Improve metadata mime type handling

Author: noirbizarre

CHANGELOG.rst

@@ -4,8 +4,9 @@ Changelog
 Current
 -------
 
-- Added ``copy`` and ``move`` operations
-- `delete` now supports directories (or prefixes for key/value stores)
+- Added ``copy()`` and ``move()`` operations
+- `delete()` now supports directories (or prefixes for key/value stores)
+- Improve `metadata()` `mime` handling
 
 0.5.1 (2018-03-12)
 ------------------

flask_fs/backends/__init__.py

@@ -3,6 +3,8 @@
 
 import six
 
+from flask_fs import files
+
 __all__ = [i.encode('ascii') for i in ('BaseBackend', 'DEFAULT_BACKEND')]
 
 
@@ -14,6 +16,7 @@ class BaseBackend(object):
     Abstract class to implement backend.
     '''
     root = None
+    DEFAULT_MIME = 'application/octet-stream'
 
     def __init__(self, name, config):
         self.name = name
@@ -66,6 +69,21 @@ def save(self, file_or_wfs, filename, overwrite=False):
         self.write(filename, file_or_wfs.read())
         return filename
 
+    def metadata(self, filename):
+        '''
+        Fetch all available metadata for a given file
+        '''
+        meta = self.get_metadata(filename)
+        # Fix backend mime misdetection
+        meta['mime'] = meta.get('mime') or files.mime(filename, self.DEFAULT_MIME)
+        return meta
+    
+    def get_metadata(self, filename):
+        '''
+        Backend specific method to retrieve metadata for a given file
+        '''
+        raise NotImplementedError('Copy operation is not implemented')
+
     def serve(self, filename):
         '''Serve a file given its filename'''
         raise NotImplementedError('serve operation is not implemented')

flask_fs/backends/gridfs.py

@@ -12,8 +12,6 @@
 from gridfs import GridFS
 from pymongo import MongoClient
 
-from flask_fs import files
-
 from . import BaseBackend
 
 log = logging.getLogger(__name__)
@@ -75,11 +73,11 @@ def serve(self, filename):
         file = self.fs.get_last_version(filename)
         return send_file(file, mimetype=file.content_type)
 
-    def metadata(self, filename):
+    def get_metadata(self, filename):
         f = self.fs.get_last_version(filename)
         return {
             'checksum': 'md5:{0}'.format(f.md5),
             'size': f.length,
-            'mime': f.content_type or files.mime(filename),
+            'mime': f.content_type,
             'modified': f.upload_date,
         }

flask_fs/backends/local.py

@@ -122,8 +122,8 @@ def serve(self, filename):
         '''Serve files for storages with direct file access'''
         return send_from_directory(self.root, filename)
 
-    def metadata(self, filename):
-        '''Fetch all availabe metadata'''
+    def get_metadata(self, filename):
+        '''Fetch all available metadata'''
         dest = self.path(filename)
         with open(dest, 'rb', buffering=0) as f:
             checksum = 'sha1:{0}'.format(sha1(f))

flask_fs/backends/s3.py

@@ -86,14 +86,15 @@ def list_files(self):
         for f in self.bucket.objects.all():
             yield f.key
 
-    def metadata(self, filename):
+    def get_metadata(self, filename):
         '''Fetch all availabe metadata'''
         obj = self.bucket.Object(filename)
         checksum = 'md5:{0}'.format(obj.e_tag[1:-1])
+        mime = obj.content_type.split(';', 1)[0] if obj.content_type else None
         return {
             'checksum': checksum,
             'size': obj.content_length,
-            'mime': obj.content_type,
+            'mime': mime,
             'modified': obj.last_modified,
         }
 

flask_fs/backends/swift.py

@@ -75,7 +75,7 @@ def list_files(self):
         for i in items:
             yield i['name']
 
-    def metadata(self, filename):
+    def get_metadata(self, filename):
         data = self.conn.head_object(self.name, filename)
         return {
             'checksum': 'md5:{0}'.format(data['etag']),

flask_fs/files.py

@@ -68,11 +68,11 @@ def lower_extension(filename):
     return filename
 
 
-def mime(filename):
+def mime(filename, default=None):
     '''
     A basic helper to guess mime type from a filename or url
     '''
-    return mimetypes.guess_type(filename)[0]
+    return mimetypes.guess_type(filename)[0] or default
 
 
 class All(object):

tests/test_backend_mixin.py

@@ -184,6 +184,13 @@ def test_metadata(self, app, faker):
         assert metadata['mime'] == 'text/plain'
         assert isinstance(metadata['modified'], datetime)
 
+    def test_metadata_unknown_mime(self, app, faker):
+        content = six.text_type(faker.sentence())
+        self.put_file('file.whatever', content)
+
+        metadata = self.backend.metadata('file.whatever')
+        assert metadata['mime'] in ('application/octet-stream', 'text/plain')
+
     def test_copy(self, faker):
         content = faker.sentence()
         self.put_file('file.test', content)

tests/test_files.py

@@ -38,3 +38,18 @@ def test_all_except():
     all_except = files.AllExcept('exe')
     assert 'csv' in all_except
     assert 'exe' not in all_except
+
+
+def test_mime_known_type():
+    assert files.mime('test.txt') == 'text/plain'
+    assert files.mime('test.csv') == 'text/csv'
+
+
+def test_mime_default_to_none():
+    assert files.mime('test') is None
+    assert files.mime('test', default=None) is None
+
+
+def test_mime_default_to_custom():
+    default = 'application/octet-stream'
+    assert files.mime('test', default=default) == default