Lucene的DocFieldProcessor类

DocFieldProcessor类的任务
1 按顺序存储所有的field和对应的fieldinfo
2 为当前这篇doc的field按照fieldname来建立hash索引
3 调用InvertedDocConsumer类（抽象），对field的内容分词和建立内存索引

DocFieldProcessor类主要用到的其他类
DocFieldProcessorPerField类的对象负责存放一个fieldinfo和其对应的field（可能是多个，他们的fieldinfo相同），next成员可以指向下一个DocFieldProcessorPerField类对象，构成链表（用于解决fieldhash冲突）
DocInverterPerField是DocFieldProcessorPerField内用到的类，负责解析同一个fieldinfo里的field，建立索引

final class DocFieldProcessor extends DocConsumer {

  final DocFieldConsumer consumer;
  final StoredFieldsConsumer storedConsumer;
  final Codec codec;

  // 存储doc里所有的DocFieldProcessorPerField（这个类里面会存放多个fieldinfo相同的field），位置按顺序存储，一个位置里面放一个
  DocFieldProcessorPerField[] fields = new DocFieldProcessorPerField[1];
  int fieldCount;

  // 存储doc里所有的DocFieldProcessorPerField（这个类里面会存放多个fieldinfo相同的field），位置按fieldname的hash来存放，一个位置可能匹配多个，以链表结构存放
  DocFieldProcessorPerField[] fieldHash = new DocFieldProcessorPerField[2];
  int hashMask = 1;
  int totalFieldCount;

  int fieldGen;
//浅拷贝 要分析的doument
  final DocumentsWriterPerThread.DocState docState;

  final Counter bytesUsed;

 

DocFieldProcessor ::processDocument(FieldInfos.Builder fieldInfos)是DocFieldProcessor类主要功能的实现函数

  public void processDocument(FieldInfos.Builder fieldInfos) throws IOException {

    consumer.startDocument();
    storedConsumer.startDocument();

    fieldCount = 0;

    final int thisFieldGen = fieldGen++;

//循环doc里的field，合并同fieldinfo的field进入一个DocFieldProcessorPerField，并且如果此轮循环有新创建的DocFieldProcessorPerField对象，则加入到顺序数组和hash数组里面
    for(IndexableField field : docState.doc) {
      final String fieldName = field.name();

      //计算fieldname在hash数组里对应的位置
      final int hashPos = fieldName.hashCode() & hashMask;

      DocFieldProcessorPerField fp = fieldHash[hashPos];
     //查找对应位置下fieldname符合当前field的DocFieldProcessorPerField对象
      while(fp != null && !fp.fieldInfo.name.equals(fieldName)) {
        fp = fp.next;
      }

      if (fp == null) {
        //表示该fieldinfo是第一次出现，把field的info信息添加到fieldinfos里面，并且创建对应的fieldinfo对象
        FieldInfo fi = fieldInfos.addOrUpdate(fieldName, field.fieldType());
        //创建一个DocFieldProcessorPerField
        fp = new DocFieldProcessorPerField(this, fi);
       
        //加入到hash链表里面
        fp.next = fieldHash[hashPos];
        fieldHash[hashPos] = fp;
        totalFieldCount++;

        //hash表快满了，重新做hash
        if (totalFieldCount >= fieldHash.length/2) {
          rehash();
        }
      } else {
      //更新fieldinfo对象
        FieldInfo fi = fieldInfos.addOrUpdate(fieldName, field.fieldType());
        assert fi == fp.fieldInfo : "should only have updated an existing FieldInfo instance";
      }

      if (thisFieldGen != fp.lastGen) {
      //说明这个fieldinfo是第一次出现，即本次循环new了DocFieldProcessorPerField
        fp.fieldCount = 0;

        //判断顺序数组是不是满了
        if (fieldCount == fields.length) {
          final int newSize = fields.length*2;
          DocFieldProcessorPerField newArray[] = new DocFieldProcessorPerField[newSize];
          System.arraycopy(fields, 0, newArray, 0, fieldCount);
          fields = newArray;
        }

        fields[fieldCount++] = fp;
        fp.lastGen = thisFieldGen;
      }
     //把field加入到fp中，同fieldinfo的field会加入到同一个fp里面
      fp.addField(field);
     
      storedConsumer.addField(docState.docID, field, fp.fieldInfo);
    }
    //遍历顺序数组里的DocFieldProcessorPerField，调用DocInverterPerField对其里面的field进行分词建立索引之类的操作
    ArrayUtil.introSort(fields, 0, fieldCount, fieldsComp);
    for(int i=0;i<fieldCount;i++) {
      final DocFieldProcessorPerField perField = fields[i];
      perField.consumer.processFields(perField.fields, perField.fieldCount);
    }
  }