fix: avoid IndexError in is_mostly_bin for short tails (#8196)

* fix: avoid IndexError in is_mostly_bin for short tails

`is_mostly_bin` looks up to 4 bytes past the 100-byte cutoff to find a
clean UTF-8 character boundary. when the input is just over 100 bytes
and starts with a continuation byte at index 100, the lookahead reads
past the end of the buffer.

cap the loop end at `len(s)`. existing fallback to `s[:100]` still
covers the case where every byte in the lookahead window is a
continuation byte.

fixes #8188

* [autofix.ci] apply automated fixes

---------

Co-authored-by: Julio César Suástegui <juliosuas@users.noreply.github.com>
Co-authored-by: autofix-ci[bot] <114827586+autofix-ci[bot]@users.noreply.github.com>

CHANGELOG.md

@@ -7,6 +7,9 @@
 
 ## Unreleased: mitmproxy next
 
+- Fix `IndexError` in `is_mostly_bin` when exporting flows to HAR with payloads
+  that have a UTF-8 continuation byte at the 100-byte cutoff.
+  ([#8196](https://github.com/mitmproxy/mitmproxy/pull/8196), @juliosuas)
 
 ## 12 April 2026: mitmproxy 12.2.2
 

mitmproxy/utils/strutils.py

@@ -132,7 +132,7 @@ def is_mostly_bin(s: bytes) -> bool:
     # Cut off at ~100 chars, but do it smartly so that if the input is UTF-8, we don't
     # chop a multibyte code point in half.
     if len(s) > 100:
-        for cut in range(100, 104):
+        for cut in range(100, min(104, len(s))):
             is_continuation_byte = (s[cut] >> 6) == 0b10
             if not is_continuation_byte:
                 # A new character starts here, so we cut off just before that.

test/mitmproxy/utils/test_strutils.py

@@ -94,6 +94,11 @@ def test_is_mostly_bin():
     assert not strutils.is_mostly_bin(b"aaaaa" + 50 * "𐍅".encode())
     # only utf8 continuation chars
     assert strutils.is_mostly_bin(150 * b"\x80")
+    # regression #8188: payloads with len 101-103 and a continuation byte at
+    # the 100-byte cutoff used to raise IndexError because the lookahead loop
+    # ran past the end of the string. Should not raise.
+    for tail in (1, 2, 3):
+        strutils.is_mostly_bin(b"a" * 100 + b"\x80" * tail)
 
 
 def test_is_xml():