ENH: Add read support for Google Cloud Storage.

Author: vision-sbm

pandas/io/common.py

@@ -168,6 +168,14 @@ def is_s3_url(url):
         return False
 
 
+def is_google_cloud_storage_url(url):
+    """Check for a gs url"""
+    try:
+        return parse_url(url).scheme == 'gs'
+    except:  # noqa
+        return False
+
+
 def get_filepath_or_buffer(filepath_or_buffer, encoding=None,
                            compression=None, mode=None):
     """
@@ -203,6 +211,14 @@ def get_filepath_or_buffer(filepath_or_buffer, encoding=None,
                                          compression=compression,
                                          mode=mode)
 
+    if is_google_cloud_storage_url(filepath_or_buffer):
+        from pandas.io import google_cloud_storage
+        return google_cloud_storage.get_filepath_or_buffer(
+            filepath_or_buffer,
+            encoding=encoding,
+            compression=compression,
+            mode=mode)
+
     if isinstance(filepath_or_buffer, (compat.string_types,
                                        compat.binary_type,
                                        mmap.mmap)):

pandas/io/google_cloud_storage.py

@@ -0,0 +1,42 @@
+""" Google Cloud Storage support for remote file interactivity """
+from io import BytesIO
+from pandas import compat
+try:
+    from google.cloud.storage import Client
+except:
+    raise ImportError("The google-cloud-storage library is required to "
+                      "read gs:// files")
+
+if compat.PY3:
+    from urllib.parse import urlparse as parse_url
+else:
+    from urlparse import urlparse as parse_url
+
+
+def _get_bucket_name(url):
+    """Returns the bucket name from the gs:// url"""
+    result = parse_url(url)
+    return result.netloc
+
+
+def _get_object_path(url):
+    """Returns the object path from the gs:// url"""
+    result = parse_url(url)
+    return result.path.lstrip('/')
+
+
+def get_filepath_or_buffer(filepath_or_buffer, encoding=None,
+                           compression=None, mode=None):
+
+    if mode is None:
+        mode = 'rb'
+
+    client = Client()
+    bucket = client.get_bucket(_get_bucket_name(filepath_or_buffer))
+    blob = bucket.blob(_get_object_path(filepath_or_buffer))
+    data = BytesIO()
+    blob.download_to_file(data)
+    data.seek(0)
+    filepath_or_buffer = data
+
+    return filepath_or_buffer, None, compression