UpgradePaliText.php 8.5 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255
  1. <?php
  2. /**
  3. * 计算章节的父子,前后关系
  4. * 输入: csv文件
  5. */
  6. namespace App\Console\Commands;
  7. use Illuminate\Console\Command;
  8. use App\Models\PaliText;
  9. use App\Models\BookTitle;
  10. use Illuminate\Support\Facades\DB;
  11. use Illuminate\Support\Facades\Log;
  12. class UpgradePaliText extends Command
  13. {
  14. /**
  15. * The name and signature of the console command.
  16. * php artisan upgrade:palitext 168
  17. * @var string
  18. */
  19. protected $signature = 'upgrade:palitext {from?} {to?}';
  20. /**
  21. * The console command description.
  22. *
  23. * @var string
  24. */
  25. protected $description = 'upgrade pali_texts paragraph infomation';
  26. /**
  27. * Create a new command instance.
  28. *
  29. * @return void
  30. */
  31. public function __construct()
  32. {
  33. parent::__construct();
  34. }
  35. /**
  36. * Execute the console command.
  37. *
  38. * @return int
  39. */
  40. public function handle()
  41. {
  42. if(\App\Tools\Tools::isStop()){
  43. return 0;
  44. }
  45. $this->info("upgrade pali text");
  46. $startTime = time();
  47. $_from = $this->argument('from');
  48. $_to = $this->argument('to');
  49. if(empty($_from) && empty($_to)){
  50. $_from = 1;
  51. $_to = 217;
  52. }else if(empty($_to)){
  53. $_to = $_from;
  54. }
  55. #载入文件列表
  56. $fileListFileName = config("mint.path.palitext_filelist");
  57. $filelist = array();
  58. if (($handle = fopen($fileListFileName, 'r')) !== false) {
  59. while (($filelist[] = fgetcsv($handle, 0, ',')) !== false) {
  60. }
  61. }
  62. $bar = $this->output->createProgressBar($_to-$_from+1);
  63. for ($from=$_from; $from <= $_to; $from++) {
  64. $inputRow = 0;
  65. $arrInserString = array();
  66. #载入csv数据
  67. $FileName = $filelist[$from-1][1];
  68. $csvFile = config("mint.path.pali_title") .'/'. $from.'_pali.csv';
  69. if (($fp = fopen($csvFile, "r")) !== false) {
  70. Log::info("csv load:" . $csvFile);
  71. while (($data = fgetcsv($fp, 0, ',')) !== false) {
  72. if ($inputRow > 0) {
  73. array_push($arrInserString, $data);
  74. }
  75. $inputRow++;
  76. }
  77. fclose($fp);
  78. } else {
  79. $this->error( "can not open csv file. filename=" . $csvFile. PHP_EOL) ;
  80. Log::error( "can not open csv file. filename=" . $csvFile) ;
  81. continue;
  82. }
  83. $title_data = PaliText::select(['book','paragraph','level','parent','toc','lenght'])
  84. ->where('book',$from)->orderby('paragraph','asc')->get();
  85. $paragraph_count = count($title_data);
  86. $paragraph_info = array();
  87. $paragraph_info[] = array($from, -1, $paragraph_count, -1, -1, -1);
  88. for ($iPar = 0; $iPar < count($title_data); $iPar++) {
  89. $title_data[$iPar]["level"] = $arrInserString[$iPar][3];
  90. }
  91. for ($iPar = 0; $iPar < count($title_data); $iPar++) {
  92. $book = $from ;
  93. $paragraph = $title_data[$iPar]["paragraph"];
  94. $true_level = (int) $title_data[$iPar]["level"];
  95. if ((int) $title_data[$iPar]["level"] == 8) {
  96. $title_data[$iPar]["level"] = 100;
  97. }
  98. $curr_level = (int) $title_data[$iPar]["level"];
  99. # 计算这个chapter的段落数量
  100. $length = -1;
  101. for ($iPar1 = $iPar + 1; $iPar1 < count($title_data); $iPar1++) {
  102. $thislevel = (int) $title_data[$iPar1]["level"];
  103. if ($thislevel <= $curr_level) {
  104. $length = (int) $title_data[$iPar1]["paragraph"] - $paragraph;
  105. break;
  106. }
  107. }
  108. if ($length == -1) {
  109. $length = $paragraph_count - $paragraph + 1;
  110. }
  111. /*
  112. 上一个段落
  113. 算法:查找上一个标题段落。而且该标题段落的下一个段落不是标题段落
  114. */
  115. $prev = -1;
  116. if ($iPar > 0) {
  117. for ($iPar1 = $iPar - 1; $iPar1 >= 0; $iPar1--) {
  118. if ($title_data[$iPar1]["level"] < 8 && $title_data[$iPar1+1]["level"]==100) {
  119. $prev = $title_data[$iPar1]["paragraph"];
  120. break;
  121. }
  122. }
  123. }
  124. /*
  125. 下一个段落
  126. 算法:查找下一个标题段落。而且该标题段落的下一个段落不是标题段落
  127. */
  128. $next = -1;
  129. if ($iPar < count($title_data) - 1) {
  130. for ($iPar1 = $iPar + 1; $iPar1 < count($title_data)-1; $iPar1++) {
  131. if ($title_data[$iPar1]["level"] <8 && $title_data[$iPar1+1]["level"]==100) {
  132. $next = $title_data[$iPar1]["paragraph"];
  133. break;
  134. }
  135. }
  136. }
  137. //查找parent
  138. $parent = -1;
  139. if ($iPar > 0) {
  140. for ($iPar1 = $iPar - 1; $iPar1 >= 0; $iPar1--) {
  141. if ($title_data[$iPar1]["level"] < $true_level) {
  142. $parent = $title_data[$iPar1]["paragraph"];
  143. break;
  144. }
  145. }
  146. }
  147. //计算章节包含总字符数
  148. $iChapter_strlen = 0;
  149. for ($i = $iPar; $i < $iPar + $length; $i++) {
  150. $iChapter_strlen += $title_data[$i]["lenght"];
  151. }
  152. $newData = [
  153. 'level' => $arrInserString[$iPar][3],
  154. 'toc' => $arrInserString[$iPar][5],
  155. 'chapter_len' => $length,
  156. 'next_chapter' => $next,
  157. 'prev_chapter' => $prev,
  158. 'parent' => $parent,
  159. 'chapter_strlen'=> $iChapter_strlen,
  160. ];
  161. if((int)$arrInserString[$iPar][3] < 8){
  162. $newData['title'] = strtolower($arrInserString[$iPar][6]);
  163. $newData['title_en'] = \App\Tools\Tools::getWordEn($newData['title']);
  164. }
  165. $path = [];
  166. $title_data[$iPar]["level"] = $newData["level"];
  167. $title_data[$iPar]["toc"] = $newData["toc"];
  168. $title_data[$iPar]["parent"] = $newData["parent"];
  169. /*
  170. *获取路径
  171. */
  172. $currParent = $parent;
  173. $iLoop = 0;
  174. while ($currParent != -1 && $iLoop<7) {
  175. # code...
  176. $pathTitle = $title_data[$currParent-1]["toc"];
  177. $pathLevel = $title_data[$currParent-1]['level'];
  178. $path[] = ["book"=>$book,"paragraph"=>$currParent,"title"=>$pathTitle,"level"=>$pathLevel];
  179. $currParent = $title_data[$currParent-1]["parent"];
  180. $iLoop++;
  181. }
  182. //插入书名
  183. if(count($path)>0){
  184. $bookPara = end($path)['paragraph'];
  185. }else{
  186. $bookPara = $paragraph;
  187. }
  188. $pcd_book = BookTitle::where('book',$book)
  189. ->where('paragraph',$bookPara)
  190. ->first();
  191. if($pcd_book){
  192. if(empty($pcd_book)){
  193. Log::error('no pcd book:'.$book.'-'.$bookPara);
  194. }
  195. $book_id = $pcd_book->sn;
  196. if(!empty($book_id)){
  197. $newData['pcd_book_id'] = $book_id;
  198. }
  199. $path[] = ["book"=>$book_id,"paragraph"=>$book_id,"title"=>$pcd_book->title,"level"=>0];
  200. }
  201. # 将路径反向
  202. $path1 = [];
  203. for ($i=count($path)-1; $i >=0 ; $i--) {
  204. # code...
  205. $path1[] = $path[$i];
  206. }
  207. $newData['path'] = $path1;
  208. PaliText::where('book',$book)
  209. ->where('paragraph',$paragraph)
  210. ->update($newData);
  211. if ($curr_level > 0 && $curr_level < 8) {
  212. $paragraph_info[] = array($book, $paragraph, $length, $prev, $next, $parent);
  213. }
  214. }
  215. $bar->advance();
  216. }
  217. $bar->finish();
  218. $this->info("instert pali text finished. in ". time()-$startTime . "s" );
  219. Log::info("instert pali text finished. in ". time()-$startTime . "s");
  220. return 0;
  221. }
  222. }