Lucene 1.9 多目录搜索的的一个bug
2008-03-11 14:41
501 查看
这个问题解决有一段时间了。刚才在Lucene的群有朋友提问了这个问题。所以方便其他朋友遇到此问题时,好参考下。特贴出来
在需要搜索多个索引目录,多个字段的时候,发现有错误,提示的大概意思是(具体忘记了),关键词已经存在字典中。研究Lucene的源代码发现,是Lucene.net 1.9 rc1的一个bug :
具体代码在Lucene.Net.Search命名空间下的Query类的Combine method :
修改前的代码如下:
public virtual Query Combine(Query[] queries)
{
System.Collections.Hashtable uniques = new System.Collections.Hashtable();
for (int i = 0; i < queries.Length; i++)
{
Query query = queries[i];
BooleanClause[] clauses = null;
// check if we can split the query into clauses
bool splittable = (query is BooleanQuery);
if (splittable)
{
BooleanQuery bq = (BooleanQuery) query;
splittable = bq.IsCoordDisabled();
clauses = bq.GetClauses();
for (int j = 0; splittable && j < clauses.Length; j++)
{
splittable = (clauses[j].GetOccur() == BooleanClause.Occur.SHOULD);
}
}
if (splittable)
{
for (int j = 0; j < clauses.Length; j++)
{
Query tmp = clauses[j].GetQuery();
uniques.Add(tmp, tmp);
}
}
else
{
uniques.Add(query, query);
}
}
// optimization: if we have just one query, just return it
if (uniques.Count == 1)
{
return (Query) uniques.GetEnumerator().Current;
}
System.Collections.IEnumerator it = uniques.GetEnumerator();
BooleanQuery result = new BooleanQuery(true);
while (it.MoveNext())
{
result.Add((Query) it.Current, BooleanClause.Occur.SHOULD);
}
return result;
}
修改后的代码(飘红的地方是更改过的):
public virtual Query Combine(Query[] queries)
{
System.Collections.Hashtable uniques = new System.Collections.Hashtable();
for (int i = 0; i < queries.Length; i++)
{
Query query = queries[i];
BooleanClause[] clauses = null;
// check if we can split the query into clauses
bool splittable = (query is BooleanQuery);
if (splittable)
{
BooleanQuery bq = (BooleanQuery) query;
splittable = bq.IsCoordDisabled();
clauses = bq.GetClauses();
for (int j = 0; splittable && j < clauses.Length; j++)
{
splittable = (clauses[j].GetOccur() == BooleanClause.Occur.SHOULD);
}
}
if (splittable)
{
for (int j = 0; j < clauses.Length; j++)
{
Query tmp = clauses[j].GetQuery();
uniques.Add(tmp, tmp);
}
}
else
{
//uniques.Add(query, query);
//modify by kwklover 2008/1/17 bug fixed for 防止加入多个重复的字典项
if (uniques.Contains(query) == false)
{
uniques.Add(query, query);
}
}
}
// optimization: if we have just one query, just return it
if (uniques.Count == 1)
{
//return (Query) uniques.GetEnumerator().Current;
//modify by kwklover 2008/1/17
System.Collections.IDictionaryEnumerator iter = uniques.GetEnumerator();
iter.MoveNext();
return iter.Value as Query;
}
System.Collections.IEnumerator it = uniques.GetEnumerator();
BooleanQuery result = new BooleanQuery(true);
while (it.MoveNext())
{
result.Add((Query) it.Current, BooleanClause.Occur.SHOULD);
}
return result;
}
另外在Lucene 2.0版已经修复了此问题。不过Lucene 2.0 和 Lucene 1.9的 API有比较大的变化
在需要搜索多个索引目录,多个字段的时候,发现有错误,提示的大概意思是(具体忘记了),关键词已经存在字典中。研究Lucene的源代码发现,是Lucene.net 1.9 rc1的一个bug :
具体代码在Lucene.Net.Search命名空间下的Query类的Combine method :
修改前的代码如下:
public virtual Query Combine(Query[] queries)
{
System.Collections.Hashtable uniques = new System.Collections.Hashtable();
for (int i = 0; i < queries.Length; i++)
{
Query query = queries[i];
BooleanClause[] clauses = null;
// check if we can split the query into clauses
bool splittable = (query is BooleanQuery);
if (splittable)
{
BooleanQuery bq = (BooleanQuery) query;
splittable = bq.IsCoordDisabled();
clauses = bq.GetClauses();
for (int j = 0; splittable && j < clauses.Length; j++)
{
splittable = (clauses[j].GetOccur() == BooleanClause.Occur.SHOULD);
}
}
if (splittable)
{
for (int j = 0; j < clauses.Length; j++)
{
Query tmp = clauses[j].GetQuery();
uniques.Add(tmp, tmp);
}
}
else
{
uniques.Add(query, query);
}
}
// optimization: if we have just one query, just return it
if (uniques.Count == 1)
{
return (Query) uniques.GetEnumerator().Current;
}
System.Collections.IEnumerator it = uniques.GetEnumerator();
BooleanQuery result = new BooleanQuery(true);
while (it.MoveNext())
{
result.Add((Query) it.Current, BooleanClause.Occur.SHOULD);
}
return result;
}
修改后的代码(飘红的地方是更改过的):
public virtual Query Combine(Query[] queries)
{
System.Collections.Hashtable uniques = new System.Collections.Hashtable();
for (int i = 0; i < queries.Length; i++)
{
Query query = queries[i];
BooleanClause[] clauses = null;
// check if we can split the query into clauses
bool splittable = (query is BooleanQuery);
if (splittable)
{
BooleanQuery bq = (BooleanQuery) query;
splittable = bq.IsCoordDisabled();
clauses = bq.GetClauses();
for (int j = 0; splittable && j < clauses.Length; j++)
{
splittable = (clauses[j].GetOccur() == BooleanClause.Occur.SHOULD);
}
}
if (splittable)
{
for (int j = 0; j < clauses.Length; j++)
{
Query tmp = clauses[j].GetQuery();
uniques.Add(tmp, tmp);
}
}
else
{
//uniques.Add(query, query);
//modify by kwklover 2008/1/17 bug fixed for 防止加入多个重复的字典项
if (uniques.Contains(query) == false)
{
uniques.Add(query, query);
}
}
}
// optimization: if we have just one query, just return it
if (uniques.Count == 1)
{
//return (Query) uniques.GetEnumerator().Current;
//modify by kwklover 2008/1/17
System.Collections.IDictionaryEnumerator iter = uniques.GetEnumerator();
iter.MoveNext();
return iter.Value as Query;
}
System.Collections.IEnumerator it = uniques.GetEnumerator();
BooleanQuery result = new BooleanQuery(true);
while (it.MoveNext())
{
result.Add((Query) it.Current, BooleanClause.Occur.SHOULD);
}
return result;
}
另外在Lucene 2.0版已经修复了此问题。不过Lucene 2.0 和 Lucene 1.9的 API有比较大的变化
相关文章推荐
- Go 1.5关键字搜索目录、文件、文件内容_修复一个小BUG
- 在学习autotools遇到一个问题:make没有按照VPATH给的目录进行搜索,请大家指导一下。
- Lucene 4.0 BytesRefHash的一个bug
- 发现了lucene一个bug
- lucene 中文分词器中的一个Bug
- 一个Lucene 2.4.0对搜索结果高亮显示的代码
- 一个lucene搜索高亮显示的例子。
- 用Lucene做一个简单的Java搜索工具
- IntelliJ IDEA Community版独特的一个bug:某些目录下的文件能显示但不被编译
- CURL库FTP传输文件,当ftpServer端默认登录的目录不是根目录/时,有一个bug会产生
- 一个emacs扩展,从指定目录或文件中搜索指定关键字
- 搜索系统13:一个查询从solr到lucene都干了些什么事?
- 采用多线程和生产者消费者模式来实现对于一个目录以及所有子目录下的文件内容的搜索,打印出包含指定关键字的行.
- 搜索和替换文件或目录的一个好类--很实用
- 一个辅助Lucene对数字类型进行索引和搜索的工具类
- lucene 中文分词器中的一个Bug
- Go 1.5keyword搜索文件夹、文件、文件内容_修复一个小BUG
- 一个用Lucene3.0 的搜索项目设计总体图
- 一个搜索引擎搜索出来的目录与其内容不同步的实例
- lucene.net多目录搜索及合并索引