UpgradePaliText.php 7.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238
  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. *
  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. $this->info("upgrade pali text");
  43. $startTime = time();
  44. $_from = $this->argument('from');
  45. $_to = $this->argument('to');
  46. if(empty($_from) && empty($_to)){
  47. $_from = 1;
  48. $_to = 217;
  49. }else if(empty($_to)){
  50. $_to = $_from;
  51. }
  52. #载入文件列表
  53. $fileListFileName = config("app.path.palitext_filelist");
  54. $filelist = array();
  55. if (($handle = fopen($fileListFileName, 'r')) !== false) {
  56. while (($filelist[] = fgetcsv($handle, 0, ',')) !== false) {
  57. }
  58. }
  59. $bar = $this->output->createProgressBar($_to-$_from+1);
  60. for ($from=$_from; $from <= $_to; $from++) {
  61. $inputRow = 0;
  62. $arrInserString = array();
  63. #载入csv数据
  64. $FileName = $filelist[$from-1][1];
  65. $csvFile = config("app.path.pali_title") .'/'. $from.'_pali.csv';
  66. if (($fp = fopen($csvFile, "r")) !== false) {
  67. Log::info("csv load:" . $csvFile);
  68. while (($data = fgetcsv($fp, 0, ',')) !== false) {
  69. if ($inputRow > 0) {
  70. array_push($arrInserString, $data);
  71. }
  72. $inputRow++;
  73. }
  74. fclose($fp);
  75. } else {
  76. $this->error( "can not open csv file. filename=" . $csvFile. PHP_EOL) ;
  77. Log::error( "can not open csv file. filename=" . $csvFile) ;
  78. continue;
  79. }
  80. $title_data = PaliText::select(['book','paragraph','level','parent','toc','lenght'])
  81. ->where('book',$from)->orderby('paragraph','asc')->get();
  82. {
  83. $paragraph_count = count($title_data);
  84. $paragraph_info = array();
  85. $paragraph_info[] = array($from, -1, $paragraph_count, -1, -1, -1);
  86. for ($iPar = 0; $iPar < count($title_data); $iPar++) {
  87. $title_data[$iPar]["level"] = $arrInserString[$iPar][3];
  88. }
  89. for ($iPar = 0; $iPar < count($title_data); $iPar++) {
  90. $book = $from ;
  91. $paragraph = $title_data[$iPar]["paragraph"];
  92. if ((int) $title_data[$iPar]["level"] == 8) {
  93. $title_data[$iPar]["level"] = 100;
  94. }
  95. $curr_level = (int) $title_data[$iPar]["level"];
  96. # 计算这个chapter的段落数量
  97. $length = -1;
  98. for ($iPar1 = $iPar + 1; $iPar1 < count($title_data); $iPar1++) {
  99. $thislevel = (int) $title_data[$iPar1]["level"];
  100. if ($thislevel <= $curr_level) {
  101. $length = (int) $title_data[$iPar1]["paragraph"] - $paragraph;
  102. break;
  103. }
  104. }
  105. if ($length == -1) {
  106. $length = $paragraph_count - $paragraph + 1;
  107. }
  108. /*
  109. 上一个段落
  110. 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
  111. */
  112. $prev = -1;
  113. if ($iPar > 0) {
  114. for ($iPar1 = $iPar - 1; $iPar1 >= 0; $iPar1--) {
  115. if ($title_data[$iPar1]["level"] < 8 && $title_data[$iPar1+1]["level"]==100) {
  116. $prev = $title_data[$iPar1]["paragraph"];
  117. break;
  118. }
  119. }
  120. }
  121. /*
  122. 下一个段落
  123. 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
  124. */
  125. $next = -1;
  126. if ($iPar < count($title_data) - 1) {
  127. for ($iPar1 = $iPar + 1; $iPar1 < count($title_data)-1; $iPar1++) {
  128. if ($title_data[$iPar1]["level"] <8 && $title_data[$iPar1+1]["level"]==100) {
  129. $next = $title_data[$iPar1]["paragraph"];
  130. break;
  131. }
  132. }
  133. }
  134. $parent = -1;
  135. if ($iPar > 0) {
  136. for ($iPar1 = $iPar - 1; $iPar1 >= 0; $iPar1--) {
  137. if ($title_data[$iPar1]["level"] < $curr_level) {
  138. $parent = $title_data[$iPar1]["paragraph"];
  139. break;
  140. }
  141. }
  142. }
  143. //计算章节包含总字符数
  144. $iChapter_strlen = 0;
  145. for ($i = $iPar; $i < $iPar + $length; $i++) {
  146. $iChapter_strlen += $title_data[$i]["lenght"];
  147. }
  148. $newData = [
  149. 'level' => $arrInserString[$iPar][3],
  150. 'toc' => $arrInserString[$iPar][5],
  151. 'chapter_len' => $length,
  152. 'next_chapter' => $next,
  153. 'prev_chapter' => $prev,
  154. 'parent' => $parent,
  155. 'chapter_strlen'=> $iChapter_strlen,
  156. ];
  157. $path = [];
  158. $title_data[$iPar]["level"] = $newData["level"];
  159. $title_data[$iPar]["toc"] = $newData["toc"];
  160. $title_data[$iPar]["parent"] = $newData["parent"];
  161. /*
  162. *获取路径
  163. */
  164. $currParent = $parent;
  165. $iLoop = 0;
  166. while ($currParent != -1 && $iLoop<7) {
  167. # code...
  168. $pathTitle = $title_data[$currParent-1]["toc"];
  169. $pathLevel = $title_data[$currParent-1]['level'];
  170. $path[] = ["book"=>$book,"paragraph"=>$currParent,"title"=>$pathTitle,"level"=>$pathLevel];
  171. $currParent = $title_data[$currParent-1]["parent"];
  172. $iLoop++;
  173. }
  174. if(count($path)>0){
  175. //插入书名
  176. $bookTitle = BookTitle::where('book',$book)
  177. ->where('paragraph',end($path)['paragraph'])
  178. ->value('title');
  179. $path[] = ["book"=>0,"paragraph"=>0,"title"=>$bookTitle,"level"=>0];
  180. }
  181. # 将路径反向
  182. $path1 = [];
  183. for ($i=count($path)-1; $i >=0 ; $i--) {
  184. # code...
  185. $path1[] = $path[$i];
  186. }
  187. $newData['path'] = $path1;
  188. PaliText::where('book',$book)
  189. ->where('paragraph',$paragraph)
  190. ->update($newData);
  191. if ($curr_level > 0 && $curr_level < 8) {
  192. $paragraph_info[] = array($book, $paragraph, $length, $prev, $next, $parent);
  193. }
  194. }
  195. }
  196. $bar->advance();
  197. }
  198. $bar->finish();
  199. $this->info("instert pali text finished. in ". time()-$startTime . "s" );
  200. Log::info("instert pali text finished. in ". time()-$startTime . "s");
  201. return 0;
  202. }
  203. }