{SOLVED} Need help with StringRegExpReplace

[fett8802]

Hello All,

So I'm trying to generate an html file based on an input. It's an added feature to a program I've been creating for my workplace. The user will write their code like this:

For bolded text: *This is bolded text*

Now, what I need my function to do is replace ONLY the *'s that are directly next to something that isn't a blank space. Also, it needs to be able to to *T as '<b>T' and t* as 't</b>'. However, it should NOT replace an * all by itself.

Now, I don't need the function written (as it is going to be doing a LOT of substitution), but I've seen other people get help on StringRegExpReplace and it's always kind of confused me.

I didn't see an exact way to do it in the help file, and was hoping some of you more knowledgeable then me on this particular function could give me some help.

Thanks for you help!

-Fett

Edited January 11, 2011 by fett8802

[Melba23]

fett8802,

I am not very good at SREs, so this can surely be bettered:

$sText = "*This is * bolded* *As i*s this*"

$sText = StringRegExpReplace($sText, "( |\A)(\*)(\S)", "$1<b>$3")
$sText = StringRegExpReplace($sText, "(\S)(\*)( |\z)", "$1</b>$3")

ConsoleWrite($sText & @CRLF)

Explanation of first SRE:

( |\A) = Capturing group of either a space or the beginning of a string
(\*)   = Capturing group of a single "*"
(\S)   = Capturing group of any non-whitespace character

$1<b>$3 = First captured group followed by "<b>" and the third group

Second SRE is essentially the same, but with "end of string" and a reversed order. You can see they ignore the * characters with no adjacent space.

No doubt a real SRE guru will be along in a minute to give you a sexy solution, but this does at least work!

M23

[fett8802]

Thanks Melba!

So, (for my understanding) this string: "( |\A)(\*)(\S)"

is calling " *a" or " *f" etc. , not all " " and "*" and "d", but all three together in a row as a group of three because of the quotes?

No wonder those strings looked uber-confusing in other people's code. Yikes.

Anyway, thanks for the lesson Melba, it is much appreciated!

-Fett

[Melba23]

fett8802,

Nothing to do with the quotes - they are just a required part of the SRER syntax. It is the justaposition of the 3 elements that is the key - it will only match when it finds all 3 together.

M23

[fett8802]

Absolutely beautiful. That solves all of my problems. You have made my day!

Hopefully I can use SRE's more in the future.

-Fett

[jchd]

This can be catchy but I'd gamble the following:

StringRegExpReplace($s, '\*(\w+)\*', '<b>$1</b>')

For instance with the input

$s = 'this * *is* text but not all * ** *** are replaced even *this is not * and *neither* this *--* but that number *666* is (but not *1.23*)!'

you get

'this * <b>is</b> text but not all * ** *** are replaced even *this is not * and <b>neither</b> this *--* but that number <b>666</b> is (but not *1.23*)!'

You may want to play with multiline option and such, if necessary.

[iamtheky]

I am finding it easier to learn some SRER by busting each step out individually first to make sure Ive covered all the bases:

$string = "* *b *1 ** *B *0 *"


$out1 = StringRegExpReplace ($string , "\*" , "<b>")    ; changes all asterix to <b> tags

msgbox (0, '' , $string & @CRLF & $out1)

$out2 = StringRegExpReplace ($out1 , "<b><b>" , "**") ; finds any doubled up <b> tags and changes them back to asterix

msgbox (0, '' , $string & @CRLF & $out2)

$out3 = StringRegExpReplace ($out2 , "<b>\z" , " *") ; checks the end for <b> tag and changes it as there would be nothing after

msgbox (0, '' , $string & @CRLF & $out3)

$out4 = StringRegExpReplace ($out3 , "<b>\h" , "* ") ; checks for <b> tags with white space after and changes them back

msgbox (0, '' , $string & @CRLF & $out4)

Edited January 11, 2011 by iamtheky

[fett8802]

Ok, so I guess my question now (for learning) is, why does it need to be "$1<b>$3" and not just "</b>"

[jchd]

Explanation of my solution:

\* match an asterisk which must be immediately followed by ...

(\w+) one or more words (they are captured as $1) immediately followed by ...

\* another asterisk

The replace part is then obvious.

It would be easy to be more specific about what constitutes a word (look at how the pattern matches various numbers).

[GEOSoft]

I am finding it easier to learn some SRER by busting each step out individually first to make sure Ive covered all the bases:

$string = "* *b *1 ** *B *0 *"


$out1 = StringRegExpReplace ($string , "\*" , "<b>")    ; changes all asterix to <b> tags

msgbox (0, '' , $string & @CRLF & $out1)

$out2 = StringRegExpReplace ($out1 , "<b><b>" , "**") ; finds any doubled up <b> tags and changes them back to asterix

msgbox (0, '' , $string & @CRLF & $out2)

$out3 = StringRegExpReplace ($out2 , "<b>\z" , " *") ; checks the end for <b> tag and changes it as there would be nothing after

msgbox (0, '' , $string & @CRLF & $out3)

$out4 = StringRegExpReplace ($out3 , "<b>\h" , "* ") ; checks for <b> tags with white space after and changes them back

msgbox (0, '' , $string & @CRLF & $out4)

your second expression would be better as

$out2 = StringRegExpReplace($out1, "(?i)(<b>)\1+", "$1")

The next one should be

$out3 = StringRegExpReplace($out2, "(?i)(<b>)\s*$", "$1")

[Melba23]

fett8802,

why does it need to be "$1<b>$3" and not just "</b>"

I assume that was adddressed to me.

The SRER searches for "something", "*", "something". If we simply replaced all of that with just "</b>", we would lose the 2 "something"s on either side. So we made sure that we captured them (by putting brackets around them ) and then we replace all 3 sections by the 2 "something"s with a new section in the middle. That sounds horribly complicated - but is actually reasonably simple if you take it slowly.

As I said earlier, I am not an expert on SREs by any means. I freely admit that I find them the hardest thing I have ever tried to learn in computing. So you should also look carefully as jchd's suggestions - he is far better at SREs than I ever will be.

M23

[jchd]

Don't believe that one second. The actual reason is that I've lost all my hair trying RegExps while Melba has kept all of it. To me this guy is just lazy, hairy but lazy

[fett8802]

The SRER searches for "something", "*", "something". If we simply replaced all of that with just "</b>", we would lose the 2 "something"s on either side. So we made sure that we captured them (by putting brackets around them ) and then we replace all 3 sections by the 2 "something"s with a new section in the middle. That sounds horribly complicated - but is actually reasonably simple if you take it slowly.

That actually makes perfect sense. So, does the number to the right of the $ denote the "something" order?

Like, ( |A)(\*)(\S) ----- ( |A) is the $1, (\*) is the $2, and (\S) is the $3? So then I could put any length in $2 as long as I denote $1 and $3? That makes sense.

Thanks Melba!

[Melba23]

fett8802,

does the number to the right of the $ denote the "something" order?

Not quite - it is the index of the capturing group (the ones in parentheses). You can also have non-capturing groups, perhaps to aid the location of the searhed-for characters in the string, but obviously there is no way to reuse them later.

Told you it was complex!

M23

[GEOSoft]

fett8802,

I assume that was adddressed to me.

The SRER searches for "something", "*", "something". If we simply replaced all of that with just "</b>", we would lose the 2 "something"s on either side. So we made sure that we captured them (by putting brackets around them ) and then we replace all 3 sections by the 2 "something"s with a new section in the middle. That sounds horribly complicated - but is actually reasonably simple if you take it slowly.

As I said earlier, I am not an expert on SREs by any means. I freely admit that I find them the hardest thing I have ever tried to learn in computing. So you should also look carefully as jchd's suggestions - he is far better at SREs than I ever will be.

M23

Excuses, excuses. jchd has the right pattern and I'm sure you know how to test it by now. New version released today.

For a beginner you were at least on the right track but as jchd pointed out it can be done with a single SRER instead of two.

[Melba23]

George,

Downloaded thanks.

I agree with you (as nearly always ) - as soon as I saw jchd's version I knew it was the way to go. But I may as well use my feeble efforts as learning points. After all you guys have forgotten what it was like when you started - I still have the fresh scar tissue on my frontal lobes!

M23

[GEOSoft]

We all have that scar. I actually don't even like SREs but in the long run it's well worth the effort to learn them. I hated them in other languages as well but I finally saw that they were by far the better way to accomplish some tasks and usually save several lines of code in the process.

[jchd]

I need to quit but I'll expand a bit more tomorrow for those interested.

Edited January 11, 2011 by jchd

[iamtheky]

thanks GEO. I am enjoying these threads that gauge the vastness of my SRER ineptitude (its hella-vast).

[GPinzone]

Does the AutoIt RegEx engine support positive look ahead and look behind? If so, you could use:

$out1 = StringRegExpReplace ($string , "\B\*(?=[^*\s])"  , "<b>" ) ; changes all asterix to <b> tags if begin on a word boundary and next to a non-asterisk or non-space.
$out1 = StringRegExpReplace ($out1   , "(?<=[^*\s])\*\B" , "</b>") ; changes all asterix to </b> tags if end on a word boundary and next to a non-asterisk or non-space.

I haven't tried the above code, but the regex would work in other languages like Perl.

It won't convert "**", "hello*there", nor "* hello there *", but I don't know if that's a good or bad thing based on the requirements.

Edited January 12, 2011 by GPinzone