Fix str.expandtabs aborting on a tab size of zero - #8562

"a\tb".expandtabs(0) takes the interpreter down:

$ rustpython -c 'print("a\tb".expandtabs(0))'
thread 'main' panicked at library/alloc/src/raw_vec/mod.rs:28:5:
capacity overflow

CPython returns 'ab': with no width to advance to, the tabs come out and nothing else moves. expandtabs(-1) is the same, because ExpandTabsArgs::tabsize sends every negative value to 0.

rustpython_common::str::expandtabs walks the string with the tab stop in tab_size and the current column in col_count, and on a tab it does tab_size - col_count. Both start at zero, so the first character makes col_count 1 while tab_size stays 0, and the subtraction underflows. The run of spaces asked for next is usize::MAX, and the allocation aborts the process. That is why a tab has to follow something: "\ta".expandtabs(0) starts the tab at column 0, subtracts 0 from 0, and comes out right by accident.

BytesInner::expandtabs already returns early for this, filtering the tabs out. The string version now does the same thing.

Reachable from any Python code, no C API and no unusual build needed:

"\t".join(parts).expandtabs(user_supplied_width)

Checked against CPython 3.14.7 over 23 subjects by 10 tab sizes by str, bytes and bytearray, 460 cases. All 460 agree now. Without the change 42 of them abort the process, and the bytes half is not among them, which is where the shape of the fix came from.

Tests are in crates/common/src/str.rs next to the function and in extra_tests/snippets/builtin_str.py alongside the expandtabs overflow cases that landed in #8524. Both cover the zero and negative sizes, ASCII and non-ASCII, tabs after a newline and after a carriage return, and a set of ordinary tab sizes so a fix that reached too far would show up.

Lib/test/string_tests.py has no case for a zero tab size, so test_str and test_bytes pass either way. They still pass here, along with test_textwrap, 523 tests.

Summary by CodeRabbit

• Bug Fixes

  • Fixed expandtabs handling for zero tab sizes, preventing excessive memory allocation and process termination.
  • Zero tab sizes now remove tab characters consistently across strings, bytes, and bytearrays.
  • Preserved existing behavior for positive tab sizes, including multiline and multibyte text.

• Tests

  • Added coverage for zero, negative, and positive tab sizes across varied input types and content.
  • Added coverage for Unicode-safe replacement using empty patterns.