Enable brotli decompression if it is available (#620)

* Enable brotli decompression if it is available

* Apply PR feedback

.github/workflows/codespell.yml

@@ -17,6 +17,6 @@ jobs:
 
     steps:
       - name: Checkout
-        uses: actions/checkout@v5
+        uses: actions/checkout@v6
       - name: Codespell
         uses: codespell-project/actions-codespell@v2

.github/workflows/docs.yml

@@ -10,7 +10,7 @@ jobs:
     runs-on: ubuntu-24.04
 
     steps:
-      - uses: actions/checkout@v5
+      - uses: actions/checkout@v6
       - uses: actions/setup-python@v6
         with:
           python-version: "3.12"

.github/workflows/main.yml

@@ -23,7 +23,7 @@ jobs:
           - "pypy-3.11"
 
     steps:
-      - uses: actions/checkout@v5
+      - uses: actions/checkout@v6
       - name: Install uv
         uses: astral-sh/setup-uv@v7
 

.github/workflows/pre-commit-detect-outdated.yml

@@ -18,7 +18,7 @@ jobs:
     name: Detect outdated pre-commit hooks
     runs-on: ubuntu-24.04
     steps:
-      - uses: actions/checkout@v5
+      - uses: actions/checkout@v6
 
       - name: Set up Python 3.12
         uses: actions/setup-python@v6

.github/workflows/pre-commit.yml

@@ -13,7 +13,7 @@ jobs:
     name: Run pre-commit
     runs-on: ubuntu-24.04
     steps:
-    - uses: actions/checkout@v5
+    - uses: actions/checkout@v6
     - uses: actions/setup-python@v6
       with:
         python-version: 3.12

tests/integration/test_aiohttp.py

@@ -1,3 +1,4 @@
+import io
 import logging
 import ssl
 import urllib.parse
@@ -462,3 +463,19 @@ def test_filter_query_parameters(tmpdir, httpbin):
         cassette_content = f.read()
         assert "password" not in cassette_content
         assert "secret" not in cassette_content
+
+
+@pytest.mark.online
+def test_use_cassette_with_io(tmpdir, caplog, httpbin):
+    url = httpbin.url + "/post"
+
+    # test without cassettes
+    data = io.BytesIO(b"hello")
+    _, response_json = request("POST", url, output="json", data=data)
+    assert response_json["data"] == "hello"
+
+    # test with cassettes
+    data = io.BytesIO(b"hello")
+    with vcr.use_cassette(str(tmpdir.join("post.yaml"))):
+        _, response_json = request("POST", url, output="json", data=data)
+        assert response_json["data"] == "hello"

tests/integration/test_filter.py

@@ -7,6 +7,7 @@ from urllib.request import Request, urlopen
 import pytest
 
 import vcr
+from vcr.filters import brotli
 
 from ..assertions import assert_cassette_has_one_response, assert_is_json_bytes
 
@@ -138,6 +139,22 @@ def test_decompress_deflate(tmpdir, httpbin):
     assert_is_json_bytes(decoded_response)
 
 
+def test_decompress_brotli(tmpdir, httpbin):
+    if brotli is None:
+        # XXX: this is never true, because brotlipy is installed with "httpbin"
+        pytest.skip("Brotli is not installed")
+
+    url = httpbin.url + "/brotli"
+    request = Request(url, headers={"Accept-Encoding": ["gzip, deflate, br"]})
+    cass_file = str(tmpdir.join("brotli_response.yaml"))
+    with vcr.use_cassette(cass_file, decode_compressed_response=True):
+        urlopen(request)
+    with vcr.use_cassette(cass_file) as cass:
+        decoded_response = urlopen(url).read()
+        assert_cassette_has_one_response(cass)
+    assert_is_json_bytes(decoded_response)
+
+
 def test_decompress_regular(tmpdir, httpbin):
     """Test that it doesn't try to decompress content that isn't compressed"""
     url = httpbin.url + "/get"

vcr/filters.py

@@ -6,6 +6,49 @@ from urllib.parse import urlencode, urlparse, urlunparse
 
 from .util import CaseInsensitiveDict
 
+try:
+    # This supports both brotli & brotlipy packages
+    import brotli
+except ImportError:
+    try:
+        import brotlicffi as brotli
+    except ImportError:
+        brotli = None
+
+
+def decompress_deflate(body):
+    try:
+        return zlib.decompress(body)
+    except zlib.error:
+        # Assume the response was already decompressed
+        return body
+
+
+def decompress_gzip(body):
+    # To (de-)compress gzip format, use wbits = zlib.MAX_WBITS | 16.
+    try:
+        return zlib.decompress(body, zlib.MAX_WBITS | 16)
+    except zlib.error:
+        # Assume the response was already decompressed
+        return body
+
+
+AVAILABLE_DECOMPRESSORS = {
+    "deflate": decompress_deflate,
+    "gzip": decompress_gzip,
+}
+
+if brotli is not None:
+
+    def decompress_brotli(body):
+        try:
+            return brotli.decompress(body)
+        except brotli.error:
+            # Assume the response was already decompressed
+            return body
+
+    AVAILABLE_DECOMPRESSORS["br"] = decompress_brotli
+
 
 def replace_headers(request, replacements):
     """Replace headers in request according to replacements.
@@ -136,45 +179,30 @@ def remove_post_data_parameters(request, post_data_parameters_to_remove):
 
 def decode_response(response):
     """
-    If the response is compressed with gzip or deflate:
+    If the response is compressed with any supported compression (gzip,
+    deflate, br if available):
       1. decompress the response body
       2. delete the content-encoding header
       3. update content-length header to decompressed length
     """
 
-    def is_compressed(headers):
-        encoding = headers.get("content-encoding", [])
-        return encoding and encoding[0] in ("gzip", "deflate")
-
-    def decompress_body(body, encoding):
-        """Returns decompressed body according to encoding using zlib.
-        to (de-)compress gzip format, use wbits = zlib.MAX_WBITS | 16
-        """
-        if not body:
-            return ""
-        if encoding == "gzip":
-            try:
-                return zlib.decompress(body, zlib.MAX_WBITS | 16)
-            except zlib.error:
-                return body  # assumes that the data was already decompressed
-        else:  # encoding == 'deflate'
-            try:
-                return zlib.decompress(body)
-            except zlib.error:
-                return body  # assumes that the data was already decompressed
-
     # Deepcopy here in case `headers` contain objects that could
     # be mutated by a shallow copy and corrupt the real response.
     response = copy.deepcopy(response)
     headers = CaseInsensitiveDict(response["headers"])
-    if is_compressed(headers):
+    content_encoding = headers.get("content-encoding")
-        encoding = headers["content-encoding"][0]
+    if not content_encoding:
-        headers["content-encoding"].remove(encoding)
+        return response
-        if not headers["content-encoding"]:
+    decompressor = AVAILABLE_DECOMPRESSORS.get(content_encoding[0])
-            del headers["content-encoding"]
+    if not decompressor:
+        return response
 
-        new_body = decompress_body(response["body"]["string"], encoding)
+    headers["content-encoding"].remove(content_encoding[0])
-        response["body"]["string"] = new_body
+    if not headers["content-encoding"]:
-        headers["content-length"] = [str(len(new_body))]
+        del headers["content-encoding"]
-        response["headers"] = dict(headers)
+
+    new_body = decompressor(response["body"]["string"])
+    response["body"]["string"] = new_body
+    headers["content-length"] = [str(len(new_body))]
+    response["headers"] = dict(headers)
     return response

vcr/request.py

@@ -20,7 +20,12 @@ class Request:
         self._was_file = hasattr(body, "read")
         self._was_iter = _is_nonsequence_iterator(body)
         if self._was_file:
-            self.body = body.read()
+            if hasattr(body, "tell"):
+                tell = body.tell()
+                self.body = body.read()
+                body.seek(tell)
+            else:
+                self.body = body.read()
         elif self._was_iter:
             self.body = list(body)
         else: