Fix StringIO Content-Length calculation for multi-byte chars

Correctly calculate Content-Length for io.StringIO objects containing
multi-byte characters by measuring the UTF-8 encoded byte length.

Added test case with emoji character to verify the fix.

Fixes: psf#6917

Author: hackspaces

Diff for: src/requests/utils.py

@@ -195,6 +195,18 @@ def super_len(o):
                     # seek back to current position to support
                     # partially read file-like objects
                     o.seek(current_position or 0)
+
+                    # Handle StringIO with multi-byte characters by correctly measuring UTF-8 byte length
+                    if isinstance(o, io.StringIO):
+                        # Save current position
+                        current_pos = o.tell()
+                        # Get entire content and calculate byte length after UTF-8 encoding
+                        o.seek(0)
+                        content = o.read()
+                        # Restore original position
+                        o.seek(current_pos)
+                        # Set total_length to the byte length, not character count
+                        total_length = len(content.encode("utf-8"))
                 except OSError:
                     total_length = 0
 

Diff for: tests/test_utils.py

@@ -1,5 +1,6 @@
 import copy
 import filecmp
+import io
 import os
 import tarfile
 import zipfile
@@ -151,6 +152,13 @@ def test_super_len_with_no_matches(self):
         """Ensure that objects without any length methods default to 0"""
         assert super_len(object()) == 0
 
+    def test_super_len_with_stringio_containing_multibyte_chars(self):
+        """Ensure StringIO with multi-byte characters reports correct byte length"""
+        # emoji takes 4 bytes in UTF-8 but is 1 character
+        s = io.StringIO("💩")
+        # Super len should return 4 (byte length) not 1 (character length)
+        assert super_len(s) == 4
+
 
 class TestToKeyValList:
     @pytest.mark.parametrize(