rename regex bug?

[dindog]

supose there is a file name : [123][abc]xyz.doc
in perl, \[.*\] is greedy, it will match [123][abc]
to get lazy match, a "?" need to add in the end of .* , in another word, \[.*?\] will match [123].

nevertheless in TC batch rename, seem like with or without the ending "?", the regex is greedy, is it a bug?

[Balderstrom]

You don't appear to be able to turn off greedy-mode, even the recommended syntax from TC's helpfile doesn't work.
e.g
Search: (?-g)\[.*\]
---> Matches: [123][abc]

Whereas if I do this in EmEditor Search/Replace:
Find: \[.*?\]
---> Matches: [123]

Note the same behaviour is seen in TC 7.56a as well.

This could explain the various problems I've had with TC's multi-rename tool in the past.

[ghisler(Author)]

I think that there is some misunderstanding how regular expressions are used in the multi-rename tool: They are not only applied once, but multiple times.

Example 1 (greedy):
Search name: [123][abc]xyz.doc
Search for: \[.*\]
Replace by: a

One match: [123][abc]
Resulting name: axyz.doc

Example 2 (non-greedy):
Search name: [123][abc]xyz.doc
Search for: \[.*\]
Replace by: a

First match: [123]
Second march: [abc]
Resulting name: aaxyz.doc

[Balderstrom]

Wouldn't it be better for that to be an option? (checkbox beside subst) With applying a regex "multiple-times" it makes it significantly harder to write a regex string that will do what you want|expect.

[ghisler(Author)]

Well, that's how search+replace works - if you don't use RegEx, you get the same thing, e.g. searching for "[" will replace all the "[", not just the first.

[fleggy]

I think the current behaviour (multiple replaces) is OK. Perhaps an option like Replace All/Replace Once (only for regexp of course) could be sometimes useful.

[Balderstrom]

@DingDog,

This seems to work in both my Editor w/ RegEx, and TC's MRT

Search: ^\[(.*?)\](.*)
Replace: $1__$2

Behaves the same whether [x] Subbst is checked or not.
Looks like you need to add the trailing (.*) to prevent the MRT's recursion.

Guess it's fine as is, just makes it slightly incompatible with other tools that use Regex, since they would allow a single replace, as well as replace all (recursion).

[fleggy]

@Balderstrom
just to be precise - Replace All is not a recursion. Replace All only finds all occurences in the source string - one by one. I don't wont to be a nitpicker but regexps are my favourite theme. Sorry for a little OT

[white]

As a workaround it is fairly easy to adjust search and replace string so replacement is done only once. You can do this by making sure the search string matches the whole string. Generally you can do this by adding "^.*?" ("^" for higher efficiency) at the beginning and ".*" at the end.

[Balderstrom]

@Balderstrom
just to be precise - Replace All is not a recursion. Replace All only finds all occurences in the source string - one by one. I don't wont to be a nitpicker but regexps are my favourite theme. Sorry for a little OT

@fleggy, I musta tested wrong at some point,

This, ^\[(.*?)\] --- works the same in TC's MRT and my Editor with proper regex implementation.

Note: It's actually not a small task to find an editor that implements Regex properly: when I was doing a search for a new Text Editor a little over two years ago - there were dozens that either 1) didn't support regex at all, 2) only a subset of regex, 3) claimed to be full regex but had major parsing bugs. For example, Notepad2 has a really poor regex implementation.

[ghisler(Author)]

In beta 8, there will be a new checkbox "1x" which allows to replace just the first match. This will work with and without regular expressions.

[white]

In beta 8, there will be a new checkbox "1x" which allows to replace just the first match. This will work with and without regular expressions.

Tested OK using TC 8.0 beta 8 32-bit.

[ghisler(Author)]

Thanks! Can anyone else try it too, please?

[Samuel]

I tested it too and it worked like a charm. With and without RegEx.

[Balderstrom]

Seems to work well, given the following file input:

Code: Select all

default - Copy.bar
default - Copy.br2
default - Copy - Copy.bar
default - Copy - Copy.br2
tcignore - Copy.txt
tcignore - Copy - Copy.txt
wincmd - Copy.ini
wincmd - Copy - Copy.ini

=====================
Search: (.*?)Copy
Replace: ($1)FOO
Options Enabled: x1, Regex
=====================
The first "Copy" is replaced only. Without x1, both are matched/replaced.

I would recommend an example using the x1 in the help doc, that recommends using (.*?) with globs as without the question mark the x1 has no effect at all. Further, if you use two globs, then again you would need two question marks otherwise the x1 has no effect and only the last 'Copy' is replaced. e.g.

Search: (.*)Copy(.*)
Replace: ($1)FOO[$2]

→ The second (or sole) Copy is replaced. [x1] no effect


Search: (.*?)Copy(.*)

→ The first (or sole) Copy is replaced. [x1] no effect.


Search: (.*?)Copy(.*?)

→ Both 'Copy' strings are replaced.
→ x1: The first 'Copy' is replaced.