add mimetypes from mediocris into serializer (#274)

Co-authored-by: Nathanael Shim <[email protected]>

Author: nateshim-indico

indico/http/serialization.py

@@ -10,6 +10,7 @@
 from collections import defaultdict
 
 import msgpack
+
 from indico.errors import IndicoDecodingError
 
 logger = logging.getLogger(__name__)
@@ -44,10 +45,10 @@ def deserialize(response, force_json=False, force_decompress=False):
         )
 
 
-
 def raw_bytes(content, *args, **kwargs):
     return content
 
+
 def msgpack_deserialization(content, charset):
     return msgpack.unpackb(content)
 
@@ -71,20 +72,52 @@ def zip_serialization(content, charset=None):
 _SERIALIZATION_FNS = defaultdict(
     lambda: text_deserialization,
     {
-        "application/octet-stream": raw_bytes,
         "application/pdf": raw_bytes,
+        "application/octet-stream": raw_bytes,
+        "application/doc": raw_bytes,
+        "application/ms-doc": raw_bytes,
+        "application/msword": raw_bytes,
+        "application/vnd.openxmlformats-officedocument.wordprocessingml.document": raw_bytes,
+        "application/vnd.oasis.opendocument.text": raw_bytes,
+        "application/zip": zip_serialization,
+        "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet": raw_bytes,
+        "application/vnd.ms-excel": raw_bytes,
+        "application/msexcel": raw_bytes,
+        "application/excel": raw_bytes,
+        "application/msexcel": raw_bytes,
+        "application/x-dos_ms_excel": raw_bytes,
+        "application/x-excel": raw_bytes,
+        "application/x-ms-excel": raw_bytes,
+        "application/x-xls": raw_bytes,
+        "application/xlc": raw_bytes,
+        "application/xls": raw_bytes,
+        "application/xlt": raw_bytes,
+        "application/vnd.ms-powerpoint": raw_bytes,
+        "application/vnd.openxmlformats-officedocument.presentationml.presentation": raw_bytes,
+        "application/mspowerpoint": raw_bytes,
+        "application/powerpoint": raw_bytes,
+        "application/vnd.ms-powerpoint": raw_bytes,
+        "application/x-mspowerpoint": raw_bytes,
+        "image/png": image_serialization,
+        "image/jpeg": image_serialization,
+        "image/jpg": image_serialization,
+        "image/tiff": image_serialization,
         "text/html": text_deserialization,
+        "text/plain": text_deserialization,
+        "text/rtf": text_deserialization,
+        "message/rfc822": raw_bytes,
+        "application/vnd.ms-outlook": raw_bytes,
+        "application/vnd.ms-office": raw_bytes,
+        "binary/octet-stream": raw_bytes,
+        "application/rtf": text_deserialization,
+        "text/csv": text_deserialization,
         "application/x-msgpack": msgpack_deserialization,
         "application/msgpack": msgpack_deserialization,
         "x-msgpack": msgpack_deserialization,
         "msgpack": msgpack_deserialization,
         "application/json": json_deserialization,
         "json": json_deserialization,
         "application/javascript": json_deserialization,
-        "image/png": image_serialization,
-        "image/jpeg": image_serialization,
-        "image/jpg": image_serialization,
-        "application/zip": zip_serialization,
         "application/x-gzip": raw_bytes,
         "application/gzip": raw_bytes,
     },

tests/unit/http/mocked_data/application_xls_

@@ -0,0 +1,2 @@
+text, targets
+hello, world

tests/unit/http/mocked_data/text_csv_

@@ -3,6 +3,7 @@
 from unittest.mock import MagicMock
 
 import pytest
+
 from indico.errors import IndicoDecodingError
 from indico.http.serialization import deserialize
 
@@ -73,6 +74,20 @@ def test_deserialize_gzip(mock_loader):
     assert isinstance(content, bytes)
 
 
+def test_deserialize_csv(mock_loader):
+    response = mock_loader("text/csv", "")
+    content = deserialize(response)
+
+    assert isinstance(content, str)
+
+
+def test_deserialize_xls(mock_loader):
+    response = mock_loader("application/xls", "")
+    content = deserialize(response)
+
+    assert isinstance(content, bytes)
+
+
 def test_deserialize_unknown(mock_loader):
     response = mock_loader("unknown", "")
     try: