Merge pull request boto#2477 from opottone/develop

Add recognized HTTP headers for S3 metadata. Fixes boto#2477, boto#2050.
stanhu · Aug 19, 2014 · c8c625a · c8c625a
2 parents 2f0e689 + 3cf9199
commit c8c625a
Show file tree

Hide file tree

Showing 4 changed files with 35 additions and 25 deletions.
diff --git a/boto/s3/bucket.py b/boto/s3/bucket.py
@@ -204,12 +204,9 @@ def _get_key_internal(self, key_name, headers, query_args_l):
             k = self.key_class(self)
             provider = self.connection.provider
             k.metadata = boto.utils.get_aws_metadata(response.msg, provider)
-            k.etag = response.getheader('etag')
-            k.content_type = response.getheader('content-type')
-            k.content_encoding = response.getheader('content-encoding')
-            k.content_disposition = response.getheader('content-disposition')
-            k.content_language = response.getheader('content-language')
-            k.last_modified = response.getheader('last-modified')
+            for field in Key.base_fields:
+                k.__dict__[field.lower().replace('-', '_')] = \
+                    response.getheader(field)
             # the following machinations are a workaround to the fact that
             # apache/fastcgi omits the content-length header on HEAD
             # requests when the content-length is zero.
@@ -219,7 +216,6 @@ def _get_key_internal(self, key_name, headers, query_args_l):
                 k.size = int(response.getheader('content-length'))
             else:
                 k.size = 0
-            k.cache_control = response.getheader('cache-control')
             k.name = key_name
             k.handle_version_headers(response)
             k.handle_encryption_headers(response)

diff --git a/boto/s3/key.py b/boto/s3/key.py
@@ -89,10 +89,16 @@ class Key(object):
     # x-amz-meta).
     base_user_settable_fields = set(["cache-control", "content-disposition",
                                     "content-encoding", "content-language",
-                                    "content-md5", "content-type"])
+                                    "content-md5", "content-type",
+                                     "x-robots-tag", "expires"])
     _underscore_base_user_settable_fields = set()
     for f in base_user_settable_fields:
       _underscore_base_user_settable_fields.add(f.replace('-', '_'))
+    # Metadata fields, whether user-settable or not, other than custom
+    # metadata fields (i.e., those beginning with a provider specific prefix
+    # like x-amz-meta).
+    base_fields = (base_user_settable_fields |
+                   set(["last-modified", "content-length", "date", "etag"]))
 
 
 
@@ -302,20 +308,8 @@ def open_read(self, headers=None, query_args='',
                 elif name.lower() == 'content-range':
                     end_range = re.sub('.*/(.*)', '\\1', value)
                     self.size = int(end_range)
-                elif name.lower() == 'etag':
-                    self.etag = value
-                elif name.lower() == 'content-type':
-                    self.content_type = value
-                elif name.lower() == 'content-encoding':
-                    self.content_encoding = value
-                elif name.lower() == 'content-language':
-                    self.content_language = value
-                elif name.lower() == 'last-modified':
-                    self.last_modified = value
-                elif name.lower() == 'cache-control':
-                    self.cache_control = value
-                elif name.lower() == 'content-disposition':
-                    self.content_disposition = value
+                elif name.lower() in Key.base_fields:
+                    self.__dict__[name.lower().replace('-', '_')] = value
             self.handle_version_headers(self.resp)
             self.handle_encryption_headers(self.resp)
             self.handle_restore_headers(self.resp)
@@ -555,6 +549,8 @@ def set_metadata(self, name, value):
             self.metadata['Content-MD5'] = value
         else:
             self.metadata[name] = value
+        if name.lower() in Key.base_user_settable_fields:
+            self.__dict__[name.lower().replace('-', '_')] = value
 
     def update_metadata(self, d):
         self.metadata.update(d)

diff --git a/boto/utils.py b/boto/utils.py
@@ -170,9 +170,7 @@ def merge_meta(headers, metadata, provider=None):
     metadata_prefix = provider.metadata_prefix
     final_headers = headers.copy()
     for k in metadata.keys():
-        if k.lower() in ['cache-control', 'content-md5', 'content-type',
-                         'content-encoding', 'content-disposition',
-                         'expires']:
+        if k.lower() in boto.s3.key.Key.base_user_settable_fields:
             final_headers[k] = metadata[k]
         else:
             final_headers[metadata_prefix + k] = metadata[k]

diff --git a/tests/integration/s3/test_key.py b/tests/integration/s3/test_key.py
@@ -403,13 +403,33 @@ def test_header_encoding(self):
         key.set_metadata('Cache-control', u'public, max-age=500')
         key.set_metadata('Test-Plus', u'A plus (+)')
         key.set_metadata('Content-disposition', u'filename=Schöne Zeit.txt')
+        key.set_metadata('Content-Encoding', 'gzip')
+        key.set_metadata('Content-Language', 'de')
+        key.set_metadata('Content-Type', 'application/pdf')
+        self.assertEqual(key.content_type, 'application/pdf')
+        key.set_metadata('X-Robots-Tag', 'all')
+        key.set_metadata('Expires', u'Thu, 01 Dec 1994 16:00:00 GMT')
         key.set_contents_from_string('foo')
 
         check = self.bucket.get_key('test_header_encoding')
+        remote_metadata = check._get_remote_metadata()
 
+        # TODO: investigate whether encoding ' ' as '%20' makes sense
         self.assertEqual(check.cache_control, 'public,%20max-age=500')
+        self.assertEqual(remote_metadata['cache-control'], 'public,%20max-age=500')
         self.assertEqual(check.get_metadata('test-plus'), 'A plus (+)')
         self.assertEqual(check.content_disposition, 'filename=Sch%C3%B6ne%20Zeit.txt')
+        self.assertEqual(remote_metadata['content-disposition'], 'filename=Sch%C3%B6ne%20Zeit.txt')
+        self.assertEqual(check.content_encoding, 'gzip')
+        self.assertEqual(remote_metadata['content-encoding'], 'gzip')
+        self.assertEqual(check.content_language, 'de')
+        self.assertEqual(remote_metadata['content-language'], 'de')
+        self.assertEqual(check.content_type, 'application/pdf')
+        self.assertEqual(remote_metadata['content-type'], 'application/pdf')
+        self.assertEqual(check.x_robots_tag, 'all')
+        self.assertEqual(remote_metadata['x-robots-tag'], 'all')
+        self.assertEqual(check.expires, 'Thu,%2001%20Dec%201994%2016:00:00%20GMT')
+        self.assertEqual(remote_metadata['expires'], 'Thu,%2001%20Dec%201994%2016:00:00%20GMT')
 
         expected = u'filename=Schöne Zeit.txt'
         if six.PY2: