Fix OSError when saving CSS files whose URL has a query string

mincss/main.py

@@ -3,11 +3,24 @@
 import io
 import os
 import time
+from urllib.parse import urlparse
 
 from mincss import __version__
 from .processor import Processor
 
 
+def filename_from_href(href):
+    """Return a filesystem-safe filename derived from a URL's href.
+
+    Using the raw last path segment of a URL (including any query string)
+    as a filename fails on Windows with OSError: [Errno 22] Invalid
+    argument, since characters like '?' and '=' are not valid in Windows
+    filenames. This drops the query string and fragment, keeping only the
+    URL's path basename.
+    """
+    return os.path.basename(urlparse(href).path)
+
+
 def run(args):
     if args.version:
         print(__version__)
@@ -37,10 +50,10 @@ def run(args):
         os.mkdir(output_dir)
     for link in p.links:
         print('FOR', link.href)
-        orig_name = link.href.split('/')[-1]
+        orig_name = filename_from_href(link.href)
         with io.open(os.path.join(output_dir, orig_name), 'w') as f:
             f.write(link.after)
-        before_name = 'before_' + link.href.split('/')[-1]
+        before_name = 'before_' + orig_name
         with io.open(os.path.join(output_dir, before_name), 'w') as f:
             f.write(link.before)
         print('Files written to', output_dir)

tests/test_main.py

@@ -0,0 +1,41 @@
+"""
+Regression test for GH issue #53 (peterbe/mincss):
+
+    OSError: [Errno 22] Invalid argument: './output\\banner-styles.css?v=...'
+
+main.run() derived the output filename for a downloaded CSS link
+directly from `link.href.split('/')[-1]`, which includes the URL's
+query string verbatim. Characters like '?' and '=' are not valid in
+Windows filenames, so writing that file raised OSError (Windows) or
+produced a surprising/wrong filename (POSIX, where '?' is legal but
+clearly not intended to be part of a filename).
+
+filename_from_href() now parses the URL properly and uses only the
+path's basename, dropping the query string and fragment.
+"""
+
+import unittest
+
+from mincss.main import filename_from_href
+
+
+class FilenameFromHrefTest(unittest.TestCase):
+    def test_strips_query_string(self):
+        href = "http://example.com/static/css/banner-styles.css?v=1516052760.0"
+        self.assertEqual(filename_from_href(href), "banner-styles.css")
+
+    def test_strips_query_string_and_fragment(self):
+        href = "http://example.com/style.css?v=1#section"
+        self.assertEqual(filename_from_href(href), "style.css")
+
+    def test_plain_url_without_query_string_unaffected(self):
+        href = "http://example.com/static/css/plain.css"
+        self.assertEqual(filename_from_href(href), "plain.css")
+
+    def test_relative_path_url(self):
+        href = "/static/css/relative.css?v=2"
+        self.assertEqual(filename_from_href(href), "relative.css")
+
+
+if __name__ == "__main__":
+    unittest.main()