Repository navigation
[BUG] Text.wrap(console, width=0) raises ValueError in chop_cells (range step 0) #4234
Description
Describe the bug
Text.wrap(console, width=0) raises ValueError: range() arg 3 must not be zero for ordinary (single-cell) text. The same call with double-width text does not crash, so the behaviour is inconsistent.
The crash is in rich/cells.py::chop_cells, whose single-cell fast path builds range(0, len(text), width):
if _is_single_cell_widths(text):
return [text[index : index + width] for index in range(0, len(text), width)]With width == 0 the step is 0, which range rejects. The general (wide-character) path below it has no such step and returns without raising, so chop_cells("hello", 0) crashes while chop_cells("你好", 0) does not.
It reaches the public API through Text.wrap -> _wrap.divide_line -> chop_cells (the word_length > width fold branch, taken by any non-empty word when width is 0). Width 0 is reachable when text is rendered into a zero-interior region (a table column squeezed to zero, a nested panel with no room, a console sized to 0); Rich otherwise degrades gracefully at tiny widths, so a hard crash here stands out.
Minimal repro (runs as-is):
import io
from rich.console import Console
from rich.text import Text
console = Console(file=io.StringIO())
Text("hello world").wrap(console, 0) # ValueError: range() arg 3 must not be zeroLower-level, showing the single-cell vs wide-char inconsistency:
from rich._wrap import divide_line
divide_line("hello", 0) # ValueError: range() arg 3 must not be zero
divide_line("你好", 0) # returns normallyTraceback:
File ".../rich/text.py", line 1239, in wrap
offsets = divide_line(str(line), width, fold=wrap_overflow == "fold")
File ".../rich/_wrap.py", line 59, in divide_line
folded_word = chop_cells(word, width=width)
File ".../rich/cells.py", line 338, in chop_cells
return [text[index : index + width] for index in range(0, len(text), width)]
ValueError: range() arg 3 must not be zero
Suggested fix
Guard chop_cells against a non-positive width (return [text] when width <= 0, since text cannot be chopped into zero-width lines). That removes the crash and also fixes the wide-character path, which for width == 0 currently returns a surprising ["", "你", "好"]. I am happy to open a PR with the guard and a regression test.
- I've checked docs and closed issues for possible solutions.
- I can't find my issue in the FAQ.
Platform
Click to expand
Reproduced on Rich master (commit 9d8f9a3, declared version 15.0.0), Python 3.10, Windows — but the code path is platform-independent (pure width arithmetic).
Activity
UzairRehman4 commentedon Oct 8, 2026
I'd like to work on this, with AI assistance (Claude Code) per AI_POLICY.md — flagging that now, before any PR, since the policy requires approval here first.
Reproduced the crash directly against current `master`:
```
Verified against all four repro cases from the issue - none crash now:
Also checked negative width, which wasn't in the original report but hits the same inconsistency the other way (single-cell path still raises the same `ValueError`, wide-character path silently returns `[]` instead of the whole text) — the `width <= 0` guard (rather than `width == 0`) covers that too.
Added `tests/test_cells.py::test_chop_cells_non_positive_width`, parametrized over 0/-1/-5 against both single-cell and wide-character text. Verified it actually catches the regression: reverted the guard, confirmed all 3 cases fail with the exact described symptoms, restored it, confirmed all 72 tests in `test_cells.py` pass. Full suite otherwise unaffected (982 passed either way).
Happy to open the PR as soon as this has the go-ahead.