I ended up creating my own HTML class, mainly to keep all our output consistent. The code is below.
<?php
/**
* A class to generate html tags
* @author Glen Solsberry
*/
class HTML {
private $isOpen;
/**
* holds all information about tags
* @var $tags array
*/
private $tags;
private $current_tag = 0;
private $depth = 0;
private $output = "";
private $separator = " ";
private $pretty_print = true;
/**
* Set the pretty_print status
* @author Glen Solsberry
*/
public function setPretty($new_value) {
$this->pretty_print = (bool)$new_value;
}
/**
* Set the "separator" (the string that will be printed before tags on a new line)
* @author Glen Solsberry
*/
public function setSeparator($new_value) {
$this->separator = $new_value;
}
/**
* add a tag to the "DOM"
* @author Glen Solsberry
*/
public function tag($tag) {
$this->updateChildrenAndParent();
$this->tags[$this->current_tag]['tag'] = $tag;
$this->tags[$this->current_tag]['open'] = true;
return $this;
}
/**
* set an attribute on the current tag
* @author Glen Solsberry
*/
public function attr($key, $value) {
$this->tags[$this->current_tag]['attrs'][$key] = $value;
return $this;
}
/**
* close the current tag, if it's open. if the tag is already closed, we work back up the chain to determine which tag needs to be closed.
* @author Glen Solsberry
*/
public function close() {
if ($this->isOpen($this->current_tag)) {
$this->tags[$this->current_tag]['open'] = false;
} else {
// work backwards till we find the first open tag, and close it
for($previous_id = $this->current_tag - 1; $previous_id >= 0; $previous_id--) {
if ($this->isOpen($previous_id)) {
$this->tags[$previous_id]['open'] = false;
break;
}
}
}
return $this;
}
/**
* Create a text node
* @author Glen Solsberry
*/
public function text($text) {
$this->updateChildrenAndParent();
$this->tags[$this->current_tag]['text'] = $text;
$this->tags[$this->current_tag]['open'] = false;
return $this;
}
/**
* Updates children and parent information, so that all tags link properly.
* Duplicates will be handled later.
* @author Glen Solsberry
*/
private function updateChildrenAndParent() {
// if the current tag is still open, then this needs to be a child of it
if ($this->isOpen($this->current_tag)) {
$this->tags[$this->current_tag]['children'][] = count($this->tags);
}
$this->current_tag = count($this->tags);
// default value for this tags parent
$this->tags[$this->current_tag]['parent'] = $this->current_tag - 1;
for($parent_id = $this->current_tag - 1; $parent_id >= 0; $parent_id--) {
// is the parent still open?
if ($this->isOpen($parent_id)) {
$this->tags[$this->current_tag]['parent'] = $parent_id;
$this->tags[$parent_id]['children'][] = $this->current_tag;
break;
}
}
}
/**
* Determines whether the passed tag_id is open
* @author Glen Solsberry
*/
private function isOpen($tag_id) {
return (bool)$this->tags[$tag_id]['open'];
}
/**
* Determines whether the passed tag_id is closed
* @author Glen Solsberry
*/
private function isClosed($tag_id) {
return (bool)!$this->tags[$tag_id]['open'];
}
/**
* Generates a single tag's html represenation. Called recursively if there are children
* @author Glen Solsberry
*/
private function generateTag($tag_id) {
$current_tag = $this->tags[$tag_id];
if (isset($current_tag['tag'])) {
$this->output .= "<" . $current_tag['tag'];
if (isset($current_tag['attrs'])) {
ksort($current_tag['attrs']);
foreach($current_tag['attrs'] as $key => $value) {
$this->output .= " " . $key . "=\"" . htmlspecialchars($value) . "\"";
}
}
$this->output .= $this->pretty_print(">");
if (isset($current_tag['children']) && count($current_tag['children']) > 0) {
$children = array_unique($current_tag['children']);
foreach($children as $position => $child_id) {
$this->depth++;
$this->generateTag($child_id);
$this->depth--;
}
$this->output .= $this->pretty_print("</" . $current_tag['tag'] . ">");
} else {
$this->output .= $this->pretty_print("</" . $current_tag['tag'] . ">");
$this->depth--;
}
} else if (isset($current_tag['text'])) {
$this->output .= $current_tag['text'];
}
}
/**
* Pretty prints the output. Uses newlines and line starters.
* @author Glen Solsberry
*/
private function pretty_print($string) {
$output = "";
if ($this->pretty_print == true && $this->depth > 0) {
$output .= str_repeat($this->separator, $this->depth);
}
$output .= $string;
if ($this->pretty_print == true) {
$output .= "\n";
}
return $output;
}
function __toString() {
$this->generateTag(0);
return $this->output;
}
}
?>
I built some tests to make sure everything comes out properly (and to be a good PHP dev)…
<?php
require_once 'PHPUnit/Framework.php';
require_once 'PHPUnit/Extensions/OutputTestCase.php';
require_once '../html.class.inc';
class HTMLTest extends PHPUnit_Extensions_OutputTestCase
{
public function testOutputLooksCorrect1() {
$html = $this->sharedFixture;
$html->tag("html");
$html->close();
$this->expectOutputString('<html></html>');
print $html;
}
public function testOutputLooksCorrect2() {
$html = $this->sharedFixture;
$html->tag("html");
$html->tag("head");
$html->close();
$html->close();
$this->expectOutputString('<html><head></head></html>');
print $html;
}
public function testOutputLooksCorrect3() {
$html = $this->sharedFixture;
$html->tag("html");
$html->tag("head");
$html->tag("link")->attr("rel", "stylesheet")->attr("href", "testOutputLooksCorrect.css")->attr("type", "text/css")->close();
$html->close();
$html->close();
$this->expectOutputString('<html><head><link href="testOutputLooksCorrect.css" rel="stylesheet" type="text/css"></link></head></html>');
print $html;
}
public function testOutputLooksCorrect4() {
$html = $this->sharedFixture;
$html->tag("html");
$html->tag("head");
$html->tag("link")->attr("href", "testOutputLooksCorrect.css")->attr("rel", "stylesheet")->attr("type", "text/css")->close();
$html->close();
$html->close();
$this->expectOutputString('<html><head><link href="testOutputLooksCorrect.css" rel="stylesheet" type="text/css"></link></head></html>');
print $html;
}
public function testOutputLooksCorrect5() {
$html = $this->sharedFixture;
$html->tag("html");
$html->tag("head");
$html->tag("link")->attr("type", "text/css")->attr("href", "testOutputLooksCorrect.css")->attr("rel", "stylesheet")->close();
$html->close();
$html->close();
$this->expectOutputString('<html><head><link href="testOutputLooksCorrect.css" rel="stylesheet" type="text/css"></link></head></html>');
print $html;
}
public function testOutputLooksCorrect10() {
$html = $this->sharedFixture;
$html->tag("html");
$html->tag("head");
$html->tag("link")->attr("rel", "stylesheet")->attr("href", "testOutputLooksCorrect.css")->attr("type", "text/css")->close();
$html->close();
$html->tag("body");
$html->close();
$html->close();
$this->expectOutputString('<html><head><link href="testOutputLooksCorrect.css" rel="stylesheet" type="text/css"></link></head><body></body></html>');
print $html;
}
public function testOutputLooksCorrect11() {
$html = $this->sharedFixture;
$html->tag("html");
$html->tag("head");
$html->tag("link")->attr("rel", "stylesheet")->attr("href", "testOutputLooksCorrect.css")->attr("type", "text/css")->close();
$html->close();
$html->tag("body");
$html->text("This is a testOutputLooksCorrect.");
$html->close();
$html->close();
$this->expectOutputString('<html><head><link href="testOutputLooksCorrect.css" rel="stylesheet" type="text/css"></link></head><body>This is a testOutputLooksCorrect.</body></html>');
print $html;
}
public function testOutputLooksCorrect12() {
$html = $this->sharedFixture;
$html->text("This is a testOutputLooksCorrect.");
$this->expectOutputString('This is a testOutputLooksCorrect.');
print $html;
}
public function testOutputLooksCorrect13() {
$html = $this->sharedFixture;
$html->tag("head")->close();
$this->expectOutputString('<head></head>');
print $html;
}
public function testOutputLooksCorrect14() {
$html = $this->sharedFixture;
$html->tag("head")->tag("title")->text("This is the title")->close()->close();
$this->expectOutputString('<head><title>This is the title</title></head>');
print $html;
}
public function testOutputLooksCorrect15() {
$html = $this->sharedFixture;
$html->tag("html");
$html->tag("head");
$html->tag("link")->attr("rel", "stylesheet")->attr("href", "testOutputLooksCorrect.css")->attr("type", "text/css")->close();
$html->tag("meta")->attr("http-equiv", "Content-Type")->attr("content", "text/html; charset=utf-8")->close();
$html->close();
$html->tag("body");
$html->text("This is a testOutputLooksCorrect.");
$html->close();
$html->close();
$this->expectOutputString('<html><head><link href="testOutputLooksCorrect.css" rel="stylesheet" type="text/css"></link><meta content="text/html; charset=utf-8" http-equiv="Content-Type"></meta></head><body>This is a testOutputLooksCorrect.</body></html>');
print $html;
}
public function testOutputLooksCorrect16() {
$html = $this->sharedFixture;
$html->tag("html");
$html->tag("head");
$html->tag("link")->attr("rel", "stylesheet")->attr("href", "testOutputLooksCorrect.css")->attr("type", "text/css")->close();
$html->tag("meta")->attr("http-equiv", "Content-Type")->attr("content", "text/html; charset=utf-8")->close();
$html->tag("script")->attr("type", "javascript")->attr("src", "blah.js")->close();
$html->close();
$html->tag("body");
$html->text("This is a testOutputLooksCorrect.");
$html->close();
$html->close();
$this->expectOutputString('<html><head><link href="testOutputLooksCorrect.css" rel="stylesheet" type="text/css"></link><meta content="text/html; charset=utf-8" http-equiv="Content-Type"></meta><script src="blah.js" type="javascript"></script></head><body>This is a testOutputLooksCorrect.</body></html>');
print $html;
}
protected function setUp() {
$html = new HTML;
$html->setPretty(0);
$this->sharedFixture = $html;
}
protected function tearDown() {
$this->sharedFixture = NULL;
}
}
?>
Everything works properly. However, the time needed is much higher than I’d like. On the order of ~3 seconds for 14k worth of HTML.
After profiling the code, it seems that most of the time is spent in the updateChildrenAndParent. Almost 75% of the entire execution is spent there. Can anyone suggest ways to speed this up, since I don’t want to slow down the site noticably.
You should be using real dom methods to parse HTML. They are written in C and will be orders of magnitude faster than anything you can code up in naively.