word_index_weight_refresh.php 5.8 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142
  1. <?php
  2. /*
  3. 计算单词权重
  4. */
  5. require_once __DIR__.'/../config.php';
  6. require_once __DIR__.'/word_index_weight_table.php';
  7. define("_PG_DB_WORD_INDEX_", _DB_ENGIN_.":host="._DB_HOST_.";port="._DB_PORT_.";dbname="._DB_NAME_.";user="._DB_USERNAME_.";password="._DB_PASSWORD_.";");
  8. define("_PG_TABLE_WORD_INDEX_", "word_indexs");
  9. define("_PG_DB_PALI_INDEX_", _DB_ENGIN_.":host="._DB_HOST_.";port="._DB_PORT_.";dbname="._DB_NAME_.";user="._DB_USERNAME_.";password="._DB_PASSWORD_.";");
  10. define("_PG_TABLE_WORD_", "words");
  11. if (isset($_GET["from"])) {
  12. $from = (int)$_GET["from"];
  13. $to = (int)$_GET["to"];
  14. } else {
  15. if ($argc != 3) {
  16. echo "无效的参数 ";
  17. exit;
  18. }
  19. $from = (int) $argv[1];
  20. $to = (int) $argv[2];
  21. if ($to > 217) {
  22. $to = 217;
  23. }
  24. }
  25. $dh_word = new PDO( _PG_DB_WORD_INDEX_, _DB_USERNAME_, _DB_PASSWORD_);
  26. $dh_word->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_WARNING);
  27. $dh_pali = new PDO( _PG_DB_PALI_INDEX_, _DB_USERNAME_, _DB_PASSWORD_);
  28. $dh_pali->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_WARNING);
  29. echo "from=$from to = $to \n";
  30. for ($i = $from; $i <= $to; $i++) {
  31. $time_start = microtime(true);
  32. echo "正在处理 book= $i ";
  33. $query = "SELECT max(paragraph) from "._PG_TABLE_WORD_." where book=?";
  34. $stmt = $dh_pali->prepare($query);
  35. $stmt->execute(array($i));
  36. $row = $stmt->fetch(PDO::FETCH_NUM);
  37. if ($row) {
  38. $max_para = $row[0];
  39. echo "段落数量:$max_para ";
  40. for ($j = 0; $j <= $max_para; $j++) {
  41. # code...
  42. $query = "SELECT id,book,wordindex,bold from "._PG_TABLE_WORD_." where book={$i} and paragraph={$j} order by id ASC";
  43. $stmt = $dh_pali->query($query);
  44. $fetch = $stmt->fetchAll(PDO::FETCH_ASSOC);
  45. $query = "SELECT wordindex,count(*) as co from "._PG_TABLE_WORD_." where book={$i} and paragraph={$j} group by wordindex";
  46. $stmt = $dh_pali->query($query);
  47. $fetch_voc = $stmt->fetchAll(PDO::FETCH_ASSOC);
  48. $vocabulary = array();
  49. foreach ($fetch_voc as $key => $value) {
  50. $vocabulary[$value["wordindex"]] = $value["co"];
  51. }
  52. for ($iWord = 0; $iWord < count($fetch); $iWord++) {
  53. # 非黑体字
  54. if ($fetch[$iWord]["bold"] == 0) {
  55. $count = $vocabulary[$fetch[$iWord]["wordindex"]];
  56. $paraWeight = pow(1.01, $count); //总分
  57. if ($paraWeight > 1.9) {
  58. $paraWeight = 1.9;
  59. }
  60. $weight = $paraWeight / $count;
  61. } else {
  62. #黑体字
  63. #查找前后相连的黑体字
  64. $begin = $iWord;
  65. while ($fetch[$begin]["bold"] == 1) {
  66. $begin--;
  67. if ($begin < 0) {
  68. break;
  69. }
  70. }
  71. $begin = $begin + 1;
  72. $end = $iWord;
  73. while ($fetch[$end]["bold"] == 1) {
  74. $end++;
  75. if ($end > count($fetch) - 1) {
  76. break;
  77. }
  78. }
  79. $end = $end - 1;
  80. $bold_count = $end - $begin + 1;
  81. if ($bold_count == 1) {
  82. $query = "SELECT * from "._PG_TABLE_WORD_INDEX_." where id=" . $fetch[$iWord]["wordindex"];
  83. $stmt_word = $dh_word->query($query);
  84. $wordinfo = $stmt_word->fetch(PDO::FETCH_ASSOC);
  85. $bookId = (int) $fetch[$iWord]["book"];
  86. if (mb_substr($wordinfo["word"], -2) == "ti") {
  87. $weight = 100 + $book_weight[$bookId];
  88. } else {
  89. $weight = 100 + $book_weight[$bookId];
  90. }
  91. //echo "单独黑体 $weight \n";
  92. } else {
  93. #连续黑体字
  94. //echo "连续黑体字";
  95. $len_sum = 0;
  96. $len_curr = 0;
  97. for ($iBold = $begin; $iBold <= $end; $iBold++) {
  98. # code...
  99. $boldid = $fetch[$iBold]["wordindex"];
  100. $query = "SELECT len from "._PG_TABLE_WORD_INDEX_." where id=" . $boldid;
  101. $stmt_bold = $dh_word->query($query);
  102. $wordbold = $stmt_bold->fetch(PDO::FETCH_ASSOC);
  103. $len_sum += $wordbold["len"];
  104. if ($iBold == $i) {
  105. $len_curr = $wordbold["len"];
  106. }
  107. }
  108. $weight = 10 + $len_curr / $len_sum;
  109. }
  110. }
  111. //echo $weight."\n";
  112. $fetch[$iWord]["weight"] = (int) ($weight * 100);
  113. }
  114. # 将整段权重写入据库
  115. $dh_pali->beginTransaction();
  116. $query = "UPDATE "._PG_TABLE_WORD_." set weight = ? where id=? ";
  117. $stmt_weight = $dh_pali->prepare($query);
  118. foreach ($fetch as $key => $value) {
  119. $stmt_weight->execute(array($value["weight"], $value["id"]));
  120. }
  121. $dh_pali->commit();
  122. if (!$stmt_weight || ($stmt_weight && $stmt_weight->errorCode() != 0)) {
  123. $error = $dh_pali->errorInfo();
  124. echo "error - $error[2]";
  125. } else {
  126. //echo "修改数据库成功 book={$i} paragraph={$j} \n";
  127. }
  128. }
  129. } else {
  130. echo "无法获取段落最大值";
  131. }
  132. echo "处理时间 :" . (microtime(true) - $time_start). "\n";
  133. }