db_update_palitext_cli.php 7.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273
  1. <?php
  2. /*
  3. 生成 巴利原文段落表
  4. */
  5. require_once __DIR__."/../config.php";
  6. require_once __DIR__."/../public/_pdo.php";
  7. set_exception_handler(function($e){
  8. fwrite(STDERR,"error-msg:".$e->getMessage().PHP_EOL);
  9. fwrite(STDERR,"error-file:".$e->getFile().PHP_EOL);
  10. fwrite(STDERR,"error-line:".$e->getLine().PHP_EOL);
  11. exit;
  12. });
  13. define("_DB_PALITEXT_", _PG_DB_PALITEXT_);
  14. define("_TABLE_",_PG_TABLE_PALI_TEXT_);
  15. fwrite(STDOUT, "Update Pali Text ".PHP_EOL);
  16. if ($argc != 3) {
  17. echo "help".PHP_EOL;
  18. echo $argv[0]." from to".PHP_EOL;
  19. echo "from = 1-217".PHP_EOL;
  20. echo "to = 1-217".PHP_EOL;
  21. exit;
  22. }
  23. $_from = (int) $argv[1];
  24. $_to = (int) $argv[2];
  25. if ($_to > 217) {
  26. $_to = 217;
  27. }
  28. $to = $_to;
  29. $filelist = array();
  30. $fileNums = 0;
  31. $log = "";
  32. if (($handle = fopen(__DIR__."/filelist.csv", 'r')) !== false) {
  33. while (($filelist[$fileNums] = fgetcsv($handle, 0, ',')) !== false) {
  34. $fileNums++;
  35. }
  36. }
  37. if ($to == 0 || $to > $fileNums) {
  38. $to = $fileNums;
  39. }
  40. PDO_Connect(_DB_PALITEXT_,_DB_USERNAME_,_DB_PASSWORD_);
  41. for ($from=$_from-1; $from < $to; $from++) {
  42. fwrite(STDOUT, "doing $_from".PHP_EOL);
  43. $FileName = $filelist[$from][1] . ".htm";
  44. $fileId = $filelist[$from][0];
  45. $fileId = $filelist[$from][0];
  46. $dirLog = _DIR_LOG_ . "/";
  47. $dirDb = "/";
  48. $inputFileName = $FileName;
  49. $outputFileNameHead = $filelist[$from][1];
  50. $bookId = $filelist[$from][2];
  51. $vriParNum = 0;
  52. $wordOrder = 1;
  53. $dirXmlBase = _DIR_PALI_CSV_ . "/";
  54. $dirPaliTextBase = _DIR_PALI_HTML_ . "/";
  55. $dirXml = $outputFileNameHead . "/";
  56. $xmlfile = $inputFileName;
  57. fwrite(STDOUT, "doing:" . $xmlfile . PHP_EOL);
  58. $log = $log . date("Y-m-d h:i:sa") . ",$from,$FileName,open\r\n";
  59. $arrInserString = array();
  60. // 打开vri html文件并读取数据
  61. $pali_text_array = array(); //vri text
  62. if (($fpPaliText = fopen($dirPaliTextBase . $xmlfile, "r")) !== false) {
  63. while (($data = fgets($fpPaliText)) !== false) {
  64. array_push($pali_text_array, $data);
  65. }
  66. fclose($fpPaliText);
  67. fwrite(STDOUT, "pali text load:" . $dirPaliTextBase . $xmlfile . PHP_EOL);
  68. } else {
  69. fwrite(STDERR, "can not pali text file. filename=" . $dirPaliTextBase . $xmlfile.PHP_EOL);
  70. }
  71. // 打开csv文件并读取数据
  72. $inputRow = 0;
  73. $tocFile = _DIR_PALI_TITLE_ . "/" . ($from + 1) . "_pali.csv";
  74. if (($fp = fopen($tocFile, "r")) !== false) {
  75. while (($data = fgetcsv($fp, 0, ',')) !== false) {
  76. if ($inputRow > 0) {
  77. $params = $data;
  78. array_push($arrInserString, $params);
  79. }
  80. $inputRow++;
  81. }
  82. fclose($fp);
  83. fwrite(STDOUT, "Toc load:" . $tocFile. PHP_EOL);
  84. } else {
  85. fwrite(STDERR, "can not open csv file. filename=" . $tocFile . PHP_EOL);
  86. }
  87. if ((count($arrInserString)) != count($pali_text_array) - 2) {
  88. $log = $log . "$from, $FileName,error,文件行数不匹配 csv = " . (count($arrInserString) - 1) . " pali_text_array=" . (count($pali_text_array) - 2) . " \r\n";
  89. }
  90. $book = $from + 1;
  91. //计算段落信息,如上一段
  92. $query = "SELECT book,paragraph,level,toc,lenght from "._TABLE_." where book = '$book' order by paragraph asc";
  93. $title_data = PDO_FetchAll($query);
  94. fwrite(STDOUT, "Paragraph Count:" . count($title_data) . " arrInserString:".count($arrInserString). PHP_EOL);
  95. $paragraph_count = count($title_data);
  96. // 开始一个事务,关闭自动提交
  97. $PDO->beginTransaction();
  98. $query = "UPDATE "._TABLE_." SET level = ? , toc = ? , chapter_len = ? , next_chapter = ?, prev_chapter=? , parent= ? , chapter_strlen = ? , path = ? ,updated_at = now() WHERE book=? and paragraph=?";
  99. $stmt = $PDO->prepare($query);
  100. $paragraph_info = array();
  101. array_push($paragraph_info, array($from, -1, $paragraph_count, -1, -1, -1));
  102. for ($iPar = 0; $iPar < count($title_data); $iPar++) {
  103. $title_data[$iPar]["level"] = $arrInserString[$iPar][3];
  104. }
  105. for ($iPar = 0; $iPar < count($title_data); $iPar++) {
  106. $book = $from + 1;
  107. $paragraph = $title_data[$iPar]["paragraph"];
  108. /*
  109. level 8 为 偈诵编号。不当作标题
  110. */
  111. if ((int)$title_data[$iPar]["level"] == 8) {
  112. $title_data[$iPar]["level"] = 100;
  113. }
  114. $curr_level = (int) $title_data[$iPar]["level"];
  115. # 计算这个chapter的段落数量
  116. $length = -1;
  117. for ($iPar1 = $iPar + 1; $iPar1 < count($title_data); $iPar1++) {
  118. $thislevel = (int) $title_data[$iPar1]["level"];
  119. if ($thislevel <= $curr_level) {
  120. $length = (int) $title_data[$iPar1]["paragraph"] - $paragraph;
  121. break;
  122. }
  123. }
  124. if ($length == -1) {
  125. $length = $paragraph_count - $paragraph + 1;
  126. }
  127. /*
  128. 上一个段落
  129. 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
  130. */
  131. $prev = -1;
  132. if ($iPar > 0) {
  133. for ($iPar1 = $iPar - 1; $iPar1 >= 0; $iPar1--) {
  134. if ($title_data[$iPar1]["level"] < 8 && $title_data[$iPar1+1]["level"]==100) {
  135. $prev = $title_data[$iPar1]["paragraph"];
  136. break;
  137. }
  138. }
  139. }
  140. /*
  141. 上一个段落
  142. 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
  143. */
  144. $next = -1;
  145. if ($iPar < count($title_data) - 1) {
  146. for ($iPar1 = $iPar + 1; $iPar1 < count($title_data)-1; $iPar1++) {
  147. if ($title_data[$iPar1]["level"] <8 && $title_data[$iPar1+1]["level"]==100) {
  148. $next = $title_data[$iPar1]["paragraph"];
  149. break;
  150. }
  151. }
  152. }
  153. $parent = -1;
  154. if ($iPar > 0) {
  155. for ($iPar1 = $iPar - 1; $iPar1 >= 0; $iPar1--) {
  156. if ($title_data[$iPar1]["level"] < $curr_level) {
  157. $parent = $title_data[$iPar1]["paragraph"];
  158. break;
  159. }
  160. }
  161. }
  162. //计算章节包含总字符数
  163. $iChapter_strlen = 0;
  164. for ($i = $iPar; $i < $iPar + $length; $i++) {
  165. $iChapter_strlen += $title_data[$i]["lenght"];
  166. }
  167. $title_data[$iPar]["level"] = $arrInserString[$iPar][3];
  168. $title_data[$iPar]["toc"] = $arrInserString[$iPar][5];
  169. $title_data[$iPar]["parent"] = $parent;
  170. /*
  171. *获取路径
  172. */
  173. $currParent = $parent;
  174. $path = [];
  175. $iLoop = 0;
  176. while ($currParent != -1 && $iLoop<7) {
  177. # code...
  178. $pathTitle = $title_data[$currParent-1]["toc"];
  179. $pathLevel = $title_data[$currParent-1]['level'];
  180. $path[] = ["book"=>$book,"paragraph"=>$currParent,"title"=>$pathTitle,"level"=>$pathLevel];
  181. $currParent = $title_data[$currParent-1]["parent"];
  182. $iLoop++;
  183. }
  184. # 将路径反向
  185. $path1 = [];
  186. for ($i=count($path)-1; $i >=0 ; $i--) {
  187. # code...
  188. $path1[] = $path[$i];
  189. }
  190. $newData = array(
  191. $arrInserString[$iPar][3],
  192. $arrInserString[$iPar][5],
  193. $length,
  194. $next,
  195. $prev,
  196. $parent,
  197. $iChapter_strlen,
  198. json_encode($path1),
  199. $book,
  200. $paragraph,
  201. );
  202. $stmt->execute($newData);
  203. if ($curr_level > 0 && $curr_level < 8) {
  204. array_push($paragraph_info, array($book, $paragraph, $length, $prev, $next, $parent));
  205. }
  206. }
  207. // 提交更改
  208. $PDO->commit();
  209. if (!$stmt || ($stmt && $stmt->errorCode() != 0)) {
  210. $error = PDO_ErrorInfo();
  211. fwrite(STDERR, "error - $error[2]". PHP_EOL);
  212. $log = $log . "$from, $FileName, error, $error[2] \r\n";
  213. } else {
  214. $count = count($title_data);
  215. fwrite(STDOUT, "updata $count paragraph info recorders.". PHP_EOL);
  216. fwrite(STDOUT, count($paragraph_info) . " Heading". PHP_EOL);
  217. }
  218. //段落信息结束
  219. }
  220. fwrite(STDOUT, "all done!".PHP_EOL);
  221. ?>