From: Aarni Koskela <akx@iki.fi>
Date: Mon, 17 Mar 2025 12:37:31 +0000 (+0200)
Subject: Avoid hot `isinstance`ing in PO file parse loop
X-Git-Url: http://git.ipfire.org/?a=commitdiff_plain;h=7a8b587cdca438396fd5aca9cf3b1a0977138eed;p=thirdparty%2Fbabel.git

Avoid hot `isinstance`ing in PO file parse loop
---

diff --git a/babel/messages/pofile.py b/babel/messages/pofile.py
index 987193e9..8220637b 100644
--- a/babel/messages/pofile.py
+++ b/babel/messages/pofile.py
@@ -300,16 +300,25 @@ class PoFileParser:
 
     def parse(self, fileobj: IO[AnyStr] | Iterable[AnyStr]) -> None:
         """
-        Reads from the file-like object `fileobj` and adds any po file
-        units found in it to the `Catalog` supplied to the constructor.
+        Reads from the file-like object (or iterable of string-likes) `fileobj`
+        and adds any po file units found in it to the `Catalog`
+        supplied to the constructor.
+
+        All of the items in the iterable must be the same type; either `str`
+        or `bytes` (decoded with the catalog charset), but not a mixture.
         """
+        needs_decode = None
 
         for lineno, line in enumerate(fileobj):
             line = line.strip()
-            if not isinstance(line, str):
-                line = line.decode(self.catalog.charset)
+            if needs_decode is None:
+                # If we don't yet know whether we need to decode,
+                # let's find out now.
+                needs_decode = not isinstance(line, str)
             if not line:
                 continue
+            if needs_decode:
+                line = line.decode(self.catalog.charset)
             if line[0] == '#':
                 if line[:2] == '#~':
                     self._process_message_line(lineno, line[2:].lstrip(), obsolete=True)
diff --git a/tests/messages/test_pofile.py b/tests/messages/test_pofile.py
index 2bcc3df8..ffc95295 100644
--- a/tests/messages/test_pofile.py
+++ b/tests/messages/test_pofile.py
@@ -1068,11 +1068,18 @@ def test_iterable_of_strings():
     """
     Test we can parse from an iterable of strings.
     """
-    catalog = pofile.read_po(['msgid "foo"', b'msgstr "Voh"'], locale="en_US")
+    catalog = pofile.read_po(['msgid "foo"', 'msgstr "Voh"'], locale="en_US")
     assert catalog.locale == Locale("en", "US")
     assert catalog.get("foo").string == "Voh"
 
 
+@pytest.mark.parametrize("order", [1, -1])
+def test_iterable_of_mismatching_strings(order):
+    # Mixing and matching byteses and strs in the same read_po call is not allowed.
+    with pytest.raises(Exception):  # noqa: B017 (will raise either TypeError or AttributeError)
+        pofile.read_po(['msgid "foo"', b'msgstr "Voh"'][::order])
+
+
 def test_issue_1087():
     buf = StringIO(r'''
 msgid ""