Remove resource leaks from tests

Close subprocesses within tests to fix resource leak warning.

Signed-off-by: Will Beason <willbeason@gmail.com>

wikiq

@@ -97,6 +97,7 @@ class WikiqPage():
         #         3          A               B            True
         #         4          A               A           False
         # Post-loop                          A          Always
+        collapsed_revs = 0
         for i, rev in enumerate(self.mwpage):
             # never yield the first time
             if i == 0:
@@ -491,7 +492,7 @@ class WikiqParser:
         # Construct dump file iterator
         dump = WikiqIterator(self.input_file, collapse_user=self.collapse_user)
 
-        # extract list of namspaces
+        # extract list of namespaces
         self.namespaces = {ns.name: ns.id for ns in dump.mwiterator.site_info.namespaces}
 
         page_count = 0
@@ -703,7 +704,6 @@ class WikiqParser:
         line = rev_data.to_tsv_row()
         print(line, file=self.output_file)
 
-
 def open_input_file(input_filename):
     if re.match(r'.*\.7z$', input_filename):
         cmd = ["7za", "x", "-so", input_filename, "*.xml"]
@@ -711,14 +711,13 @@ def open_input_file(input_filename):
         cmd = ["zcat", input_filename]
     elif re.match(r'.*\.bz2$', input_filename):
         cmd = ["bzcat", "-dk", input_filename]
+    else:
+        raise ValueError("Unrecognized file type: %s" % input_filename)
 
     try:
-        input_file = Popen(cmd, stdout=PIPE).stdout
+        return Popen(cmd, stdout=PIPE).stdout
     except NameError:
-        input_file = open(input_filename, 'r')
-
-    return input_file
-
+        return open(input_filename, 'r')
 
 def get_output_filename(input_filename, parquet=False):
     output_filename = re.sub(r'\.(7z|gz|bz2)?$', '', input_filename)