PluginProbe
Plugin Check (PCP) / trunk
Plugin Check (PCP) vtrunk
2.1.0 trunk 0.1 0.2.0 0.2.1 0.2.2 0.2.3 1.0.0 1.0.1 1.0.2 1.1.0 1.2.0 1.3.0 1.3.1 1.4.0 1.5.0 1.6.0 1.7.0 1.8.0 1.9.0 2.0.0 ci-artifacts
plugin-check / includes / Scanner / Prefix_Scanner.php

Prefix_Scanner.php in Plugin Check (PCP) trunk, at includes/Scanner/Prefix_Scanner.php

815 lines 29.0 KB
No matching file
Up and down to move Enter to open Esc to close
Raw Download Zip
1 <?php
2 /**
3 * Class WordPress\Plugin_Check\Scanner\Prefix_Scanner
4 *
5 * @package plugin-check
6 */
7
8 namespace WordPress\Plugin_Check\Scanner;
9
10 use PhpParser\Node;
11 use PhpParser\Node\Const_;
12
13 /**
14 * Class Prefix_Scanner
15 *
16 * This class is responsible for analyzing and logging PHP files for specific
17 * prefixes used in WordPress plugin and theme development. It identifies potential
18 * issues with prefix usage to maintain code consistency and avoid conflicts.
19 */
20 class Prefix_Scanner extends PHP_Parser {
21
22 /**
23 * List of common prefixes that should be avoided to ensure consistent naming conventions and avoid conflicts.
24 *
25 * @since 1.7.0
26 * @var array
27 */
28 private array $not_valid_prefixes;
29
30 /**
31 * Possible prefixes.
32 *
33 * @since 1.7.0
34 * @var array
35 */
36 private array $possible_prefixes = array();
37
38 /**
39 * Processes items.
40 *
41 * @since 1.7.0
42 * @var array
43 */
44 private array $already_processed_items = array();
45
46 /**
47 * Number of namespaces encountered.
48 *
49 * @since 1.7.0
50 * @var int
51 */
52 private int $namespaces_count = 0;
53
54 /**
55 * Final potential prefixes.
56 *
57 * @since 1.7.0
58 * @var array
59 */
60 public array $final_prefixes = array();
61
62 /**
63 * Array of actions.
64 *
65 * @since 1.7.0
66 * @var array
67 */
68 private array $wp_actions;
69
70 /**
71 * Array of filters.
72 *
73 * @since 1.7.0
74 * @var array
75 */
76 private array $wp_filters;
77
78 /**
79 * Array of constants.
80 *
81 * @since 1.7.0
82 * @var array
83 */
84 private array $wp_constants;
85
86 /**
87 * Array of global variables.
88 *
89 * @since 1.7.0
90 * @var array
91 */
92 private array $wp_globals;
93
94 /**
95 * Array of options.
96 *
97 * @since 1.7.0
98 * @var array
99 */
100 private array $wp_options;
101
102 /**
103 * Array of transients.
104 *
105 * @since 1.7.0
106 * @var array
107 */
108 private array $wp_transients;
109
110 /**
111 * Array of site transients.
112 *
113 * @since 1.7.0
114 * @var array
115 */
116 private array $wp_site_transients;
117
118 /**
119 * Constructor.
120 *
121 * @since 1.7.0
122 */
123 public function __construct() {
124 parent::__construct();
125 $this->not_valid_prefixes = include dirname( __DIR__ ) . '/Vars/prefix-not-valid.php';
126
127 $wp_actions = include dirname( __DIR__ ) . '/Vars/wp-actions.php';
128 $common_actions = include dirname( __DIR__ ) . '/Vars/common-actions.php';
129 $this->wp_actions = array_merge( $wp_actions, $common_actions );
130
131 $wp_filters = include dirname( __DIR__ ) . '/Vars/wp-filters.php';
132 $common_filters = include dirname( __DIR__ ) . '/Vars/common-filters.php';
133 $this->wp_filters = array_merge( $wp_filters, $common_filters );
134
135 $wp_globals = include dirname( __DIR__ ) . '/Vars/wp-globals.php';
136 $common_globals = include dirname( __DIR__ ) . '/Vars/common-globals.php';
137 $this->wp_globals = array_merge( $wp_globals, $common_globals );
138
139 $this->wp_constants = include dirname( __DIR__ ) . '/Vars/wp-constants.php';
140 $this->wp_options = include dirname( __DIR__ ) . '/Vars/wp-options.php';
141 $this->wp_transients = include dirname( __DIR__ ) . '/Vars/wp-transients.php';
142 $this->wp_site_transients = include dirname( __DIR__ ) . '/Vars/wp-site-transients.php';
143 }
144
145 /**
146 * Loads files.
147 *
148 * @since 1.7.0
149 *
150 * @param array $files Array of file paths.
151 * @return void
152 */
153 public function load_files( $files ) {
154 parent::load_files( $files );
155
156 if ( ! empty( $files ) ) {
157 foreach ( $files as $file ) {
158 $this->load( $file );
159 }
160
161 $this->check_prefixes( 'prefixes_code' );
162 }
163 }
164
165 /**
166 * Loads file.
167 *
168 * @since 1.7.0
169 *
170 * @param string $file File path.
171 * @return null
172 */
173 public function load( $file ) {
174 parent::load( $file );
175
176 return null;
177 }
178
179 /**
180 * Executes a series of logging operations including tracking function calls, method calls, global declarations, and abstraction declarations.
181 *
182 * @since 1.7.0
183 *
184 * @return void
185 */
186 public function find() {
187 $this->add_function_calls_to_log();
188 $this->add_method_calls_to_log();
189 $this->add_globals_declarations_to_log();
190 $this->add_abstractions_declarations_to_log();
191 }
192
193 /**
194 * Detects calls to specific functions (specially a set of WordPress functions) and adds them to the log with the argument that is of interest regarding prefixes.
195 *
196 * @since 1.7.0
197 *
198 * @return void
199 */
200 private function add_function_calls_to_log() {
201 // Find some calls to specific functions.
202 $function_calls = $this->node_finder->findInstanceOf( $this->stmts, Node\Expr\FuncCall::class );
203 if ( ! empty( $function_calls ) ) {
204 foreach ( $function_calls as $function_call ) {
205 if ( $this->has_function_name( $function_call ) ) {
206 switch ( $this->get_call_name( $function_call ) ) {
207 case 'define':
208 if ( isset( $function_call->args[0] ) ) {
209 $this->add_call_expression_to_log( $function_call, $function_call->args[0], $this->wp_constants, 0 );
210 }
211 break;
212 case 'add_option':
213 case 'add_site_option':
214 case 'update_option':
215 case 'update_site_option':
216 if ( isset( $function_call->args[0] ) ) {
217 $this->add_call_expression_to_log( $function_call, $function_call->args[0], $this->wp_options, 0 );
218 }
219 break;
220 case 'add_network_option':
221 case 'update_network_option':
222 case 'register_setting':
223 if ( isset( $function_call->args[1] ) ) {
224 $this->add_call_expression_to_log( $function_call, $function_call->args[1], $this->wp_options, 1 );
225 }
226 break;
227 case 'set_transient':
228 if ( isset( $function_call->args[0] ) ) {
229 $this->add_call_expression_to_log( $function_call, $function_call->args[0], $this->wp_transients, 0 );
230 }
231 break;
232 case 'set_site_transient':
233 if ( isset( $function_call->args[0] ) ) {
234 $this->add_call_expression_to_log( $function_call, $function_call->args[0], $this->wp_site_transients, 0 );
235 }
236 break;
237 case 'add_shortcode':
238 case 'register_post_type':
239 case 'register_taxonomy':
240 if ( isset( $function_call->args[0] ) ) {
241 $this->add_call_expression_to_log( $function_call, $function_call->args[0], array(), 0 );
242 }
243 break;
244 case 'do_action':
245 if ( isset( $function_call->args[0] ) ) {
246 $this->add_call_expression_to_log( $function_call, $function_call->args[0], $this->wp_actions, 0 );
247 }
248 break;
249 case 'apply_filters':
250 if ( isset( $function_call->args[0] ) ) {
251 $this->add_call_expression_to_log( $function_call, $function_call->args[0], $this->wp_filters, 0 );
252 }
253 break;
254 case 'wp_schedule_event':
255 if ( isset( $function_call->args[2] ) ) {
256 $this->add_call_expression_to_log( $function_call, $function_call->args[2], array(), 2 );
257 }
258 break;
259 case 'add_menu_page':
260 case 'add_management_page':
261 case 'add_options_page':
262 case 'add_theme_page':
263 case 'add_plugins_page':
264 case 'add_users_page':
265 case 'add_dashboard_page':
266 case 'add_posts_page':
267 case 'add_media_page':
268 case 'add_links_page':
269 case 'add_pages_page':
270 case 'add_comments_page':
271 if ( isset( $function_call->args[3] ) ) {
272 $possible_string = $this->get_possible_string_for_element( $function_call->args[3] );
273 if ( ! empty( $possible_string ) && strlen( $possible_string ) > 8 && preg_match( '/[?\/&]/m', $possible_string ) ) {
274 break; // Ignore it if this is enough long url-looking string.
275 }
276 $this->add_call_expression_to_log( $function_call, $function_call->args[3], array(), 3 );
277 }
278 break;
279 case 'add_submenu_page':
280 if ( isset( $function_call->args[4] ) ) {
281 $possible_string = $this->get_possible_string_for_element( $function_call->args[4] );
282 if ( ! empty( $possible_string ) && strlen( $possible_string ) > 8 && preg_match( '/[?\/&]/m', $possible_string ) ) {
283 break; // Ignore it if this is enough long url-looking string.
284 }
285 $this->add_call_expression_to_log( $function_call, $function_call->args[4], array(), 4 );
286 }
287 break;
288 case 'wp_enqueue_script':
289 case 'wp_register_script':
290 case 'wp_enqueue_style':
291 case 'wp_register_style':
292 if ( isset( $function_call->args[0] ) ) {
293 // Check only obvious cases where this is not the name of a library.
294 $possible_string = $this->get_possible_string_for_element( $function_call->args[0] );
295 if ( ! empty( $possible_string ) ) {
296 $re1 = '/^(my|custom|style|script|css|file)(?!lint|ize)/m'; // Obvious names.
297 $re2 = '/^(js)(?!lint|on|ize|hint|api|olor|-cookie)[-_]/m'; // Obvious names, but more careful with js.
298 $re3 = '/^(admin|wp)(?!-components|-categories|-tags|-custom-fields|-custom-header|-comments|-users|-forms|-gallery|-widgets|-mediaelement|-tinymce|-ajax-response|-lists|-codemirror|-color-picker|-api|-embed|-jquery-ui-dialog|-auth-check|-pointer|-embed-template-ie|-admin|-colopicker|-editor-font|-reusable-blocks|-blocks|-block-directory|-format-library|-block-editor-content|-editor-classic-layout-styles|-reset-editor-styles|-patterns|-block-library-theme|-theme-plugin-editor|-i18n|-a11y|-backbone|-sanitize|-date|-util|-hooks|-api-request|-plupload)[-_]?/m'; // Beginning by admin or wp but ignoring core names.
299 if ( preg_match( $re1, $possible_string ) || preg_match( $re2, $possible_string ) || preg_match( $re3, $possible_string ) ) {
300 $this->add_call_expression_to_log( $function_call, $function_call->args[0], array(), 0 );
301 }
302 }
303 }
304 break;
305 }
306 } else {
307 $this->log()->add_call_expr( $function_call, 0, 'prefixes_code', true );
308 }
309 }
310 }
311 }
312
313 /**
314 * Detects calls to specific methods and adds them to the log.
315 *
316 * @since 1.7.0
317 *
318 * @return void
319 */
320 private function add_method_calls_to_log() {
321 $method_calls = $this->node_finder->findInstanceOf( $this->stmts, Node\Expr\MethodCall::class );
322 if ( ! empty( $method_calls ) ) {
323 foreach ( $method_calls as $method_call ) {
324 if ( ! empty( $method_call->name ) && ! empty( $method_call->var ) ) {
325 $methodname = $this->get_possible_string_for_element( $method_call->name );
326 $varname = $this->get_variable_name( $method_call->var );
327
328 // WooCommerce add/update metadata (Heuristic since it checks if the variable is called product).
329 if ( str_contains( $varname, 'product' ) && in_array( $methodname, array( 'update_meta_data', 'add_meta_data' ), true ) ) {
330 if ( isset( $method_call->args[0] ) ) {
331 $this->add_call_expression_to_log( $method_call, $method_call->args[0], array(), 0 );
332 }
333 }
334 }
335 }
336 }
337 }
338
339 /**
340 * Detects calls to globals variables and adds them to the log.
341 *
342 * @since 1.7.0
343 *
344 * @return void
345 */
346 private function add_globals_declarations_to_log() {
347 // Find global variables definitions.
348
349 // Eg: global $example variable.
350 $globals = $this->node_finder->findInstanceOf( $this->stmts, Node\Stmt\Global_::class );
351 if ( ! empty( $globals ) ) {
352 foreach ( $globals as $global ) {
353 if ( ! empty( $global->vars ) ) {
354 foreach ( $global->vars as $global_variable ) {
355 if ( ! empty( $global_variable->name ) && is_string( $global_variable->name ) ) {
356 if ( ! in_array( $global_variable->name, $this->wp_globals, true ) ) { // Ignore known core global variables.
357 $this->log()->add_var_expr( $global_variable, 'prefixes_code', true );
358 }
359 }
360 }
361 }
362 }
363 }
364
365 // Eg: $GLOBALS['example'] variable.
366 $assigns = $this->node_finder->findInstanceOf( $this->stmts, Node\Expr\Assign::class );
367 if ( ! empty( $assigns ) ) {
368 foreach ( $assigns as $assign ) {
369 if ( ! empty( $assign->var ) && is_a( $assign->var, 'PhpParser\Node\Expr\ArrayDimFetch' ) ) {
370 $dimfetch = $assign->var;
371 if ( 'GLOBALS' === $this->get_variable_name( $dimfetch ) ) {
372 $dims = $this->extract_dims_objects( $dimfetch );
373 // Ignores all cases where we can't find the dim elements.
374 if ( isset( $dims[0] ) ) {
375 $this->add_call_expression_to_log( $dimfetch, $dims[0], $this->wp_globals );
376 }
377 }
378 }
379 }
380 }
381
382 // Find all consts: const EXAMPLE = 'example'.
383 $namespaces = $this->node_finder->findInstanceOf( $this->stmts, Node\Stmt\Namespace_::class );
384 if ( empty( $namespaces ) ) {
385 $consts = $this->node_finder->findInstanceOf( $this->stmts, Const_::class );
386 if ( ! empty( $consts ) ) {
387 foreach ( $consts as $const ) {
388 $is_inside_element_type = null;
389 $contextual_stmts = $this->get_contextual_stmts_for_element( $const, $is_inside_element_type )['context'];
390 if ( ! empty( $contextual_stmts ) && ! in_array( $is_inside_element_type, array( 'PhpParser\Node\Stmt\Class_', 'PhpParser\Node\Stmt\Interface_' ), true ) ) { // Ignore const inside a class.
391 if ( isset( $const->name ) && method_exists( $const->name, '__toString' ) && ! empty( $const->name->__toString() ) ) {
392 $this->log()->add_var_expr( $const, 'prefixes_code', true );
393 }
394 }
395 }
396 }
397 }
398 }
399
400 /**
401 * Logs abstraction declarations including namespaces, classes, functions (excluding anonymous and nested ones),
402 * interfaces, and traits found in the provided statements. If a namespace is defined, only namespace declarations
403 * are logged and the process stops early. Otherwise, all abstraction types are processed and logged accordingly.
404 *
405 * @since 1.7.0
406 *
407 * @return void
408 */
409 private function add_abstractions_declarations_to_log() {
410 // Find namespace (if namespace is defined there is no need to continue).
411 $namespaces = $this->node_finder->findInstanceOf( $this->stmts, Node\Stmt\Namespace_::class );
412 if ( ! empty( $namespaces ) ) {
413 foreach ( $namespaces as $namespace ) {
414 $this->log()->add_namespace( $namespace, 'prefixes_code', true );
415 }
416 } else {
417 // Find all class nodes.
418 $classes = $this->node_finder->findInstanceOf( $this->stmts, Node\Stmt\Class_::class );
419 $this->log()->add_abstraction_declarations( $classes, 'prefixes_code', true );
420
421 // Find all functions (ignores anonymous, and inside abstractions).
422 $functions = $this->node_finder->findInstanceOf( $this->stmts, Node\Stmt\Function_::class );
423 $this->log()->add_abstraction_declarations( $functions, 'prefixes_code', true );
424
425 // Find all interfaces.
426 $interfaces = $this->node_finder->findInstanceOf( $this->stmts, Node\Stmt\Interface_::class );
427 $this->log()->add_abstraction_declarations( $interfaces, 'prefixes_code', true );
428
429 // Find all traits.
430 $traits = $this->node_finder->findInstanceOf( $this->stmts, Node\Stmt\Trait_::class );
431 $this->log()->add_abstraction_declarations( $traits, 'prefixes_code', true );
432 }
433 }
434
435 /**
436 * Adds a call expression or variable expression to the log based on the given parameters.
437 * Handles various cases depending on the type of the call and the potential string element provided.
438 * Also manages additional context logging if the expression is found on a different line.
439 *
440 * @since 1.7.0
441 *
442 * @param object $call The call expression node to be logged.
443 * @param mixed $expr The expression associated with the call.
444 * @param array $filter An optional array of strings to exclude from logging.
445 * @param int $arg_position The argument position to be used in logging, default is 0.
446 * @return void
447 */
448 private function add_call_expression_to_log( $call, $expr, $filter = array(), $arg_position = 0 ) {
449 if ( ! isset( $expr ) ) {
450 return;
451 }
452 $possible_string_accurate = false;
453 $found_in_same_line = true;
454 $possible_string = $this->get_possible_string_for_element( $expr, $found_in_same_line, $possible_string_accurate );
455
456 if ( empty( $possible_string ) ) {
457 $this->log()->add_call_expr( $call, 0, 'prefixes_code', true );
458 return;
459 }
460
461 if ( ! in_array( $possible_string, $filter, true ) ) { // Ignore some strings.
462 $call_class = get_class( $call );
463 if ( in_array( $call_class, array( 'PhpParser\Node\Expr\FuncCall', 'PhpParser\Node\Expr\MethodCall' ), true ) ) {
464 $this->log()->add_call_expr( $call, $arg_position, 'prefixes_code', true, $possible_string_accurate );
465 } elseif ( in_array( $call_class, array( 'PhpParser\Node\Expr\ArrayDimFetch' ), true ) ) {
466 $this->log()->add_var_expr( $call, 'prefixes_code', true, $possible_string_accurate );
467 }
468 }
469 }
470
471 /**
472 * Analyzes a log of items to check naming prefixes, identifying and normalizing
473 * common prefixes, and logging any potential issues or observations regarding
474 * prefix usage.
475 *
476 * @since 1.7.0
477 *
478 * @param string $logid The identifier of the log to analyze.
479 * @return void
480 */
481 private function check_prefixes( string $logid ) {
482 $log = $this->log()->get( $logid );
483
484 // Removes exceptions and logs items whose prefix cannot be read.
485 $log = $this->check_prefixes_process_exceptions_from_log( $log );
486
487 // Lowercase all the names.
488 $log = $this->lowercase_item_names( $log );
489
490 $this->possible_prefixes = array();
491 $this->already_processed_items = array();
492 $this->namespaces_count = 0;
493
494 // Check anything having a name. Extract possible prefixes.
495 foreach ( $log as $key => $log_item ) {
496 $item_identifier = $log_item['name'] . ':' . $log_item['type'];
497 if ( isset( $log_item['name'] ) && ! in_array( $item_identifier, $this->already_processed_items, true ) ) {
498 $this->already_processed_items[] = $item_identifier;
499
500 $should_check_other_prefixes = $this->check_prefixes_add_special_prefixes( $log_item, $key );
501
502 if ( $should_check_other_prefixes ) {
503 $this->check_prefixes_add_possible_prefixes( $log_item, $key, '_' );
504 $this->check_prefixes_add_possible_prefixes( $log_item, $key, '-' );
505 $this->check_prefixes_add_possible_prefixes( $log_item, $key, '\\' );
506 }
507 }
508 }
509
510 $this->already_processed_items = array_unique( $this->already_processed_items );
511
512 // Repetitions of each possible prefix.
513 if ( ! empty( $this->possible_prefixes ) ) {
514 $logsize = count( $log ) - ( count( $log ) - count( $this->already_processed_items ) ) - $this->namespaces_count;
515
516 $return = $this->normalize_prefixes_array( $this->possible_prefixes, $log );
517 $common_prefixes = $return['commonprefixes'];
518 $minimum_quantity_for_considering_prefix = $this->calculate_minimum_prefix_quantity( $logsize );
519
520 // Checking for prefixed used commonly.
521 if ( ! empty( $common_prefixes ) ) {
522 // In some situations they use similar prefixes like "PRTPR" and "PRT_PR" we process those cases together as they were a similar prefix regarding the prefixes_analysis_prefixed log.
523 $similar_prefixes = $this->get_too_similar_prefixes_array( $common_prefixes );
524 $prefixes_already_processed = array();
525 foreach ( $common_prefixes as $prefix => $incidences ) {
526 $incidences_count = count( $incidences );
527 if ( ! empty( $similar_prefixes[ $prefix ] ) ) {
528 $incidences_count = $similar_prefixes[ $prefix ]['numberincidences'];
529 }
530 if ( $incidences_count >= $minimum_quantity_for_considering_prefix || reset( $incidences ) === 'namespace' ) {
531 if ( ! in_array( $prefix, $prefixes_already_processed, true ) ) {
532 if ( ! empty( $similar_prefixes[ $prefix ] ) ) {
533 $this->final_prefixes = array_merge( $this->final_prefixes, $similar_prefixes[ $prefix ]['prefixes'] );
534 } else {
535 $this->final_prefixes[] = $prefix;
536 }
537 }
538 }
539 if ( ! empty( $similar_prefixes[ $prefix ] ) ) {
540 $prefixes_already_processed = array_merge( $prefixes_already_processed, $similar_prefixes[ $prefix ]['prefixes'] );
541 } else {
542 $prefixes_already_processed[] = $prefix;
543 }
544 }
545 }
546
547 // Discard prefixes that are less than 4 characters long.
548 $this->final_prefixes = array_filter(
549 $this->final_prefixes,
550 function ( $item ) {
551 return strlen( $item ) >= 4;
552 }
553 );
554
555 // Discard not valid prefixes.
556 $this->final_prefixes = array_diff( $this->final_prefixes, $this->not_valid_prefixes );
557 }
558 }
559
560 /**
561 * Checks if the given log item's name has a specific special prefix and updates the possible prefixes list accordingly.
562 *
563 * @since 1.7.0
564 *
565 * @param array $log_item The log item array containing information such as 'name' and 'type'.
566 * @param mixed $key The key used to associate the log item's type with the special prefix in the internal structure.
567 * @return bool Returns false if a matching special prefix is found and processed, otherwise true.
568 */
569 private function check_prefixes_add_special_prefixes( array $log_item, $key ) {
570 $special_prefixes = array( '__', '_', '-' );
571
572 foreach ( $special_prefixes as $special_prefix ) {
573 if ( str_starts_with( $log_item['name'], $special_prefix ) &&
574 ( 'abstraction' === $log_item['type'] || $special_prefix === $log_item['name'] )
575 ) {
576 $this->possible_prefixes[ $special_prefix ][ $key ] = $log_item['type'];
577 return false;
578 }
579 }
580
581 return true;
582 }
583
584 /**
585 * Processes and filters a log array to detect and handle items without names,
586 * and exceptions based on specific conditions. Alters the $log array by
587 * removing items that meet exclusion criteria while maintaining a record
588 * of unidentified prefix items.
589 *
590 * @since 1.7.0
591 *
592 * @param array $log The log array to process, containing items with keys such as 'name', 'type', and 'text'.
593 * @return array The filtered log array after processing and exclusion of specific items.
594 */
595 private function check_prefixes_process_exceptions_from_log( array $log ) {
596 $log_without_name = array();
597 foreach ( $log as $key => $item ) {
598 // Log anything that doesn't has name.
599 // No name or is a function defined by a variable.
600 if ( ! isset( $item['name'] ) || 'function_call-PhpParser\Node\Expr\Variable' === $item['type'] || false === $item['name'] ) {
601 $log_without_name[ $key ] = $item;
602 unset( $log[ $key ] );
603 }
604
605 // Exceptions.
606 if ( ! empty( $item['name'] ) && 'abstraction' === $item['type'] && str_starts_with( $item['text'], 'class ' ) ) {
607 if ( preg_match( '/^wc_gateway_/i', $item['name'] ) ) {
608 unset( $log[ $key ] ); // Ignoring classes that begins with WC_Gateway as that's recommended in the WC documentation.
609 }
610 }
611 }
612
613 if ( ! empty( $log_without_name ) ) {
614 $this->log()->add( '0', '# ██ ⚠ There are some elements we are not able to detect their prefixes.', 'prefixes_analysis_noname' );
615 $this->log()->merge( $log_without_name, 'prefixes_analysis_noname' );
616 }
617
618 return $log;
619 }
620
621 /**
622 * Converts the 'name' value of each item in the provided array to lowercase if it exists.
623 *
624 * @since 1.7.0
625 *
626 * @param array $log An array of items, where each item may contain a 'name' key.
627 * @return array The modified array with the 'name' values converted to lowercase.
628 */
629 private function lowercase_item_names( array $log ) {
630 foreach ( $log as $key => $item ) {
631 if ( isset( $item['name'] ) ) {
632 $log[ $key ]['name'] = strtolower( $item['name'] );
633 }
634 }
635 return $log;
636 }
637
638 /**
639 * Determines the minimum prefix quantity based on the provided log count,
640 * using varying thresholds and calculations depending on the value of the input.
641 *
642 * @since 1.7.0
643 *
644 * @param int $log_count The total number of logs for which the minimum prefix quantity is to be calculated.
645 * @return int The calculated minimum prefix quantity.
646 */
647 private function calculate_minimum_prefix_quantity( int $log_count ) {
648 if ( $log_count <= 2 ) {
649 return 1;
650 } elseif ( $log_count <= 4 ) {
651 return 2;
652 } elseif ( $log_count <= 8 ) {
653 return 3;
654 } elseif ( $log_count < 20 ) {
655 return intval( $log_count / 4 );
656 }
657 return (int) ( log( $log_count ) * 2 );
658 }
659
660 /**
661 * Processes the input item to identify potential prefixes and add them to the list of possible prefixes.
662 * Handles namespace counting, formats complete names, splits prefixes using a specified separator,
663 * and checks for multi-part prefixes.
664 *
665 * @since 1.7.0
666 *
667 * @param array $item Array containing details, including 'name' and 'type', used to process and identify prefixes.
668 * @param string $key Key associated with the prefix being processed.
669 * @param string $separator Optional separator used to split the item name for extracting prefixes. Default is '_'.
670 * @return void
671 */
672 private function check_prefixes_add_possible_prefixes( $item, $key, $separator = '_' ) {
673 $complete_name_formatted = str_replace( array( '-', '\\' ), '_', $item['name'] );
674
675 if ( 'namespace' === $item['type'] ) {
676 ++$this->namespaces_count;
677 }
678
679 $this->possible_prefixes[ $complete_name_formatted ][ $key ] = $item['type'];
680
681 $extract_prefixes = explode( $separator, $complete_name_formatted );
682
683 // Remove special characters at the beginning.
684 if ( empty( $extract_prefixes[0] ) ) {
685 array_shift( $extract_prefixes );
686 }
687
688 $prefix_parts_count = count( $extract_prefixes );
689
690 if ( $prefix_parts_count > 1 ) {
691 $this->check_prefixes_add_possible_prefix( $item, $key, $extract_prefixes[0] );
692 if ( $prefix_parts_count > 2 ) {
693 $this->check_prefixes_add_possible_prefix( $item, $key, $extract_prefixes[0] . '_' . $extract_prefixes[1] );
694 }
695 }
696 }
697
698 /**
699 * Updates the possible prefixes by adding the type of the provided item
700 * under the specified prefix and key.
701 *
702 * @since 1.7.0
703 *
704 * @param array $item The item containing data, including its type, to be added.
705 * @param string $key The key under which the item type is stored for the prefix.
706 * @param string $prefix The prefix to which the item type is associated.
707 * @return void
708 */
709 private function check_prefixes_add_possible_prefix( $item, $key, $prefix ) {
710 $this->possible_prefixes[ $prefix ][ $key ] = $item['type'];
711 }
712
713 /**
714 * Normalizes the given array of common prefixes by performing several tasks including:
715 * filtering invalid or warning prefixes, removing redundant incidences,
716 * ordering prefixes by length, and removing prefixes that are contained within others.
717 *
718 * @since 1.7.0
719 *
720 * @param array $common_prefixes An associative array where keys are prefixes and values are arrays of incidences.
721 * @param array $log A log array that tracks incidences and messages related to prefixes.
722 * @return array An associative array containing the updated 'commonprefixes' and 'log' after processing.
723 */
724 private function normalize_prefixes_array( $common_prefixes, $log ) {
725 if ( ! empty( $common_prefixes ) ) {
726
727 // Remove incidences for just one element, unless it's a namespace, or there is just really one element.
728 if ( sizeof( $log ) > 1 ) {
729 foreach ( $common_prefixes as $prefix => $array ) {
730 if ( sizeof( $array ) <= 1 ) {
731 if ( reset( $array ) !== 'namespace' ) {
732 unset( $common_prefixes[ $prefix ] );
733 }
734 }
735 }
736 }
737
738 // Order by bigger prefixes.
739 $keys = array_map( 'strlen', array_keys( $common_prefixes ) );
740 array_multisort( $keys, SORT_DESC, $common_prefixes );
741
742 // Remove prefixes that are inside other prefixes.
743 foreach ( $common_prefixes as $prefix => $incidences ) {
744 foreach ( $common_prefixes as $prefix_2 => $incidences_2 ) {
745 if ( $prefix !== $prefix_2 ) {
746 if ( str_starts_with( $prefix, $prefix_2 ) ) { // If $prefix starts with $prefix2.
747 // Remove short prefixes incidences that are within longer prefixes.
748 $within = ! array_diff_assoc( $incidences_2, $incidences );
749 if ( $within ) {
750 unset( $common_prefixes[ $prefix_2 ] );
751 } else {
752 // Remove longer prefixes incidences that are within shorter prefixes.
753 $within = ! array_diff_assoc( $incidences, $incidences_2 );
754 if ( $within ) {
755 unset( $common_prefixes[ $prefix ] );
756 }
757 }
758 }
759 }
760 }
761 }
762 }
763
764 return array(
765 'commonprefixes' => $common_prefixes,
766 'log' => $log,
767 );
768 }
769
770 /**
771 * Identifies and groups prefixes from the provided list that are too similar
772 * based on normalization criteria. It returns an associative array containing
773 * the group of similar prefixes along with their total incidences.
774 *
775 * @since 1.7.0
776 *
777 * @param array $common_prefixes An associative array where keys are prefixes and values are arrays of incidences for each prefix.
778 * @return array An associative array where each key is a prefix and the corresponding value is an array containing similar prefixes and the total number of incidences.
779 */
780 private function get_too_similar_prefixes_array( $common_prefixes ) {
781 $similar_prefixes = array();
782 if ( ! empty( $common_prefixes ) ) {
783 foreach ( $common_prefixes as $prefix => $incidences ) {
784 foreach ( $common_prefixes as $prefix_search => $incidences_search ) {
785 if ( $prefix !== $prefix_search && $this->get_prefix_name_normalized( $prefix ) === $this->get_prefix_name_normalized( $prefix_search ) ) {
786 if ( empty( $similar_prefixes[ $prefix ] ) ) {
787 $similar_prefixes[ $prefix ] = array(
788 'prefixes' => array( $prefix ),
789 'numberincidences' => count( $incidences ),
790 );
791 }
792 $similar_prefixes[ $prefix ]['prefixes'][] = $prefix_search;
793 $similar_prefixes[ $prefix ]['numberincidences'] += count( $incidences_search );
794 }
795 }
796 }
797 }
798 return $similar_prefixes;
799 }
800
801 /**
802 * Normalizes a given prefix by removing underscores and hyphens.
803 *
804 * @since 1.7.0
805 *
806 * @param string $prefix The original prefix to be normalized.
807 * @return string The normalized prefix.
808 */
809 private function get_prefix_name_normalized( $prefix ) {
810 $normalized = preg_replace( '/[\-_]/', '', $prefix );
811
812 return empty( $normalized ) ? $prefix : $normalized;
813 }
814 }
815