UpgradePaliText.php 6.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229
  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 Illuminate\Support\Facades\DB;
  10. use Illuminate\Support\Facades\Log;
  11. class UpgradePaliText extends Command
  12. {
  13. /**
  14. * The name and signature of the console command.
  15. *
  16. * @var string
  17. */
  18. protected $signature = 'upgrade:palitext {from?} {to?}';
  19. /**
  20. * The console command description.
  21. *
  22. * @var string
  23. */
  24. protected $description = 'upgrade pali_texts paragraph infomation';
  25. /**
  26. * Create a new command instance.
  27. *
  28. * @return void
  29. */
  30. public function __construct()
  31. {
  32. parent::__construct();
  33. }
  34. /**
  35. * Execute the console command.
  36. *
  37. * @return int
  38. */
  39. public function handle()
  40. {
  41. $this->info("upgrade pali text");
  42. $startTime = time();
  43. $_from = $this->argument('from');
  44. $_to = $this->argument('to');
  45. if(empty($_from) && empty($_to)){
  46. $_from = 1;
  47. $_to = 217;
  48. }else if(empty($_to)){
  49. $_to = $_from;
  50. }
  51. #载入文件列表
  52. $fileListFileName = config("app.path.palitext_filelist");
  53. $filelist = array();
  54. if (($handle = fopen($fileListFileName, 'r')) !== false) {
  55. while (($filelist[] = fgetcsv($handle, 0, ',')) !== false) {
  56. }
  57. }
  58. $bar = $this->output->createProgressBar($_to-$_from+1);
  59. for ($from=$_from; $from <= $_to; $from++) {
  60. $inputRow = 0;
  61. $arrInserString = array();
  62. #载入csv数据
  63. $FileName = $filelist[$from-1][1];
  64. $csvFile = config("app.path.pali_title") .'/'. $from.'_pali.csv';
  65. if (($fp = fopen($csvFile, "r")) !== false) {
  66. Log::info("csv load:" . $csvFile);
  67. while (($data = fgetcsv($fp, 0, ',')) !== false) {
  68. if ($inputRow > 0) {
  69. array_push($arrInserString, $data);
  70. }
  71. $inputRow++;
  72. }
  73. fclose($fp);
  74. } else {
  75. $this->error( "can not open csv file. filename=" . $csvFile. PHP_EOL) ;
  76. Log::error( "can not open csv file. filename=" . $csvFile) ;
  77. continue;
  78. }
  79. $title_data = PaliText::select(['book','paragraph','level','parent','toc','lenght'])
  80. ->where('book',$from)->orderby('paragraph','asc')->get();
  81. {
  82. $paragraph_count = count($title_data);
  83. $paragraph_info = array();
  84. $paragraph_info[] = array($from, -1, $paragraph_count, -1, -1, -1);
  85. for ($iPar = 0; $iPar < count($title_data); $iPar++) {
  86. $title_data[$iPar]["level"] = $arrInserString[$iPar][3];
  87. }
  88. for ($iPar = 0; $iPar < count($title_data); $iPar++) {
  89. $book = $from ;
  90. $paragraph = $title_data[$iPar]["paragraph"];
  91. if ((int) $title_data[$iPar]["level"] == 8) {
  92. $title_data[$iPar]["level"] = 100;
  93. }
  94. $curr_level = (int) $title_data[$iPar]["level"];
  95. # 计算这个chapter的段落数量
  96. $length = -1;
  97. for ($iPar1 = $iPar + 1; $iPar1 < count($title_data); $iPar1++) {
  98. $thislevel = (int) $title_data[$iPar1]["level"];
  99. if ($thislevel <= $curr_level) {
  100. $length = (int) $title_data[$iPar1]["paragraph"] - $paragraph;
  101. break;
  102. }
  103. }
  104. if ($length == -1) {
  105. $length = $paragraph_count - $paragraph + 1;
  106. }
  107. /*
  108. 上一个段落
  109. 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
  110. */
  111. $prev = -1;
  112. if ($iPar > 0) {
  113. for ($iPar1 = $iPar - 1; $iPar1 >= 0; $iPar1--) {
  114. if ($title_data[$iPar1]["level"] < 8 && $title_data[$iPar1+1]["level"]==100) {
  115. $prev = $title_data[$iPar1]["paragraph"];
  116. break;
  117. }
  118. }
  119. }
  120. /*
  121. 下一个段落
  122. 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
  123. */
  124. $next = -1;
  125. if ($iPar < count($title_data) - 1) {
  126. for ($iPar1 = $iPar + 1; $iPar1 < count($title_data)-1; $iPar1++) {
  127. if ($title_data[$iPar1]["level"] <8 && $title_data[$iPar1+1]["level"]==100) {
  128. $next = $title_data[$iPar1]["paragraph"];
  129. break;
  130. }
  131. }
  132. }
  133. $parent = -1;
  134. if ($iPar > 0) {
  135. for ($iPar1 = $iPar - 1; $iPar1 >= 0; $iPar1--) {
  136. if ($title_data[$iPar1]["level"] < $curr_level) {
  137. $parent = $title_data[$iPar1]["paragraph"];
  138. break;
  139. }
  140. }
  141. }
  142. //计算章节包含总字符数
  143. $iChapter_strlen = 0;
  144. for ($i = $iPar; $i < $iPar + $length; $i++) {
  145. $iChapter_strlen += $title_data[$i]["lenght"];
  146. }
  147. $newData = [
  148. 'level' => $arrInserString[$iPar][3],
  149. 'toc' => $arrInserString[$iPar][5],
  150. 'chapter_len' => $length,
  151. 'next_chapter' => $next,
  152. 'prev_chapter' => $prev,
  153. 'parent' => $parent,
  154. 'chapter_strlen'=> $iChapter_strlen,
  155. ];
  156. $path = [];
  157. $title_data[$iPar]["level"] = $newData["level"];
  158. $title_data[$iPar]["toc"] = $newData["toc"];
  159. $title_data[$iPar]["parent"] = $newData["parent"];
  160. /*
  161. *获取路径
  162. */
  163. $currParent = $parent;
  164. $iLoop = 0;
  165. while ($currParent != -1 && $iLoop<7) {
  166. # code...
  167. $pathTitle = $title_data[$currParent-1]["toc"];
  168. $pathLevel = $title_data[$currParent-1]['level'];
  169. $path[] = ["book"=>$book,"paragraph"=>$currParent,"title"=>$pathTitle,"level"=>$pathLevel];
  170. $currParent = $title_data[$currParent-1]["parent"];
  171. $iLoop++;
  172. }
  173. # 将路径反向
  174. $path1 = [];
  175. for ($i=count($path)-1; $i >=0 ; $i--) {
  176. # code...
  177. $path1[] = $path[$i];
  178. }
  179. $newData['path'] = $path1;
  180. PaliText::where('book',$book)
  181. ->where('paragraph',$paragraph)
  182. ->update($newData);
  183. if ($curr_level > 0 && $curr_level < 8) {
  184. $paragraph_info[] = array($book, $paragraph, $length, $prev, $next, $parent);
  185. }
  186. }
  187. }
  188. $bar->advance();
  189. }
  190. $bar->finish();
  191. $this->info("instert pali text finished. in ". time()-$startTime . "s" );
  192. Log::info("instert pali text finished. in ". time()-$startTime . "s");
  193. return 0;
  194. }
  195. }