db_insert_palitext_cli.php 4.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148
  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 = $fileNums - 1;
  34. }
  35. PDO_Connect(_PG_DB_PALITEXT_,_DB_USERNAME_,_DB_PASSWORD_);
  36. for ($from=$_from-1; $from < $to; $from++) {
  37. # code...
  38. $FileName = $filelist[$from][1] . ".htm";
  39. $fileId = $filelist[$from][0];
  40. $fileId = $filelist[$from][0];
  41. $dirLog = _DIR_LOG_ . "/";
  42. $inputFileName = $FileName;
  43. $outputFileNameHead = $filelist[$from][1];
  44. $bookId = $filelist[$from][2];
  45. $vriParNum = 0;
  46. $wordOrder = 1;
  47. $dirXmlBase = _DIR_PALI_CSV_ . "/";
  48. $dirPaliTextBase = _DIR_PALI_HTML_ . "/";
  49. $dirXml = $outputFileNameHead . "/";
  50. $xmlfile = $inputFileName;
  51. echo "doing:" . $xmlfile . PHP_EOL;
  52. $log = $log . "$from,$FileName,open\r\n";
  53. $arrInserString = array();
  54. // 打开vri html文件并读取数据
  55. $pali_text_array = array();
  56. if (($fpPaliText = fopen($dirPaliTextBase . $xmlfile, "r")) !== false) {
  57. while (($data = fgets($fpPaliText)) !== false) {
  58. if (substr($data, 0, 2) === "<p") {
  59. array_push($pali_text_array, $data);
  60. }
  61. }
  62. fclose($fpPaliText);
  63. echo "pali text load:" . $dirPaliTextBase . $xmlfile . PHP_EOL;
  64. } else {
  65. echo "can not pali text file. filename=" . $dirPaliTextBase . $xmlfile;
  66. }
  67. $inputRow = 0;
  68. if (($fp = fopen($dirXmlBase . $dirXml . $outputFileNameHead . "_pali.csv", "r")) !== false) {
  69. while (($data = fgetcsv($fp, 0, ',')) !== false) {
  70. if ($inputRow > 0) {
  71. if (($inputRow - 1) < count($pali_text_array)) {
  72. $data[5] = $pali_text_array[$inputRow - 1];
  73. }
  74. $arrInserString[] = $data;
  75. }
  76. $inputRow++;
  77. }
  78. fclose($fp);
  79. echo "单词表load:" . $dirXmlBase . $dirXml . $outputFileNameHead . ".csv".PHP_EOL;
  80. } else {
  81. echo "can not open csv file. filename=" . $dirXmlBase . $dirXml . $outputFileNameHead . ".csv";
  82. continue;
  83. }
  84. if (($inputRow - 1) != count($pali_text_array)) {
  85. $log = $log . "$from, $FileName,error,文件行数不匹配 inputRow=$inputRow pali_text_array=" . count($pali_text_array) . " \r\n";
  86. echo "line count error".PHP_EOL;
  87. }
  88. #删除 旧数据
  89. $query = "DELETE FROM "._PG_TABLE_PALI_TEXT_." WHERE book=?";
  90. PDO_Execute($query,array($from+1));
  91. // 开始一个事务,关闭自动提交
  92. $PDO->beginTransaction();
  93. $query = "INSERT INTO "._PG_TABLE_PALI_TEXT_." ( book , paragraph , level , class , toc , text , html , lenght ) VALUES ( ? , ? , ? , ? , ? , ? , ? , ? )";
  94. $stmt = $PDO->prepare($query);
  95. foreach ($arrInserString as $oneParam) {
  96. if ($oneParam[3] < 100) {
  97. $toc = $oneParam[6];
  98. } else {
  99. $toc = "";
  100. }
  101. $newData = array($from + 1, $oneParam[2], $oneParam[3], $oneParam[4], $toc, $oneParam[6], $oneParam[5], mb_strlen($oneParam[6], "UTF-8"));
  102. $stmt->execute($newData);
  103. }
  104. // 提交更改
  105. $PDO->commit();
  106. if (!$stmt || ($stmt && $stmt->errorCode() != 0)) {
  107. $error = PDO_ErrorInfo();
  108. echo "error - $error[2]".PHP_EOL;
  109. $log = $log . "$from, $FileName, error, $error[2] \r\n";
  110. } else {
  111. $count = count($arrInserString);
  112. echo "updata $count recorders.".PHP_EOL;
  113. }
  114. /*
  115. $myLogFile = fopen($dirLog . "db_insert_palitext.log", "a");
  116. fwrite($myLogFile, $log);
  117. fclose($myLogFile);
  118. */
  119. }
  120. echo "all done!".PHP_EOL;
  121. ?>