Use of [^>]*? skipping past > in regex

tvink

Hi,

I am using a regex (see below) to find a particular input tag on an HTML page. I want to find a tag with value="myvalue" ahead of the name="myname". I have [^>]*? in the regex to allow the engine to scan up to the end of each tag and fail there.

Unfortunately, the engine is finding a value= several tags ahead of the one with name="myname" and is gobbeling up all of the characters in between, matching characters spaning several tags.
The scan is passing several ">" characters and does not care.

Here is my regex:

\s?<input[^{>]?name="myname"[^>]?value=".?"[^>]?>}

Here is an example of the target HTML

<input type="checkbox" name="forgotPassword" value="checkbox">
</font></p>
<p align="left"><font size="-1" face="Arial, Helvetica, sans-serif">Email
Address:
<input type="text" name="loginEmail" >

==============================
insteading of failing to mach ( which I thought should happen, the regex is matching everything from the value= up to the end of the string.

I am using preg_match() and have the s option endbled to allow the scan to span newlines. ( I do not want to loose the s option )

So how do I make the regex fail if it scans to a ">" character before it gets to "name=" ?

Thanks in advance,
Tony

bubblenut

Try using the U option as well for ungreedy matching. I'm going to have a think about this one a bit later when I get home because I think it's going to be pretty tricky to do well.
HTH though
Bubble

tvink

I'll try the U option when I get home too.

It is worth a shot but I ungreedyed everything with the ?, I think.

Tony

Weedpacket

Some points:

[^>]* will stop on and refuse to match a >, the ? doesn't contribute anything.
Matching
```
\s?<input([^>]*?)name="myname"([^>]*?)value="(.*?)"([^>]?*)> 
```
and using print_r() to see which bits are being matched by which subexpressions might help - the > characters that are being matched might be in the ".*?" section.
Your last subexpression has the ? and * reversed but since that's a parse error I guess you mistyped it copying it into the post.
Your sample HTML fails to match like you say it should. Changing "myname" to "forgotPassword" matches only the <input> tag.

tvink

I think I have my sollution but I do not understand why...

( by the way I also botched yesterdays post and had the regex backwards )

Here is the regex that works

\s?<input[^{>]value="[^>]"[^{>]name="loginEmail".?>}}

Here is the one that does not work

\s?<input[^{>]value=".?"[^{>]name="loginEmail".?>}}

Here is the test string I am matching

<input type="checkbox" name="forgotPassword" value="checkbox">
</font></p>
<p align="left"><font size="-1" face="Arial, Helvetica, sans-serif">Email
Address:
<input value="dfdf" type="text" name="loginEmail" >

The one that works matches only the last tag

The one that does not work matches everything. The difference is the term in value= ( "[^>]" ) works and ( ".?" )does not. In the bad one the .*? is matching everything like you would expect for a greedy match. But why does not the ? make it not greedy and stop on the quote?

Hmmm

Would be interesting to know...
Sorry about the messed up regex in yesterdays post.

Weedpacket

The reason why the one that works doesn't match the first input tag is because that tag has the value= and name= attributes in a different order from that which the pattern is looking for. If that's going to be a problem I'd suggest matching <input> tags in one piece (/<input\s([^>]*)>/) and then using preg_match_all to pull the attribute name=value pairs out into an array.

The one that doesn't work - well, for me, it matches the last tag only (the first .*? matching dfdf, the other the space after "LoginEmail").