Improvements to syncutil and repo-cp

* check overwrite and bail out right away if object exists and should
  not be overwritten
* improve logging configuration and handling
* ignore whitespace when reading pids from file

eulfedora/syncutil.py

@@ -69,6 +69,13 @@ def sync_object(src_obj, dest_repo, export_context='migrate',
     # NOTE: currently exceptions are expected to be handled by the
     # calling method; see repo-cp script for an example
 
+    # if overwrite is not requested, check first and bail out
+    dest_obj = dest_repo.get_object(src_obj.pid)
+    if not overwrite and dest_obj.exists:
+        logger.info('%s exists in destination repo and no overwrite; skipping',
+                    src_obj.pid)
+        return False
+
     if show_progress and progressbar:
         # calculate rough estimate of object size
         size_estimate = estimate_object_size(src_obj,
@@ -121,14 +128,9 @@ def data(self):
             export_data = (re.sub(checksum_re, '', chunk)
                            for chunk in export_data)
 
-
-    dest_obj = dest_repo.get_object(src_obj.pid)
-    if dest_obj.exists:
-        if overwrite:
-            dest_repo.purge_object(src_obj.pid)
-        else:
-            # exception ?
-            return False
+    if overwrite and dest_obj.exists:
+        print 'overwriting'
+        dest_repo.purge_object(src_obj.pid)
 
     result = dest_repo.ingest(export_data)
     if pbar:
@@ -462,8 +464,6 @@ def encoded_datastream(self):
                     yield decoded_content
 
 
-
-
 def binarycontent_sections(chunk):
     '''Split a chunk of data into sections by start and end binary
     content tags.'''
@@ -515,6 +515,7 @@ def estimate_object_size(obj, archive=True):
 
     return size_estimate
 
+
 def base64_size(input_size):
     # from http://stackoverflow.com/questions/1533113/calculate-the-size-to-a-base-64-encoded-message
     adjustment = 3 - (input_size % 3) if (input_size % 3) else 0
@@ -533,6 +534,7 @@ def humanize_file_size(size):
     p = math.floor(math.log(size, 2)/10)
     return "%.2f%s" % (size/math.pow(1024, p), units[int(p)])
 
+
 def endswith_partial(text, partial_str):
     '''Check if the text ends with any partial version of the
     specified string.'''

scripts/repo-cp

@@ -138,7 +138,8 @@ def repo_copy():
             pids = args.pids
         elif args.file:
             with open(args.file) as pidlistfile:
-                pids = pidlistfile.read().splitlines()
+                # allow whitespace on front or end of pid, for convenience
+                pids = [p.strip() for p in pidlistfile.read().splitlines()]
         else:
             print('Specify either one or more pids or a file with a list of pids')
             parser.print_help()
@@ -147,6 +148,7 @@ def repo_copy():
     allow_overwrite = cfg.has_option(args.dest, 'allow_overwrite') and \
         cfg.getboolean(args.dest, 'allow_overwrite')
 
+    # configure logging based on verbosity level requested
     logging.config.dictConfig(get_logging_config(args.verbosity))
 
     for pid in pids:
@@ -156,7 +158,13 @@ def repo_copy():
             if isinstance(src_repo, Airlock):
                 with open(pid, 'rb') as export:
                     result = dest_repo.ingest(export)
-                    print('%s copied' % result)
+                    if result:
+                        print('%s copied' % result)
+                    else:
+                        # false means not copied but no error; currently means
+                        # skipped because object exists and overwrite not allowed
+                        print('%s skipped' % pid)
+
                 continue
 
             src_obj = src_repo.get_object(pid)
@@ -178,7 +186,12 @@ def repo_copy():
                     overwrite=allow_overwrite, show_progress=args.progress,
                     requires_auth=args.requires_auth, omit_checksums=args.omit_checksums,
                     verify=args.verify)
-                print('%s copied' % result)
+                if result:
+                    print('%s copied' % result)
+                else:
+                    # false means not copied but no error; currently means
+                    # skipped because object exists and overwrite not allowed
+                    print('%s skipped' % pid)
         except ChecksumMismatch:
             print('ChecksumMismatch on %s' % pid)
 
@@ -199,12 +212,12 @@ def get_logging_config(level):
         'disable_existing_loggers': True,
         'formatters': {
             'simple': {
-                'format': '%(levelname)s: %(message)s',
+                'format': '%(message)s',
             },
         },
         'handlers': {
             'console': {
-                'level': level,
+                'level': 'DEBUG',
                 'class': 'logging.StreamHandler',
                 'formatter': 'simple'
             },
@@ -218,6 +231,7 @@ def get_logging_config(level):
         }
     }
 
+
 def is_airlock(cfg, name):
     # if a name is not in the config file and is a directory,
     # consider it an airlock