Skip to content

Commit 703a2c7

Browse files
whyisjakewhyisjake
committed
Ensure that a user can publish_posts before making a post sticky.
Props: danielbachhuber, whyisjake, peterwilson, xknown. Prevent stored XSS through wp_targeted_link_rel(). Props: vortfu, whyisjake, peterwilsoncc, xknown, SergeyBiryukov, flaviozavan. Update `wp_kses_bad_protocol()` to recognize `:` on uri attributes, `wp_kses_bad_protocol()` makes sure to validate that uri attributes don’t contain invalid/or not allowed protocols. While this works fine in most cases, there’s a risk that by using the colon html5 named entity, one is able to bypass this function. Brings r46895 to the 5.3 branch. Props: xknown, nickdaugherty, peterwilsoncc. Prevent stored XSS in the block editor. Brings r46896 to the 5.3 branch. Prevent escaped unicode characters become unescaped in unsafe HTML during JSON decoding. Props: aduth, epiqueras. git-svn-id: https://develop.svn.wordpress.org/branches/5.2@46901 602fd350-edb4-49c9-b593-d223f7449a82
1 parent 813cb5a commit 703a2c7

8 files changed

Lines changed: 364 additions & 53 deletions

File tree

‎src/wp-includes/blocks.php‎

Lines changed: 227 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -74,11 +74,11 @@ function has_blocks( $post = null ) {
7474
* @since 5.0.0
7575
* @see parse_blocks()
7676
*
77-
* @param string $block_type Full Block type to look for.
77+
* @param string $block_name Full Block type to look for.
7878
* @param int|string|WP_Post|null $post Optional. Post content, post ID, or post object. Defaults to global $post.
7979
* @return bool Whether the post content contains the specified block.
8080
*/
81-
function has_block( $block_type, $post = null ) {
81+
function has_block( $block_name, $post = null ) {
8282
if ( ! has_blocks( $post ) ) {
8383
return false;
8484
}
@@ -90,7 +90,30 @@ function has_block( $block_type, $post = null ) {
9090
}
9191
}
9292

93-
return false !== strpos( $post, '<!-- wp:' . $block_type . ' ' );
93+
/*
94+
* Normalize block name to include namespace, if provided as non-namespaced.
95+
* This matches behavior for WordPress 5.0.0 - 5.3.0 in matching blocks by
96+
* their serialized names.
97+
*/
98+
if ( false === strpos( $block_name, '/' ) ) {
99+
$block_name = 'core/' . $block_name;
100+
}
101+
102+
// Test for existence of block by its fully qualified name.
103+
$has_block = false !== strpos( $post, '<!-- wp:' . $block_name . ' ' );
104+
105+
if ( ! $has_block ) {
106+
/*
107+
* If the given block name would serialize to a different name, test for
108+
* existence by the serialized form.
109+
*/
110+
$serialized_block_name = strip_core_block_namespace( $block_name );
111+
if ( $serialized_block_name !== $block_name ) {
112+
$has_block = false !== strpos( $post, '<!-- wp:' . $serialized_block_name . ' ' );
113+
}
114+
}
115+
116+
return $has_block;
94117
}
95118

96119
/**
@@ -113,6 +136,207 @@ function get_dynamic_block_names() {
113136
return $dynamic_block_names;
114137
}
115138

139+
/**
140+
* Given an array of attributes, returns a string in the serialized attributes
141+
* format prepared for post content.
142+
*
143+
* The serialized result is a JSON-encoded string, with unicode escape sequence
144+
* substitution for characters which might otherwise interfere with embedding
145+
* the result in an HTML comment.
146+
*
147+
* @since 5.3.1
148+
*
149+
* @param array $attributes Attributes object.
150+
* @return string Serialized attributes.
151+
*/
152+
function serialize_block_attributes( $block_attributes ) {
153+
$encoded_attributes = json_encode( $block_attributes );
154+
$encoded_attributes = preg_replace( '/--/', '\\u002d\\u002d', $encoded_attributes );
155+
$encoded_attributes = preg_replace( '/</', '\\u003c', $encoded_attributes );
156+
$encoded_attributes = preg_replace( '/>/', '\\u003e', $encoded_attributes );
157+
$encoded_attributes = preg_replace( '/&/', '\\u0026', $encoded_attributes );
158+
// Regex: /\\"/
159+
$encoded_attributes = preg_replace( '/\\\\"/', '\\u0022', $encoded_attributes );
160+
161+
return $encoded_attributes;
162+
}
163+
164+
/**
165+
* Returns the block name to use for serialization. This will remove the default
166+
* "core/" namespace from a block name.
167+
*
168+
* @since 5.3.1
169+
*
170+
* @param string $block_name Original block name.
171+
* @return string Block name to use for serialization.
172+
*/
173+
function strip_core_block_namespace( $block_name = null ) {
174+
if ( is_string( $block_name ) && 0 === strpos( $block_name, 'core/' ) ) {
175+
return substr( $block_name, 5 );
176+
}
177+
178+
return $block_name;
179+
}
180+
181+
/**
182+
* Returns the content of a block, including comment delimiters.
183+
*
184+
* @since 5.3.1
185+
*
186+
* @param string $block_name Block name.
187+
* @param array $attributes Block attributes.
188+
* @param string $content Block save content.
189+
* @return string Comment-delimited block content.
190+
*/
191+
function get_comment_delimited_block_content( $block_name = null, $block_attributes, $block_content ) {
192+
if ( is_null( $block_name ) ) {
193+
return $block_content;
194+
}
195+
196+
$serialized_block_name = strip_core_block_namespace( $block_name );
197+
$serialized_attributes = empty( $block_attributes ) ? '' : serialize_block_attributes( $block_attributes ) . ' ';
198+
199+
if ( empty( $block_content ) ) {
200+
return sprintf( '<!-- wp:%s %s/-->', $serialized_block_name, $serialized_attributes );
201+
}
202+
203+
return sprintf(
204+
'<!-- wp:%s %s-->%s<!-- /wp:%s -->',
205+
$serialized_block_name,
206+
$serialized_attributes,
207+
$block_content,
208+
$serialized_block_name
209+
);
210+
}
211+
212+
/**
213+
* Returns the content of a block, including comment delimiters, serializing all
214+
* attributes from the given parsed block.
215+
*
216+
* This should be used when preparing a block to be saved to post content.
217+
* Prefer `render_block` when preparing a block for display. Unlike
218+
* `render_block`, this does not evaluate a block's `render_callback`, and will
219+
* instead preserve the markup as parsed.
220+
*
221+
* @since 5.3.1
222+
*
223+
* @param WP_Block_Parser_Block $block A single parsed block object.
224+
* @return string String of rendered HTML.
225+
*/
226+
function serialize_block( $block ) {
227+
$block_content = '';
228+
229+
$index = 0;
230+
foreach ( $block['innerContent'] as $chunk ) {
231+
$block_content .= is_string( $chunk ) ? $chunk : serialize_block( $block['innerBlocks'][ $index++ ] );
232+
}
233+
234+
if ( ! is_array( $block['attrs'] ) ) {
235+
$block['attrs'] = array();
236+
}
237+
238+
return get_comment_delimited_block_content(
239+
$block['blockName'],
240+
$block['attrs'],
241+
$block_content
242+
);
243+
}
244+
245+
/**
246+
* Returns a joined string of the aggregate serialization of the given parsed
247+
* blocks.
248+
*
249+
* @since 5.3.1
250+
*
251+
* @param WP_Block_Parser_Block[] $blocks Parsed block objects.
252+
* @return string String of rendered HTML.
253+
*/
254+
function serialize_blocks( $blocks ) {
255+
return implode( '', array_map( 'serialize_block', $blocks ) );
256+
}
257+
258+
/**
259+
* Filters and sanitizes block content to remove non-allowable HTML from
260+
* parsed block attribute values.
261+
*
262+
* @since 5.3.1
263+
*
264+
* @param string $text Text that may contain block content.
265+
* @param array[]|string $allowed_html An array of allowed HTML elements
266+
* and attributes, or a context name
267+
* such as 'post'.
268+
* @param string[] $allowed_protocols Array of allowed URL protocols.
269+
* @return string The filtered and sanitized content result.
270+
*/
271+
function filter_block_content( $text, $allowed_html = 'post', $allowed_protocols = array() ) {
272+
$result = '';
273+
274+
$blocks = parse_blocks( $text );
275+
foreach ( $blocks as $block ) {
276+
$block = filter_block_kses( $block, $allowed_html, $allowed_protocols );
277+
$result .= serialize_block( $block );
278+
}
279+
280+
return $result;
281+
}
282+
283+
/**
284+
* Filters and sanitizes a parsed block to remove non-allowable HTML from block
285+
* attribute values.
286+
*
287+
* @since 5.3.1
288+
*
289+
* @param WP_Block_Parser_Block $block The parsed block object.
290+
* @param array[]|string $allowed_html An array of allowed HTML
291+
* elements and attributes, or a
292+
* context name such as 'post'.
293+
* @param string[] $allowed_protocols Allowed URL protocols.
294+
* @return array The filtered and sanitized block object result.
295+
*/
296+
function filter_block_kses( $block, $allowed_html, $allowed_protocols = array() ) {
297+
$block['attrs'] = filter_block_kses_value( $block['attrs'], $allowed_html, $allowed_protocols );
298+
299+
if ( is_array( $block['innerBlocks'] ) ) {
300+
foreach ( $block['innerBlocks'] as $i => $inner_block ) {
301+
$block['innerBlocks'][ $i ] = filter_block_kses( $inner_block, $allowed_html, $allowed_protocols );
302+
}
303+
}
304+
305+
return $block;
306+
}
307+
308+
/**
309+
* Filters and sanitizes a parsed block attribute value to remove non-allowable
310+
* HTML.
311+
*
312+
* @since 5.3.1
313+
*
314+
* @param mixed $value The attribute value to filter.
315+
* @param array[]|string $allowed_html An array of allowed HTML elements
316+
* and attributes, or a context name
317+
* such as 'post'.
318+
* @param string[] $allowed_protocols Array of allowed URL protocols.
319+
* @return array The filtered and sanitized result.
320+
*/
321+
function filter_block_kses_value( $value, $allowed_html, $allowed_protocols = array() ) {
322+
if ( is_array( $value ) ) {
323+
foreach ( $value as $key => $inner_value ) {
324+
$filtered_key = filter_block_kses_value( $key, $allowed_html, $allowed_protocols );
325+
$filtered_value = filter_block_kses_value( $inner_value, $allowed_html, $allowed_protocols );
326+
327+
if ( $filtered_key !== $key ) {
328+
unset( $value[ $key ] );
329+
}
330+
331+
$value[ $filtered_key ] = $filtered_value;
332+
}
333+
} elseif ( is_string( $value ) ) {
334+
return wp_kses( $value, $allowed_html, $allowed_protocols );
335+
}
336+
337+
return $value;
338+
}
339+
116340
/**
117341
* Parses blocks out of a content string, and renders those appropriate for the excerpt.
118342
*

‎src/wp-includes/default-filters.php‎

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -245,6 +245,7 @@
245245
add_filter( 'tiny_mce_before_init', '_mce_set_direction' );
246246
add_filter( 'teeny_mce_before_init', '_mce_set_direction' );
247247
add_filter( 'pre_kses', 'wp_pre_kses_less_than' );
248+
add_filter( 'pre_kses', 'wp_pre_kses_block_attributes', 10, 3 );
248249
add_filter( 'sanitize_title', 'sanitize_title_with_dashes', 10, 3 );
249250
add_action( 'check_comment_flood', 'check_comment_flood_db', 10, 4 );
250251
add_filter( 'comment_flood_filter', 'wp_throttle_comment_flood', 10, 3 );

‎src/wp-includes/formatting.php‎

Lines changed: 71 additions & 33 deletions
Original file line numberDiff line numberDiff line change
@@ -3043,8 +3043,26 @@ function wp_rel_nofollow_callback( $matches ) {
30433043
*/
30443044
function wp_targeted_link_rel( $text ) {
30453045
// Don't run (more expensive) regex if no links with targets.
3046-
if ( stripos( $text, 'target' ) !== false && stripos( $text, '<a ' ) !== false ) {
3047-
$text = preg_replace_callback( '|<a\s([^>]*target\s*=[^>]*)>|i', 'wp_targeted_link_rel_callback', $text );
3046+
if ( stripos( $text, 'target' ) === false || stripos( $text, '<a ' ) === false || is_serialized( $text ) ) {
3047+
return $text;
3048+
}
3049+
3050+
$script_and_style_regex = '/<(script|style).*?<\/\\1>/si';
3051+
3052+
preg_match_all( $script_and_style_regex, $text, $matches );
3053+
$extra_parts = $matches[0];
3054+
$html_parts = preg_split( $script_and_style_regex, $text );
3055+
3056+
foreach ( $html_parts as &$part ) {
3057+
$part = preg_replace_callback( '|<a\s([^>]*target\s*=[^>]*)>|i', 'wp_targeted_link_rel_callback', $part );
3058+
}
3059+
3060+
$text = '';
3061+
for ( $i = 0; $i < count( $html_parts ); $i++ ) {
3062+
$text .= $html_parts[ $i ];
3063+
if ( isset( $extra_parts[ $i ] ) ) {
3064+
$text .= $extra_parts[ $i ];
3065+
}
30483066
}
30493067

30503068
return $text;
@@ -3062,8 +3080,17 @@ function wp_targeted_link_rel( $text ) {
30623080
* @return string HTML A Element with rel noreferrer noopener in addition to any existing values
30633081
*/
30643082
function wp_targeted_link_rel_callback( $matches ) {
3065-
$link_html = $matches[1];
3066-
$rel_match = array();
3083+
$link_html = $matches[1];
3084+
$original_link_html = $link_html;
3085+
3086+
// Consider the html escaped if there are no unescaped quotes
3087+
$is_escaped = ! preg_match( '/(^|[^\\\\])[\'"]/', $link_html );
3088+
if ( $is_escaped ) {
3089+
// Replace only the quotes so that they are parsable by wp_kses_hair, leave the rest as is
3090+
$link_html = preg_replace( '/\\\\([\'"])/', '$1', $link_html );
3091+
}
3092+
3093+
$atts = wp_kses_hair( $link_html, wp_allowed_protocols() );
30673094

30683095
/**
30693096
* Filters the rel values that are added to links with `target` attribute.
@@ -3075,35 +3102,21 @@ function wp_targeted_link_rel_callback( $matches ) {
30753102
*/
30763103
$rel = apply_filters( 'wp_targeted_link_rel', 'noopener noreferrer', $link_html );
30773104

3078-
// Avoid additional regex if the filter removes rel values.
3079-
if ( ! $rel ) {
3080-
return "<a $link_html>";
3081-
}
3082-
3083-
// Value with delimiters, spaces around are optional.
3084-
$attr_regex = '|rel\s*=\s*?(\\\\{0,1}["\'])(.*?)\\1|i';
3085-
preg_match( $attr_regex, $link_html, $rel_match );
3086-
3087-
if ( empty( $rel_match[0] ) ) {
3088-
// No delimiters, try with a single value and spaces, because `rel = va"lue` is totally fine...
3089-
$attr_regex = '|rel\s*=(\s*)([^\s]*)|i';
3090-
preg_match( $attr_regex, $link_html, $rel_match );
3091-
}
3092-
3093-
if ( ! empty( $rel_match[0] ) ) {
3094-
$parts = preg_split( '|\s+|', strtolower( $rel_match[2] ) );
3095-
$parts = array_map( 'esc_attr', $parts );
3096-
$needed = explode( ' ', $rel );
3097-
$parts = array_unique( array_merge( $parts, $needed ) );
3098-
$delimiter = trim( $rel_match[1] ) ? $rel_match[1] : '"';
3099-
$rel = 'rel=' . $delimiter . trim( implode( ' ', $parts ) ) . $delimiter;
3100-
$link_html = str_replace( $rel_match[0], $rel, $link_html );
3101-
} elseif ( preg_match( '|target\s*=\s*?\\\\"|', $link_html ) ) {
3102-
$link_html .= " rel=\\\"$rel\\\"";
3103-
} elseif ( preg_match( '#(target|href)\s*=\s*?\'#', $link_html ) ) {
3104-
$link_html .= " rel='$rel'";
3105-
} else {
3106-
$link_html .= " rel=\"$rel\"";
3105+
// Return early if no rel values to be added or if no actual target attribute
3106+
if ( ! $rel || ! isset( $atts['target'] ) ) {
3107+
return "<a $original_link_html>";
3108+
}
3109+
3110+
if ( isset( $atts['rel'] ) ) {
3111+
$all_parts = preg_split( '/\s/', "{$atts['rel']['value']} $rel", -1, PREG_SPLIT_NO_EMPTY );
3112+
$rel = implode( ' ', array_unique( $all_parts ) );
3113+
}
3114+
3115+
$atts['rel']['whole'] = 'rel="' . esc_attr( $rel ) . '"';
3116+
$link_html = join( ' ', array_column( $atts, 'whole' ) );
3117+
3118+
if ( $is_escaped ) {
3119+
$link_html = preg_replace( '/[\'"]/', '\\\\$0', $link_html );
31073120
}
31083121

31093122
return "<a $link_html>";
@@ -4807,6 +4820,31 @@ function wp_pre_kses_less_than_callback( $matches ) {
48074820
return $matches[0];
48084821
}
48094822

4823+
/**
4824+
* Remove non-allowable HTML from parsed block attribute values when filtering
4825+
* in the post context.
4826+
*
4827+
* @since 5.3.1
4828+
*
4829+
* @param string $string Content to be run through KSES.
4830+
* @param array[]|string $allowed_html An array of allowed HTML elements
4831+
* and attributes, or a context name
4832+
* such as 'post'.
4833+
* @param string[] $allowed_protocols Array of allowed URL protocols.
4834+
* @return string Filtered text to run through KSES.
4835+
*/
4836+
function wp_pre_kses_block_attributes( $string, $allowed_html, $allowed_protocols ) {
4837+
/*
4838+
* `filter_block_content` is expected to call `wp_kses`. Temporarily remove
4839+
* the filter to avoid recursion.
4840+
*/
4841+
remove_filter( 'pre_kses', 'wp_pre_kses_block_attributes', 10 );
4842+
$string = filter_block_content( $string, $allowed_html, $allowed_protocols );
4843+
add_filter( 'pre_kses', 'wp_pre_kses_block_attributes', 10, 3 );
4844+
4845+
return $string;
4846+
}
4847+
48104848
/**
48114849
* WordPress implementation of PHP sprintf() with filters.
48124850
*

0 commit comments

Comments
 (0)