上一篇中講到lucene在合併倒排表時候的交集演算法操作.本文繼續對倒排表求並集的演算法:

lucene處理交集時採用的資料結構是一個倒排表的陣列,陣列的元素是一個個的迭代器來表現每個倒排表.

而在求並集的時候則是採用了佇列資料結構.在DisjunctionSumScorer類的建構函式中對佇列進行了初始化操作:

Iterator si = subScorers.iterator(); scorerDocQueue = new ScorerDocQueue(nrScorers); while (si.hasNext()) { Scorer se = (Scorer) si.next(); if (se.nextDoc() != NO_MORE_DOCS) { // doc() method will be used in scorerDocQueue. scorerDocQueue.insert(se); } } 上述程式碼將一個個倒排表加入佇列中,ScorerDocQueue則是一般的佇列實現.

lucene在查詢的時候,BooleanScorer2呼叫nextDoc()會呼叫DisjunctionSumScorer類中一個非常重要的方法:advanceAfterCurrent()

do { // repeat until minimum nr of matchers currentDoc = scorerDocQueue.topDoc(); currentScore = scorerDocQueue.topScore(); nrMatchers = 1; do { // Until all subscorers are after currentDoc if (!scorerDocQueue.topNextAndAdjustElsePop()) { if (scorerDocQueue.size() == 0) { break; // nothing more to advance, check for last match. } } if (scorerDocQueue.topDoc() != currentDoc) { break; // All remaining subscorers are after currentDoc. } currentScore += scorerDocQueue.topScore(); nrMatchers++; } while (true); if (nrMatchers >= minimumNrMatchers) { return true; } else if (scorerDocQueue.size() < minimumNrMatchers) { return false; } } while (true); 上述程式碼段使用兩個do...while ,最後得到滿足最少N個條件的文件

int i = 1; HeapedScorerDoc node = heap[i]; // save top node int j = i << 1; // find smaller child int k = j + 1; if ((k <= size) && (heap[k].doc < heap[j].doc)) { j = k; } while ((j <= size) && (heap[j].doc < node.doc)) { heap[i] = heap[j]; // shift up child i = j; j = i << 1; k = j + 1; if (k <= size && (heap[k].doc < heap[j].doc)) { j = k; } } heap[i] = node; // install saved node topHSD = heap[1];

      當nextDoc()到最後一個doc時,說明並集操作結束

      注:因為差集相對簡單,不再累述