From 9f90967b0a165c40a8a497b5dab75cc89ecdbc9a Mon Sep 17 00:00:00 2001
From: "Jeong, YunWon" <69878+youknowone@users.noreply.github.com>
Date: Fri, 14 Aug 2026 23:32:36 +0900
Subject: [PATCH] str: count a stepped slice's characters with a ceiling
 division (#8525)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

`do_stepped_slice` and `do_stepped_slice_reverse` took the character count of
a non-ASCII result as `(range.len() / step) + 1`, which overshoots by one
whenever the span is an exact multiple of the step: `"aéc"[::3]` is one
character and reported two, `"가나다라"[::2]` two and reported three. ASCII
subjects were unaffected, since that arm collects into an `AsciiString` whose
length comes from the data.

The count is stored as the string's character length, so the result then
claimed a character its buffer does not hold, and `reversed()` on it indexed
past the end and panicked:

    s = "".join(["a", "é", "c"])
    list(reversed(s[::3]))      # index out of bounds: the len is 1 but the index is 1

Use `div_ceil`, which is the number of elements the underlying range yields.

A differential over 44331 subscript and slice shapes -- every combination of
start, stop and step over strings straddling the ASCII split, the surrogate
range and the astral plane -- now matches CPython 3.14 exactly, where it
differed on 3100 lines before.

Assisted-by: Claude
---
 crates/vm/src/builtins/str.rs                 |  8 ++---
 .../snippets/builtin_str_unicode_slice.py     | 29 +++++++++++++++++++
 2 files changed, 33 insertions(+), 4 deletions(-)

diff --git a/crates/vm/src/builtins/str.rs b/crates/vm/src/builtins/str.rs
index 07325159a..ae8ba0ed0 100644
--- a/crates/vm/src/builtins/str.rs
+++ b/crates/vm/src/builtins/str.rs
@@ -1870,14 +1870,14 @@ impl SliceableSequenceOp for PyStr {
                 .collect::<AsciiString>()
                 .into(),
             PyKindStr::Utf8(s) => {
-                let char_len = (range.len() / step) + 1;
+                let char_len = range.len().div_ceil(step);
                 let mut out = String::with_capacity(2 * char_len);
                 out.extend(s.chars().skip(range.start).take(range.len()).step_by(step));
                 // SAFETY: char_len is accurate
                 unsafe { Self::new_with_char_len(out, char_len) }
             }
             PyKindStr::Wtf8(w) => {
-                let char_len = (range.len() / step) + 1;
+                let char_len = range.len().div_ceil(step);
                 let mut out = Wtf8Buf::with_capacity(2 * char_len);
                 out.extend(
                     w.code_points()
@@ -1900,7 +1900,7 @@ impl SliceableSequenceOp for PyStr {
                 .collect::<AsciiString>()
                 .into(),
             PyKindStr::Utf8(s) => {
-                let char_len = (range.len() / step) + 1;
+                let char_len = range.len().div_ceil(step);
                 // not ascii, so the codepoints have to be at least 2 bytes each
                 let mut out = String::with_capacity(2 * char_len);
                 out.extend(
@@ -1914,7 +1914,7 @@ impl SliceableSequenceOp for PyStr {
                 unsafe { Self::new_with_char_len(out, char_len) }
             }
             PyKindStr::Wtf8(w) => {
-                let char_len = (range.len() / step) + 1;
+                let char_len = range.len().div_ceil(step);
                 // not ascii, so the codepoints have to be at least 2 bytes each
                 let mut out = Wtf8Buf::with_capacity(2 * char_len);
                 out.extend(
diff --git a/extra_tests/snippets/builtin_str_unicode_slice.py b/extra_tests/snippets/builtin_str_unicode_slice.py
index 252f84b1c..1d35c6c48 100644
--- a/extra_tests/snippets/builtin_str_unicode_slice.py
+++ b/extra_tests/snippets/builtin_str_unicode_slice.py
@@ -59,3 +59,32 @@ assert hebrew_text[30:10:-3] == "אםהֱאּ "
 assert len(hebrew_text[30:10:-3]) == 7
 assert hebrew_text[30:10:-1] == "א ,םיִהֹלֱא אָרָּב ,"
 assert len(hebrew_text[30:10:-1]) == 20
+
+
+# A stepped slice whose span is an exact multiple of the step ends on the last
+# character it collects rather than one past it, so the character count is the
+# span divided by the step and not one more. The subject goes through a
+# variable because a constant subscript is folded at compile time and would
+# never reach the runtime slice at all.
+def stepped(s, step):
+    return s[::step]
+
+
+for subject, step, expected in [
+    ("a\u00e9c", 3, "a"),
+    ("가나다라", 2, "가다"),
+    ("가나다라마바", 3, "가라"),
+    ("가나다라", -2, "라나"),
+    ("가나다라마바", -3, "바다"),
+    ("\U0001f600\U0001f601\U0001f602\U0001f603", 2, "\U0001f600\U0001f602"),
+]:
+    sliced = stepped(subject, step)
+    assert sliced == expected, (subject, step, sliced)
+    assert len(sliced) == len(expected), (subject, step, len(sliced))
+    # An overstated count makes the string claim characters its buffer does not
+    # hold, which reversed() then reads past.
+    assert list(reversed(sliced)) == list(expected)[::-1]
+
+assert len(stepped(hebrew_text, 2)) == 30
+assert len(stepped(hebrew_text, 4)) == 15
+assert len(stepped(hebrew_text, -2)) == 30