A few minor corrections

Author: JustAnotherArchivist

wpull/application/options.py

@@ -1119,7 +1119,7 @@ def _add_warc_args(self):
             '--warc-split-meta',
             action='store_true',
             default=False,
-            help=_('split the log WARC along with the data WARC'),
+            help=_('when used with --warc-max-size, split the meta WARC as well, not just the data WARC'),
         )
 
     def _add_recursive_args(self):

wpull/url.py

@@ -130,7 +130,6 @@ def parse(cls, url, default_scheme='http', encoding='utf-8'):
             return None
 
         url = url.strip()
-        # TODO: \r\n\t should be removed from anywhere in the URL.
         if frozenset(url) & C0_CONTROL_SET:
             raise ValueError('URL contains control codes: {}'.format(ascii(url)))
 

wpull/warc/recorder.py

@@ -218,8 +218,7 @@ def _flush_log(self, closing=False):
             )
             log_record.set_common_fields('resource', 'text/plain')
 
-            log_record.fields['WARC-Target-URI'] = \
-                'urn:X-wpull:log'
+            log_record.fields['WARC-Target-URI'] = 'urn:X-wpull:log'
 
             if self._params.max_size is not None: # Only create a separate meta WARC if max_size is set
                 self._start_new_warc_file(meta=True)

wpull/warc/recorder_test.py

@@ -336,8 +336,9 @@ def test_warc_recorder_max_size_split_meta(self):
             params=WARCRecorderParams(
                 compress=False,
                 extra_fields=[('Extra-field', 'my_extra_field')],
-                cdx=True, max_size=1,
-                split_meta = True,
+                cdx=True,
+                max_size=1,
+                split_meta=True,
             )
         )