diff --git a/scrapy/utils/url.py b/scrapy/utils/url.py index c9abb12d5..c29ed4461 100644 --- a/scrapy/utils/url.py +++ b/scrapy/utils/url.py @@ -85,11 +85,9 @@ def add_http_if_no_scheme(url): def guess_scheme(url): - """Add an URL scheme if missing: file:// for filepath-like input or http:// otherwise.""" - parts = urlparse(url) - if parts.scheme: - return url - # Note: this does not match Windows filepath + """Add an URL scheme if missing: file:// for filepath-like input or + http:// otherwise.""" + # POSIX path if re.match(r'''^ # start with... ( \. # ...a single dot, @@ -99,7 +97,10 @@ def guess_scheme(url): )? # optional match of ".", ".." or ".blabla" / # at least one "/" for a file path, . # and something after the "/" - ''', parts.path, flags=re.VERBOSE): + ''', url, flags=re.VERBOSE): + return any_to_uri(url) + # Windows drive-letter path + elif re.match(r'''^[a-z]:\\''', url, flags=re.IGNORECASE): return any_to_uri(url) else: return add_http_if_no_scheme(url) diff --git a/tests/test_command_shell.py b/tests/test_command_shell.py index d664b6ade..acf8e9f71 100644 --- a/tests/test_command_shell.py +++ b/tests/test_command_shell.py @@ -94,7 +94,7 @@ class ShellTest(ProcessTest, SiteTest, unittest.TestCase): @defer.inlineCallbacks def test_local_file(self): - filepath = join(tests_datadir, 'test_site/index.html') + filepath = join(tests_datadir, 'test_site', 'index.html') _, out, _ = yield self.execute([filepath, '-c', 'item']) assert b'{}' in out