Switch default fallback encoding detection lib to charset-normalizer

CHANGES/5930.feature

@@ -0,0 +1 @@
+Switch chardet to charset-normalizer when aiohttp need to guess the body encoding

CONTRIBUTORS.txt

@@ -7,6 +7,7 @@ Adam Horacek
 Adam Mills
 Adrian Krupa
 Adrián Chaves
+Ahmed Tahri
 Alan Tse
 Alec Hanefeld
 Alejandro Gómez

README.rst

@@ -164,14 +164,14 @@ Requirements
 
 - Python >= 3.7
 - async-timeout_
-- chardet_
+- charset-normalizer_
 - multidict_
 - yarl_
 
 Optionally you may install the cChardet_ and aiodns_ libraries (highly
 recommended for sake of speed).
 
-.. _chardet: https://pypi.python.org/pypi/chardet
+.. _charset-normalizer: https://pypi.org/project/charset-normalizer
 .. _aiodns: https://pypi.python.org/pypi/aiodns
 .. _multidict: https://pypi.python.org/pypi/multidict
 .. _yarl: https://pypi.python.org/pypi/yarl

aiohttp/client_reqrep.py

@@ -70,7 +70,7 @@
 try:
     import cchardet as chardet
 except ImportError:  # pragma: no cover
-    import chardet  # type: ignore[no-redef]
+    import charset_normalizer as chardet  # type: ignore[no-redef]
 
 
 __all__ = ("ClientRequest", "ClientResponse", "RequestInfo", "Fingerprint")

docs/client_reference.rst

@@ -1374,10 +1374,10 @@ Response object
       specified *encoding* parameter.
 
       If *encoding* is ``None`` content encoding is autocalculated
-      using ``Content-Type`` HTTP header and *chardet* tool if the
+      using ``Content-Type`` HTTP header and *charset-normalizer* tool if the
       header is not provided by server.
 
-      :term:`cchardet` is used with fallback to :term:`chardet` if
+      :term:`cchardet` is used with fallback to :term:`charset-normalizer` if
       *cchardet* is not available.
 
       Close underlying connection if data reading gets an error,
@@ -1389,14 +1389,14 @@ Response object
 
       :return str: decoded *BODY*
 
-      :raise LookupError: if the encoding detected by chardet or cchardet is
+      :raise LookupError: if the encoding detected by cchardet is
                           unknown by Python (e.g. VISCII).
 
       .. note::
 
          If response has no ``charset`` info in ``Content-Type`` HTTP
-         header :term:`cchardet` / :term:`chardet` is used for content
-         encoding autodetection.
+         header :term:`cchardet` / :term:`charset-normalizer` is used for
+         content encoding autodetection.
 
          It may hurt performance. If page encoding is known passing
          explicit *encoding* parameter might help::
@@ -1411,7 +1411,7 @@ Response object
       a ``read`` call will be done,
 
       If *encoding* is ``None`` content encoding is autocalculated
-      using :term:`cchardet` or :term:`chardet` as fallback if
+      using :term:`cchardet` or :term:`charset-normalizer` as fallback if
       *cchardet* is not available.
 
       if response's `content-type` does not match `content_type` parameter
@@ -1449,11 +1449,11 @@ Response object
       Automatically detect content encoding using ``charset`` info in
       ``Content-Type`` HTTP header. If this info is not exists or there
       are no appropriate codecs for encoding then :term:`cchardet` /
-      :term:`chardet` is used.
+      :term:`charset-normalizer` is used.
 
       Beware that it is not always safe to use the result of this function to
       decode a response. Some encodings detected by cchardet are not known by
-      Python (e.g. VISCII).
+      Python (e.g. VISCII). *charset-normalizer* is not concerned by that issue.
 
       :raise RuntimeError: if called before the body has been read,
                            for :term:`cchardet` usage

docs/glossary.rst

@@ -45,11 +45,12 @@
       Any object that can be called. Use :func:`callable` to check
       that.
 
-   chardet
+   charset-normalizer
 
-       The Universal Character Encoding Detector
+       The Real First Universal Charset Detector.
+       Open, modern and actively maintained alternative to Chardet.
 
-       https://pypi.python.org/pypi/chardet/
+       https://pypi.org/project/charset-normalizer/
 
    cchardet
 

docs/index.rst

@@ -34,7 +34,7 @@ Library Installation
    $ pip install aiohttp
 
 You may want to install *optional* :term:`cchardet` library as faster
-replacement for :term:`chardet`:
+replacement for :term:`charset-normalizer`:
 
 .. code-block:: bash
 
@@ -51,7 +51,7 @@ This option is highly recommended:
 Installing speedups altogether
 ------------------------------
 
-The following will get you ``aiohttp`` along with :term:`chardet`,
+The following will get you ``aiohttp`` along with :term:`charset-normalizer`,
 :term:`aiodns` and ``Brotli`` in one bundle. No need to type
 separate commands anymore!
 
@@ -148,11 +148,11 @@ Dependencies
 
 - Python 3.7+
 - *async_timeout*
-- *chardet*
+- *charset-normalizer*
 - *multidict*
 - *yarl*
 - *Optional* :term:`cchardet` as faster replacement for
-  :term:`chardet`.
+  :term:`charset-normalizer`.
 
   Install it explicitly via:
 

docs/spelling_wordlist.txt

@@ -124,6 +124,8 @@ cchardet
 ceil
 charset
 charsetdetect
+normalizer
+Chardet
 chunked
 chunking
 cls

requirements/base.txt

@@ -6,7 +6,7 @@ async-timeout==4.0.0a3
 asynctest==0.13.0; python_version<"3.8"
 Brotli==1.0.9
 cchardet==2.1.7
-chardet==4.0.0
+charset-normalizer==2.0.4
 frozenlist==1.1.1
 gunicorn==20.1.0
 typing_extensions==3.7.4.3

requirements/dev.txt

@@ -46,7 +46,7 @@ cfgv==3.2.0
     # via
     #   -r requirements/lint.txt
     #   pre-commit
-chardet==4.0.0
+charset-normalizer==2.0.4
     # via
     #   -r requirements/base.txt
     #   requests

setup.py

@@ -50,7 +50,7 @@
     raise RuntimeError("Unable to determine version.")
 
 install_requires = [
-    "chardet>=2.0,<5.0",
+    "charset-normalizer>=2.0,<3.0",
     "multidict>=4.5,<7.0",
     "async_timeout>=4.0a2,<5.0",
     'asynctest==0.13.0; python_version<"3.8"',