Clean up pr387: add unit tests to expose headers issue, and also to test from_buffer. Fix from_buffer by populating the headers, and simplifying the arg structure.

Author: chrismattmann

tika/tests/tests_unpack.py

@@ -26,6 +26,18 @@ def test_ascii(self):
             parsed = unpack.from_file(f.name)
             self.assertEqual(parsed["content"].strip(), self.text_ascii)
 
+    def test_from_buffer(self):
+        parsed = unpack.from_buffer('what?')
+        self.assertIsNotNone(parsed)
+        self.assertIsNotNone(parsed["metadata"])
+        self.assertEqual(parsed["metadata"]["Content-Length"], "5")
+
+    def test_from_buffer_with_headers(self):
+        parsed = unpack.from_buffer('what?', headers={'Param': 'whatever'})
+        self.assertIsNotNone(parsed)
+        self.assertIsNotNone(parsed["metadata"])
+        self.assertEqual(parsed["metadata"]["Content-Length"], "5")
+
 
 if __name__ == '__main__':
     unittest.main()

tika/unpack.py

@@ -52,9 +52,13 @@ def from_buffer(string, serverEndpoint=ServerEndpoint, headers=None, requestOpti
     :param serverEndpoint: Tika server URL (Optional)
     :return: parsed content
     '''
+
+    headers = headers or {}
+    headers.update({'Accept': 'application/x-tar'})
+
     status, response = callServer('put', serverEndpoint, '/unpack/all', string,
-                                  {'Accept': 'application/x-tar'}, False,
-                                  rawResponse=True, headers=headers, requestOptions=requestOptions)
+                                  headers, False,
+                                  rawResponse=True, requestOptions=requestOptions)
 
     return _parse((status, response))