db_insert_palitext_cli.php 4.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164
  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. define("_PG_DB_PALITEXT_", _DB_ENGIN_.":host="._DB_HOST_.";port="._DB_PORT_.";dbname="._DB_NAME_.";user="._DB_USERNAME_.";password="._DB_PASSWORD_.";");
  8. define("_PG_TABLE_PALI_TEXT_","pali_texts");
  9. echo "Insert Pali Text To DB".PHP_EOL;
  10. if ($argc != 3) {
  11. echo "help".PHP_EOL;
  12. echo $argv[0]." from to".PHP_EOL;
  13. echo "from = 1-217".PHP_EOL;
  14. echo "to = 1-217".PHP_EOL;
  15. exit;
  16. }
  17. $_from = (int) $argv[1];
  18. $_to = (int) $argv[2];
  19. if ($_to > 217) {
  20. $_to = 217;
  21. }
  22. $to = $_to;
  23. $filelist = array();
  24. $fileNums = 0;
  25. $log = "";
  26. echo "doing $_from";
  27. if (($handle = fopen(__DIR__."/filelist.csv", 'r')) !== false) {
  28. while (($filelist[$fileNums] = fgetcsv($handle, 0, ',')) !== false) {
  29. $fileNums++;
  30. }
  31. }
  32. if ($to == 0 || $to >= $fileNums) {
  33. $to = $fileNums - 1;
  34. }
  35. $dns = _PG_DB_PALITEXT_;
  36. $dbh = new PDO($dns, _DB_USERNAME_, _DB_PASSWORD_, array(PDO::ATTR_PERSISTENT => true));
  37. $dbh->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_EXCEPTION);
  38. for ($from=$_from-1; $from < $to; $from++) {
  39. # code...
  40. $FileName = $filelist[$from][1] . ".htm";
  41. $fileId = $filelist[$from][0];
  42. $fileId = $filelist[$from][0];
  43. $dirLog = _DIR_LOG_ . "/";
  44. $inputFileName = $FileName;
  45. $outputFileNameHead = $filelist[$from][1];
  46. $bookId = $filelist[$from][2];
  47. $vriParNum = 0;
  48. $wordOrder = 1;
  49. $dirXmlBase = _DIR_PALI_CSV_ . "/";
  50. $dirPaliTextBase = _DIR_PALI_HTML_ . "/";
  51. $dirXml = $outputFileNameHead . "/";
  52. $xmlfile = $inputFileName;
  53. echo "doing:" . $xmlfile . PHP_EOL;
  54. $log = $log . "$from,$FileName,open\r\n";
  55. $arrInserString = array();
  56. // 打开vri html文件并读取数据
  57. $pali_text_array = array();
  58. if (($fpPaliText = fopen($dirPaliTextBase . $xmlfile, "r")) !== false) {
  59. while (($data = fgets($fpPaliText)) !== false) {
  60. if (substr($data, 0, 2) === "<p") {
  61. array_push($pali_text_array, $data);
  62. }
  63. }
  64. fclose($fpPaliText);
  65. echo "pali text load:" . $dirPaliTextBase . $xmlfile . PHP_EOL;
  66. } else {
  67. echo "can not pali text file. filename=" . $dirPaliTextBase . $xmlfile;
  68. }
  69. $inputRow = 0;
  70. if (($fp = fopen($dirXmlBase . $dirXml . $outputFileNameHead . "_pali.csv", "r")) !== false) {
  71. while (($data = fgetcsv($fp, 0, ',')) !== false) {
  72. if ($inputRow > 0) {
  73. if (($inputRow - 1) < count($pali_text_array)) {
  74. $data[5] = $pali_text_array[$inputRow - 1];
  75. }
  76. $arrInserString[] = $data;
  77. }
  78. $inputRow++;
  79. }
  80. fclose($fp);
  81. echo "单词表load:" . $dirXmlBase . $dirXml . $outputFileNameHead . ".csv".PHP_EOL;
  82. } else {
  83. echo "can not open csv file. filename=" . $dirXmlBase . $dirXml . $outputFileNameHead . ".csv";
  84. continue;
  85. }
  86. if (($inputRow - 1) != count($pali_text_array)) {
  87. $log = $log . "$from, $FileName,error,文件行数不匹配 inputRow=$inputRow pali_text_array=" . count($pali_text_array) . " \r\n";
  88. echo "line count error".PHP_EOL;
  89. }
  90. #删除 旧数据
  91. $query = "DELETE FROM "._PG_TABLE_PALI_TEXT_." WHERE book=?";
  92. $stmt = $dbh->prepare($query);
  93. try{
  94. $stmt->execute(array($from+1));
  95. }catch(PDOException $e){
  96. fwrite(STDERR,$e->getMessage());
  97. continue;
  98. }
  99. // 开始一个事务,关闭自动提交
  100. $dbh->beginTransaction();
  101. $query = "INSERT INTO "._PG_TABLE_PALI_TEXT_." ( book , paragraph , level , class , toc , text , html , lenght ) VALUES ( ? , ? , ? , ? , ? , ? , ? , ? )";
  102. $stmt = $dbh->prepare($query);
  103. foreach ($arrInserString as $oneParam) {
  104. if ($oneParam[3] < 100) {
  105. $toc = $oneParam[6];
  106. } else {
  107. $toc = "";
  108. }
  109. $newData = array($from + 1, $oneParam[2], $oneParam[3], $oneParam[4], $toc, $oneParam[6], $oneParam[5], mb_strlen($oneParam[6], "UTF-8"));
  110. try{
  111. $stmt->execute($newData);
  112. }catch(PDOException $e){
  113. fwrite(STDERR,$e->getMessage());
  114. fwrite(STDERR,implode(",",$newData));
  115. break;
  116. }
  117. }
  118. // 提交更改
  119. $dbh->commit();
  120. if (!$stmt || ($stmt && $stmt->errorCode() != 0)) {
  121. $error = $dbh->errorInfo();
  122. echo "error - $error[2]".PHP_EOL;
  123. $log = $log . "$from, $FileName, error, $error[2] \r\n";
  124. } else {
  125. $count = count($arrInserString);
  126. echo "updata $count recorders.".PHP_EOL;
  127. }
  128. /*
  129. $myLogFile = fopen($dirLog . "db_insert_palitext.log", "a");
  130. fwrite($myLogFile, $log);
  131. fclose($myLogFile);
  132. */
  133. }
  134. echo "all done!".PHP_EOL;
  135. ?>