| 1 |
<?php |
| 2 |
/** |
| 3 |
* robots.txt helper — keeps well-behaved crawlers away from filtering result pages. |
| 4 |
* |
| 5 |
* Why this exists: filter combinations produce a practically unlimited number of |
| 6 |
* URLs, and bots (search engines, AI crawlers, scrapers) that once discover them |
| 7 |
* keep requesting them for months — even after the filter links are hidden or the |
| 8 |
* plugin is removed. Hiding the links (see «Disable filter links for crawlers») |
| 9 |
* stops new discovery; the only signal that also stops compliant crawlers from |
| 10 |
* re-fetching URLs they already know is a robots.txt Disallow. That is Google's own |
| 11 |
* recommendation for faceted navigation: disallow the URL patterns rather than rely |
| 12 |
* on noindex (a noindex page still costs a full crawl request). |
| 13 |
* |
| 14 |
* Two URL schemes, two rule sets: |
| 15 |
* |
| 16 |
* QUERY MODE (free build, or PRO with pretty permalinks off) — every filter URL is |
| 17 |
* fully described by its parameter names (?color=red&min_price=10), so robots.txt |
| 18 |
* can express "any URL carrying a filter parameter" precisely: |
| 19 |
* Disallow: /*?color= |
| 20 |
* Disallow: /*&color= |
| 21 |
* (two patterns per parameter: first in the query string or after another one — |
| 22 |
* deliberately NOT the looser /*?*color= form, which would also match unrelated |
| 23 |
* parameters that merely end in the same word, e.g. ?product_color=). |
| 24 |
* |
| 25 |
* PRETTY MODE (PRO) — filters travel as path segments (/shop/color-red/size-xl/) |
| 26 |
* and SEO Rules may make some of those pages indexable, so the rules must only |
| 27 |
* name the classes of URL the plugin never indexes (SeoFrontend::isNoindex()): |
| 28 |
* - numeric/date ranges and the in-filter search term (still query params); |
| 29 |
* - several values of one filter (…/color-red-or-blue/); |
| 30 |
* - more filters in one URL than the highest Indexing Depth allows; |
| 31 |
* - and, when no post type has an Indexing Depth at all (nothing is |
| 32 |
* indexable), every filter segment outright. |
| 33 |
* Single-filter pages stay crawlable otherwise (commented suggestions show how |
| 34 |
* to block them too). Path patterns match a segment that STARTS with a filter |
| 35 |
* prefix, so a product/category slug beginning with the same word would match |
| 36 |
* as well — the Settings box shows the rules for review before anything is |
| 37 |
* added. |
| 38 |
* |
| 39 |
* The generated block is |
| 40 |
* - appended to WordPress' virtual /robots.txt while the option is on (the filter |
| 41 |
* has no effect when a physical robots.txt file exists — WordPress never serves |
| 42 |
* the virtual one then); |
| 43 |
* - shown on Settings → General so it can be copied into a physical file. |
| 44 |
*/ |
| 45 |
|
| 46 |
if ( ! defined('ABSPATH') ) { |
| 47 |
exit; |
| 48 |
} |
| 49 |
|
| 50 |
if ( ! function_exists( 'flrt_robots_helper_available' ) ) { |
| 51 |
/** |
| 52 |
* Whether the robots.txt helper (option, rules box, robots_txt hook) is active |
| 53 |
* in this build. Both builds since 1.9.6; kept as a single switch/filter. |
| 54 |
* |
| 55 |
* @return bool |
| 56 |
*/ |
| 57 |
function flrt_robots_helper_available() |
| 58 |
{ |
| 59 |
return (bool) apply_filters( 'wpc_robots_helper_available', true ); |
| 60 |
} |
| 61 |
} |
| 62 |
|
| 63 |
if ( ! function_exists( 'flrt_robots_pretty_mode' ) ) { |
| 64 |
/** |
| 65 |
* True when filters travel as path segments (PRO with pretty permalinks). |
| 66 |
* |
| 67 |
* @return bool |
| 68 |
*/ |
| 69 |
function flrt_robots_pretty_mode() |
| 70 |
{ |
| 71 |
return defined( 'FLRT_PERMALINKS_ENABLED' ) && FLRT_PERMALINKS_ENABLED; |
| 72 |
} |
| 73 |
} |
| 74 |
|
| 75 |
if ( ! function_exists( 'flrt_robots_registry' ) ) { |
| 76 |
/** |
| 77 |
* The URL-prefix registry (option wpc_filter_permalinks, 'entity#ename' => slug) |
| 78 |
* in URL order, as [ ['entity' => …, 'slug' => …], … ] with empty slugs dropped. |
| 79 |
* |
| 80 |
* @return array[] |
| 81 |
*/ |
| 82 |
function flrt_robots_registry() |
| 83 |
{ |
| 84 |
$registry = get_option( 'wpc_filter_permalinks', [] ); |
| 85 |
|
| 86 |
if ( ! is_array( $registry ) ) { |
| 87 |
$registry = maybe_unserialize( $registry ); |
| 88 |
} |
| 89 |
if ( ! is_array( $registry ) ) { |
| 90 |
$registry = []; |
| 91 |
} |
| 92 |
|
| 93 |
$entries = []; |
| 94 |
|
| 95 |
foreach ( $registry as $entityAndEname => $slug ) { |
| 96 |
// Same normalisation as UrlManager::getParamName() |
| 97 |
$slug = sanitize_title( (string) $slug ); |
| 98 |
if ( $slug === '' ) { |
| 99 |
continue; |
| 100 |
} |
| 101 |
$parts = explode( '#', (string) $entityAndEname, 2 ); |
| 102 |
$entries[] = [ 'entity' => $parts[0], 'slug' => $slug ]; |
| 103 |
} |
| 104 |
|
| 105 |
return $entries; |
| 106 |
} |
| 107 |
} |
| 108 |
|
| 109 |
if ( ! function_exists( 'flrt_robots_filter_params' ) ) { |
| 110 |
/** |
| 111 |
* Query-string parameter names the plugin's filter URLs can carry. |
| 112 |
* |
| 113 |
* Built the same way UrlManager/RequestParser build and read them: numeric |
| 114 |
* entities travel as min_{slug}/max_{slug}, date entities as {slug}_from/{slug}_to |
| 115 |
* and — in query mode only — everything else as the bare slug. The in-filter |
| 116 |
* search term («srch») always travels with filter URLs too. |
| 117 |
* |
| 118 |
* @param bool|null $pretty_mode null = detect; true = only the parameters that |
| 119 |
* stay in the query string with pretty permalinks |
| 120 |
* @return string[] sorted, unique parameter names |
| 121 |
*/ |
| 122 |
function flrt_robots_filter_params( $pretty_mode = null ) |
| 123 |
{ |
| 124 |
if ( $pretty_mode === null ) { |
| 125 |
$pretty_mode = flrt_robots_pretty_mode(); |
| 126 |
} |
| 127 |
|
| 128 |
$params = []; |
| 129 |
|
| 130 |
foreach ( flrt_robots_registry() as $entry ) { |
| 131 |
$slug = $entry['slug']; |
| 132 |
|
| 133 |
switch ( $entry['entity'] ) { |
| 134 |
case 'post_meta_num': |
| 135 |
case 'tax_numeric': |
| 136 |
$params[] = 'min_' . $slug; |
| 137 |
$params[] = 'max_' . $slug; |
| 138 |
break; |
| 139 |
case 'post_date': |
| 140 |
case 'post_meta_date': |
| 141 |
$params[] = $slug . '_from'; |
| 142 |
$params[] = $slug . '_to'; |
| 143 |
break; |
| 144 |
default: |
| 145 |
if ( ! $pretty_mode ) { |
| 146 |
$params[] = $slug; |
| 147 |
} |
| 148 |
} |
| 149 |
} |
| 150 |
|
| 151 |
if ( ! empty( flrt_robots_registry() ) ) { |
| 152 |
$params[] = 'srch'; |
| 153 |
} |
| 154 |
|
| 155 |
$params = array_values( array_unique( $params ) ); |
| 156 |
sort( $params, SORT_STRING ); |
| 157 |
|
| 158 |
return apply_filters( 'wpc_robots_filter_params', $params ); |
| 159 |
} |
| 160 |
} |
| 161 |
|
| 162 |
if ( ! function_exists( 'flrt_robots_path_prefixes' ) ) { |
| 163 |
/** |
| 164 |
* Prefixes of the filters that travel as path segments, in URL order (the |
| 165 |
* registry order IS the canonical segment order — any other order 404s). |
| 166 |
* |
| 167 |
* @return string[] |
| 168 |
*/ |
| 169 |
function flrt_robots_path_prefixes() |
| 170 |
{ |
| 171 |
$prefixes = []; |
| 172 |
|
| 173 |
foreach ( flrt_robots_registry() as $entry ) { |
| 174 |
if ( in_array( $entry['entity'], [ 'post_meta_num', 'tax_numeric', 'post_date', 'post_meta_date' ], true ) ) { |
| 175 |
continue; |
| 176 |
} |
| 177 |
$prefixes[] = $entry['slug']; |
| 178 |
} |
| 179 |
|
| 180 |
return array_values( array_unique( $prefixes ) ); |
| 181 |
} |
| 182 |
} |
| 183 |
|
| 184 |
if ( ! function_exists( 'flrt_robots_indexing_depth' ) ) { |
| 185 |
/** |
| 186 |
* The highest Indexing Depth configured for any post type (PRO option |
| 187 |
* wpc_indexing_deep_settings), i.e. the most filters a single indexable URL |
| 188 |
* may carry anywhere on the site. 0 = no filter page is indexable. |
| 189 |
* |
| 190 |
* @return int |
| 191 |
*/ |
| 192 |
function flrt_robots_indexing_depth() |
| 193 |
{ |
| 194 |
$depth = 0; |
| 195 |
$options = get_option( 'wpc_indexing_deep_settings', [] ); |
| 196 |
|
| 197 |
if ( is_array( $options ) ) { |
| 198 |
foreach ( $options as $value ) { |
| 199 |
$depth = max( $depth, (int) $value ); |
| 200 |
} |
| 201 |
} |
| 202 |
|
| 203 |
return (int) apply_filters( 'wpc_robots_indexing_depth', $depth ); |
| 204 |
} |
| 205 |
} |
| 206 |
|
| 207 |
if ( ! function_exists( 'flrt_robots_combinations' ) ) { |
| 208 |
/** |
| 209 |
* All ordered k-combinations of $items (order preserved), or null when there |
| 210 |
* would be more than $limit of them. |
| 211 |
* |
| 212 |
* @return array[]|null |
| 213 |
*/ |
| 214 |
function flrt_robots_combinations( array $items, $k, $limit ) |
| 215 |
{ |
| 216 |
$n = count( $items ); |
| 217 |
if ( $k < 1 || $k > $n ) { |
| 218 |
return []; |
| 219 |
} |
| 220 |
|
| 221 |
// C(n, k) without overflow: stop counting once past the limit |
| 222 |
$count = 1; |
| 223 |
for ( $i = 1; $i <= $k; $i++ ) { |
| 224 |
$count = $count * ( $n - $k + $i ) / $i; |
| 225 |
if ( $count > $limit ) { |
| 226 |
return null; |
| 227 |
} |
| 228 |
} |
| 229 |
|
| 230 |
$result = []; |
| 231 |
$indexes = range( 0, $k - 1 ); |
| 232 |
|
| 233 |
while ( true ) { |
| 234 |
$combo = []; |
| 235 |
foreach ( $indexes as $i ) { |
| 236 |
$combo[] = $items[ $i ]; |
| 237 |
} |
| 238 |
$result[] = $combo; |
| 239 |
|
| 240 |
// advance |
| 241 |
$i = $k - 1; |
| 242 |
while ( $i >= 0 && $indexes[ $i ] === $n - $k + $i ) { |
| 243 |
$i--; |
| 244 |
} |
| 245 |
if ( $i < 0 ) { |
| 246 |
break; |
| 247 |
} |
| 248 |
$indexes[ $i ]++; |
| 249 |
for ( $j = $i + 1; $j < $k; $j++ ) { |
| 250 |
$indexes[ $j ] = $indexes[ $j - 1 ] + 1; |
| 251 |
} |
| 252 |
} |
| 253 |
|
| 254 |
return $result; |
| 255 |
} |
| 256 |
} |
| 257 |
|
| 258 |
if ( ! function_exists( 'flrt_robots_txt_rules' ) ) { |
| 259 |
/** |
| 260 |
* The robots.txt block for the current filters, or '' when there are no |
| 261 |
* filters yet. A repeated "User-agent: *" group is fine: crawlers merge groups |
| 262 |
* that address the same user agent. |
| 263 |
* |
| 264 |
* @return string |
| 265 |
*/ |
| 266 |
function flrt_robots_txt_rules() |
| 267 |
{ |
| 268 |
$pretty = flrt_robots_pretty_mode(); |
| 269 |
$params = flrt_robots_filter_params( $pretty ); |
| 270 |
$prefixes = $pretty ? flrt_robots_path_prefixes() : []; |
| 271 |
|
| 272 |
if ( empty( $params ) && empty( $prefixes ) ) { |
| 273 |
return ''; |
| 274 |
} |
| 275 |
|
| 276 |
$lines = []; |
| 277 |
$lines[] = '# Filter Everything: do not crawl filtering result pages'; |
| 278 |
$lines[] = 'User-agent: *'; |
| 279 |
|
| 280 |
// 1) Query-string parameters — in query mode that is every filter; with |
| 281 |
// pretty permalinks only the ranges/dates/search that never index. |
| 282 |
foreach ( $params as $param ) { |
| 283 |
$lines[] = 'Disallow: /*?' . $param . '='; |
| 284 |
$lines[] = 'Disallow: /*&' . $param . '='; |
| 285 |
} |
| 286 |
|
| 287 |
// 2) Path segments (pretty permalinks) |
| 288 |
if ( $pretty && ! empty( $prefixes ) ) { |
| 289 |
$depth = flrt_robots_indexing_depth(); |
| 290 |
|
| 291 |
if ( $depth < 1 ) { |
| 292 |
// Nothing on this site indexes filter pages: block every filter |
| 293 |
// segment. (A product or category slug that starts with one of |
| 294 |
// these prefixes would match the same pattern — review the list.) |
| 295 |
$lines[] = '# No Indexing Depth is set, so no filtering result page is indexable: block every filter segment'; |
| 296 |
foreach ( $prefixes as $prefix ) { |
| 297 |
$lines[] = 'Disallow: /*/' . $prefix . '-'; |
| 298 |
} |
| 299 |
} else { |
| 300 |
// 2a) several values of one filter — never indexable |
| 301 |
$lines[] = '# Several values of one filter (…/color-red-or-blue/) are never indexable'; |
| 302 |
foreach ( $prefixes as $prefix ) { |
| 303 |
foreach ( [ '-or-', '-and-' ] as $separator ) { |
| 304 |
$lines[] = 'Disallow: /*/' . $prefix . '-*' . $separator; |
| 305 |
} |
| 306 |
} |
| 307 |
|
| 308 |
// 2b) more filters in one URL than the Indexing Depth allows |
| 309 |
$size = $depth + 1; |
| 310 |
$combos = flrt_robots_combinations( $prefixes, $size, 300 ); |
| 311 |
|
| 312 |
if ( $combos === null ) { |
| 313 |
$lines[] = sprintf( '# URLs with more than %d filters are never indexable, but there are too many prefix combinations to list them here', $depth ); |
| 314 |
} elseif ( ! empty( $combos ) ) { |
| 315 |
$lines[] = sprintf( '# URLs with more than %d filters (your highest Indexing Depth) are never indexable', $depth ); |
| 316 |
foreach ( $combos as $combo ) { |
| 317 |
$pattern = ''; |
| 318 |
foreach ( $combo as $prefix ) { |
| 319 |
$pattern .= '/*/' . $prefix . '-'; |
| 320 |
} |
| 321 |
// "/*/a-/*/b-" → "/*/a-*/b-": the wildcard between segments |
| 322 |
$lines[] = 'Disallow: ' . str_replace( '-/*/', '-*/', $pattern ); |
| 323 |
} |
| 324 |
} |
| 325 |
|
| 326 |
// 2c) single-filter pages stay crawlable — they are the ones SEO |
| 327 |
// Rules index; leave the opt-in as commented lines |
| 328 |
$lines[] = '# Single-filter pages stay crawlable because your SEO Rules may index them; to block them too, remove the # below'; |
| 329 |
foreach ( $prefixes as $prefix ) { |
| 330 |
$lines[] = '# Disallow: /*/' . $prefix . '-'; |
| 331 |
} |
| 332 |
} |
| 333 |
} |
| 334 |
|
| 335 |
$rules = implode( "\n", $lines ); |
| 336 |
|
| 337 |
return (string) apply_filters( 'wpc_robots_txt_rules', $rules, $params, $prefixes ); |
| 338 |
} |
| 339 |
} |
| 340 |
|
| 341 |
if ( ! function_exists( 'flrt_robots_txt_is_physical' ) ) { |
| 342 |
/** |
| 343 |
* True when a physical robots.txt file exists in the WordPress root. WordPress |
| 344 |
* serves the virtual robots.txt (and runs the robots_txt filter) only when it |
| 345 |
* does not. |
| 346 |
* |
| 347 |
* @return bool |
| 348 |
*/ |
| 349 |
function flrt_robots_txt_is_physical() |
| 350 |
{ |
| 351 |
return file_exists( ABSPATH . 'robots.txt' ); |
| 352 |
} |
| 353 |
} |
| 354 |
|
| 355 |
if ( ! function_exists( 'flrt_robots_txt_append_rules' ) ) { |
| 356 |
/** |
| 357 |
* robots_txt filter callback: appends the filter rules to WordPress' virtual |
| 358 |
* robots.txt while the option is on. |
| 359 |
* |
| 360 |
* @param string $output robots.txt content so far |
| 361 |
* @param bool $public blog_public option; when false WordPress already emits |
| 362 |
* "Disallow: /" for everyone and our rules are redundant |
| 363 |
* @return string |
| 364 |
*/ |
| 365 |
function flrt_robots_txt_append_rules( $output, $public ) |
| 366 |
{ |
| 367 |
if ( ! $public || ! flrt_robots_helper_available() ) { |
| 368 |
return $output; |
| 369 |
} |
| 370 |
|
| 371 |
if ( flrt_get_option( 'robots_txt_block_filters' ) !== 'on' ) { |
| 372 |
return $output; |
| 373 |
} |
| 374 |
|
| 375 |
$rules = flrt_robots_txt_rules(); |
| 376 |
if ( $rules === '' ) { |
| 377 |
return $output; |
| 378 |
} |
| 379 |
|
| 380 |
return rtrim( (string) $output ) . "\n\n" . $rules . "\n"; |
| 381 |
} |
| 382 |
} |
| 383 |
|
| 384 |
add_filter( 'robots_txt', 'flrt_robots_txt_append_rules', 20, 2 ); |
| 385 |
|
