UpgradePaliText.php 6.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225
  1. <?php
  2. namespace App\Console\Commands;
  3. use Illuminate\Console\Command;
  4. use App\Models\PaliText;
  5. use Illuminate\Support\Facades\DB;
  6. use Illuminate\Support\Facades\Log;
  7. class UpgradePaliText extends Command
  8. {
  9. /**
  10. * The name and signature of the console command.
  11. *
  12. * @var string
  13. */
  14. protected $signature = 'upgrade:palitext {from?} {to?}';
  15. /**
  16. * The console command description.
  17. *
  18. * @var string
  19. */
  20. protected $description = 'upgrade pali_texts paragraph infomation';
  21. /**
  22. * Create a new command instance.
  23. *
  24. * @return void
  25. */
  26. public function __construct()
  27. {
  28. parent::__construct();
  29. }
  30. /**
  31. * Execute the console command.
  32. *
  33. * @return int
  34. */
  35. public function handle()
  36. {
  37. $this->info("upgrade pali text");
  38. $startTime = time();
  39. $_from = $this->argument('from');
  40. $_to = $this->argument('to');
  41. if(empty($_from) && empty($_to)){
  42. $_from = 1;
  43. $_to = 217;
  44. }else if(empty($_to)){
  45. $_to = $_from;
  46. }
  47. #载入文件列表
  48. $fileListFileName = config("app.path.palitext_filelist");
  49. $filelist = array();
  50. if (($handle = fopen($fileListFileName, 'r')) !== false) {
  51. while (($filelist[] = fgetcsv($handle, 0, ',')) !== false) {
  52. }
  53. }
  54. $bar = $this->output->createProgressBar($_to-$_from+1);
  55. for ($from=$_from; $from <= $_to; $from++) {
  56. $inputRow = 0;
  57. $arrInserString = array();
  58. #载入csv数据
  59. $FileName = $filelist[$from-1][1];
  60. $csvFile = config("app.path.pali_title") .'/'. $from.'_pali.csv';
  61. if (($fp = fopen($csvFile, "r")) !== false) {
  62. Log::info("csv load:" . $csvFile);
  63. while (($data = fgetcsv($fp, 0, ',')) !== false) {
  64. if ($inputRow > 0) {
  65. array_push($arrInserString, $data);
  66. }
  67. $inputRow++;
  68. }
  69. fclose($fp);
  70. } else {
  71. $this->error( "can not open csv file. filename=" . $csvFile. PHP_EOL) ;
  72. Log::error( "can not open csv file. filename=" . $csvFile) ;
  73. continue;
  74. }
  75. $title_data = PaliText::where('book',$from)->orderby('paragraph','asc')->get();
  76. {
  77. $paragraph_count = count($title_data);
  78. $paragraph_info = array();
  79. $paragraph_info[] = array($from, -1, $paragraph_count, -1, -1, -1);
  80. for ($iPar = 0; $iPar < count($title_data); $iPar++) {
  81. $title_data[$iPar]["level"] = $arrInserString[$iPar][3];
  82. }
  83. for ($iPar = 0; $iPar < count($title_data); $iPar++) {
  84. $book = $from ;
  85. $paragraph = $title_data[$iPar]["paragraph"];
  86. if ((int) $title_data[$iPar]["level"] == 8) {
  87. $title_data[$iPar]["level"] = 100;
  88. }
  89. $curr_level = (int) $title_data[$iPar]["level"];
  90. # 计算这个chapter的段落数量
  91. $length = -1;
  92. for ($iPar1 = $iPar + 1; $iPar1 < count($title_data); $iPar1++) {
  93. $thislevel = (int) $title_data[$iPar1]["level"];
  94. if ($thislevel <= $curr_level) {
  95. $length = (int) $title_data[$iPar1]["paragraph"] - $paragraph;
  96. break;
  97. }
  98. }
  99. if ($length == -1) {
  100. $length = $paragraph_count - $paragraph + 1;
  101. }
  102. /*
  103. 上一个段落
  104. 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
  105. */
  106. $prev = -1;
  107. if ($iPar > 0) {
  108. for ($iPar1 = $iPar - 1; $iPar1 >= 0; $iPar1--) {
  109. if ($title_data[$iPar1]["level"] < 8 && $title_data[$iPar1+1]["level"]==100) {
  110. $prev = $title_data[$iPar1]["paragraph"];
  111. break;
  112. }
  113. }
  114. }
  115. /*
  116. 下一个段落
  117. 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
  118. */
  119. $next = -1;
  120. if ($iPar < count($title_data) - 1) {
  121. for ($iPar1 = $iPar + 1; $iPar1 < count($title_data)-1; $iPar1++) {
  122. if ($title_data[$iPar1]["level"] <8 && $title_data[$iPar1+1]["level"]==100) {
  123. $next = $title_data[$iPar1]["paragraph"];
  124. break;
  125. }
  126. }
  127. }
  128. $parent = -1;
  129. if ($iPar > 0) {
  130. for ($iPar1 = $iPar - 1; $iPar1 >= 0; $iPar1--) {
  131. if ($title_data[$iPar1]["level"] < $curr_level) {
  132. $parent = $title_data[$iPar1]["paragraph"];
  133. break;
  134. }
  135. }
  136. }
  137. //计算章节包含总字符数
  138. $iChapter_strlen = 0;
  139. for ($i = $iPar; $i < $iPar + $length; $i++) {
  140. $iChapter_strlen += $title_data[$i]["lenght"];
  141. }
  142. $newData = [
  143. 'level' => $arrInserString[$iPar][3],
  144. 'toc' => $arrInserString[$iPar][5],
  145. 'chapter_len' => $length,
  146. 'next_chapter' => $next,
  147. 'prev_chapter' => $prev,
  148. 'parent' => $parent,
  149. 'chapter_strlen'=> $iChapter_strlen,
  150. ];
  151. $path = [];
  152. $title_data[$iPar]["level"] = $newData["level"];
  153. $title_data[$iPar]["toc"] = $newData["toc"];
  154. $title_data[$iPar]["parent"] = $newData["parent"];
  155. /*
  156. *获取路径
  157. */
  158. $currParent = $parent;
  159. $iLoop = 0;
  160. while ($currParent != -1 && $iLoop<7) {
  161. # code...
  162. $pathTitle = $title_data[$currParent-1]["toc"];
  163. $pathLevel = $title_data[$currParent-1]['level'];
  164. $path[] = ["book"=>$book,"paragraph"=>$currParent,"title"=>$pathTitle,"level"=>$pathLevel];
  165. $currParent = $title_data[$currParent-1]["parent"];
  166. $iLoop++;
  167. }
  168. # 将路径反向
  169. $path1 = [];
  170. for ($i=count($path)-1; $i >=0 ; $i--) {
  171. # code...
  172. $path1[] = $path[$i];
  173. }
  174. $newData['path'] = $path1;
  175. PaliText::where('book',$book)
  176. ->where('paragraph',$paragraph)
  177. ->update($newData);
  178. if ($curr_level > 0 && $curr_level < 8) {
  179. $paragraph_info[] = array($book, $paragraph, $length, $prev, $next, $parent);
  180. }
  181. }
  182. }
  183. $bar->advance();
  184. }
  185. $bar->finish();
  186. $this->info("instert pali text finished. in ". time()-$startTime . "s" .PHP_EOL);
  187. Log::info("instert pali text finished. in ". time()-$startTime . "s");
  188. return 0;
  189. }
  190. }