POSIX regex file expression question

Lemming

hey guys,

am setting up some file-extension filters and am just wondering where or not I'm getting this right

for example:

.*\..*vb.*

will pick up filename1.vbs, or filename2.vbe, or filename.blah.vbs.txt, yes?



similarly ....

.*\.d*l

will pick up filename1.dll and filename2.dpl, but not filename3.dogl, yes?

Typedef

Hmm.

Here is the line I use to reject window$ executable attachments on mailservers if it's any help

/etc/postfix/body_checks


/(filename|name)=".*\.(au|bat|chm|cmd|com|css|dll|dot|exe|hlp|hta|exe|hlp|jse|lnk|ocx|pak|pif|pps|scr|sct|shs|src|vbe|vbs|vxd|wsh"/ REJECT

Lemming

Originally posted by Typedef

/(filename|name)=".*\.(au|bat|chm|cmd|com|css|dll|dot|exe|hlp|hta|exe|hlp|jse|lnk|ocx|pak|pif|pps|scr|sct|shs|src|vbe|vbs|vxd|wsh"/ REJECT

Hmm ... would I be right in saying that that filename setup checks for filename.extension?

If it does, whilst it's a one-stop-filter-shopping-list, what I'm trying to do is create as generic a list as possible, rather than have to list every file-type and/or filetype combination (eg. name.txt.vbs) that I want blocked (since some file extensions are similar - give or take a character) something like:

.*\.(vbs|vbe|dll|dpl|asp|tsp|etc etc).

So what I'm asking is the following:

"Was my syntax correct from my initial post"?

Typedef

That'll pick up a filename.txt.vbs (I just tested it).

/(filename|name)=".*\.(vb*)"/ REJECT

should work for vb(x) I think.

Lemming

Originally posted by Typedef
That'll pick up a filename.txt.vbs (I just tested it).

/(filename|name)=".*\.(vb*)"/ REJECT

should work for vb(x) I think.

Cheers type

one last question (I hope anyway) for ye. The syntax * means one character, whilst if you have .* does this mean that you have 0 or more characters ? Or does it mean that you have one or more?

If you follow the distinction I'm trying to make in that question?

So , say for example I want to filter hta, htt, htm, html would I just have to have .*\.(ht.*) ? or would I have to specify a seperate filter for .html (since it's four characters as opposed to three) ?

krinDar

* means 0 or more occurrences of the previous RE.
. (<period>) matchs any character except new line.
Therefore the RE '.*' matches 0 or more occurences of any characters.

The RE you give will match what you want, but be careful as it will match *any* file that has
'.ht' anywhere in the name e.g important.ht.exe

Check out regexp(5)

Lemming

Originally posted by krinDar
* means 0 or more occurrences of the previous RE.
. (<period>) matchs any character except new line.
Therefore the RE '.*' matches 0 or more occurences of any characters.

Ok, so let me just clarify this.

\.vb* will give me an RE that checks vb(vb n times)

\.vb.* will give me an RE that checks all paterns with .vb in them? (eg. vbs/vbe)

Check out regexp(5)

MyBox# man 5 regexp
No entry for regexp in section 5 of the manual

pants ....

Typedef

man re_syntax

* a sequence of 0 or more matches of the atom
+ a sequence of 1 or more matches of the atom
? a sequence of 0 or 1 matches of hte atom
{m} a sequence of exactly m matches of the atom

. matches any single character
\k (where ks is a non-alphnumeric character) matches that chcaracter taken as an ordinary character, a.g. \\ matches a blackslash character.

rtfm ; )

krinDar

Originally posted by Lemming
Ok, so let me just clarify this.

\.vb* will give me an RE that checks vb(vb n times)

\.vb.* will give me an RE that checks all paterns with .vb in them? (eg. vbs/vbe)

You can use either, they both do the same thing really.

Lemming

Originally posted by Typedef
man re_syntax


rtfm ; )

hehe ... helps if you know that said man page exists to rtfm in the first place

Typedef

: )

Lemming

Originally posted by krinDar
You can use either, they both do the same thing really.

Hmm .. well the understanding I had was that they don't

.*\.vb* checks for name.vb, or name.vbvb, or name.vbvbvb(n times)

whereas

.*\.vb.* checks for name.vb(x), or name.vb(xy) etc.

By the man page so graciously suggested that I rftm (courtesy of Type ), it would appear that to do this you actually would type:

.*\.vb. to do a search for name.vb(x), but something like .*\.vb.. or .*\.vb.* to search for anything more than .vb(x)


yes ? No ?

Typedef

I think so.

Kev

Originally posted by Lemming


.*\.vb* checks for name.vb, or name.vbvb, or name.vbvbvb(n times)

that would check for name.vb or name.vbbbbb or name.v

Typedef

use *\.vb.

Lemming

Cheers guys

now to see what other Regexp situations I can come up ......

Lemming

And she's back .........

lets say I want to filter for .com or .cmd extensions,

instead of putting in:
.*\.cmd
.*\.com
or .*\.(cmd|com)

can I do the following:

.*\.c{1,}o?m{1,}d?

to get the same effect?

From what I /think/ it will interpret is the 'c' character once, 'o' occurs 0-1 times, 'm' occurs once, and then 'd' occuers 0-1 times therefore leaving me with the following possibilities:

.cm
.cmd
.com
.comd


that right ?

Kev

{1,} mean 1 or more times. the + modifier is also a shortcut for this and looks nicer.

so it would match multiple c's and d's

if you want to filter for just com and cmd.

.*\.c[om]d$

the $ mean match the end of the string.

Lemming

Originally posted by Kev
{1,} mean 1 or more times. the + modifier is also a shortcut for this and looks nicer.

so it would match multiple c's and d's

if you want to filter for just com and cmd.

.*\.c[om]d$

the $ mean match the end of the string.

oops .. that should have been c{1}o?m{1}d?

but anyway ..... doesn't the [] only allow the matching of one character? So I could match either 'o' or 'm', but not 'om' ??

Kev

yes it would only match one o or m, if you just want to match cmd or com then you only need one, for more use + or {x,y}

also {1} is redundant.

c{1}o?m{1}d? will match all of

comd
com
cmd
cm