Reinstate --purge option for grok-pull

With very large repo collections it makes sense to run purge operations
infrequently, to avoid having to walk the entire tree after each
manifest check.

Signed-off-by: Konstantin Ryabitsev <konstantin@linuxfoundation.org>

4 files changed, 18 insertions, 4 deletions

diff --git a/grokmirror.conf b/grokmirror.conf
index 99496f2..968557d 100644
--- a/grokmirror.conf
+++ b/grokmirror.conf
@@ -117,7 +117,10 @@ projectslist_symlinks = no
 post_update_hook =
 #
 # Should we purge repositories that are not present in the remote
-# manifest?  See also: purgeprotect.
+# manifest? If set to "no" this can be overridden via the -p flag to
+# grok-pull (useful if you have a very large collection of repos
+# and don't want to walk the entire tree on each manifest run).
+# See also: purgeprotect.
 purge = yes
 #
 # This prevents catastrophic mirror purges when our upstream gives us a
diff --git a/grokmirror/pull.py b/grokmirror/pull.py
index fcd8667..6d90ea2 100755
--- a/grokmirror/pull.py
+++ b/grokmirror/pull.py
@@ -845,7 +845,7 @@ def fill_todo_from_manifest(config, active_actions, nomtime=False, forcepurge=Fa
         # Finally, clone ourselves.
         actions.append((gitdir, 'init'))
 
-    if config['pull'].get('purge') == 'yes':
+    if config['pull'].getboolean('purge', False):
         to_purge = set()
         found_repos = 0
         for founddir in grokmirror.find_all_gitdirs(toplevel, exclude_objstore=True):
@@ -1188,6 +1188,9 @@ def parse_args():
     op.add_option('-n', '--no-mtime-check', dest='nomtime',
                   action='store_true', default=False,
                   help='Run without checking manifest mtime.')
+    op.add_option('-p', '--purge', dest='purge',
+                  action='store_true', default=False,
+                  help='Remove any git trees that are no longer in manifest.')
     op.add_option('', '--force-purge', dest='forcepurge',
                   action='store_true', default=False,
                   help='Force purge despite significant repo deletions.')
@@ -1205,7 +1208,7 @@ def parse_args():
     return opts, args
 
 
-def grok_pull(cfgfile, verbose=False, nomtime=False, forcepurge=False, runonce=False):
+def grok_pull(cfgfile, verbose=False, nomtime=False, purge=False, forcepurge=False, runonce=False):
     global logger
 
     config = grokmirror.load_config_file(cfgfile)
@@ -1218,6 +1221,10 @@ def grok_pull(cfgfile, verbose=False, nomtime=False, forcepurge=False, runonce=F
     else:
         loglevel = logging.INFO
 
+    if purge:
+        # Override the pull.purge setting
+        config['pull']['purge'] = 'yes'
+
     logger = grokmirror.init_logger('pull', logfile, loglevel, verbose)
 
     return pull_mirror(config, nomtime, forcepurge, runonce)
@@ -1227,7 +1234,7 @@ def command():
     opts, args = parse_args()
 
     retval = grok_pull(
-        opts.config, opts.verbose, opts.nomtime, opts.forcepurge, opts.runonce)
+        opts.config, opts.verbose, opts.nomtime, opts.purge, opts.forcepurge, opts.runonce)
 
     sys.exit(retval)
 
diff --git a/man/grok-pull.1 b/man/grok-pull.1
index 2c91a68..792ed88 100644
--- a/man/grok-pull.1
+++ b/man/grok-pull.1
@@ -83,6 +83,9 @@ Run continuously (no effect if refresh is not set)
 .BI \-c \ CONFIG\fP,\fB \ \-\-config\fB= CONFIG
 Location of the configuration file
 .TP
+.B \-p\fP,\fB  \-\-purge
+Remove any git trees that are no longer in manifest.
+.TP
 .B \-\-force\-purge
 Force purge operation despite significant repo deletions
 .UNINDENT
diff --git a/man/grok-pull.1.rst b/man/grok-pull.1.rst
index 6920bf5..f146b64 100644
--- a/man/grok-pull.1.rst
+++ b/man/grok-pull.1.rst
@@ -49,6 +49,7 @@ OPTIONS
   -o, --continuous      Run continuously (no effect if refresh is not set)
   -c CONFIG, --config=CONFIG
                         Location of the configuration file
+  -p, --purge           Remove any git trees that are no longer in manifest.
   --force-purge         Force purge operation despite significant repo deletions
 
 EXAMPLES