From 9890b671f2e7656a183492bc1cddf3da0c4850f4 Mon Sep 17 00:00:00 2001
From: Rob Gibson <nosbig@gmail.com>
Date: Sat, 21 May 2022 06:13:08 +0000
Subject: [PATCH] Added XSD Support for Pruning Old Backups

---
 .gitignore |  9 ++++++
 backup.py  | 81 ++++++++++++++++++++++++++++++++++++++++++++++++++++++
 config.xml | 34 +++++++++++++++++++++++
 config.xsd | 14 ++++++++++
 4 files changed, 138 insertions(+)
 create mode 100644 config.xml

diff --git a/.gitignore b/.gitignore
index ff3b78a..fd69eda 100644
--- a/.gitignore
+++ b/.gitignore
@@ -25,3 +25,12 @@ __pycache__/
 *.sqlite3
 *.deb
 .idea/
+files/*
+repo/*
+repo\:testrepo/*
+config
+hints.21
+index.21
+integrity.21
+nonce
+README
\ No newline at end of file
diff --git a/backup.py b/backup.py
index 7c6b934..f14dc79 100755
--- a/backup.py
+++ b/backup.py
@@ -20,6 +20,7 @@ import re
 import subprocess
 import sys
 import tempfile
+from xmlrpc.client import Server
 # TODO: virtual env?
 from lxml import etree  # A lot safer and easier to use than the stdlib xml module.
 try:
@@ -452,6 +453,79 @@ class Backup(object):
                     objPrinter(r, indent = 3)
                     print()
         return()
+    
+    def prune(self):
+        # TODO: support "--strip-components N"?
+        self.logger.info('START: prune')
+        for server in self.repos:
+            _env = os.environ.copy()
+            if self.repos[server]['remote'].lower()[0] in ('1', 't'):
+                _env['BORG_RSH'] = self.repos[server].get('rsh', None)
+            _env['LANG'] = 'en_US.UTF-8'
+            _env['LC_CTYPE'] = 'en_US.UTF-8'
+            _user = self.repos[server].get('user', pwd.getpwuid(os.geteuid()).pw_name)
+            for repo in self.repos[server]['repos']:
+                _loc_env = _env.copy()
+                if 'password' not in repo:
+                    print('Password not supplied for {0}:{1}.'.format(server, repo['name']))
+                    _loc_env['BORG_PASSPHRASE'] = getpass.getpass('Password (will NOT echo back): ')
+                else:
+                    _loc_env['BORG_PASSPHRASE'] = repo['password']
+                self.logger.info('[{0}]: BEGIN PRUNE: {1}'.format(server, repo['name']))
+                # This is where we actually do the thing.
+                _cmd = [self.borgbin,
+                        '--log-json',
+                        '--{0}'.format(self.args['loglevel']),
+                        'prune',
+                        '--stats']
+                if self.repos[server]['keepYearly'][0].isnumeric() and int(self.repos[server]['keepYearly'][0]) > 0:
+                    _cmd.extend(['--keep-yearly', self.repos[server]['keepYearly'].lower()[0]])
+                if self.repos[server]['keepMonthly'][0].isnumeric() and int(self.repos[server]['keepMonthly'][0]) > 0:
+                    _cmd.extend(['--keep-monthly', self.repos[server]['keepMonthly'].lower()[0]])
+                if self.repos[server]['keepWeekly'][0].isnumeric() and int(self.repos[server]['keepWeekly'][0]) > 0:
+                    _cmd.extend(['--keep-weekly', self.repos[server]['keepWeekly'].lower()[0]])
+                if self.repos[server]['keepDaily'][0].isnumeric() and int(self.repos[server]['keepDaily'][0]) > 0:
+                    _cmd.extend(['--keep-daily', self.repos[server]['keepDaily'].lower()[0]])
+                if self.repos[server]['keepHourly'][0].isnumeric() and int(self.repos[server]['keepHourly'][0]) > 0:
+                    _cmd.extend(['--keep-hourly', self.repos[server]['keepHourly'].lower()[0]])
+                if self.repos[server]['keepMinutely'][0].isnumeric() and int(self.repos[server]['keepMinutely'][0]) > 0:
+                    _cmd.extend(['--keep-minutely', self.repos[server]['keepMinutely'].lower()[0]])
+                if self.repos[server]['keepSecondly'][0].isnumeric() and int(self.repos[server]['keepSecondly'][0]) > 0:
+                    _cmd.extend(['--keep-secondly', self.repos[server]['keepSecondly'].lower()[0]])
+                if self.repos[server]['remote'].lower()[0] in ('1', 't'):
+                    repo_tgt = '{0}@{1}'.format(_user, server)
+                else:
+                    repo_tgt = os.path.abspath(os.path.expanduser(server))
+                _cmd.append('{0}:{1}'.format(repo_tgt,
+                                                  repo['name']))
+                self.logger.debug('VARS: {0}'.format(vars()))
+                # We don't use self.cmdExec() here because we want to explicitly
+                # pass the env and format the log line differently.
+                self.logger.debug('[{0}]: Running command: {1}'.format(repo['name'],
+                                                                       ' '.join(_cmd)))
+                if not self.args['dryrun']:
+                    _out = subprocess.run(_cmd,
+                                          env = _loc_env,
+                                          stdout = subprocess.PIPE,
+                                          stderr = subprocess.PIPE)
+                    _stdout = _out.stdout.decode('utf-8').strip()
+                    _stderr = _out.stderr.decode('utf-8').strip()
+                    _returncode = _out.returncode
+                    self.logger.debug('[{0}]: (RESULT) {1}'.format(repo['name'], _stdout))
+                    self.logger.debug('[{0}]: STDERR: ({2})\n{1}'.format(repo['name'],
+                                                                         _stderr,
+                                                                         ' '.join(
+                                                                                 _cmd)))
+                    if _returncode != 0:
+                        self.logger.error(
+                                '[{0}]: FAILED: {1}'.format(repo['name'], ' '.join(_cmd)))
+                    if _stderr != '' and self.cron and _returncode != 0:
+                        self.logger.warning('Command {0} failed: {1}'.format(' '.join(_cmd),
+                                                                             _stderr))
+                    del (_loc_env['BORG_PASSPHRASE'])
+                self.logger.info('[{0}]: END PRUNE'.format(repo['name']))
+        self.logger.info('END: prune')
+        return()
 
     def printer(self):
         # TODO: better alignment. https://stackoverflow.com/a/5676884
@@ -684,6 +758,11 @@ def parseArgs():
                                      parents = [commonargs,
                                                 remoteargs,
                                                 fileargs])
+    pruneargs = subparsers.add_parser('prune',
+                                     help = ('Prune backups to schedule.'),
+                                     parents = [commonargs,
+                                                remoteargs,
+                                                fileargs])
     cvrtargs = subparsers.add_parser('convert',
                                      help = ('Convert the legacy JSON format to the new XML format and quit'))
     ### OPERATION-SPECIFIC OPTIONS ###
@@ -837,6 +916,8 @@ def main():
         bak.createRepo()
     elif args['oper'] == 'restore':
         bak.restore()
+    elif args['oper'] == 'prune':
+        bak.prune()
     return()
 
 
diff --git a/config.xml b/config.xml
new file mode 100644
index 0000000..c7023e4
--- /dev/null
+++ b/config.xml
@@ -0,0 +1,34 @@
+<?xml version="1.0" encoding="UTF-8" ?>
+<borg xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
+      xmlns="http://git.square-r00t.net/BorgExtend/tree/storage/backups/borg/"
+      xsi:schemaLocation="http://git.square-r00t.net/BorgExtend/plain/config.xsd"
+      xmlns:xi="http://www.w3.org/2001/XInclude">
+    <!-- You can have multiple server elements, but each one *MUST* have a unique "target" attribute. -->
+    <!-- "target" = either the local filesystem path (absolute or relative to execution) or the remote host
+         "remote" = 1/true if "target" is a remote host or 0/false if it's a local filepath
+         "rsh" = (remote host only) the ssh command to use. The default is given below.
+         "user" = (remote host only) the ssh user to use.
+         "dummy" = a boolean; if you need to create a "dummy" server, set this to "true".
+                   It will *not* be parsed or executed upon.
+                   It won't even be created by an init operation or show up in a repolist operation. -->
+    <server target="repo" remote="false" keepSecondly="3" keepMinutely="2">
+        <!-- You can (and probably will) have multiple repos for each server. -->
+        <!-- "name" = the repositoriy name.
+             "password" = the repository's password for the key. If not specified, you will be prompted
+                          to enter it interactively and securely.
+             "dummy" = see server[@dummy] explanation.
+             "compression" = see https://borgbackup.readthedocs.io/en/stable/usage/create.html (-C option) -->
+        <repo name="testrepo" password="SuperSecretPassword" compression="lzma,9">
+            <!-- Each path entry is a path to back up.
+                 See https://borgbackup.readthedocs.io/en/stable/usage/create.html
+                 Note that globbing, etc. is *disabled* for security reasons, so you will need to specify all
+                 directories explicitly. -->
+		<path>/home/nosbig/git-repos/nosbig/BorgExtend/files</path>
+            <!-- Each exclude entry should be a subdirectory of a <path> (otherwise it wouldn't match, obviously). -->
+		<exclude>/home/nosbig/git-repos/nosbig/BorgExtend/files/b.txt</exclude>
+        </repo>
+        <!-- You can also include other snippets. Either absolute paths, paths relative to your backup.xml file,
+             or a URL. -->
+        <xi:include href="sample.config.snippet.xml"/>
+    </server>
+</borg>
diff --git a/config.xsd b/config.xsd
index 25d0130..c1cdeed 100644
--- a/config.xsd
+++ b/config.xsd
@@ -118,6 +118,20 @@
                         <!-- This specifies if a server is a "dummy" configuration.
                              Useful for testing and placeholder. -->
                         <xs:attribute name="dummy" type="xs:boolean" use="optional" default="false"/>
+                        <!-- Used to define the number of yearly backups to keep. -->
+                        <xs:attribute name="keepYearly" type="xs:int" use="optional" />
+                        <!-- Used to define the number of monthly backups to keep. -->
+                        <xs:attribute name="keepMonthly" type="xs:int" use="optional" />
+                        <!-- Used to define the number of weekly backups to keep. -->
+                        <xs:attribute name="keepWeekly" type="xs:int" use="optional" />
+                        <!-- Used to define the number of daily backups to keep. -->
+                        <xs:attribute name="keepDaily" type="xs:int" use="optional" />
+                        <!-- Used to define the number of hourly backups to keep. -->
+                        <xs:attribute name="keepHourly" type="xs:int" use="optional" />
+                        <!-- Used to define the number of minutely backups to keep. -->
+                        <xs:attribute name="keepMinutely" type="xs:int" use="optional" />
+                        <!-- Used to define the number of secondly backups to keep. -->
+                        <xs:attribute name="keepSecondly" type="xs:int" use="optional" />
                     </xs:complexType>
                 </xs:element>
                 <!-- END SERVER -->