1074 replies to this topic

[TLM]

Thanks sinkfaze, your needle works

Why would you need word boundaries?

Well the fact that I'm still learning or \K is not a well documented option might have something to do with it ..
I get this when I highlight \K in RegExr

So \b = word boundaries only, got ya! I figured it was "anything" boundaries or characters could be substituted.


Plz correct me if I'm wrong ( which is likely ):

\[\K[^\]]+

literal open sq bracket ( or maybe class start/open )
*unknown option \K* ( is it untill?? )
open class
not literal closed sq bracket
class close/end
all instances
Sound close??

[sinkfaze]

\K is a different method of lookbehind. The traditional method (?<=...) cannot support quantifiers of varying size (i.e. *, ?, and +) whereas \K can. That aside of it just being more simple to use, IMO.

Everything to the left of \K will not be returned but must be matched, so the regex finds the first instance of a literal open bracket (which is not returned) then finds everything immediately after it that is not a literal closed bracket.

[TLM]

it just being more simple to use, IMO.

Agreed 100%

Everything to the left of \K will not be returned but must be matched, so the regex finds the first instance of a literal open bracket (which is not returned) then finds everything immediately after it that is not a literal closed bracket.

This totally clears it up for me!!
Thanks so much

BTW this is what I get in Expresso for \K

Any idea of a helper that supports it?

Thanks again..

[TLM]

\K is a different method of lookbehind. The traditional method (?<=...) cannot support quantifiers of varying size (i.e. *, ?, and +) whereas \K can. That aside of it just being more simple to use, IMO.

Everything to the left of \K will not be returned but must be matched, so the regex finds the first instance of a literal open bracket (which is not returned) then finds everything immediately after it that is not a literal closed bracket.

The escape sequence \K is similar to a look-behind assertion because it causes any previously-matched characters to be omitted from the final matched string. For example, foo\Kbar matches "foobar" but reports that it has matched "bar".

I'm so blind! Now I see why its not supported anywhere else..

As punishment for my lack of fastidiousness,
I will now force myself to understand the look ahead/behind assertions

file=c:\testfile.file
msgbox % regExReplace( file, "\.|[\w]:\\(?=.*[\w])|(?!.*[\.])[\w]+" )

Let me go get my whip .
Something tells me:1. I do not need the 1st alternative to add the single literal period to the file extension.
2. I can run into problems if the file name has an extra period in it ( actually this is ok).
3. ( redundant ) Theres a way to add the period to the negative look ahead alternative.
4. Most importantly, will not support more than 1 dir depth (grrr)Feel free to correct me if I'm wrong and as always your wisdom and guidance are greatly appreciated
thnx agn..

edit:

This seems to fix dir depth

file=c:\download\dir1\dir2\testfile.file
msgbox % regExReplace( file, "^[\w]:\\|\.|[\w]+[\\](?=.*[\w]+)|(?!.*[\.])[\w]+" )

but still leaves the question ,"Is the extra alternative needed for single period?".
Also what if a dir contains a period :?

[Frankie]

I'm trying to match a HTML tag parameter. I've tested my needle here: <!-- m -->http://www.gethifi.com/tools/regex<!-- m -->

As you can see in my example, the needle has quotes. For simplicity I put it on a separate line, outside of an expression. It should match part of an <input> tag. More specifically the value of the id parameter.

HTML =
(
<div id="gaia_loginbox"> 
<table class="form-noindent" cellspacing="3" cellpadding="5" width="100`%" border="0"> 
  <tr> 
  <td valign="top" style="text-align:center" nowrap="nowrap"
        bgcolor="#e8eefa"> 
  <input type="hidden" name="ltmpl"
             value="default"> 
  <input type="hidden" name="ltmplcache"
             value="2"> 
  <input type="hidden" id="pstMsg"
        name="pstMsg"  value="" /> 
  <input type="hidden" id="dnConn"
        name="dnConn"  value="" /> 
  <div class="loginBox"> 
  <table id="gaia_table" align="center" border="0" cellpadding="1" cellspacing="0"> 
  <tr> 
<td colspan="2" align="center"> 
  <font size="-1"> 
  Sign in with your
  </font> 
  <table> 
  <tr> 
  <td valign="top"> 
)
N = <input.+id="(.+)"
While Pos := RegExMatch(HTML, N, Match, Pos + 1)
	Msgbox "%Match1%"`n`n"%Match%"
return

It will do other things once it finds them, and it usually gets it's input from a COM object, but I want to get this part working first before I worry about that...

[Frankie]

Thanks! That works great.

EDIT: It worked when I plugged it into my driver, but not in my actual script. Here's what I've done so far. I'm sure there's a more COM-efficient way to do this, but I prefer to parse the HTML.

Relevant code is the ChangeLabels: sub.

OnExit, ExitSub
Gui, Add, Button, gChangeLabels Default, Label Website Fields

; URL Prompt, is displayed first
Gui, 2: Margin, 0, 0
Gui, 2: Add, Picture, w300 h100, Web.jpg
Gui, 2: Font, s14
Gui, 2: Add, Edit, vURL +BackgroundTrans x30 y30 w240, http:://
Gui, 2: Add, Button, x500 y150 gEnterURL Default, Hidden_Button
Gui, 2: Show, w300 h100, Enter the URL
return

EnterURL:
Gui, 2: Submit
pwb := ComObjCreate("InternetExplorer.Application")
pwb.Silent := true
pwb.Navigate(URL)
Gui, 1: Show
return

StartOver:
pwb.Quit()
pwb =
Gui, Hide
Gui, 2: Show, w300 h100, Enter the URL
return

[color=red]ChangeLabels:[/color]
While pwb.Busy = true
	Sleep, 100
;Traytip, t, t ; Edit2: Commented this out, this just showed that it gets past checking if pwb is busy
HTML := pwb.Document.body.innerhtml
N = <input[^\r\n]*?id="\K[^"]*
pos := 0
While Pos := RegExMatch(HTML, N, Match, Pos + 1)
	Msgbox "%Match%"
return

ExitSub:
pwb.Quit()
pwb = 
ExitApp
return

If you want to test, should work on autohotkey.net.

[sinkfaze]

This works for me:

HTML =
(
<div id="gaia_loginbox">
<table class="form-noindent" cellspacing="3" cellpadding="5" width="100`%" border="0">
  <tr>
  <td valign="top" style="text-align:center" nowrap="nowrap"
        bgcolor="#e8eefa">
  <input type="hidden" name="ltmpl"
             value="default">
  <input type="hidden" name="ltmplcache"
             value="2">
  <input type="hidden" id="pstMsg"
        name="pstMsg"  value="" />
  <input type="hidden" id="dnConn"
        name="dnConn"  value="" />
  <div class="loginBox">
  <table id="gaia_table" align="center" border="0" cellpadding="1" cellspacing="0">
  <tr>
<td colspan="2" align="center">
  <font size="-1">
  Sign in with your
  </font>
  <table>
  <tr>
  <td valign="top">
)
Pos=1
N = [color=red]is)[/color]<input.*?id="\K[^"]+
While	Pos :=	RegExMatch(HTML, N, M, Pos + StrLen(M))
	Msgbox "%M%"
return

[TLM]

Quick query for extracting direct image URL's from FaceBook img elements.

Wanted to make sure this was strong enough:

eL = <img class="spotlight" alt="" aria-describedby="fbPhotoTheaterCaption" aria-busy="false" src="http://a3.sphotos.ak.fbcdn.net/hphotos-ak-snc6/200761_143748095692151_202001708838478_286803_1275819_n.jpg">

msgbox % regExReplace( eL, "[color=#ff6600][\W\w]+src=`""|`"">[/color]" )

The format is always the same plus I'm not searching individual chrs at all.
Should be ok right??

Let me know what you think and I'll try breaking it in the meantime..

thnx

[Frankie]

If for whatever reason there was even a space after the last quote before the > it would fail. Besides that (not sure how reliable facebook's pages are), it looks fine.

Nice regex btw. I need to work on my pattern matching skill.

[TLM]

If for whatever reason there was even a space after the last quote before the > it would fail. Besides that (not sure how reliable facebook's pages are), it looks fine.
Nice regex btw. I need to work on my pattern matching skill.

Good point although the format is pretty much unless Zukerberg pulls a fast one :lol:.
jic, I could potentially use a reverse search for img type but meh this'l get spaces I think:

clipboard := regExReplace( clipboard, "[\W\w]+src=`""|[\s]*`"">" ) ; places img URL on clipboard

Thnx for the vote of confidence, regex is a must have skill set for me now!
Keep at it, it'l click in trust me

[Gerakon]

I'm trying to do a regex replace. I'm trying to replace the last backslash with a number that will eventually be from a variable (though in the example it's currently just 001. From the documentation I've looked at, the ? should make the expression less greedy, but it doesn't seem to be working.

NewStr := RegExReplace("\\192.168.1.2\backup$\Home\Desktop", "\\.*?$", "\001\")
MsgBox, %NewStr%

The result I'm looking for is
\\192.168.1.2\backup$\Home\001\Desktop

Thanks,
Gerakon

[Tuncay]

This works for your example, but I have not tested in other cases. Can be very tricky. Especially I dont know if the backslash is at another position or not existing.
It works the way that the part before and after the backslashes is saved in a variable (inside the regex system) and used as backreference in replace field.

NewStr := RegExReplace("\\192.168.1.2\backup$\Home\Desktop", "^(.*)\\(.*?)$", "$1\001\$2")
MsgBox, %NewStr%

[guest3456]

i'm a complete newb at regex, don't understand a lick

i need to match a string with either "$" in it or "/", they can be anywhere

can anyone help?

[nimda]

\$\/

is your needle