MensBeam
/
HTML-Parser
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
Browse Source

Beginning Rewrite

ns
Dustin Wilson 6 years ago
parent
06cffb2e25
commit
a89f6c9f09
11 changed files with 4167 additions and 10018 deletions
Whitespace
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Split View
Diff Options
Show Stats Download Patch File Download Diff File
  1. 2
      .gitignore
  2. 28
      composer.json
  3. 22
      composer.lock
  4. 10023
      html5.php
  5. 74
      lib/DOM.php
  6. 439
      lib/DataStream.php
  7. 79
      lib/Exception.php
  8. 86
      lib/ParseError.php
  9. 3259
      lib/Parser.php
  10. 90
      lib/Stack.php
  11. 83
      lib/Token.php

2
.gitignore
View File

@ -62,3 +62,5 @@ $RECYCLE.BIN/
# .nfs files are created when an open file is removed but is still being accessed
.nfs*
/vendor/

28
composer.json
View File

@ -0,0 +1,28 @@
{
"name": "dw/html5",
"description": "Parses HTML5 text into a php DOMDocument",
"type": "library",
"require": {
"php": "^7.0",
"ext-intl": "*",
"ext-mcrypt": "*",
"ext-hash": "*"
},
"license": "MIT",
"authors": [
{
"name": "Dustin Wilson",
"email": "dustin@dustinwilson.com",
"homepage": "https://dustinwilson.com/"
}
],
"autoload": {
"psr-4": {
"dW\\HTML5\\": "lib/"
},
"classmap": ["lib/Token.php"]
},
"autoload-dev": {
"files": ["lib/Token.php"]
}
}

22
composer.lock
View File

@ -0,0 +1,22 @@
{
"_readme": [
"This file locks the dependencies of your project to a known state",
"Read more about it at https://getcomposer.org/doc/01-basic-usage.md#composer-lock-the-lock-file",
"This file is @generated automatically"
],
"content-hash": "9ae6b8a7d4d830c471b21ccefc1ee8fb",
"packages": [],
"packages-dev": [],
"aliases": [],
"minimum-stability": "stable",
"stability-flags": [],
"prefer-stable": false,
"prefer-lowest": false,
"platform": {
"php": "^7.0",
"ext-intl": "*",
"ext-mcrypt": "*",
"ext-hash": "*"
},
"platform-dev": []
}

10023
html5.php
View File

File diff suppressed because it is too large

74
lib/DOM.php
View File

@ -0,0 +1,74 @@
<?php
declare(strict_types=1);
namespace dW\HTML5;
class DOM {
public static function getAncestor(mixed $needle, \DOMElement $context): \DOMElement {
return static::ancestor($needle, $context, true);
}
public static function hasAncestor(mixed $needle, \DOMElement $context): bool {
return static::ancestor($needle, $context, false);
}
public static function getDescendant(mixed $needle, \DOMElement $context): \DOMNode {
return static::descendant($needle, $context, true);
}
public static function hasDescendant(mixed $needle, \DOMElement $context): bool {
return static::descendant($needle, $context, false);
}
public static function descendant(mixed $needle, \DOMElement $context, bool $returnNode = true): \DOMNode {
if ($context->hasChildNodes() === false) {
return ($returnNode === true) ? null : false;
}
$context = $context->firstChild;
do {
$result = static::compare($needle, $context);
if (!is_null($result)) {
return ($returnNode === true) ? $result : true;
}
$result = static::descendant($needle, $context);
if (!is_null($result)) {
return ($returnNode === true) ? $result : true;
}
} while ($context = $context->nextSibling);
return ($returnNode === true) ? null : false;
}
protected static function ancestor(mixed $needle, \DOMElement $context, bool $returnNode = true) {
while ($context = $context->parentNode) {
$result = static::compare($needle, $context);
if (!is_null($result)) {
return ($returnNode === true) ? $result : true;
}
}
return ($returnNode === true) ? null : false;
}
protected static function compare(mixed $needle, \DOMNode $context): \DOMNode {
if (is_string($needle)) {
if ($context instanceof \DOMElement && $context->nodeName == $needle) {
return $context;
}
} elseif ($needle instanceof \DOMNode) {
if ($context->isSameNode($needle)) {
return $context;
}
} elseif ($needle instanceof \Closure) {
if ($needle($context) === true) {
return $context;
}
} else {
throw new Exception(Exception::DOM_DOMELEMENT_STRING_OR_CLOSURE_EXPECTED, gettype($needle));
}
return null;
}
}

439
lib/DataStream.php
View File

File diff suppressed because one or more lines are too long

79
lib/Exception.php
View File

@ -0,0 +1,79 @@
<?php
declare(strict_types=1);
namespace dW\HTML5;
class Exception extends \Exception {
const INVALID_CODE = 10000;
const UNKNOWN_ERROR = 10001;
const INCORRECT_PARAMETERS_FOR_MESSAGE = 10002;
const PARSER_DOMDOCUMENT_EXPECTED = 10101;
const PARSER_DOMELEMENT_DOMDOCUMENT_DOMDOCUMENTFRAG_EXPECTED = 10102;
const PARSER_DOMNODE_EXPECTED = 10103;
const STACK_INVALID_INDEX = 10201;
const STACK_DOMNODE_ONLY = 10202;
const DATASTREAM_NODATA = 10301;
const DATASTREAM_INVALID_DATA_CONSUMPTION_LENGTH = 10302;
const DOM_DOMELEMENT_STRING_OR_CLOSURE_EXPECTED = 10401;
protected static $messages = [10000 => 'Invalid error code',
10001 => 'Unknown error; escaping',
10002 => 'Incorrect number of parameters for Exception message; %s expected',
10101 => 'DOMDocument expected; found %s',
10102 => 'DOMElement, DOMDocument, or DOMDocumentFrag expected; found %s',
10103 => 'DOMNode expected; found %s',
10201 => '%s is an invalid index',
10202 => 'Instances of DOMNode are the only types allowed in an HTML5Stack',
10301 => 'Data string expected; found %s',
10302 => '%s is an invalid data consumption length; a value of 1 or above is expected',
10401 => 'The first argument must either be an instance of \DOMElement, a string, or a closure; found %s'];
public function __construct(int $code, ...$args) {
if (!isset(static::$messages[$code])) {
throw new Exception(static::INVALID_CODE);
}
$message = static::$messages[$code];
$previous = null;
// Grab a previous exception if there is one.
if ($args[0] instanceof \Throwable) {
$previous = array_shift($args);
} elseif (end($args) instanceof \Throwable) {
$previous = array_pop($args);
}
// Count the number of replacements needed in the message.
$count = substr_count($message, '%s');
// If the number of replacements don't match the arguments then oops.
if (count($args) !== $count) {
throw new Exception(static::INCORRECT_PARAMETERS_FOR_MESSAGE, $count);
}
if ($count > 0) {
// Convert newlines and tabs in the arguments to words to better express what they
// are.
/*$args = array_map(function($value) {
switch ($value) {
case "\n": return 'Newline';
break;
case "\t": return 'Tab';
break;
default: return $value;
}
}, $args);*/
// Go through each of the arguments and run sprintf on the strings.
$message = call_user_func_array('sprintf', array_merge([$message], $args));
}
parent::__construct($message, $code, $previous);
}
}

86
lib/ParseError.php
View File

@ -0,0 +1,86 @@
<?php
declare(strict_types=1);
namespace dW\HTML5;
class ParseError {
// DataStream object passed to it used to get information used in error
// reporting.
public static $data;
const TAG_NAME_EXPECTED = 0;
const UNEXPECTED_EOF = 1;
const UNEXPECTED_CHARACTER = 2;
const ATTRIBUTE_EXISTS = 3;
const UNEXPECTED_TAG_END = 4;
const UNEXPECTED_START_TAG = 5;
const UNEXPECTED_END_TAG = 6;
const UNEXPECTED_DOCTYPE = 7;
const INVALID_DOCTYPE = 8;
const INVALID_CONTROL_OR_NONCHARACTERS = 9;
protected static $messages = ['Tag name expected; found %s',
'Unexpected end-of-file; %s expected',
'Unexpected "%s" character; %s expected',
'%s attribute already exists; discarding',
'Unexpected tag end; %s expected',
'Unexpected %s start tag; %s expected',
'Unexpected %s end tag; %s expected',
'Unexpected DOCTYPE; %s expected',
'Invalid DOCTYPE',
'Invalid Control or Non-character; removing'];
public static function errorHandler($code, $message, $file, $line, array $context) {
if ($code === E_USER_WARNING) {
$errMsg = sprintf("HTML5 Parse Error: \"%s\" in %s", $message, static::$data->filePath);
if (static::$data->length !== 0) {
$errMsg .= sprintf(" on line %s, column %s\n", static ::$data->line, static::$data->column);
} else {
$errMsg .= "\n";
}
echo $errMsg;
}
}
public static function trigger(int $code, DataStream $data, ...$args): bool {
if (!isset(static::$messages[$code])) {
throw new Exception(Exception::INVALID_CODE);
}
static::$data = $data;
// Set the error handler and honor already-set error reporting rules.
set_error_handler('\\dW\\HTML5\\ParseError::errorHandler', error_reporting());
$message = static::$messages[$code];
// Count the number of replacements needed in the message.
$count = substr_count($message, '%s');
// If the number of replacements don't match the arguments then oops.
if (count($args) !== $count) {
throw new Exception(static::INCORRECT_PARAMETERS_FOR_MESSAGE, $count);
}
if ($count > 0) {
// Convert newlines and tabs in the arguments to words to better express what they
// are.
$args = array_map(function($value) {
switch ($value) {
case "\n": return 'Newline';
break;
case "\t": return 'Tab';
break;
default: return $value;
}
}, $args);
// Go through each of the arguments and run sprintf on the strings.
$message = call_user_func_array('sprintf', array_merge([$message], $args));
}
$output = trigger_error($message, E_USER_WARNING);
restore_error_handler();
return $output;
}
}

3259
lib/Parser.php
View File

File diff suppressed because it is too large

90
lib/Stack.php
View File

@ -0,0 +1,90 @@
<?php
declare(strict_types=1);
namespace dW\HTML5;
class Stack implements \ArrayAccess {
protected $storage = [];
// Temporarily change this from DOMNode to HTML5StartTagToken for the purposes of
// testing the tokenizer.
public function offsetSet($offset, $value) {
if ($offset < 0) {
throw new Exception(Exception::STACK_INVALID_INDEX);
}
if (is_null($offset)) {
$this->storage[] = $value;
} else {
$this->storage[$offset] = $value;
}
}
public function offsetExists($offset) {
return isset($this->storage[$offset]);
}
public function offsetUnset($offset) {
if ($offset < 0) {
throw new Exception(Exception::STACK_INVALID_INDEX);
}
unset($this->storage[$offset]);
}
public function offsetGet($offset) {
if ($offset < 0) {
throw new Exception(Exception::STACK_INVALID_INDEX);
}
return $this->storage[$offset];
}
public function pop() {
return array_pop($this->storage);
}
public static function search(mixed $needle): int {
if (!$needle) {
return false;
}
if ($needle instanceof DOMElement) {
foreach ($this->storage as $key=>$value) {
if ($value->isSameNode($needle)) {
return $key;
}
}
} elseif (is_string($needle)) {
foreach ($this->storage as $key=>$value) {
if ($value->nodeName === $needle) {
return $key;
}
}
}
return false;
}
public function __get($property) {
switch ($property) {
case 'length': return count($this->storage);
break;
case 'currentNode':
$currentNode = end($this->storage);
return ($currentNode) ? $currentNode : null;
break;
case 'adjustedCurrentNode':
# The adjusted current node is the context element if the parser was created by
# the HTML fragment parsing algorithm and the stack of open elements has only one
# element in it (fragment case); otherwise, the adjusted current node is the
# current node.
return (Parser::$self->fragmentCase && $this->length === 1) ? Parser::$self->fragmentContext : $this->currentNode;
break;
case 'currentNodeName':
$currentNode = $this->currentNode;
return ($currentNode && $currentNode->nodeType) ? $currentNode->nodeName : null;
break;
default: return null;
}
}
}

83
lib/Token.php
View File

@ -0,0 +1,83 @@
<?php
declare(strict_types=1);
namespace dW\HTML5;
abstract class Token {}
abstract class DataToken extends Token {
public $data;
public function __construct($data) {
$this->data = (string)$data;
}
}
abstract class TagToken extends Token {
public $name;
public function __construct($name) {
$this->name = (string)$name;
}
}
class EOFToken extends Token {}
class DOCTYPEToken extends Token {
public $forceQuirks = false;
public $public;
public $system;
public function __construct($name = null, $public = null, $system = null) {
$this->name = (string)$name;
$this->public = (string)$public;
$this->system = (string)$system;
}
}
class CharacterToken extends DataToken {}
class CommentToken extends DataToken {
public function __construct($data = '') {
parent::__construct($data);
}
}
class StartTagToken extends TagToken {
public $namespace;
public $selfClosing;
protected $_attributes;
public function __construct($name, bool $selfClosing = false, string $namespace = \dW\HTML5\Parser::HTML_NAMESPACE) {
$this->selfClosing = $selfClosing;
$this->namespace = $namespace;
parent::__construct($name);
}
public function getAttribute(string $name): \DOMAttr {
return ($this->_attributes[$name]) ? $this->_attributes[$name] : null;
}
public function hasAttribute(string $name): bool {
return (isset($this->_attributes[$name]));
}
public function removeAttribute(string $name) {
unset($this->_attributes[$name]);
}
public function setAttribute($name, $value) {
$this->_attributes[(string)$name] = (string)$value;
}
public function __get($property) {
if ($property === 'attributes') {
return $this->_attributes;
}
return null;
}
}
class EndTagToken extends TagToken {}
Write Preview
Loading…
Cancel
Save