Address.php 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289
  1. <?php
  2. /**
  3. * 重庆赤晓店信息科技有限公司
  4. * https://www.chixiaodian.com
  5. * Copyright (c) 2023 赤店商城 All rights reserved.
  6. */
  7. namespace app\utils\address;
  8. class Address
  9. {
  10. /*
  11. ** 智能解析
  12. */
  13. public static function smart($string, $user = true)
  14. {
  15. if ($user) {
  16. $decompose = self::decompose($string);
  17. $re = $decompose;
  18. } else {
  19. $re['addr'] = $string;
  20. }
  21. // \Yii::error([__METHOD__, $fuzz, $fuzz['a3']]);
  22. // 定义正则表达式模式
  23. $pattern = '/(.*?(省|自治区|特别行政区|市))(.*?(自治州|市|盟|地区))?(.*?(市|区|县|自治县|旗|自治旗|特区))(.*)/';
  24. // 使用 preg_match() 函数匹配地址字符串
  25. if (preg_match($pattern, $string, $matches)) {
  26. // 匹配成功,提取省、市、区或县、详细地址
  27. $province = $matches[1];
  28. $city = $matches[3];
  29. $region = $matches[5];
  30. $addr = str_replace($province.$city.$region, '', $string);
  31. $fuzz = array(
  32. 'a1' => $province,
  33. 'a2' => $city,
  34. 'a3' => $region,
  35. 'street' => $addr,
  36. );
  37. } else {
  38. $fuzz = self::fuzz($re['addr']);
  39. //// \Yii::error([__METHOD__, $fuzz, $re]);
  40. $fuzz['a3'] = mb_substr($fuzz['a3'], (int)mb_strpos($fuzz['a3'], '市') + 1);
  41. }
  42. // $parse = self::parse($fuzz['a1'], $fuzz['a2'], $fuzz['a3']);
  43. $re['province'] = $fuzz['a1'];
  44. $re['city'] = $fuzz['a2'];
  45. $re['region'] = $fuzz['a3'];
  46. $re['street'] = ($fuzz['street']) ? $fuzz['street'] : '';
  47. $re['street'] = str_replace([$re['region'], $re['city'], $re['province']], ['', '', ''], $re['street']);
  48. return $re;
  49. }
  50. /*
  51. ** 分离手机号(座机),身份证号,姓名等用户信息
  52. */
  53. public static function decompose($string)
  54. {
  55. $compose = array();
  56. $search = array('收货地址', '详细地址', '地址', '收货人', '收件人', '收货', '所在地区', '邮编', '电话', '手机号码','身份证号码', '身份证号', '身份证', ':', ':', ';', ';', ',', ',', '。');
  57. $replace = array(' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ', ' ');
  58. $string = str_replace($search, $replace, $string);
  59. $string = preg_replace('/\s{1,}/', ' ', $string);
  60. $string = preg_replace('/0-|0?(\d{3})-(\d{4})-(\d{4})/', '$1$2$3', $string);
  61. preg_match('/\d{18}|\d{17}X/i', $string, $match);
  62. if ($match && $match[0]) {
  63. $compose['idn'] = strtoupper($match[0]);
  64. $string = str_replace($match[0], '', $string);
  65. }
  66. preg_match('/\d{7,11}|\d{3,4}-\d{6,8}/', $string, $match);
  67. if ($match && $match[0]) {
  68. $compose['mobile'] = $match[0];
  69. $string = str_replace($match[0], '', $string);
  70. }
  71. preg_match('/\d{6}/', $string, $match);
  72. if ($match && $match[0]) {
  73. $compose['postcode'] = $match[0];
  74. $string = str_replace($match[0], '', $string);
  75. }
  76. $string = trim(preg_replace('/ {2,}/', ' ', $string));
  77. $split_arr = explode(' ', $string);
  78. if (count($split_arr) > 1) {
  79. $compose['name'] = $split_arr[0];
  80. foreach ($split_arr as $value) {
  81. if (strlen($value) < strlen($compose['name'])) {
  82. $compose['name'] = $value;
  83. }
  84. }
  85. $string = trim(str_replace($compose['name'], '', $string));
  86. }
  87. $compose['addr'] = $string;
  88. return $compose;
  89. }
  90. /*
  91. ** 根据统计规律分析出二三级地址
  92. */
  93. public static function fuzz($addr)
  94. {
  95. $addr_origin = $addr;
  96. $addr = str_replace([' ', ','], ['', ''], $addr);
  97. $addr = str_replace('自治区', '省', $addr);
  98. $addr = str_replace('自治州', '州', $addr);
  99. $addr = str_replace('小区', '', $addr);
  100. $addr = str_replace('校区', '', $addr);
  101. $a1 = '';
  102. $a2 = '';
  103. $a3 = '';
  104. $street = '';
  105. if (mb_strpos($addr, '县') !== false && mb_strpos($addr, '县') <= floor((mb_strlen($addr) / 3) * 2) || (mb_strpos($addr, '区') !== false && mb_strpos($addr, '区') <= floor((mb_strlen($addr) / 3) * 2)) || mb_strpos($addr, '旗') !== false && mb_strpos($addr, '旗') <= floor((mb_strlen($addr) / 3) * 2)) {
  106. if (mb_strstr($addr, '旗')) {
  107. $deep3_keyword_pos = mb_strpos($addr, '旗');
  108. $a3 = mb_substr($addr, $deep3_keyword_pos - 1, 2);
  109. }
  110. if (mb_strstr($addr, '区')) {
  111. $deep3_keyword_pos = mb_strpos($addr, '区');
  112. if (mb_strstr($addr, '市')) {
  113. $city_pos = mb_strpos($addr, '市');
  114. $zone_pos = mb_strpos($addr, '区');
  115. $a3 = mb_substr($addr, $city_pos + 1, $zone_pos - $city_pos);
  116. } else {
  117. $a3 = mb_substr($addr, $deep3_keyword_pos - 2, 3);
  118. }
  119. }
  120. if (mb_strstr($addr, '县')) {
  121. $deep3_keyword_pos = mb_strpos($addr, '县');
  122. if (mb_strstr($addr, '市')) {
  123. $city_pos = mb_strpos($addr, '市');
  124. $zone_pos = mb_strpos($addr, '县');
  125. $a3 = mb_substr($addr, $city_pos + 1, $zone_pos - $city_pos);
  126. } else {
  127. if (mb_strstr($addr, '自治县')) {
  128. $a3 = mb_substr($addr, $deep3_keyword_pos - 6, 7);
  129. if (in_array(mb_substr($a3, 0, 1), ['省', '市', '州'])) {
  130. $a3 = mb_substr($a3, 1);
  131. }
  132. } else {
  133. $a3 = mb_substr($addr, $deep3_keyword_pos - 2, 3);
  134. }
  135. }
  136. }
  137. $street = mb_substr($addr_origin, $deep3_keyword_pos + 1);
  138. } else {
  139. if (mb_strripos($addr, '市')) {
  140. if (mb_substr_count($addr, '市') == 1) {
  141. $deep3_keyword_pos = mb_strripos($addr, '市');
  142. $a3 = mb_substr($addr, $deep3_keyword_pos - 2, 3);
  143. $street = mb_substr($addr_origin, $deep3_keyword_pos + 1);
  144. } else if (mb_substr_count($addr, '市') >= 2) {
  145. $deep3_keyword_pos = mb_strripos($addr, '市');
  146. $a3 = mb_substr($addr, $deep3_keyword_pos - 2, 3);
  147. $street = mb_substr($addr_origin, $deep3_keyword_pos + 1);
  148. }
  149. } else {
  150. $a3 = '';
  151. $street = $addr;
  152. }
  153. }
  154. if (mb_strpos($addr, '市') || mb_strstr($addr, '盟') || mb_strstr($addr, '州')) {
  155. if ($tmp_pos = mb_strpos($addr, '市')) {
  156. $a2 = mb_substr($addr, $tmp_pos - 2, 3);
  157. } else if ($tmp_pos = mb_strpos($addr, '盟')) {
  158. $a2 = mb_substr($addr, $tmp_pos - 2, 3);
  159. } else if ($tmp_pos = mb_strpos($addr, '州')) {
  160. if ($tmp_pos = mb_strpos($addr, '自治州')) {
  161. $a2 = mb_substr($addr, $tmp_pos - 4, 5);
  162. } else {
  163. $a2 = mb_substr($addr, $tmp_pos - 2, 3);
  164. }
  165. }
  166. } else {
  167. $a2 = '';
  168. }
  169. $a2;
  170. if ($a2 == $a3) {
  171. $a3 = trim($a3, '市');
  172. }
  173. $r = array(
  174. 'a1' => $a1,
  175. 'a2' => $a2,
  176. 'a3' => $a3,
  177. 'street' => $street,
  178. );
  179. return $r;
  180. }
  181. /*
  182. ** 智能解析出省市区+街道地址
  183. */
  184. public static function parse($a1, $a2, $a3)
  185. {
  186. require 'data/a3.php';
  187. require 'data/a2.php';
  188. require 'data/a1.php';
  189. $r = array();
  190. if ($a3 != '') {
  191. $area3_matches = array();
  192. foreach ($a3_data as $id => $v) {
  193. if (mb_strpos($v['name'], $a3) !== false) {
  194. $area3_matches[$id] = $v;
  195. }
  196. }
  197. if ($area3_matches && count($area3_matches) > 1) {
  198. if ($a2) {
  199. foreach ($a2_data as $id => $v) {
  200. if (mb_strpos($v['name'], $a2) !== false) {
  201. $area2_matches[$id] = $v;
  202. }
  203. }
  204. if ($area2_matches) {
  205. foreach ($area3_matches as $id => $v) {
  206. if (isset($area2_matches[$v['pid']])) {
  207. $r['city'] = $area2_matches[$v['pid']]['name'];
  208. $r['region'] = $v['name'];
  209. $sheng_id = $area2_matches[$v['pid']]['pid'];
  210. $r['province'] = $a1_data[$sheng_id]['name'];
  211. }
  212. }
  213. }
  214. } else {
  215. $r['province'] = '';
  216. $r['city'] = '';
  217. $r['region'] = $a3;
  218. }
  219. } else if ($area3_matches && count($area3_matches) == 1) {
  220. foreach ($area3_matches as $id => $v) {
  221. $city_id = $v['pid'];
  222. $r['region'] = $v['name'];
  223. }
  224. $city = $a2_data[$city_id];
  225. $province = $a1_data[$city['pid']];
  226. $r['province'] = $province['name'];
  227. $r['city'] = $city['name'];
  228. } else if (empty($area3_matches) && $a2 == $a3) {
  229. foreach ($a2_data as $id => $v) {
  230. if (mb_strpos($v['name'], $a2) !== false) {
  231. $area2_matches[$id] = $v;
  232. $sheng_id = $v['pid'];
  233. $r['city'] = $v['name'];
  234. }
  235. }
  236. $r['province'] = $a1_data[$sheng_id]['name'];
  237. $r['region'] = '';
  238. }
  239. }
  240. return $r;
  241. }
  242. }