Let's say I wanna build me own XML/HTML tag parser...

s_shrum

For my templating engine, I want to build in a tag reader...not nessecarily XML, but if you understand where I am coming from better with XML, all the better.

I would like to take something like:

<author>Sean Shrum</author>
<created>03.03.10</created>
<foo>bar</foo>

And build a hash (associative array) of:

$tag_array['author'] = Sean Shrum
$tag_array['created'] = 03.03.10
$tag_array['foo'] = bar

Anyone have a code snippet that would take a html string and break it down like this?

I am still playing with the idea of use the META tag but I'm not sure that this would be the right thing to do.

TIA

Sean Shrum

xblue

you probably could do it with regular expression.
some questions:
- are the tags always named 'author', 'created' and 'foo' or do you need these to be totally variable?
- may tags have attributes? if so, would you want them to be ignored?
- may tags be nested? if so, what would you want to match?
- what about repeated tags? if there is another <author>Mr Foo</author>, it would overwrite the array entry with your name since they cannot have the same key.

Mordecai

I'd also like to know how to do this. Right now, I'm working on an HTML parser that can only read one HTML tag per line. I'm doing this with an elaborate set of switches... 🙂

foreach ($lines as $line) {
  $line = trim($line);
  $spaced = split(" ",$line,2);
  $tag = ereg_replace("<|>","",$spaced[0]);
  $params = split(" ",$spaced[1]);
  switch ($tag) {
    // html tags
    case "body":
      for ($i=0;$i<=count($params);$i++) {
        $param_name = split("=",$params[$i]);
        $value = ereg_replace("#|\"|>","",$param_name[1]);
        switch ($param_name[0]) {
        // parameters on the tag
        break;
    }
  }
}

Mordecai

Err.. I'm stupid.
I'm just playing around with this, and it dawns upon me: why not turn the beginnings and ends of html tags into returns?

Lo and behold, it works:
$contents = ereg_replace("<|>","\n",$contents);

However, you still have xblue's things to think about, but at least now you know how to read them as HTML tags.

superwormy

I have a script at home that I'm trying to do this with... so far what I've found out is that the Regex way of doing it is VERY slow.

What I've tried now is to have two big WHILE loops, the first one goes through and records all the tags, basically everything inbetween < > characters. It also records the ' depth ' of the tag, so basically if you were to indent it, how many times do you hit the tab key.

The last loops builds a big expression and then uses eval() to evaulate the expression and build the arrays.

Right now, its still pretty slow, cause the more tags you have the more times it needs to loop through the document.

Anyone else have ideas on how to do this better?

s_shrum

That's a good point...if I do this sort of thing on a large number of large documents, this process could bog down a server and take forever to return results.

I think I'm liking my idea of just looking in the <head> for the META tags. This way I can end looping on "</head>"

Besides the META construct supports the hash idea with 'name' and 'content' and most HTML editors have a META editor (ala Frontpage).

Before I dive into the deep end, can anyone see a downside to using the META tags for this sort of thing. This is mainly for holding content info anyways like author, created date, etc.

...
<head>
<META name='author' content='Sean Shrum'>
<META name='created' content='March 11, 2003'>
</head>
...

TIA

Sean Shrum

superwormy

I would say as long as its pretty controlled content like that, author and date and stuff like that, meta tags would definitally be the way to go.

$begin = stristr ("<meta name='", $content);
$begin = substr ($begin, 12);
$tostrip = stristr ("'", $begin);

$name = str_replace ($tostrip, "", $begin);

tomhath

I'd stay away from both regexes and while loops. Tags can be nested; as soon as you see a nested data structure you should be thinking recursion, not patterns or iterations.