This code often fails to work

teguh123

$pattern="'(" .$pre.")(.?)(" .$end .")'si";
//$pattern is '(.?<body>)(.?)(</body>.?)'si
preg_match($pattern, $content, $matches);

Basically I want to analyze a website content. I want to split the content into 3 part. The one above <body>, the body inner html, and the on below </body>

The <body> is expected to be found. So it's not going to be <body bla bla onclick=> something. It has to be <body>. I made sure of it for simplicity.

I wonder why often it doesn't work.

Also I wonder why I wrote .? shouldn't it be .?

NogDog

How about:

$parts = preg_split('#</?body.*?>#si', $text);

teguh123

Okay good. But why does you use .*? to express "any strings of characters"

.* seems to be enough.

Also shouldn't it be

$parts=preg_split('#</?body[^>]*?>#si', $content);

I mean we need to know that whatever is after body is not > until the first > after body right?

When i change .? to . things sort of work out though.

halojoy

Have you tested NogDog code and your code.
This is a best way to know what works, I think
🙂

teguh123

halojoy;10942592 wrote:
Have you tested NogDog code and your code.
This is a best way to know what works, I think
🙂

Both works.

My code works if I change that to
//$pattern is '(.?<body>)(.?)(</body>.*?)'si

preg_match($pattern, $content, $matches);

if $pattern is changed to (.<body>)(.)(</body>.*)'si

without ? after .*

Weedpacket

teguh123 wrote:
The <body> is expected to be found. So it's not going to be <body bla bla onclick=> something. It has to be <body>. I made sure of it for simplicity.

Using [man]DOM[/man] you wouldn't have to worry about that; the parsing would already be done for you.

NogDog

The "?" after the "" makes the "" un-greedy. Without it, the "" is greedy and will match on the longest possible string that matches, while with the "?" modifier it matches only the minimum number of characters to fulfill the match. Since the <body> tag can have attributes, it is not sufficient to match simply on <body>. You could get the same functional result of only matching to the end of the closing ">" of the body tag with:

<body[^>]*>

I just find it a bit simpler to do this, instead:

<body.*?>

As Weedpacket suggests, though, depending on what your actual goal is with this part of the application, using the [man]DOM[/man] extension may likely be more effective and robust than doing all your own parsing via regular expressions.

teguh123

Using dom or preg_split will mean changing the program significantly. You guys do open my eyes though and I am grateful.

What I am still confused is why my code work if I use . and doesn't if I use .?

I thought it should be the other way around. Greedy is a strange concept.

I used this now

'(.<body[^>]?>)(.)(</body>.)'si

and it's working

The previous one
'(.?<body[^>]?>)(.?)(</body>.?)'si

often doesn't work. I wonder why.

halojoy

Greedy can be this

we are looking for <.*> and hope to catch <body>

and our string is: <body>xxxxxxxxxxxxxx</body>xxxxxxx

but if Greedy we will catch
<body>xxxxxxxxxxxxxx</body>
as this also matches
<.*>