bloomreach#88 (multiple) --exclude for local src files in dsync.

README.md

@@ -143,6 +143,7 @@ Synchronize the contents of two directories. The directory can either be local o
 *   -f/--force: override existing file instead of showing error message.
 *   -n/--dry-run: emulate the operation without real sync.
 *   --delete-removed: delete files not in source directory.
+*   --exclude: exclude files with the given pattern(s). Only works for local files to be uploaded to s3 directory. This option can be provided several times.
 
 #### `s4cmd sync [source] [target]`
 
@@ -263,6 +264,9 @@ before given parameter.
 Condition on files where their last modified dates are
 after given parameter.
 
+##### `--exclude`
+(local) Filenames, matching the given pattern (https://docs.python.org/3.4/library/fnmatch.html) will not be synced to s3 directory. You may provide this param several times.
+
 
 ## S3 API Pass-through Options
 

s4cmd.py

@@ -994,8 +994,14 @@ def dsync_files(self, source, target):
       for src, dest in sync_list:
         pool.download(src, dest)
     elif not src_s3_url and dst_s3_url:
+      expats = []
+      if self.opt.exclude:
+        expats = self.opt.exclude
+      info("excluded patterns: %s", expats)
+
       for src, dest in sync_list:
-        pool.upload(src, dest)
+        if not any(fnmatch.fnmatch(src, expat) for expat in expats):
+          pool.upload(src, dest)
     elif src_s3_url and dst_s3_url:
       for src, dest in sync_list:
         pool.copy(src, dest)
@@ -1815,6 +1821,18 @@ def check_dict(self, opt, value):
   TYPE_CHECKER['datetime'] = check_datetime
   TYPE_CHECKER['dict'] = check_dict
 
+  # enable multiple option values (e.g. --exclude '*.thumbnail.jpg' --exclude '*.layout.jpg')
+  ACTIONS = optparse.Option.ACTIONS + ("extend",)
+  STORE_ACTIONS = optparse.Option.STORE_ACTIONS + ("extend",)
+  TYPED_ACTIONS = optparse.Option.TYPED_ACTIONS + ("extend",)
+  ALWAYS_TYPED_ACTIONS = optparse.Option.ALWAYS_TYPED_ACTIONS + ("extend",)
+
+  def take_action(self, action, dest, opt, value, values, parser):
+    if action == "extend":
+      values.ensure_value(dest, []).append(value)
+    else:
+      optparse.Option.take_action(self, action, dest, opt, value, values, parser)
+
 def main():
   try:
       if not sys.argv[0]: sys.argv[0] = ''  # Workaround for running with optparse from egg
@@ -1910,6 +1928,9 @@ def main():
           '--last-modified-after',
           help='Condition on files where their last modified dates are after given parameter.',
           type='datetime', default=None)
+      parser.add_option(
+          '--exclude', help='Exclude all files from src dsync (file-)list matching the specified pattern',
+          dest='exclude', action="extend", type='string', default=None)
 
       # Extra S3 API arguments
       BotoClient.add_options(parser)