Regex is not being as "hungry" as it is supposed to

Regex is not being as "hungry" as it is supposed to

[suvayu ali]

Hi everyone,

I have a strange problem, recently some of my regexps started becoming
less "hungry" than usual. For example I have a file like this,

 #1 153030 120 = 423988
 #2 152845 181 = 1500234
 #3 155160 247 = 767821
 #4 155160 310 = 11131347
 #5 155160 319 = 13979167

I wanted to remove the 3rd column of numbers along with the = sign. So
I called `query-replace-regexp' with this regexp replacement,

`[0-9]+ = ->'

But this matches the numbers like this,

120 =
1 =
7 =
0 =
9 =

where as I am expecting it to be this,

120 =
181 =
247 =
310 =
319 =

This is just one instance, I have faced similar other cases. I get the
proper expected match only for the first match, for the subsequent
matches the regexp becomes less "hungry".

Is this a consequence of some conflicting settings or a possible bug?
I see this behaviour with GNU Emacs 24.0.50.1, emacs-snapshot from an
Ubuntu Lucid ppa and emacs compiled from source on Fedora 12. (but the
settings are identical for both these setups)

Thanks in advance for any comments/hints about this issue. :)

-- 
Suvayu

Open source is the future. It sets us free.

Re: Regex is not being as "hungry" as it is supposed to

[Deniz Dogan]

2010/8/6 suvayu ali <fatkasuvayu+linux@gmail.com>:
> Hi everyone,
>
> I have a strange problem, recently some of my regexps started becoming
> less "hungry" than usual. For example I have a file like this,
>
>  #1 153030 120 = 423988
>  #2 152845 181 = 1500234
>  #3 155160 247 = 767821
>  #4 155160 310 = 11131347
>  #5 155160 319 = 13979167
>
> I wanted to remove the 3rd column of numbers along with the = sign. So
> I called `query-replace-regexp' with this regexp replacement,
>
> `[0-9]+ = ->'
>
> But this matches the numbers like this,
>
> 120 =
> 1 =
> 7 =
> 0 =
> 9 =
>
> where as I am expecting it to be this,
>
> 120 =
> 181 =
> 247 =
> 310 =
> 319 =
>
> This is just one instance, I have faced similar other cases. I get the
> proper expected match only for the first match, for the subsequent
> matches the regexp becomes less "hungry".
>
> Is this a consequence of some conflicting settings or a possible bug?
> I see this behaviour with GNU Emacs 24.0.50.1, emacs-snapshot from an
> Ubuntu Lucid ppa and emacs compiled from source on Fedora 12. (but the
> settings are identical for both these setups)
>
> Thanks in advance for any comments/hints about this issue. :)
>
> --
> Suvayu
>
> Open source is the future. It sets us free.
>
>

That's weird, it happens to me to on:
GNU Emacs 24.0.50.1 (i386-mingw-nt6.1.7600) of 2010-07-12 on 3249CTO

I noticed two things when tinkering around with this problem.

1. M-x re-builder correctly highlights the portions we both expect.
2. It's only the *highlighting* in query-replace-regexp that
   fails. When performing the actual replacing, it works just as
   expected.

I'm sensing that this is a bug of some sort and you should report it.

-- 
Deniz Dogan

Re: Regex is not being as "hungry" as it is supposed to

[Andreas Röhler]

Am 06.08.2010 06:04, schrieb Deniz Dogan:
> 2010/8/6 suvayu ali<fatkasuvayu+linux@gmail.com>:
>> Hi everyone,
>>
>> I have a strange problem, recently some of my regexps started becoming
>> less "hungry" than usual. For example I have a file like this,
>>
>>   #1 153030 120 = 423988
>>   #2 152845 181 = 1500234
>>   #3 155160 247 = 767821
>>   #4 155160 310 = 11131347
>>   #5 155160 319 = 13979167
>>
>> I wanted to remove the 3rd column of numbers along with the = sign. So
>> I called `query-replace-regexp' with this regexp replacement,
>>
>> `[0-9]+ = ->'
>>
>> But this matches the numbers like this,
>>
>> 120 =
>> 1 =
>> 7 =
>> 0 =
>> 9 =
>>
>> where as I am expecting it to be this,
>>
>> 120 =
>> 181 =
>> 247 =
>> 310 =
>> 319 =
>>
>> This is just one instance, I have faced similar other cases. I get the
>> proper expected match only for the first match, for the subsequent
>> matches the regexp becomes less "hungry".
>>
>> Is this a consequence of some conflicting settings or a possible bug?
>> I see this behaviour with GNU Emacs 24.0.50.1, emacs-snapshot from an
>> Ubuntu Lucid ppa and emacs compiled from source on Fedora 12. (but the
>> settings are identical for both these setups)
>>
>> Thanks in advance for any comments/hints about this issue. :)
>>
>> --
>> Suvayu
>>
>> Open source is the future. It sets us free.
>>
>>
>
> That's weird, it happens to me to on:
> GNU Emacs 24.0.50.1 (i386-mingw-nt6.1.7600) of 2010-07-12 on 3249CTO
>
> I noticed two things when tinkering around with this problem.
>
> 1. M-x re-builder correctly highlights the portions we both expect.
> 2. It's only the *highlighting* in query-replace-regexp that
>     fails. When performing the actual replacing, it works just as
>     expected.
>
> I'm sensing that this is a bug of some sort and you should report it.
>

same at

GNU Emacs 23.1.92.1 (i686-pc-linux-gnu, GTK+ Version 2.12.0) of 2010-02-19

replacement works as desribed, just the highlighting fails.

Re: Regex is not being as "hungry" as it is supposed to

[Tassilo Horn]

Andreas Röhler <andreas.roehler@easy-emacs.de> writes:

> same at
>
> GNU Emacs 23.1.92.1 (i686-pc-linux-gnu, GTK+ Version 2.12.0) of 2010-02-19
>
> replacement works as desribed, just the highlighting fails.

Ditto with the current emacs 24 bzr trunk.  I've reported the bug as
#6808.

Bye,
Tassilo

Re: Regex is not being as "hungry" as it is supposed to

[suvayu ali]

On 5 August 2010 23:49, Tassilo Horn <tassilo@member.fsf.org> wrote:
> Andreas Röhler <andreas.roehler@easy-emacs.de> writes:
>
>> same at
>>
>> GNU Emacs 23.1.92.1 (i686-pc-linux-gnu, GTK+ Version 2.12.0) of 2010-02-19
>>
>> replacement works as desribed, just the highlighting fails.

Usually when I see the highlighting is wrong I don't proceed with the
replace. I guess that is why I didn't pick up on the fact that the
highlighting is the problem. :-p

After seeing all the responses I proceeded with the replace, and I see
the same behaviour. Thank you every one for confirming this.

>
> Ditto with the current emacs 24 bzr trunk.  I've reported the bug as
> #6808.
>

Thanks a lot Tassilo! :)

> Bye,
> Tassilo
>

-- 
Suvayu

Open source is the future. It sets us free.

Re: Regex is not being as "hungry" as it is supposed to

[Andreas Röhler]

Am 06.08.2010 09:03, schrieb suvayu ali:
> On 5 August 2010 23:49, Tassilo Horn<tassilo@member.fsf.org>  wrote:
>> Andreas Röhler<andreas.roehler@easy-emacs.de>  writes:
>>
>>> same at
>>>
>>> GNU Emacs 23.1.92.1 (i686-pc-linux-gnu, GTK+ Version 2.12.0) of 2010-02-19
>>>
>>> replacement works as desribed, just the highlighting fails.
>
> Usually when I see the highlighting is wrong I don't proceed with the
> replace. I guess that is why I didn't pick up on the fact that the
> highlighting is the problem. :-p
>
> After seeing all the responses I proceeded with the replace, and I see
> the same behaviour. Thank you every one for confirming this.
>
>>
>> Ditto with the current emacs 24 bzr trunk.  I've reported the bug as
>> #6808.
>>
>
> Thanks a lot Tassilo! :)
>
>> Bye,
>> Tassilo
>>
>

BTW, as running into difficulties with regexps is common:
quite often helps ancoring it

^\(.+ \)[0-9]+ =

\1

circumvents the bug

Re: Regex is not being as "hungry" as it is supposed to

[suvayu ali]

On 6 August 2010 00:36, Andreas Röhler <andreas.roehler@easy-emacs.de> wrote:
>
> BTW, as running into difficulties with regexps is common:
> quite often helps ancoring it
>
> ^\(.+ \)[0-9]+ =
>
> \1
>
> circumvents the bug
>

Thank you for this suggestion! :) Whenever working with sed I usually
do that. I guess its good practice when using regexps interactively in
emacs too.

-- 
Suvayu

Open source is the future. It sets us free.