2828# tarfile returned object can be used as is in earlier versions.
2929_text_wrapper = TextIOWrapper if version_info .major >= 3 else lambda x : x
3030
31+
3132def from_file (filename , serverEndpoint = ServerEndpoint ):
3233 '''
3334 Parse from file
@@ -36,7 +37,7 @@ def from_file(filename, serverEndpoint=ServerEndpoint):
3637 :return:
3738 '''
3839 tarOutput = parse1 ('unpack' , filename , serverEndpoint ,
39- responseMimeType = 'application/x-tar' ,
40+ responseMimeType = 'application/x-tar' ,
4041 services = {'meta' : '/meta' , 'text' : '/tika' ,
4142 'all' : '/rmeta/xml' , 'unpack' : '/unpack/all' },
4243 rawResponse = True )
@@ -52,8 +53,8 @@ def from_buffer(string, serverEndpoint=ServerEndpoint):
5253 '''
5354 status , response = callServer ('put' , serverEndpoint , '/unpack/all' , string ,
5455 {'Accept' : 'application/x-tar' }, False ,
55- rawResponse = True )
56-
56+ rawResponse = True )
57+
5758 return _parse ((status , response ))
5859
5960
@@ -96,7 +97,10 @@ def _parse(tarOutput):
9697
9798 contentMember = tarFile .getmember ("__TEXT__" )
9899 if not contentMember .issym () and contentMember .isfile ():
99- content = _text_wrapper (tarFile .extractfile (contentMember ), encoding = 'utf8' ).read ()
100+ if version_info .major >= 3 :
101+ content = _text_wrapper (tarFile .extractfile (contentMember ), encoding = 'utf8' ).read ()
102+ else :
103+ content = tarFile .extractfile (contentMember ).read ().decode ('utf8' )
100104
101105 # get the remaining files as attachments
102106 attachments = {}
0 commit comments