word_index_weight_refresh.php 5.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150
  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_EXCEPTION);
  27. $dh_pali = new PDO( _PG_DB_PALI_INDEX_, _DB_USERNAME_, _DB_PASSWORD_);
  28. $dh_pali->setAttribute(PDO::ATTR_ERRMODE, PDO::ERRMODE_EXCEPTION);
  29. echo "from=$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. try {
  35. //code...
  36. $stmt = $dh_pali->prepare($query);
  37. $stmt->execute(array($i));
  38. $row = $stmt->fetch(PDO::FETCH_NUM);
  39. }catch(PDOException $e){
  40. fwrite(STDERR,"error:".$e->getMessage());
  41. continue;
  42. }
  43. if ($row) {
  44. $max_para = $row[0];
  45. echo "段落数量:$max_para ";
  46. for ($j = 0; $j <= $max_para; $j++) {
  47. # code...
  48. $query = "SELECT id,book,wordindex,bold from "._PG_TABLE_WORD_." where book={$i} and paragraph={$j} order by id ASC";
  49. $stmt = $dh_pali->query($query);
  50. $fetch = $stmt->fetchAll(PDO::FETCH_ASSOC);
  51. $query = "SELECT wordindex,count(*) as co from "._PG_TABLE_WORD_." where book={$i} and paragraph={$j} group by wordindex";
  52. $stmt = $dh_pali->query($query);
  53. $fetch_voc = $stmt->fetchAll(PDO::FETCH_ASSOC);
  54. $vocabulary = array();
  55. foreach ($fetch_voc as $key => $value) {
  56. $vocabulary[$value["wordindex"]] = $value["co"];
  57. }
  58. for ($iWord = 0; $iWord < count($fetch); $iWord++) {
  59. # 非黑体字
  60. if ($fetch[$iWord]["bold"] == 0) {
  61. $count = $vocabulary[$fetch[$iWord]["wordindex"]];
  62. $paraWeight = pow(1.01, $count); //总分
  63. if ($paraWeight > 1.9) {
  64. $paraWeight = 1.9;
  65. }
  66. $weight = $paraWeight / $count;
  67. } else {
  68. #黑体字
  69. #查找前后相连的黑体字
  70. $begin = $iWord;
  71. while ($fetch[$begin]["bold"] == 1) {
  72. $begin--;
  73. if ($begin < 0) {
  74. break;
  75. }
  76. }
  77. $begin = $begin + 1;
  78. $end = $iWord;
  79. while ($fetch[$end]["bold"] == 1) {
  80. $end++;
  81. if ($end > count($fetch) - 1) {
  82. break;
  83. }
  84. }
  85. $end = $end - 1;
  86. $bold_count = $end - $begin + 1;
  87. if ($bold_count == 1) {
  88. $query = "SELECT * from "._PG_TABLE_WORD_INDEX_." where id=" . $fetch[$iWord]["wordindex"];
  89. $stmt_word = $dh_word->query($query);
  90. $wordinfo = $stmt_word->fetch(PDO::FETCH_ASSOC);
  91. $bookId = (int) $fetch[$iWord]["book"];
  92. if (mb_substr($wordinfo["word"], -2) == "ti") {
  93. $weight = 100 + $book_weight[$bookId];
  94. } else {
  95. $weight = 100 + $book_weight[$bookId];
  96. }
  97. //echo "单独黑体 $weight \n";
  98. } else {
  99. #连续黑体字
  100. //echo "连续黑体字";
  101. $len_sum = 0;
  102. $len_curr = 0;
  103. for ($iBold = $begin; $iBold <= $end; $iBold++) {
  104. # code...
  105. $boldid = $fetch[$iBold]["wordindex"];
  106. $query = "SELECT len from "._PG_TABLE_WORD_INDEX_." where id=" . $boldid;
  107. $stmt_bold = $dh_word->query($query);
  108. $wordbold = $stmt_bold->fetch(PDO::FETCH_ASSOC);
  109. $len_sum += $wordbold["len"];
  110. if ($iBold == $i) {
  111. $len_curr = $wordbold["len"];
  112. }
  113. }
  114. $weight = 10 + $len_curr / $len_sum;
  115. }
  116. }
  117. //echo $weight."\n";
  118. $fetch[$iWord]["weight"] = (int) ($weight * 100);
  119. }
  120. # 将整段权重写入据库
  121. $dh_pali->beginTransaction();
  122. $query = "UPDATE "._PG_TABLE_WORD_." set weight = ? where id=? ";
  123. $stmt_weight = $dh_pali->prepare($query);
  124. foreach ($fetch as $key => $value) {
  125. $stmt_weight->execute(array($value["weight"], $value["id"]));
  126. }
  127. $dh_pali->commit();
  128. if (!$stmt_weight || ($stmt_weight && $stmt_weight->errorCode() != 0)) {
  129. $error = $dh_pali->errorInfo();
  130. echo "error - $error[2]";
  131. } else {
  132. //echo "修改数据库成功 book={$i} paragraph={$j} \n";
  133. }
  134. }
  135. } else {
  136. echo "无法获取段落最大值";
  137. }
  138. echo "处理时间 :" . (microtime(true) - $time_start). "\n";
  139. }