0
0
mirror of https://github.com/ezyang/htmlpurifier.git synced 2025-01-05 14:11:52 +00:00

- Restore substr_count compatibility method; it's not just PHP 4

- Update missing includes
- Fix generate-standalone.php fatal error
- Make LexerTest resilient against variant versions of libxml

git-svn-id: http://htmlpurifier.org/svnroot/htmlpurifier/trunk@1573 48356398-32a2-884e-a903-53898d9a118a
This commit is contained in:
Edward Z. Yang 2008-02-20 01:28:19 +00:00
parent b5f1c76ee8
commit cb793cd9b9
4 changed files with 49 additions and 20 deletions
library
maintenance
tests/HTMLPurifier

View File

@ -136,6 +136,7 @@ require 'HTMLPurifier/HTMLModule/List.php';
require 'HTMLPurifier/HTMLModule/NonXMLCommonAttributes.php'; require 'HTMLPurifier/HTMLModule/NonXMLCommonAttributes.php';
require 'HTMLPurifier/HTMLModule/Object.php'; require 'HTMLPurifier/HTMLModule/Object.php';
require 'HTMLPurifier/HTMLModule/Presentation.php'; require 'HTMLPurifier/HTMLModule/Presentation.php';
require 'HTMLPurifier/HTMLModule/Proprietary.php';
require 'HTMLPurifier/HTMLModule/Ruby.php'; require 'HTMLPurifier/HTMLModule/Ruby.php';
require 'HTMLPurifier/HTMLModule/Scripting.php'; require 'HTMLPurifier/HTMLModule/Scripting.php';
require 'HTMLPurifier/HTMLModule/StyleAttribute.php'; require 'HTMLPurifier/HTMLModule/StyleAttribute.php';

View File

@ -81,7 +81,7 @@ class HTMLPurifier_Lexer_DirectLex extends HTMLPurifier_Lexer
$cursor > 0 && // cursor is further than zero $cursor > 0 && // cursor is further than zero
$loops % $synchronize_interval === 0 // time to synchronize! $loops % $synchronize_interval === 0 // time to synchronize!
) { ) {
$current_line = 1 + substr_count($html, $nl, 0, $cursor); $current_line = 1 + $this->substrCount($html, $nl, 0, $cursor);
} }
$position_next_lt = strpos($html, '<', $cursor); $position_next_lt = strpos($html, '<', $cursor);
@ -106,7 +106,7 @@ class HTMLPurifier_Lexer_DirectLex extends HTMLPurifier_Lexer
); );
if ($maintain_line_numbers) { if ($maintain_line_numbers) {
$token->line = $current_line; $token->line = $current_line;
$current_line += substr_count($html, $nl, $cursor, $position_next_lt - $cursor); $current_line += $this->substrCount($html, $nl, $cursor, $position_next_lt - $cursor);
} }
$array[] = $token; $array[] = $token;
$cursor = $position_next_lt + 1; $cursor = $position_next_lt + 1;
@ -150,7 +150,7 @@ class HTMLPurifier_Lexer_DirectLex extends HTMLPurifier_Lexer
// Check if it's a comment // Check if it's a comment
if ( if (
strncmp('!--', $segment, 3) === 0 substr($segment, 0, 3) === '!--'
) { ) {
// re-determine segment length, looking for --> // re-determine segment length, looking for -->
$position_comment_end = strpos($html, '-->', $cursor); $position_comment_end = strpos($html, '-->', $cursor);
@ -168,11 +168,13 @@ class HTMLPurifier_Lexer_DirectLex extends HTMLPurifier_Lexer
$segment = substr($html, $cursor, $strlen_segment); $segment = substr($html, $cursor, $strlen_segment);
$token = new $token = new
HTMLPurifier_Token_Comment( HTMLPurifier_Token_Comment(
substr($segment, 3) substr(
$segment, 3, $strlen_segment - 3
)
); );
if ($maintain_line_numbers) { if ($maintain_line_numbers) {
$token->line = $current_line; $token->line = $current_line;
$current_line += substr_count($html, $nl, $cursor, $strlen_segment); $current_line += $this->substrCount($html, $nl, $cursor, $strlen_segment);
} }
$array[] = $token; $array[] = $token;
$cursor = $end ? $position_comment_end : $position_comment_end + 3; $cursor = $end ? $position_comment_end : $position_comment_end + 3;
@ -187,7 +189,7 @@ class HTMLPurifier_Lexer_DirectLex extends HTMLPurifier_Lexer
$token = new HTMLPurifier_Token_End($type); $token = new HTMLPurifier_Token_End($type);
if ($maintain_line_numbers) { if ($maintain_line_numbers) {
$token->line = $current_line; $token->line = $current_line;
$current_line += substr_count($html, $nl, $cursor, $position_next_gt - $cursor); $current_line += $this->substrCount($html, $nl, $cursor, $position_next_gt - $cursor);
} }
$array[] = $token; $array[] = $token;
$inside_tag = false; $inside_tag = false;
@ -211,7 +213,7 @@ class HTMLPurifier_Lexer_DirectLex extends HTMLPurifier_Lexer
); );
if ($maintain_line_numbers) { if ($maintain_line_numbers) {
$token->line = $current_line; $token->line = $current_line;
$current_line += substr_count($html, $nl, $cursor, $position_next_gt - $cursor); $current_line += $this->substrCount($html, $nl, $cursor, $position_next_gt - $cursor);
} }
$array[] = $token; $array[] = $token;
$cursor = $position_next_gt + 1; $cursor = $position_next_gt + 1;
@ -240,7 +242,7 @@ class HTMLPurifier_Lexer_DirectLex extends HTMLPurifier_Lexer
} }
if ($maintain_line_numbers) { if ($maintain_line_numbers) {
$token->line = $current_line; $token->line = $current_line;
$current_line += substr_count($html, $nl, $cursor, $position_next_gt - $cursor); $current_line += $this->substrCount($html, $nl, $cursor, $position_next_gt - $cursor);
} }
$array[] = $token; $array[] = $token;
$inside_tag = false; $inside_tag = false;
@ -272,7 +274,7 @@ class HTMLPurifier_Lexer_DirectLex extends HTMLPurifier_Lexer
} }
if ($maintain_line_numbers) { if ($maintain_line_numbers) {
$token->line = $current_line; $token->line = $current_line;
$current_line += substr_count($html, $nl, $cursor, $position_next_gt - $cursor); $current_line += $this->substrCount($html, $nl, $cursor, $position_next_gt - $cursor);
} }
$array[] = $token; $array[] = $token;
$cursor = $position_next_gt + 1; $cursor = $position_next_gt + 1;
@ -300,6 +302,22 @@ class HTMLPurifier_Lexer_DirectLex extends HTMLPurifier_Lexer
return $array; return $array;
} }
/**
* PHP 5.0.x compatible substr_count that implements offset and length
*/
protected function substrCount($haystack, $needle, $offset, $length) {
static $oldVersion;
if ($oldVersion === null) {
$oldVersion = version_compare(PHP_VERSION, '5.1', '<');
}
if ($oldVersion) {
$haystack = substr($haystack, $offset, $length);
return substr_count($haystack, $needle);
} else {
return substr_count($haystack, $needle, $offset, $length);
}
}
/** /**
* Takes the inside of an HTML tag and makes an assoc array of attributes. * Takes the inside of an HTML tag and makes an assoc array of attributes.
* *

View File

@ -110,7 +110,6 @@ function replace_includes_callback($matches) {
shell_exec('php generate-includes.php'); shell_exec('php generate-includes.php');
chdir(dirname(__FILE__) . '/../library/'); chdir(dirname(__FILE__) . '/../library/');
create_blank('HTMLPurifier.php');
echo 'Creating full file...'; echo 'Creating full file...';
$contents = replace_includes(file_get_contents('HTMLPurifier.includes.php')); $contents = replace_includes(file_get_contents('HTMLPurifier.includes.php'));

View File

@ -510,6 +510,26 @@ class HTMLPurifier_LexerTest extends HTMLPurifier_Harness
} }
function test_tokenizeHTML_() { function test_tokenizeHTML_() {
$extra = array(
// PH5P doesn't seem to like style tags
'PH5P' => false,
// DirectLex defers to RemoveForeignElements for textification
'DirectLex' => array(
new HTMLPurifier_Token_Start('style', array('type' => 'text/css')),
new HTMLPurifier_Token_Comment("\ndiv {}\n"),
new HTMLPurifier_Token_End('style'),
),
);
if (!defined('LIBXML_VERSION') || LIBXML_VERSION < 20628) {
// libxml's behavior is wrong prior to this version, so make
// appropriate accomodations
// :NOTE: LIBXML_VERSION is missing in early versions of PHP
// prior to 1.30 of php-src/ext/libxml/libxml.c (version-wise,
// this translates to 5.0.x. In such cases, we assume that an old
// version of libxml is being used, although that *might* not
// be the case (it's very unlikely though)
$extra['DOMLex'] = $extra['DirectLex'];
}
$this->assertTokenization( $this->assertTokenization(
'<style type="text/css"><!-- '<style type="text/css"><!--
div {} div {}
@ -519,16 +539,7 @@ div {}
new HTMLPurifier_Token_Text("\ndiv {}\n"), new HTMLPurifier_Token_Text("\ndiv {}\n"),
new HTMLPurifier_Token_End('style'), new HTMLPurifier_Token_End('style'),
), ),
array( $extra
// PH5P doesn't seem to like style tags
'PH5P' => false,
// DirectLex defers to RemoveForeignElements for textification
'DirectLex' => array(
new HTMLPurifier_Token_Start('style', array('type' => 'text/css')),
new HTMLPurifier_Token_Comment("\ndiv {}\n"),
new HTMLPurifier_Token_End('style'),
),
)
); );
} }