← All changes | classes/models/FrmSpamCheckDenylist.php +136 -9 6.34 → trunk View file →

		@@ -17,8 +17,33 @@
17	17
18	18	const COMPARE_EQUALS = 'equals';
19	19
20	20	/**
	21	+ * How many leading characters of a denylist line are used as its index key.
	22	+ * Four measured best on the shipped denylists: shorter keys are not selective
	23	+ * enough, and longer ones push more lines below the length the index needs.
	24	+ *
	25	+ * @since 6.35
	26	+ */
	27	+ const PREFIX_LENGTH = 4;
	28	+
	29	+ /**
	30	+ * Values shorter than this are not indexed. Comparing them is already cheap
	31	+ * enough that building the index would cost more than it saves.
	32	+ *
	33	+ * @since 6.35
	34	+ */
	35	+ const MIN_LENGTH_TO_INDEX = 1024;
	36	+
	37	+ /**
	38	+ * Values longer than this are not indexed, so the index cannot grow without
	39	+ * bound on an unusually large submission.
	40	+ *
	41	+ * @since 6.35
	42	+ */
	43	+ const MAX_LENGTH_TO_INDEX = 524288;
	44	+
	45	+ /**
21	46	* @var array|null
22	47	*/
23	48	protected $posted_fields;
24	49
		@@ -95,9 +120,9 @@
95	120	'file' => FrmAppHelper::plugin_path() . '/denylist/domain-partial.txt',
96	121	),
97	122	array(
98	123	'file' => FrmAppHelper::plugin_path() . '/denylist/splorp-wp-comment.txt',
99		- 'skip' => FrmAppHelper::current_user_can( 'frm_create_entries' ),
	124	+ 'skip' => current_user_can( 'frm_create_entries' ),
100	125	'skip_field_types' => array( 'file' ),
101	126	),
102	127	array(
103	128	'words' => array(
		@@ -179,8 +204,13 @@
179	204
180	205	$this->fill_default_denylist_data( $denylist );
181	206	$denylist['allowed_words'] = $allowed_words;
182	207
	208	+ if ( ! $this->add_values_to_check( $denylist ) ) {
	209	+ // Nothing in this submission needs to be checked against this denylist.
	210	+ continue;
	211	+ }
	212	+
183	213	if ( ! empty( $denylist['words'] ) ) {
184	214	foreach ( $denylist['words'] as $word ) {
185	215	if ( $this->single_line_check_values( $word, $denylist ) ) {
186	216	self::add_spam_keyword_to_option( $word );
		@@ -231,8 +261,103 @@
231	261	);
232	262	}
233	263
234	264	/**
	265	+ * Extracts the submitted values this denylist needs to check, and the string
	266	+ * forms those values are compared against.
	267	+ *
	268	+ * The values depend on the denylist configuration, not on the word or file line
	269	+ * being compared, so they are extracted once here and carried on the denylist.
	270	+ * The shipped denylist files hold tens of thousands of lines and a large form
	271	+ * posts more than a thousand values, so extracting per line is quadratic.
	272	+ *
	273	+ * @since 6.35
	274	+ *
	275	+ * @param array $denylist Denylist data, with the defaults already filled in.
	276	+ *
	277	+ * @return bool False if this submission has no values for this denylist to check.
	278	+ */
	279	+ protected function add_values_to_check( &$denylist ) {
	280	+ $values_to_check = $this->get_values_to_check( $denylist );
	281	+
	282	+ if ( ! $values_to_check ) {
	283	+ return false;
	284	+ }
	285	+
	286	+ $values_string = $this->convert_values_to_string( $values_to_check );
	287	+
	288	+ $denylist['values_to_check'] = $values_to_check;
	289	+ $denylist['values_string'] = $values_string;
	290	+ $denylist['values_string_lower'] = $this->convert_to_lowercase( $values_string );
	291	+ $denylist['values_prefix_index'] = $this->get_values_prefix_index( $denylist );
	292	+
	293	+ return true;
	294	+ }
	295	+
	296	+ /**
	297	+ * Indexes every PREFIX_LENGTH character window of the values.
	298	+ *
	299	+ * A line can only be inside the values if its own first PREFIX_LENGTH
	300	+ * characters are somewhere in them, so a line whose prefix is missing from
	301	+ * this index cannot match and does not need to be compared at all. The shipped
	302	+ * denylists hold tens of thousands of lines and each comparison reads the whole
	303	+ * values string, so ruling a line out with one array lookup is worth the index.
	304	+ *
	305	+ * Returns an empty array when the index would not answer for this denylist, or
	306	+ * would not pay for itself. Every line is then compared as before.
	307	+ *
	308	+ * @since 6.35
	309	+ *
	310	+ * @param array $denylist Denylist data, holding the values strings.
	311	+ *
	312	+ * @return array Index of value prefixes, or an empty array for no index.
	313	+ */
	314	+ protected function get_values_prefix_index( $denylist ) {
	315	+ if ( ! empty( $denylist['is_regex'] ) || self::COMPARE_CONTAINS !== $denylist['compare'] ) {
	316	+ // A regex line is a pattern rather than a literal, so its leading
	317	+ // characters are not text to look for. Only "contains" is indexable.
	318	+ return array();
	319	+ }
	320	+
	321	+ $values = $denylist['values_string_lower'];
	322	+ $length = strlen( $values );
	323	+
	324	+ if ( $length < self::MIN_LENGTH_TO_INDEX || $length > self::MAX_LENGTH_TO_INDEX ) {
	325	+ return array();
	326	+ }
	327	+
	328	+ $index = array();
	329	+ $last = $length - self::PREFIX_LENGTH;
	330	+
	331	+ for ( $i = 0; $i <= $last; $i++ ) {
	332	+ $index[ substr( $values, $i, self::PREFIX_LENGTH ) ] = true;
	333	+ }
	334	+
	335	+ return $index;
	336	+ }
	337	+
	338	+ /**
	339	+ * Checks the values index to rule a line out before comparing it.
	340	+ *
	341	+ * A `false` here does not mean the line matches, only that the index cannot
	342	+ * rule it out, so the caller still has to compare it.
	343	+ *
	344	+ * @since 6.35
	345	+ *
	346	+ * @param string $line The lowercased denylist line.
	347	+ * @param array $args Check args, holding the index when there is one.
	348	+ *
	349	+ * @return bool True when the line cannot be inside the values.
	350	+ */
	351	+ protected function line_is_ruled_out( $line, $args ) {
	352	+ if ( empty( $args['values_prefix_index'] ) || strlen( $line ) < self::PREFIX_LENGTH ) {
	353	+ return false;
	354	+ }
	355	+
	356	+ return ! isset( $args['values_prefix_index'][ substr( $line, 0, self::PREFIX_LENGTH ) ] );
	357	+ }
	358	+
	359	+ /**
235	360	* Gets words from setting.
236	361	*
237	362	* @param string $setting_key Setting key.
238	363	*
		@@ -254,9 +379,10 @@
254	379	/**
255	380	* Checks the values against each single word.
256	381	*
257	382	* @param string $line Single line.
258		- * @param array $args Check args.
	383	+ * @param array $args Check args. Carries the values to check when they have
	384	+ * already been extracted by {@see FrmSpamCheckDenylist::add_values_to_check()}.
259	385	*
260	386	* @return bool
261	387	*/
262	388	protected function single_line_check_values( $line, $args ) {
		@@ -266,21 +392,19 @@
266	392	if ( ! empty( $args['allowed_words'] ) && in_array( $line, $args['allowed_words'], true ) ) {
267	393	return false;
268	394	}
269	395
270		- $values_to_check = $this->get_values_to_check( $args );
271		-
272		- if ( ! $values_to_check ) {
	396	+ if ( ! isset( $args['values_to_check'] ) && ! $this->add_values_to_check( $args ) ) {
273	397	// Nothing needs to be checked.
274	398	return false;
275	399	}
276	400
277	401	if ( ! empty( $args['is_regex'] ) ) {
278		- return preg_match( '/' . trim( $line, '/' ) . '/i', $this->convert_values_to_string( $values_to_check ) );
	402	+ return preg_match( '/' . trim( $line, '/' ) . '/i', $args['values_string'] );
279	403	}
280	404
281	405	if ( self::COMPARE_EQUALS === $args['compare'] ) {
282		- foreach ( $values_to_check as $value ) {
	406	+ foreach ( $args['values_to_check'] as $value ) {
283	407	$value = $this->convert_to_lowercase( $value );
284	408
285	409	if ( $line === $value ) {
286	410	return true;
		@@ -289,10 +413,13 @@
289	413
290	414	return false;
291	415	}
292	416
293		- $values_str = strtolower( $this->convert_values_to_string( $values_to_check ) );
294		- return str_contains( $values_str, $line );
	417	+ if ( $this->line_is_ruled_out( $line, $args ) ) {
	418	+ return false;
	419	+ }
	420	+
	421	+ return str_contains( $args['values_string_lower'], $line );
295	422	}
296	423
297	424	/**
298	425	* Converts values to string to check.