--
The path that must be currently mounted for sync to proceed.
-->
<mountCheck>/</mountCheck>
<!--
The speed to cap socket bandwidth at (in KiB). Decimals are okay.
-->
<bwlimit>7000</bwlimit>
<upstream>
<!--
The following example uses "rsync://arch.mirror.constant.com/archlinux/"
(https://www.archlinux.org/mirrors/constant.com/1008/)
-->
<!--
One of:
* rsync
* ftp
-->
<syncType>rsync</syncType>
<!--
ONLY the domain goes here.
-->
<domain>arch.mirror.constant.com</domain>
<!--
If not specified,the protocol's default port will be used.
-->
<port>873</port>
<!--
The *remote* path part of the URI. The leading / is necessary. A trailing one will be assumed.
-->
<path>/archlinux/</path>
</upstream>
<!--
Multiple upstreams can be specified. They are tried in order specified and if connection fails or times out,
the next one will be tried until no more specified upstreams exist.
-->
<upstream>
<syncType>rsync</syncType>
<domain>arch.mirror.square-r00t.net</domain>
<path>/arch/</path>
</upstream>
<upstream>
<!--
Only passive (PASV) mode for FTP is supported.
-->
<syncType>ftp</syncType>
<domain>sub.domain.tld</domain>
<port>21</port>
<path>/distros/archlinux/</path>
</upstream>
</distro>
<distro name="centos"/>
</mirror>