From e0c594c91a9ecd5be8a04d90949df3276ee0befc Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Michal=20=C4=8Ciha=C5=99?= <michal@cihar.com>
Date: Thu, 23 May 2024 09:00:01 +0200
Subject: [PATCH] Use charset-normalizer for encoding detection

Fixes #222
---
 aeidon/encodings.py | 21 +++++----------------
 aeidon/util.py      | 14 +++++++-------
 setup-aeidon.py     |  4 ++--
 3 files changed, 14 insertions(+), 25 deletions(-)

diff --git a/aeidon/encodings.py b/aeidon/encodings.py
index 41d5873e..17daac91 100644
--- a/aeidon/encodings.py
+++ b/aeidon/encodings.py
@@ -177,23 +177,12 @@ def detect(path):
     bom_encoding = detect_bom(path)
     if bom_encoding is not None:
         return bom_encoding
-    from chardet import universaldetector
-    detector = universaldetector.UniversalDetector()
-    with open(path, "rb") as f:
-        detector.reset()
-        for line in f:
-            detector.feed(line)
-            if detector.done: break
-    detector.close()
-    code = detector.result["encoding"]
-    if code is None: return None
-    try:
-        # chardet returns what seem to be IANA names. They need to be
-        # translated to their Python equivalents. Some of the encodings
-        # returned by chardet are not supported by Python.
-        return translate_code(code)
-    except ValueError:
+    from charset_normalizer import from_path
+    detector = from_path(path)
+    result = detector.best()
+    if result is None:
         return None
+    return result.encoding
 
 def detect_bom(path):
     """Return corresponding encoding if BOM found, else ``None``."""
diff --git a/aeidon/util.py b/aeidon/util.py
index 69c15259..15c8bb39 100644
--- a/aeidon/util.py
+++ b/aeidon/util.py
@@ -113,11 +113,11 @@ def atomic_open(path, mode="w", *args, **kwargs):
 
 @aeidon.deco.once
 def chardet_available():
-    """Return ``True`` if :mod:`chardet` module is available."""
+    """Return ``True`` if :mod:`charset_normalizer` module is available."""
     try:
-        import chardet # noqa
+        import charset_normalizer # noqa
         return True
-    except Exception:
+    except ImportError:
         return False
 
 def compare_versions(x, y):
@@ -207,11 +207,11 @@ def flatten(lst):
     return flat_lst
 
 def get_chardet_version():
-    """Return :mod:`chardet` version number as string or ``None``."""
+    """Return :mod:`charset_normalizer` version number as string or ``None``."""
     try:
-        import chardet
-        return chardet.__version__
-    except Exception:
+        import charset_normalizer
+        return "charset_normalizer {}".format(charset_normalizer.__version__)
+    except ImportError:
         return None
 
 @aeidon.deco.once
diff --git a/setup-aeidon.py b/setup-aeidon.py
index 15723b9f..a6879350 100755
--- a/setup-aeidon.py
+++ b/setup-aeidon.py
@@ -24,8 +24,8 @@
     license="GPL",
     packages=find_packages(exclude=["gaupol*", "*.test"]),
     package_data={"aeidon": ["data/*/*"]},
-    python_requires=">=3.2.0",
-    install_requires=["chardet>=2.2.1"],
+    python_requires=">=3.5.0",
+    install_requires=["charset-normalizer>2.0"],
 )
 
 shutil.rmtree("aeidon/data")