Guess file content type when not specified (#655)

kujenga · martindurant · web-flow · commit f9781069d7bb · 2025-01-06T11:38:31.000-05:00
Co-authored-by: Martin Durant &lt;martin.durant@alumni.utoronto.ca&gt;
diff --git a/gcsfs/core.py b/gcsfs/core.py
@@ -1,10 +1,12 @@
 """
 Google Cloud Storage pythonic interface
 """
+
 import asyncio
 import io
 import json
 import logging
+import mimetypes
 import os
 import posixpath
 import re
@@ -1391,7 +1393,7 @@ async def _put_file(
         rpath,
         metadata=None,
         consistency=None,
-        content_type="application/octet-stream",
+        content_type=None,
         chunksize=50 * 2**20,
         callback=None,
         fixed_key_metadata=None,
@@ -1401,6 +1403,10 @@ async def _put_file(
         # enforce blocksize should be a multiple of 2**18
         if os.path.isdir(lpath):
             return
+        if content_type is None:
+            content_type, _ = mimetypes.guess_type(lpath)
+            if content_type is None:
+                content_type = "application/octet-stream"
         callback = callback or NoOpCallback()
         consistency = consistency or self.consistency
         checker = get_consistency_checker(consistency)
@@ -1755,7 +1761,8 @@ def __init__(
             the number we wrote; 'md5' does a full checksum. Any value other
             than 'size' or 'md5' or 'crc32c' is assumed to mean no checking.
         content_type: str
-            default is `application/octet-stream`. See the list of available
+            default when unspecified is provided by mimetypes.guess_type or
+            otherwise `application/octet-stream`. See the list of available
             content types at https://www.iana.org/assignments/media-types/media-types.txt
         metadata: dict
             Custom metadata, in key/value pairs, added at file creation
@@ -1798,7 +1805,8 @@ def __init__(
         else:
             det = {}
         self.content_type = content_type or det.get(
-            "contentType", "application/octet-stream"
+            "contentType",
+            mimetypes.guess_type(self.path)[0] or "application/octet-stream",
         )
         self.metadata = metadata or det.get("metadata", {})
         self.fixed_key_metadata = _convert_fixed_key_metadata(det, from_google=True)
diff --git a/gcsfs/tests/test_core.py b/gcsfs/tests/test_core.py
@@ -890,6 +890,36 @@ def test_array(gcs):
         assert out == b"A" * 1000
 
 
+def test_content_type_set(gcs):
+    fn = TEST_BUCKET + "/content_type"
+    with gcs.open(fn, "wb", content_type="text/html") as f:
+        f.write(b"<html>")
+    assert gcs.info(fn)["contentType"] == "text/html"
+
+
+def test_content_type_guess(gcs):
+    fn = TEST_BUCKET + "/content_type.txt"
+    with gcs.open(fn, "wb") as f:
+        f.write(b"zz")
+    assert gcs.info(fn)["contentType"] == "text/plain"
+
+
+def test_content_type_default(gcs):
+    fn = TEST_BUCKET + "/content_type.abcdef"
+    with gcs.open(fn, "wb") as f:
+        f.write(b"zz")
+    assert gcs.info(fn)["contentType"] == "application/octet-stream"
+
+
+def test_content_type_put_guess(gcs):
+    dst = TEST_BUCKET + "/content_type_put_guess"
+    with tmpfile(extension="txt") as fn:
+        with open(fn, "w") as f:
+            f.write("zz")
+        gcs.put(fn, f"gs://{dst}", b"")
+    assert gcs.info(dst)["contentType"] == "text/plain"
+
+
 def test_attrs(gcs):
     if not gcs.on_google:
         # https://github.com/fsspec/gcsfs/pull/479
@@ -1194,7 +1224,6 @@ def test_dir_marker(gcs):
 
 
 def test_mkdir_with_path(gcs):
-
     with pytest.raises(FileNotFoundError):
         gcs.mkdir(f"{TEST_BUCKET + 'new'}/path", create_parents=False)
     assert not gcs.exists(f"{TEST_BUCKET + 'new'}")
