#439 "Whole word" option does not work in Replace

[Anonymous]

Problem:
Trying to replace all occurrences of "debug_client" with "DebugClient", when the "Whole word" option is selected in the Search&Replace dialog, causes also partial word replacements. E.g. both of the following were replaced:

genericapi: debug_client
genericapi_all: debug_client_all

Note that in the 2nd line, debug_client does not appear as a whole word.

Patch:
The _replace() method in SearchAndReplace.java file noWordSep is assigned with wordBreakChars. So i changed it to assign noWordSep

[Anonymous]

Patch

[Anonymous]

Hey I am creating this patch with the version 4.5.2

[Jarek Czekalski]

This is a patch for bug #3543524
https://sourceforge.net/tracker/index.php?func=detail&aid=3522497&group_id=588&atid=100588

[Jarek Czekalski]

Malahti, you found and fixed an evident bug. Thanks for that. But I wonder if this change is good:

- if (noWordSep == null)
+ if (noWordSep == null || noWordSep.isEmpty())

That means that user will not be able to treat _ as a non-word character. Even if he leaves "Extra word characters" empty, the underscore will still be treated as being part of the word. _ is a default "Extra word character" (see jedit.props), isn't it enough?

[Anonymous]

The reason I included "noWordSep.isEmpty()" is that before changing WordBreakChars to noWordSep in _replace() the value that noWordSep get in setNoWordSep is an empty string(because wordBreakChars is empty in jedit.props). At that time if "noWordSep == null" condition only is checked, then noWordSep is assigned with the empty string. To avoid that i included "noWordSep.isEmpty()" condition as a precautionary. If I am wrong clarify me. Thanks

[Jarek Czekalski]

Are you saying: "it's a mistake"?

[Anonymous]

Ya it is a mistake. No need for that condition. Do i have to submit the patch again without this condition?

[Jarek Czekalski]

I'll change it manually, no need to resubmit.

Regarding wordBreakChars: it is set by mode. So for java files you get ,+-=<>/?^&*
by buffer.getProperty("wordBreakChars") in BeanShell console. That means that a in a+b was not treated as a separate word and your patch will fix it as well.

[Jarek Czekalski]

r21965