[Bug Report] Replace does nothing when trying to replace ",*" with nothing #2108

Activity

1. ticket-craft

   added

   priority: medium

   on Oct 21, 2026
2. ipswe commented on Oct 21, 2026

   

   ipswe

   on Oct 21, 2026

   Last edited by ipswe

   Author

   More actions

   Version 0.9.0 on Windows 11
3. luigiPinna

   added a commit that references this issue on Oct 22, 2026

   Find and Replace: replace the text a wildcard pattern matches (storyt…

   ...

   333d2e4
4. iluebbe commented on Oct 22, 2026

   

   iluebbe

   on Oct 22, 2026

   More actions

   #2127 checked

   Fixes it. Replace with a * or ? pattern now replaces the text the pattern matches; ,* with nothing turns 10.0.0.1, fe80::1 into 10.0.0.1, as in Excel.

   Setup: PR head a8de4640 (the PR has since been rebased to 5fb58db1 with the same code change), main at 06920de, Windows 11. gridcraft-cli run with cell.set + edit.replace (the main binary and the PR build), compared with Excel for Microsoft 365 (16.0 build 20430) through COM (Range.Replace, part match).

   Cell	Find → Replace	main	PR	Excel
   10.0.0.1, fe80::1	,* → nothing	unchanged (replaced: 1)	10.0.0.1	10.0.0.1
   hello world	h*o → X	unchanged	X world	X world
   hello world	l* → X	unchanged	heX	heX
   xaxbyazb	a?b → X	unchanged	xXyX	xXyX
   a*b*c	~* → +	unchanged	a+b+c	a+b+c
   Hello hello	h?llo → X	unchanged	X X	X X
   Hello hello	h?llo → X, Match case	unchanged	Hello X	Hello X
   a1, b2; c3	,*; → |	unchanged	a1| c3	a1| c3
   ab-cd-ef	-*- → +	unchanged	ab+ef	ab+ef

   Other cases on the PR build (not measured in Excel): * → X replaces the whole text, ~? replaces a literal ?, b? on abcabc gives a_a_, a plain x → - still replaces literally (aXbxc → a-b-c). 200 cells of 32,767 × a with a*b: 0.58 s on the PR build, 0.68 s on main, so no slowdown.

   No problems found in the change.

   Merging: the rebased head 5fb58db1 is based on main at 06920de, so it merges without conflicts. The earlier head conflicted only in tests and contributors/people.toml.

   Tests: the three new tests pass; cargo test -p gridcraft-engine 348 passed, 0 failed; cargo clippy -p gridcraft-engine --all-targets -- -D warnings and cargo fmt --all -- --check are clean (on the PR head, not on the merge).

   ❤️React with ❤️1Reacted by PJ
5. luigiPinna commented on Oct 22, 2026

   

   luigiPinna

   on Oct 22, 2026

   More actions

   @iluebbe thanks for checking this against Excel through COM — the table covers more cases than the tests do. The rebased head 5fb58db1 carries the same change as a8de4640, so your results hold for it.
6. iluebbe commented on Oct 28, 2026

   

   iluebbe

   on Oct 28, 2026

   More actions

   #2127 checked again

   Still fixes it. Since the first check, the wildcard Replace folds case as Find does (9515054), and main was merged in (last at 11bb673). Every wildcard Replace case tried gives Excel's result. The two differences left are ~ patterns that main handles the same way and that #1455 covers.

   Setup: PR head a6c2872, merged locally with main at 70327cd; main at 70327cd (0.13.0 release binaries); Windows 11. gridcraft-cli run with cell.set, edit.replace and edit.find (one new workbook per Replace case), compared with Excel for Microsoft 365 (16.0 build 20527) through COM: Range.Replace and Range.Find/FindNext on a new workbook, closed without saving.

   Earlier finding	Now
   The nine cases of the first table match Excel: ,* with nothing on 10.0.0.1, fe80::1 gives 10.0.0.1; h*o, l*, a?b, ~*, h?llo with and without Match case, ,*;, -*-	Still so, measured again on the merge
   No problems found in the change	Still none

   New cases (part of the cell unless noted). On main every matching pattern with * or ? leaves the cell unchanged, though it counts the cell as replaced. The Match entire cell rows already worked on main:

   Cell	Find → Replace	PR	Excel
   HELLO world	h*o → X	X world	X world
   HELLO world	h*o → X, Match case	unchanged	unchanged
   a?b?c	~? → +	a+b+c	a+b+c
   a~b*c	~~? → X	aX*c	aX*c
   a~b	a~~* → X	X	X
   banana	*a → X	XXX	XXX
   a,b,c	*, → X	XXc	XXc
   a1b2a3b4	a*b* → X	X	X
   hello world	* → X	X	X
   abc	? → X	XXX	XXX
   aXYb	a**b → X	X	X
   hello world	h*d / h*o / ?ELLO* → X, Match entire cell	X / unchanged / X	X / unchanged / X
   hello world	?ELLO* → X, Match entire cell and Match case	unchanged	unchanged
   12345 (number)	2?4 → X / 2*4 → 0	text 1X5 / number 105	text 1X5 / number 105
   ="ab"&"cd"	b*c → X	="aXd" (shows aXd)	="aXd"
   ="xyz"	x?z → Q	="Q"	="Q"
   ΚΑΛΟΣ	λ?ς → X (9515054)	ΚΑX	ΚΑX
   a~b~c	~~ → +	unchanged, as on main	a+b+c
   ab~c	b*~ → X	unchanged, as on main	aX

   Find (edit.find, unchanged by the PR) over ten cells, with h*o, h?llo (with and without Match case), ~*, ~?, a?b (entire cell), b*c (formulas and values), 3 (formulas and values), 2?4, h* and * (entire cell): the same cells as Excel, on main and on the PR. ~~ finds nothing, while Excel finds a~b.

   Problems: none in the change. The last two Replace rows and the ~~ Find differ from Excel as they do on main. A pattern without * or ? is still taken literally, and a trailing ~ escapes the * that matches appends (find.rs#L6-L14). Excel reads ~~ as one ~ and drops a trailing ~. Both rules are listed in #1455, and the PR description leaves them there.

   Merging: merges without conflicts with main at 70327cd.

   Tests: on the merge with main at 70327cd:

   • The four new tests pass, including replace_with_wildcards_folds_case_as_find_does from 9515054.
   • cargo test -p gridcraft-engine: 504 lib and 83 integration tests pass. io::tests::save_fails_upfront_when_the_temp_name_is_taken was gated locally, because it doesn't compile on Windows (Windows: the engine's tests compile and the CLI passes clippy again #2426).
   • cargo clippy -p gridcraft-engine --all-targets -- -D warnings is clean under the stricter workspace lints main gained in Enforce Rust coding standards via Clippy #2711.
   • cargo fmt --all -- --check is clean.