From 961952ed56deeeccdf49f40a517775ca3f622d2e Mon Sep 17 00:00:00 2001
From: brent s <bts@square-r00t.net>
Date: Tue, 16 Jun 2020 15:44:50 -0400
Subject: [PATCH] time to test.

---
 sample.config.xml => example.config.xml |  18 ++-
 repomirror/config.py                    |   2 +-
 repomirror/constants.py                 |  22 ++--
 repomirror/fetcher/_base.py             |   7 +-
 repomirror/fetcher/rsync.py             |  16 +--
 repomirror/logger.py                    |   4 +-
 repomirror/sync.py                      | 152 ++++++++++++++++++------
 repomirror/test.py                      |  22 ----
 reposync.py                             |   4 +-
 9 files changed, 155 insertions(+), 92 deletions(-)
 rename sample.config.xml => example.config.xml (91%)
 delete mode 100755 repomirror/test.py
 mode change 100644 => 100755 reposync.py

diff --git a/sample.config.xml b/example.config.xml
similarity index 91%
rename from sample.config.xml
rename to example.config.xml
index db68d96..73ec7ac 100644
--- a/sample.config.xml
+++ b/example.config.xml
@@ -1,6 +1,7 @@
 <?xml version="1.0" encoding="UTF-8" ?>
 <!--
   This is by default placed in ~/.config/repomirror.xml
+  Remember to replace any necessary special characters (https://stackoverflow.com/a/1091953/733214).
 -->
 <mirror xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
         xmlns="https://git.square-r00t.net/RepoMirror/"
@@ -71,7 +72,7 @@
           <long>delay-updates</long>
           <long>copy-links</long>
           <long>safe-links</long>
-          <long>delete-extended</long>
+          <long>delete-excluded</long>
           <long value=".*">exclude</long>
         <rsyncArgs>
       These arguments should be sane for most, if not all, rsync-driven repository mirroring. The last one (exclude) may
@@ -86,9 +87,9 @@
       <long>archive</long>
       <long>delete-after</long>
       <!--
-        An argument with a value (info=2).
+        An argument with a value (info=progress2).
       -->
-      <long value="2">info</long>
+      <long value="progress2">info</long>
       <!--
         A "short" option (single hyphen).
       -->
@@ -121,11 +122,6 @@
         Required; the *remote* path part of the URI. The leading / is necessary. A trailing one will be assumed.
       -->
       <path>/archlinux/</path>
-      <!--
-        The speed to cap socket bandwidth at (in KiB). Decimals are okay.
-        Only valid for rsync; ignored for FTP. If not provided, the default is to not throttle.
-      -->
-      <bwlimit>7000</bwlimit>
     </upstream>
     <!--
       Multiple upstreams can be specified. They are tried in order specified and if connection fails or times out,
@@ -153,7 +149,9 @@
       <path>/centos/</path>
     </upstream>
     <dest>/srv/repos/arch/.</dest>
-    <lastLocalCheck timeFormat="MICROSECOND_EPOCH">/srv/http/centos.lastcheck</lastLocalCheck>
-    <lastLocalSync timeFormat="UNIX_EPOCH">/srv/repos/arch/lastsync</lastLocalSync>
+    <lastLocalCheck timeFormat="MICROSECOND_EPOCH">/srv/repos/centos/CHECKED</lastLocalCheck>
+    <lastLocalSync timeFormat="UNIX_EPOCH">/srv/repos/centos/TIME</lastLocalSync>
+    <lastRemoteUpdate timeFormat="%a %d %b %H:%M:%S UTC %Y">/timestamp.txt</lastRemoteUpdate>
+    <lastRemoteSync timeFormat="UNIX_EPOCH">/TIME</lastRemoteSync>
   </distro>
 </mirror>
diff --git a/repomirror/config.py b/repomirror/config.py
index bd31f21..b552b83 100644
--- a/repomirror/config.py
+++ b/repomirror/config.py
@@ -86,7 +86,7 @@ class Config(object):
                            '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:
+            with open(self.xml_path, 'rb') as fh:
                 self.raw = fh.read()
         self.xml = None
         self.xsd = None
diff --git a/repomirror/constants.py b/repomirror/constants.py
index bd025e6..97f4beb 100644
--- a/repomirror/constants.py
+++ b/repomirror/constants.py
@@ -1,12 +1,14 @@
 PROTO_DEF_PORTS = {'ftp': 21,
                    'rsync': 873}
-RSYNC_DEF_ARGS = ['recursive',
-                  'times',
-                  'links',
-                  'hard-links',
-                  'delete-after',
-                  'delay-updates',
-                  'copy-links',
-                  'safe-links',
-                  'delete-extended',
-                  'exclude=.*']
+RSYNC_DEF_ARGS = ['--recursive',
+                  '--times',
+                  '--links',
+                  '--hard-links',
+                  '--delete-after',
+                  '--delay-updates',
+                  '--copy-links',
+                  '--safe-links',
+                  '--delete-excluded',
+                  '--exclude=.*']
+# How many days an upstream should have last synced by before it's considered stale.
+DAYS_WARN = 2
diff --git a/repomirror/fetcher/_base.py b/repomirror/fetcher/_base.py
index b70c553..c4094a1 100644
--- a/repomirror/fetcher/_base.py
+++ b/repomirror/fetcher/_base.py
@@ -23,10 +23,13 @@ class BaseFetcher(object):
             os.chown(self.dest, **self.owner)
 
     def check(self):
-        for k, v in self.filechecks['remote']:
+        for k, v in self.filechecks['remote'].items():
             if v:
                 tstmp_raw = self.fetch_content(v.path).decode('utf-8').strip()
-                tstmp = datetime.datetime.strptime(tstmp_raw, v.fmt)
+                if '%s' in v.fmt:
+                    tstmp = datetime.datetime.fromtimestamp(int(tstmp_raw))
+                else:
+                    tstmp = datetime.datetime.strptime(tstmp_raw, v.fmt)
                 self.timestamps[k] = tstmp
         _logger.debug('Updated timestamps: {0}'.format(self.timestamps))
         return(None)
diff --git a/repomirror/fetcher/rsync.py b/repomirror/fetcher/rsync.py
index b3ae612..3ec3b74 100644
--- a/repomirror/fetcher/rsync.py
+++ b/repomirror/fetcher/rsync.py
@@ -32,9 +32,9 @@ class RSync(_base.BaseFetcher):
         super().__init__(domain, port, path, dest, owner = owner, filechecks = filechecks, *args, **kwargs)
         _logger.debug('Instantiated RSync fetcher')
         if rsync_args:
-            self.rsync_args = rsync_args
+            self.rsync_args = rsync_args.args[:]
         else:
-            self.rsync_args = constants.RSYNC_DEF_ARGS
+            self.rsync_args = constants.RSYNC_DEF_ARGS[:]
         _logger.debug('RSync args given: {0}'.format(self.rsync_args))
         if log:
             # Do I want to do this in subprocess + logging module? Or keep this?
@@ -43,7 +43,7 @@ class RSync(_base.BaseFetcher):
             _log_path = None
             for h in _logger.handlers:
                 if isinstance(h, logging.handlers.RotatingFileHandler):
-                    _log_path = h.baseFileName
+                    _log_path = h.baseFilename
                     break
             self.rsync_args.extend(['--verbose',
                                     '--log-file-format="[RSYNC {0}:{1}]:%l:%f%L"'.format(self.domain, self.port),
@@ -61,11 +61,12 @@ class RSync(_base.BaseFetcher):
                    *self.rsync_args,
                    path,
                    dest]
+        _logger.debug('Running command: {0}'.format(' '.join(cmd_str)))
         cmd = subprocess.run(cmd_str,
                              stdout = subprocess.PIPE,
                              stderr = subprocess.PIPE)
-        stdout = cmd.stdout.read().decode('utf-8').strip()
-        stderr = cmd.stderr.read().decode('utf-8').strip()
+        stdout = cmd.stdout.decode('utf-8').strip()
+        stderr = cmd.stderr.decode('utf-8').strip()
         if stdout != '':
             _logger.debug('STDOUT: {0}'.format(stdout))
         if stderr != '' or cmd.returncode != 0:
@@ -81,11 +82,12 @@ class RSync(_base.BaseFetcher):
                    *self.rsync_args,
                    url,
                    tf]
+        _logger.debug('Running command: {0}'.format(' '.join(cmd_str)))
         cmd = subprocess.run(cmd_str,
                              stdout = subprocess.PIPE,
                              stderr = subprocess.PIPE)
-        stdout = cmd.stdout.read().decode('utf-8').strip()
-        stderr = cmd.stderr.read().decode('utf-8').strip()
+        stdout = cmd.stdout.decode('utf-8').strip()
+        stderr = cmd.stderr.decode('utf-8').strip()
         if stdout != '':
             _logger.debug('STDOUT: {0}'.format(stdout))
         if stderr != '' or cmd.returncode != 0:
diff --git a/repomirror/logger.py b/repomirror/logger.py
index 980c7e3..78a990c 100644
--- a/repomirror/logger.py
+++ b/repomirror/logger.py
@@ -12,9 +12,9 @@ except ImportError:
 def preplog(logfile = None):
     if not logfile:
         if os.geteuid() == 0:
-            logfile = '/var/log/repo/main.log'
+            logfile = '/var/log/repo/_main.log'
         else:
-            logfile = '~/.cache/repo.log'
+            logfile = '~/.cache/repomirror.log'
     # Prep the log file.
     logfile = os.path.abspath(os.path.expanduser(logfile))
     os.makedirs(os.path.dirname(logfile), exist_ok = True, mode = 0o0700)
diff --git a/repomirror/sync.py b/repomirror/sync.py
index c243d15..daf7cf0 100644
--- a/repomirror/sync.py
+++ b/repomirror/sync.py
@@ -4,10 +4,14 @@ import pwd
 import grp
 import os
 import socket
+import warnings
+##
+import psutil
 ##
 from . import config
 from . import constants
 from . import fetcher
+from . import logger
 
 
 _logger = logging.getLogger()
@@ -16,10 +20,10 @@ _logger = logging.getLogger()
 def get_owner(owner_xml):
     owner = {}
     user = owner_xml.find('user')
-    if user:
+    if user is not None:
         user = user.text
     group = owner_xml.find('group')
-    if group:
+    if group is not None:
         group = group.text
     if user:
         user_obj = pwd.getpwnam(user)
@@ -60,7 +64,7 @@ class Args(object):
 
 class Mount(object):
     def __init__(self, mpchk_xml):
-        self.path = os.path.abspath(os.path.expanduser(mpchk_xml))
+        self.path = os.path.abspath(os.path.expanduser(mpchk_xml.text))
         self.is_mounted = None
         self._check_mount()
 
@@ -90,21 +94,27 @@ class TimestampFile(object):
         _logger.debug('Set timestamp format string to {0}'.format(self.fmt))
         self.owner_xml = owner_xml
         self.owner = {}
-        if self.owner_xml:
+        if self.owner_xml is not None:
             self.owner = get_owner(self.owner_xml)
         _logger.debug('Owner set is {0}'.format(self.owner))
         self.path = os.path.abspath(os.path.expanduser(ts_xml.text))
         _logger.debug('Path resolved to {0}'.format(self.path))
 
     def read(self, parentdir = None):
+        timestamp = None
         if parentdir:
             path = os.path.join(os.path.abspath(os.path.expanduser(parentdir)),
                                 self.path.lstrip('/'))
         else:
             path = self.path
-        with open(path, 'r') as fh:
-            timestamp = datetime.datetime.strptime(fh.read().strip(), self.fmt)
-        _logger.debug('Read timestamp {0} from {1}'.format(str(timestamp), self.path))
+        if os.path.isfile(path):
+            with open(path, 'r') as fh:
+                ts_raw = fh.read().strip()
+            if '%s' in self.fmt:
+                timestamp = datetime.datetime.fromtimestamp(int(ts_raw))
+            else:
+                timestamp = datetime.datetime.strptime(ts_raw, self.fmt)
+            _logger.debug('Read timestamp {0} from {1}'.format(str(timestamp), self.path))
         return(timestamp)
 
     def write(self):
@@ -136,13 +146,10 @@ class Upstream(object):
         self.filechecks = filechecks
         self.has_new = False
         # These are optional.
-        for i in ('port', 'bwlimit'):
-            e = self.xml.find(i)
-            if e:
-                setattr(self, i, int(e.text))
-            else:
-                setattr(self, i, None)
-        if not getattr(self, 'port'):
+        port = self.xml.find('port')
+        if port is not None:
+            self.port = int(port.text)
+        else:
             self.port = constants.PROTO_DEF_PORTS[self.sync_type]
         self.available = None
         if self.sync_type == 'rsync':
@@ -176,8 +183,8 @@ class Upstream(object):
 class Distro(object):
     def __init__(self, distro_xml):
         self.xml = distro_xml
-        self.name = distro_xml.attrib['name']
-        self.dest = os.path.abspath(os.path.expanduser(distro_xml.find('dest').text))
+        self.name = self.xml.attrib['name']
+        self.dest = os.path.abspath(os.path.expanduser(self.xml.find('dest').text))
         self.mount = Mount(self.xml.find('mountCheck'))
         self.filechecks = {'local': {'check': None,
                                      'sync': None},
@@ -187,20 +194,21 @@ class Distro(object):
         self.rsync_args = None
         self.owner = None
         self.upstreams = []
+        self.lockfile = '/var/run/repomirror/{0}.lck'.format(self.name)
         # These are optional.
         self.owner_xml = self.xml.find('owner')
-        if self.owner_xml:
+        if self.owner_xml is not None:
             self.owner = get_owner(self.owner_xml)
         self.rsync_xml = self.xml.find('rsyncArgs')
-        if self.rsync_xml:
+        if self.rsync_xml is not None:
             self.rsync_args = Args(self.rsync_xml)
         for i in ('Check', 'Sync'):
             e = self.xml.find('lastLocal{0}'.format(i))
-            if e:
+            if e is not None:
                 self.filechecks['local'][i.lower()] = TimestampFile(e)
         for i in ('Sync', 'Update'):
             e = self.xml.find('lastRemote{0}'.format(i))
-            if e:
+            if e is not None:
                 self.filechecks['remote'][i.lower()] = TimestampFile(e)
         for u in self.xml.findall('upstream'):
             self.upstreams.append(Upstream(u,
@@ -210,18 +218,69 @@ class Distro(object):
                                            filechecks = self.filechecks))
 
     def check(self):
-        for k, v in self.filechecks['local']:
+        for k, v in self.filechecks['local'].items():
             if v:
                 tstmp = v.read()
                 self.timestamps[k] = tstmp
         _logger.debug('Updated timestamps: {0}'.format(self.timestamps))
-
-    def sync(self):
-        self.check()
+        local_checks = sorted([i for i in self.timestamps.values() if i])
+        for u in self.upstreams:
+            if not u.available:
+                continue
+            u.fetcher.check()
+            remote_checks = sorted([i for i in u.fetcher.timestamps.values() if i])
+            if not any((local_checks, remote_checks)) or not remote_checks:
+                u.has_new = True
+            else:
+                update = u.fetcher.timestamps.get('update')
+                sync = u.fetcher.timestamps.get('sync')
+                if update:
+                    if local_checks and local_checks[-1] < update:
+                        u.has_new = True
+                    elif not local_checks:
+                        u.has_new = True
+                if sync:
+                    td = datetime.datetime.utcnow() - sync
+                    if td.days > constants.DAYS_WARN:
+                        _logger.warning(('Upstream {0} has not synced for {1}} or more days; this '
+                                         'repository may be out of date.').format(u.fetcher.url, constants.DAYS_WARN))
+                        warnings.warn('Upstream may be out of date')
+        return(None)
+
+    def sync(self):
+        self.check()
+        my_pid = os.getpid()
+        if os.path.isfile(self.lockfile):
+            with open(self.lockfile, 'r') as fh:
+                pid = int(fh.read().strip())
+            if my_pid == pid:  # This logically should not happen, but something might have gone stupid.
+                _logger.warning('Someone call the Ghostbusters because this machine is haunted.')
+                return(False)
+            else:
+                warnmsg = 'The sync process for {0} is locked with file {1} and PID {2}'.format(self.name,
+                                                                                                self.lockfile,
+                                                                                                pid)
+                try:
+                    proc = psutil.Process(pid)
+                    warnmsg += '.'
+                except (psutil.NoSuchProcess, FileNotFoundError, AttributeError):
+                    proc = None
+                    warnmsg += ' but that PID no longer exists.'
+                _logger.warning(warnmsg)
+                if proc:
+                    _logger.warning('PID information: {0}'.format(vars(proc)))
+                warnings.warn(warnmsg)
+                return(False)
+        if not self.mount.is_mounted:
+            _logger.error(('The mountpoint {0} for distro {1} is not mounted; '
+                           'refusing to sync').format(self.mount.path, self.name))
+            return(False)
+        os.makedirs(os.path.dirname(self.lockfile), mode = 0o0755, exist_ok = True)
+        with open(self.lockfile, 'w') as fh:
+            fh.write('{0}\n'.format(str(my_pid)))
         for u in self.upstreams:
             if not u.available:
                 continue
-            u.fetcher.check(self.filechecks['local'])
             if u.has_new:
                 u.sync()
                 if self.filechecks['local']['sync']:
@@ -229,11 +288,20 @@ class Distro(object):
                 break
         if self.filechecks['local']['check']:
             self.filechecks['local']['check'].write()
-        return(None)
+        os.remove(self.lockfile)
+        return(True)
 
 
 class Sync(object):
     def __init__(self, cfg = None, dummy = False, distro = None, logdir = None, *args, **kwargs):
+        if logdir:
+            self.logdir = logdir
+        else:
+            self.logdir = os.path.dirname(logger.filehandler.baseFilename)
+        self._orig_log_old = logger.filehandler.baseFilename
+        self._orig_log = logger.preplog(os.path.join(self.logdir, '_main.log'))
+        logger.filehandler.close()
+        logger.filehandler.baseFilename = self._orig_log
         try:
             _args = dict(locals())
             del(_args['self'])
@@ -243,16 +311,28 @@ class Sync(object):
                 self.distro = distro
             else:
                 self.distro = []
-            self._distro_objs = []
-            self.logdir = logdir
-            self.xml = config.Config(cfg)
-            self._distro_populate()
-        except Exception:
+            self.cfg = config.Config(cfg)
+        except Exception as e:
             _logger.error('FATAL ERROR. Stacktrace follows.', exc_info = True)
-
-    def _distro_populate(self):
-        pass
+            raise e
 
     def sync(self):
-        for d in self._distro_objs:
-            d.sync()
+        if self.distro:
+            for d in self.distro:
+                e = self.cfg.xml.xpath('//distro[@name="{0}"]'.format(d))
+                if e is None:
+                    _logger.error('Could not find specified distro {0}; skipping'.format(d))
+                    continue
+                logger.filehandler.close()
+                logger.filehandler.baseFilename = os.path.join(self.logdir, '{0}.log'.format(e.attrib['name']))
+                distro = Distro(e[0])
+                distro.sync()
+        else:
+            for e in self.cfg.xml.findall('distro'):
+                logger.filehandler.close()
+                logger.filehandler.baseFilename = os.path.join(self.logdir, '{0}.log'.format(e.attrib['name']))
+                distro = Distro(e)
+                distro.sync()
+        logger.filehandler.close()
+        logger.filehandler.baseFilename = self._orig_log
+        return(None)
diff --git a/repomirror/test.py b/repomirror/test.py
deleted file mode 100755
index 82ca827..0000000
--- a/repomirror/test.py
+++ /dev/null
@@ -1,22 +0,0 @@
-#!/usr/bin/env python3
-
-import os
-import shutil
-##
-import logger
-import fetcher
-
-dest = '/tmp/ipxe_ftp'
-path = 'ipxe'
-
-
-def main():
-    if os.path.isdir(dest):
-        shutil.rmtree(dest)
-    f = fetcher.FTP('10.11.12.12', 21, path, dest)
-    f.fetch()
-
-
-if __name__ == '__main__':
-    main()
-
diff --git a/reposync.py b/reposync.py
old mode 100644
new mode 100755
index 3b5ccf1..edf3569
--- a/reposync.py
+++ b/reposync.py
@@ -15,7 +15,7 @@ import repomirror
 if os.geteuid() == 0:
     _def_logdir = '/var/log/repo'
 else:
-    _def_logdir = '~/.cache/logs'
+    _def_logdir = '~/.cache/repologs'
 
 
 def parseArgs():
@@ -24,7 +24,7 @@ def parseArgs():
                       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'))
+                              'Default: ~/.config/repomirror.xml'))
     # args.add_argument('-n', '--dry-run',
     #                   action = 'store_true',
     #                   dest = 'dummy',