From 43f8f17336234ce61658adff3655037471609874 Mon Sep 17 00:00:00 2001
From: carlo cancellieri <ccancellieri@hotmail.com>
Date: Fri, 8 Oct 2021 20:38:01 +0200
Subject: [PATCH] introduce Google Cloud Bucket support and private file proxy

---
 ckanext/cloudstorage/controller.py | 26 +++++++++++++++++++++++---
 ckanext/cloudstorage/storage.py    | 26 +++++++++++++++++++++++---
 2 files changed, 46 insertions(+), 6 deletions(-)

diff --git a/ckanext/cloudstorage/controller.py b/ckanext/cloudstorage/controller.py
index 992b0c1..4cc7a7d 100644
--- a/ckanext/cloudstorage/controller.py
+++ b/ckanext/cloudstorage/controller.py
@@ -5,10 +5,14 @@
 from pylons import c
 from pylons.i18n import _
 
+from ckan.common import request, response
 from ckan import logic, model
 from ckan.lib import base, uploader
 import ckan.lib.helpers as h
 
+import ckanext.cloudstorage.storage as _storage
+storage = _storage.CloudStorage
+is_proxy_download=storage.proxy_download.fget(storage)
 
 class StorageController(base.BaseController):
     def resource_download(self, id, resource_id, filename=None):
@@ -48,12 +52,28 @@ def resource_download(self, id, resource_id, filename=None):
         # if the client requests with a Content-Type header (e.g. Text preview)
         # we have to add the header to the signature
         try:
-            content_type = getattr(c.pylons.request, "content_type", None)
+            content_type = getattr(request, "content_type", None)
         except AttributeError:
             content_type = None
-        uploaded_url = upload.get_url_from_filename(resource['id'], filename,
-                                                    content_type=content_type)
+        
+        # If the repository is private you may want to use ckan accout to proxy
+        # protected contents
+        # ckanext.cloudstorage.proxy_download = [False|True]
+        # Default: False
+        if is_proxy_download:
+            # remote object
+            obj = upload.get_object(resource['id'],filename)
+            # metaadta
+            extra = obj.extra
+            if extra:
+                # let's leverage on external mimetype if present
+                response.headers['Content-Type'] = extra.get('content_type',content_type)
+            # return stream back
+            return upload.get_object_as_stream(obj)
 
+        uploaded_url = upload.get_url_from_filename(resource['id'], filename,
+                                                            content_type=content_type)
+        
         # The uploaded file is missing for some reason, such as the
         # provider being down.
         if uploaded_url is None:
diff --git a/ckanext/cloudstorage/storage.py b/ckanext/cloudstorage/storage.py
index 903390b..4e959db 100644
--- a/ckanext/cloudstorage/storage.py
+++ b/ckanext/cloudstorage/storage.py
@@ -151,7 +151,16 @@ def guess_mimetype(self):
         return p.toolkit.asbool(
             config.get('ckanext.cloudstorage.guess_mimetype', False)
         )
-
+    
+    @property
+    def proxy_download(self):
+        """
+        If the ckan may stream the object (will use service account to download
+        from private storages)
+        """
+        return p.toolkit.asbool(
+            config.get('ckanext.cloudstorage.proxy_download', False)
+        )
 
 class ResourceCloudStorage(CloudStorage):
     def __init__(self, resource):
@@ -244,7 +253,7 @@ def upload(self, id, max_size=10):
                     content_settings=content_settings
                 )
             else:
-
+            
                 # TODO: This might not be needed once libcloud is upgraded
                 if isinstance(self.file_upload, SpooledTemporaryFile):
                     self.file_upload.next = self.file_upload.next()
@@ -340,7 +349,9 @@ def get_url_from_filename(self, rid, filename, content_type=None):
         try:
             return self.driver.get_object_cdn_url(obj)
         except NotImplementedError:
-            if 'S3' in self.driver_name:
+            if 'S3' in self.driver_name \
+                or 'GOOGLE_STORAGE' in self.driver_name:
+
                 return urlparse.urljoin(
                     'https://' + self.driver.connection.host,
                     '{container}/{path}'.format(
@@ -354,6 +365,15 @@ def get_url_from_filename(self, rid, filename, content_type=None):
                 return obj.extra['url']
             raise
 
+    def get_object(self, rid, filename):
+        # Find the key the file *should* be stored at.
+        path = self.path_from_filename(rid, filename)
+        # Find the object for the given key.
+        return self.container.get_object(path)
+
+    def get_object_as_stream(self, obj):
+        return self.driver.download_object_as_stream(obj) 
+        
     @property
     def package(self):
         return model.Package.get(self.resource['package_id'])