fix: handle multibyte characters in physical plan tree renderer - #24664
fix: handle multibyte characters in physical plan tree renderer#24664rustyconover wants to merge 2 commits into
Conversation
saadtajwar
left a comment
saadtajwar
left a comment
There was a problem hiding this comment.
Gotta love Rust 😆 this makes sense & LGTM!
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## main #24664 +/- ##
=======================================
Coverage 81.44% 81.44%
=======================================
Files 1118 1118
Lines 399550 399568 +18
Branches 399550 399568 +18
=======================================
+ Hits 325398 325414 +16
+ Misses 55154 55153 -1
- Partials 18998 19001 +3 ☔ View full report in Codecov by Harness. 🚀 New features to boost your workflow:
|
nuno-faria
left a comment
nuno-faria
left a comment
There was a problem hiding this comment.
Thanks @rustyconover. Could you also add a sqllogictest at explain_tree.slt? Something like this which currently panics on main:
create table t (a varchar);
explain select * from t where a = '🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤🌤';
kosiew
left a comment
kosiew
left a comment
There was a problem hiding this comment.
Thanks for working on this. The character-based handling looks like a good fix for the UTF-8 boundary issue, and the focused multibyte tests cover the regression nicely.
I left one non-blocking suggestion about terminal display width. It is pre-existing behavior and does not need to be addressed as part of this PR.
| let chars: Vec<char> = source.chars().collect(); | ||
|
|
||
| while character_pos < chars.len() { | ||
| // Treating each char as width 1 for simplification |
There was a problem hiding this comment.
Nice fix. As a follow-up, it might be worth considering terminal display-cell width here, for example with unicode-width, rather than counting Unicode scalar values. This fixes the UTF-8 boundary panic, but wide emoji and CJK characters can still occupy more than one terminal column, which may cause the rendered tree to exceed its box or misalign connectors. This looks like pre-existing behavior to me, so I would not consider it blocking for this PR.
5 participants
Which issue does this PR close?
Rationale for this change
Rendering a physical plan in tree format can panic when an operator name or detail contains multibyte UTF-8 characters. The renderer measures text in characters but previously used those character positions as byte offsets when slicing strings. Those offsets are not necessarily valid UTF-8 boundaries.
What changes are included in this PR?
Use character-based iteration consistently when truncating and wrapping text in the physical plan tree renderer. Also use the number of characters, rather than the number of bytes, when checking for remaining text.
Are these changes tested?
Yes. Unit tests cover both wrapping and truncating text containing multibyte characters.
The following checks pass:
cargo fmt --all -- --checkcargo test -p datafusion-physical-plan --libcargo clippy -p datafusion-physical-plan --all-targets --all-features -- -D warningsAre there any user-facing changes?
EXPLAINoutput using the tree renderer no longer panics when rendered plan details contain multibyte UTF-8 characters. There are no public API changes.