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

Fix a few tree tests

ns
J. King 3 years ago
parent
bb4002abcb
commit
4e5fd35775
4 changed files with 45 additions and 73 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. 8
      lib/OpenElementsStack.php
  2. 42
      lib/ParseError.php
  3. 2
      lib/ParseErrorEmitter.php
  4. 66
      lib/TreeBuilder.php

8
lib/OpenElementsStack.php
View File

@ -78,15 +78,9 @@ class OpenElementsStack extends Stack {
$this->_storage = array_values($this->_storage);
}
public function generateImpliedEndTags($exclude = []) {
public function generateImpliedEndTags(array $exclude = []) {
$tags = ['caption', 'colgroup', 'dd', 'dt', 'li', 'optgroup', 'option', 'p', 'rb', 'rp', 'rt', 'rtc', 'tbody', 'td', 'tfoot', 'th', 'thead', 'tr'];
if (is_string($exclude)) {
$exclude = [$exclude];
}
assert(is_array($exclude), new Exception(Exception::STACK_STRING_ARRAY_EXPECTED));
if (count($exclude) > 0) {
$modified = false;
foreach ($exclude as $e) {

42
lib/ParseError.php
View File

@ -5,6 +5,7 @@ namespace dW\HTML5;
class ParseError {
protected $data;
// tokenization parse errors; these have been standardized
const ENCODING_ERROR = 100;
const UNEXPECTED_NULL_CHARACTER = 101;
const UNEXPECTED_QUESTION_MARK_INSTEAD_OF_TAG_NAME = 102;
@ -53,9 +54,19 @@ class ParseError {
const CONTROL_CHARACTER_REFERENCE = 145;
const SURROGATE_IN_INPUT_STREAM = 146;
const NONCHARACTER_IN_INPUT_STREAM = 147;
const CONTROL_CHARACTER_IN_INPUT_STREAM = 148;
const CONTROL_CHARACTER_IN_INPUT_STREAM = 148;
// tree construction parse errors; these have not been standardized, but html5lib's error names are likely to become standard in future
const EXPECTED_DOCTYPE_BUT_GOT_START_TAG = 200;
const EXPECTED_DOCTYPE_BUT_GOT_END_TAG = 201;
const EXPECTED_DOCTYPE_BUT_GOT_CHARS = 202;
const UNEXPECTED_END_TAG = 203; // html5lib also uses 'adoption-agency-1.2' and 'adoption-agency-1.3' for this
const MESSAGES = [
self::EXPECTED_DOCTYPE_BUT_GOT_START_TAG => 'Expected DOCTYPE but got start tag',
self::EXPECTED_DOCTYPE_BUT_GOT_END_TAG => 'Expected DOCTYPE but got end tag',
self::EXPECTED_DOCTYPE_BUT_GOT_CHARS => 'Expected DOCTYPE but got characters',
self::UNEXPECTED_END_TAG => 'Unexpected end tag',
self::ENCODING_ERROR => 'Corrupt encoding near byte position %s',
self::UNEXPECTED_NULL_CHARACTER => 'Unexpected null character',
self::UNEXPECTED_QUESTION_MARK_INSTEAD_OF_TAG_NAME => 'Unexpected "?" character instead of tag name',
@ -108,55 +119,26 @@ class ParseError {
];
const REPORT_OFFSETS = [
self::ENCODING_ERROR => 0,
self::UNEXPECTED_NULL_CHARACTER => -1,
self::UNEXPECTED_QUESTION_MARK_INSTEAD_OF_TAG_NAME => 0,
self::EOF_BEFORE_TAG_NAME => 0,
self::INVALID_FIRST_CHARACTER_OF_TAG_NAME => 0,
self::MISSING_END_TAG_NAME => -1,
self::EOF_IN_TAG => 0,
self::EOF_IN_SCRIPT_HTML_COMMENT_LIKE_TEXT => 0,
self::UNEXPECTED_EQUALS_SIGN_BEFORE_ATTRIBUTE_NAME => -1,
self::DUPLICATE_ATTRIBUTE => -1,
self::UNEXPECTED_CHARACTER_IN_ATTRIBUTE_NAME => -1,
self::MISSING_ATTRIBUTE_VALUE => -1,
self::UNEXPECTED_CHARACTER_IN_UNQUOTED_ATTRIBUTE_VALUE => -1,
self::MISSING_WHITESPACE_BETWEEN_ATTRIBUTES => 0,
self::UNEXPECTED_SOLIDUS_IN_TAG => 0,
self::CDATA_IN_HTML_CONTENT => -1,
self::INCORRECTLY_OPENED_COMMENT => 0,
self::ABRUPT_CLOSING_OF_EMPTY_COMMENT => -1,
self::EOF_IN_COMMENT => 0,
self::NESTED_COMMENT => 0,
self::INCORRECTLY_CLOSED_COMMENT => -1,
self::EOF_IN_DOCTYPE => 0,
self::MISSING_WHITESPACE_BEFORE_DOCTYPE_NAME => 0,
self::MISSING_DOCTYPE_NAME => -1,
self::INVALID_CHARACTER_SEQUENCE_AFTER_DOCTYPE_NAME => 0,
self::MISSING_WHITESPACE_AFTER_DOCTYPE_PUBLIC_KEYWORD => -1,
self::MISSING_DOCTYPE_PUBLIC_IDENTIFIER => -1,
self::MISSING_QUOTE_BEFORE_DOCTYPE_PUBLIC_IDENTIFIER => 0,
self::ABRUPT_DOCTYPE_PUBLIC_IDENTIFIER => -1,
self::MISSING_WHITESPACE_BETWEEN_DOCTYPE_PUBLIC_AND_SYSTEM_IDENTIFIERS => -1,
self::MISSING_WHITESPACE_AFTER_DOCTYPE_SYSTEM_KEYWORD => -1,
self::MISSING_DOCTYPE_SYSTEM_IDENTIFIER => -1,
self::MISSING_QUOTE_BEFORE_DOCTYPE_SYSTEM_IDENTIFIER => 0,
self::ABRUPT_DOCTYPE_SYSTEM_IDENTIFIER => -1,
self::UNEXPECTED_CHARACTER_AFTER_DOCTYPE_SYSTEM_IDENTIFIER => 0,
self::EOF_IN_CDATA => 0,
self::END_TAG_WITH_ATTRIBUTES => -1,
self::END_TAG_WITH_TRAILING_SOLIDUS => -1,
self::MISSING_SEMICOLON_AFTER_CHARACTER_REFERENCE => 0,
self::UNKNOWN_NAMED_CHARACTER_REFERENCE => 0,
self::ABSENCE_OF_DIGITS_IN_NUMERIC_CHARACTER_REFERENCE => 0,
self::NULL_CHARACTER_REFERENCE => 0,
self::CHARACTER_REFERENCE_OUTSIDE_UNICODE_RANGE => 0,
self::SURROGATE_CHARACTER_REFERENCE => 0,
self::NONCHARACTER_CHARACTER_REFERENCE => 0,
self::CONTROL_CHARACTER_REFERENCE => 0,
self::SURROGATE_IN_INPUT_STREAM => 0,
self::NONCHARACTER_IN_INPUT_STREAM => 0,
self::CONTROL_CHARACTER_IN_INPUT_STREAM => 0,
];
public function setHandler() {

2
lib/ParseErrorEmitter.php
View File

@ -10,7 +10,7 @@ trait ParseErrorEmitter {
$data = ($this instanceof Data) ? $this : ($this->data ?? null);
assert($data instanceof Data);
assert($this->errorHandler instanceof ParseError);
list($line, $column) = $data->whereIs(ParseError::REPORT_OFFSETS[$code]);
list($line, $column) = $data->whereIs(ParseError::REPORT_OFFSETS[$code] ?? 0);
return $this->errorHandler->emit($data->filePath, $line, $column, $code, ...$arg);
}
}

66
lib/TreeBuilder.php
View File

@ -109,12 +109,6 @@ class TreeBuilder {
$this->insertionMode = self::INITIAL_MODE;
$this->quirksMode = self::QUIRKS_MODE_OFF;
static::$instance = $this;
}
public function __destruct() {
static::$instance = null;
}
public function emitToken(Token $token) {
@ -388,14 +382,16 @@ class TreeBuilder {
# set the Document to quirks mode.
// DEVIATION: There is no iframe srcdoc document because there are no nested
// browsing contexts in this implementation.
switch (get_class($token)) {
case 'StartTagToken': $this->error(ParseError::UNEXPECTED_START_TAG, $token->name);
break;
case 'EndTagToken': $this->error(ParseError::UNEXPECTED_END_TAG, $token->name);
break;
case 'EOFToken': $this->error(ParseError::UNEXPECTED_EOF);
break;
default: throw new Exception(Exception::UNKNOWN_ERROR);
if ($token instanceof StartTagToken) {
$this->error(ParseError::EXPECTED_DOCTYPE_BUT_GOT_START_TAG);
} elseif ($token instanceof EndTagToken) {
$this->error(ParseError::EXPECTED_DOCTYPE_BUT_GOT_END_TAG);
} elseif ($token instanceof CharacterToken) {
$this->error(ParseError::EXPECTED_DOCTYPE_BUT_GOT_CHARS);
} elseif ($token instanceof EOFToken) {
$this->error(ParseError::UNEXPECTED_EOF);
} else {
throw new \Exception("Unexpected token type".get_class($token));
}
$this->quirksMode = self::QUIRKS_MODE_ON;
@ -431,7 +427,7 @@ class TreeBuilder {
# Create an element for the token in the HTML namespace, with the Document as
# the intended parent. Append it to the Document object. Put this element in the
# stack of open elements.
$element = static::insertStartTagToken($token, $this->DOM);
$element = $this->insertStartTagToken($token, $this->DOM);
# Switch the insertion mode to "before head".
$this->insertionMode = self::BEFORE_HEAD_MODE;
@ -490,7 +486,7 @@ class TreeBuilder {
# A start tag whose tag name is "head"
elseif ($token->name === 'head') {
# Insert an HTML element for the token.
$element = static::insertStartTagToken($token);
$element = $this->insertStartTagToken($token);
# Set the head element pointer to the newly created head element.
$this->headElement = $element;
@ -507,7 +503,7 @@ class TreeBuilder {
# Anything else
else {
# Insert an HTML element for a "head" start tag token with no attributes.
$element = static::insertStartTagToken(new StartTagToken('head'));
$element = $this->insertStartTagToken(new StartTagToken('head'));
# Set the head element pointer to the newly created head element.
$this->headElement = $element;
@ -551,7 +547,7 @@ class TreeBuilder {
elseif ($token->name === 'base' || $token->name === 'basefont' || $token->name === 'bgsound' || $token->name === 'link') {
# Insert an HTML element for the token. Immediately pop the current node off the
# stack of open elements.
static::insertStartTagToken($token);
$this->insertStartTagToken($token);
$this->stack->pop();
# Acknowledge the token’s *self-closing flag*, if it is set.
@ -561,7 +557,7 @@ class TreeBuilder {
elseif ($token->name === 'meta') {
# Insert an HTML element for the token. Immediately pop the current node off the
# stack of open elements.
static::insertStartTagToken($token);
$this->insertStartTagToken($token);
$this->stack->pop();
# Acknowledge the token’s *self-closing flag*, if it is set.
@ -597,7 +593,7 @@ class TreeBuilder {
// flag is always disabled.
elseif ($token->name === 'noscript') {
# Insert an HTML element for the token.
static::insertStartTagToken($token);
$this->insertStartTagToken($token);
# Switch the insertion mode to "in head noscript".
$this->insertionMode = self::IN_HEAD_NOSCRIPT_MODE;
}
@ -615,7 +611,7 @@ class TreeBuilder {
// intended parent isn't used when determining anything;
// Parser::createAndInsertElement will get the adjusted insertion location
// anyway.
static::insertStartTagToken($token);
$this->insertStartTagToken($token);
# 3. Mark the element as being "parser-inserted" and unset the element’s
# "non-blocking" flag.
@ -637,7 +633,7 @@ class TreeBuilder {
# A start tag whose tag name is "template"
elseif ($token->name === 'template') {
# Insert an HTML element for the token.
static::insertStartTagToken($token);
$this->insertStartTagToken($token);
# Insert a marker at the end of the list of active formatting elements.
$this->activeFormattingElementsList->insertMarker();
# Set the frameset-ok flag to "not ok".
@ -703,7 +699,7 @@ class TreeBuilder {
# 2. If the current node is not a template element, then this is a parse error.
if ($this->stack->currentNodeName !== 'template') {
$this->error(ParseError::UNEXPECTED_END_TAG, 'template');
$this->error(ParseError::UNEXPECTED_END_TAG);
}
# 3. Pop elements from the stack of open elements until a template element has been popped from the stack.
@ -722,7 +718,7 @@ class TreeBuilder {
# Any other end tag
else {
# Parse error.
$this->error(ParseError::UNEXPECTED_END_TAG, $token->name);
$this->error(ParseError::UNEXPECTED_END_TAG);
}
}
# Anything else
@ -1326,7 +1322,7 @@ class TreeBuilder {
}
# Switch the insertion mode to "after body".
self::$insertionMode = self::AFTER_BODY_MODE;
$this->insertionMode = self::AFTER_BODY_MODE;
// The only thing different between body and html here is that when processing
// an html end tag the token is reprocessed.
@ -1789,7 +1785,7 @@ class TreeBuilder {
# Insert a foreign element for the token, in the same namespace as the adjusted
# current node.
static::insertStartTagToken($token, null, $this->stack->adjustedCurrentNode->namespaceURI);
$this->insertStartTagToken($token, null, $this->stack->adjustedCurrentNode->namespaceURI);
# If the token has its self-closing flag set, then run the appropriate steps
# from the following list:
@ -1944,7 +1940,7 @@ class TreeBuilder {
];
}
public static function insertCharacterToken(CharacterToken $token) {
public function insertCharacterToken(CharacterToken $token) {
# 1. Let data be the characters passed to the algorithm, or, if no characters
# were explicitly specified, the character of the character token being
# processed.
@ -1952,7 +1948,7 @@ class TreeBuilder {
# 2. Let the adjusted insertion location be the appropriate place for inserting
# a node.
$location = static::$instance->appropriatePlaceForInsertingNode();
$location = $this->appropriatePlaceForInsertingNode();
$adjustedInsertionLocation = $location['node'];
$insertBefore = $location['insert before'];
@ -1998,7 +1994,7 @@ class TreeBuilder {
$adjustedInsertionLocation = $position;
$insertBefore = false;
} else {
$location = static::$instance->appropriatePlaceForInsertingNode();
$location = $this->appropriatePlaceForInsertingNode();
$adjustedInsertionLocation = $location['node'];
$insertBefore = $location['insert before'];
}
@ -2016,7 +2012,7 @@ class TreeBuilder {
}
}
public static function insertStartTagToken(StartTagToken $token, \DOMNode $intendedParent = null, string $namespace = null): Element {
public function insertStartTagToken(StartTagToken $token, \DOMNode $intendedParent = null, string $namespace = null): Element {
if (!is_null($namespace)) {
$namespace = $token->namespace;
}
@ -2042,9 +2038,9 @@ class TreeBuilder {
// DEVIATION: There is no point to setting the synchronous custom elements flag
// and custom element definition; there is no scripting in this implementation.
if ($namespace === Parser::HTML_NAMESPACE) {
$element = static::$instance->DOM->createElement($token->name);
$element = $this->DOM->createElement($token->name);
} else {
$element = static::$instance->DOM->createElementNS($namespace, $token->name);
$element = $this->DOM->createElementNS($namespace, $token->name);
}
# 8. Append each attribute in the given token to element.
@ -2108,7 +2104,7 @@ class TreeBuilder {
# 1. Let the adjusted insertion location be the appropriate place for inserting
# a node.
$location = static::$instance->appropriatePlaceForInsertingNode($intendedParent);
$location = $this->appropriatePlaceForInsertingNode($intendedParent);
$adjustedInsertionLocation = $location['node'];
$insertBefore = $location['insert before'];
@ -2136,7 +2132,7 @@ class TreeBuilder {
// DEVIATION: Unnecessary because there is no scripting in this implementation.
# 4. Push element onto the stack of open elements so that it is the new current node.
static::$instance->stack[] = $element;
$this->stack[] = $element;
# Return element.
return $element;
@ -2148,7 +2144,7 @@ class TreeBuilder {
# invoked in response to a start tag token.
# 1. Insert an HTML element for the token.
static::insertStartTagToken($token);
$this->insertStartTagToken($token);
# 2. If the algorithm that was invoked is the generic raw text element parsing
# algorithm, switch the tokenizer to the RAWTEXT state; otherwise the algorithm

Write Preview
Loading…
Cancel
Save