From 273ec55679297e489947e3e373180c68042f366f Mon Sep 17 00:00:00 2001
From: John Parton <john.parton.iv@gmail.com>
Date: Wed, 4 Oct 2023 22:23:15 -0500
Subject: [PATCH] Move mapping logic of non-standard encoding names to rust
 code.

---
 python/chardetng_py/shortcuts.py | 24 ++----------------------
 src/lib.rs                       | 21 +++++++++++++++++++--
 2 files changed, 21 insertions(+), 24 deletions(-)

diff --git a/python/chardetng_py/shortcuts.py b/python/chardetng_py/shortcuts.py
index dfcb4a1..775342a 100644
--- a/python/chardetng_py/shortcuts.py
+++ b/python/chardetng_py/shortcuts.py
@@ -1,21 +1,9 @@
 """Functions for dealing with byte strings of unknown encoding."""
 
-from typing import Dict, Final, Union
+from typing import Union
 
 from chardetng_py.detector import EncodingDetector
 
-ALIASES: Final[Dict[str, str]] = {
-    "windows-874": "cp874",
-}
-"""Python prefers to use "cpXXX" for legacy encodings, while :code:`encoding_rs`
-and :code:`chardetng` use whatwg names.
-
-References
-----------
-https://docs.python.org/3/library/codecs.html#standard-encodings
-https://encoding.spec.whatwg.org/#legacy-single-byte-encodings
-"""
-
 
 def detect(
     byte_str: Union[bytes, bytearray],
@@ -49,12 +37,4 @@ def detect(
     encoding_detector = EncodingDetector()
     encoding_detector.feed(byte_str, last=True)
 
-    encoding: str = encoding_detector.guess(tld=tld, allow_utf8=allow_utf8)
-
-    # chardetng uses 'windows-874' as an encoding, which Python does not understand
-    # I believe that windows-874 and cp874 are basically the same encoding
-    if encoding in ALIASES:
-        # TODO Log/warn?
-        return ALIASES[encoding]
-
-    return encoding
+    return encoding_detector.guess(tld=tld, allow_utf8=allow_utf8)
diff --git a/src/lib.rs b/src/lib.rs
index 1f35dd2..2d20442 100644
--- a/src/lib.rs
+++ b/src/lib.rs
@@ -1,6 +1,21 @@
 use chardetng::EncodingDetector;
 use pyo3::prelude::*;
 
+// See https://github.com/john-parton/chardetng-py/issues/11
+// Python prefers to use "cpXXX" for legacy encodings, while :code:`encoding_rs`
+// and :code:`chardetng` use whatwg names.
+
+// References
+// ----------
+// https://docs.python.org/3/library/codecs.html#standard-encodings
+// https://encoding.spec.whatwg.org/#legacy-single-byte-encodings
+fn _fix_encoding_name(encoding: &str) -> &str {
+    if encoding == "windows-874" {
+        "cp874"
+    } else {
+        encoding
+    }
+}
 
 #[doc = include_str!("../chardetng_docs/EncodingDetector.md")]
 #[pyclass(name="EncodingDetector")]
@@ -27,7 +42,9 @@ impl EncodingDetectorWrapper {
     #[pyo3(signature=(*, tld, allow_utf8))]
     fn guess(&self, tld: Option<&[u8]>, allow_utf8: bool) -> &'static str {
 
-        self.encoding_detector.guess(tld, allow_utf8).name()
+        _fix_encoding_name(
+            self.encoding_detector.guess(tld, allow_utf8).name()
+        )
     }
 
     #[doc = include_str!("../chardetng_docs/guess_assess.md")]
@@ -37,7 +54,7 @@ impl EncodingDetectorWrapper {
         let (encoding, higher_score) = self.encoding_detector.guess_assess(tld, allow_utf8);
 
         (
-            encoding.name(),
+            _fix_encoding_name(encoding.name()),
             higher_score
         )
     }