gsutil: Pass proxy information via Boto options.

gsutil.py

Index: gsutil.py
diff --git a/gsutil.py b/gsutil.py
index 6578d7844171202eb787bf1414c4a371e15a75a0..8b338df680c7ae3ddb8bd1d05459b7f501eeea1c 100755
--- a/gsutil.py
+++ b/gsutil.py
@@ -15,6 +15,7 @@ import shutil
 import subprocess
 import sys
 import urllib2
+import urlparse
 import zipfile
 
 
@@ -121,12 +122,42 @@ def ensure_gsutil(version, target):
   return gsutil_bin
 
 
+def get_boto_proxy_options_from_environment():
+  """Derive proxy options for gsutil from the environment.
+
+  Parses the contents of the 'http_proxy' (and variations) environment variable
+  and converts it to the corresponding Boto options that gsutil accepts.
+  """
+  proxy_variable = None
+  for env_var in ('http_proxy', 'HTTP_PROXY', 'https_proxy', 'HTTPS_PROXY'):

[agable, 2015/01/08 20:39:01]

Shouldn't we prioritize an https proxy over an http one?

[Raphael Kubo da Costa (rakuco), 2015/01/08 22:29:02]

I don't have a strong opinion here, as the whole thing isn't standardized. I'll
change the order.

+    if env_var in os.environ:
+      proxy_variable = env_var
+      break
+  if proxy_variable is None:
+    return []
+
+  proxy_url = urlparse.urlsplit(os.environ[proxy_variable])
+  if proxy_url.hostname is None or proxy_url.port is None:
+    raise EnvironmentError(
+        'The "%s" environment is set, but has an invalid format. Its '
+        'format must be "http://[user[:password]]host:port".' % proxy_variable)

[agable, 2015/01/08 20:39:01]

[user[:password]@]host:port

+  boto_options = ['-o', 'Boto:proxy=%s' % proxy_url.hostname,
+                  '-o', 'Boto:proxy_port=%d' % proxy_url.port]
+  if proxy_url.username:
+    boto_options.extend(['-o', 'Boto:proxy_user=%s' % proxy_url.username])
+  if proxy_url.password:
+    boto_options.extend(['-o', 'Boto:proxy_pass=%s' % proxy_url.password])
+  return boto_options
+
+
 def run_gsutil(force_version, fallback, target, args):
   if force_version:
     gsutil_bin = ensure_gsutil(force_version, target)
+    proxy_options = get_boto_proxy_options_from_environment()
   else:
     gsutil_bin = fallback
-  cmd = [sys.executable, gsutil_bin] + args
+    proxy_options = []  # Not available in the fallback gsutil.
+  cmd = [sys.executable, gsutil_bin] + proxy_options + args
   try:
     call(cmd)
   except SubprocessError as e: