在Solr（J）中搜索嵌套文档 [英] Searching for nested Documents in Solr(J)
问题描述
我为PDF文档实现了一个简单的添加工具。我首先创建所有文档的主文档（SolrInputDocument）。它获取了诸如author，filehash，keywords，'content_type = document'等字段。
I implemented a simple adding tool for PDF documents. I first create a master document (SolrInputDocument) of all documents. It gets fields like author, filehash, keywords, 'content_type=document' and so on.
之后我为每个页面生成一个SolrInputDocument，这个对象得到一个像'parentID_p01'的id，页面作为字段值，'content_type = page'usw。
After that I generate a SolrInputDocument for every page, this object gets an id like 'parentID_p01', the page as a field value, 'content_type=page' usw.
最后，我使用addChildDocument（）将所有页面文档添加到我的主文档中。
Finally, I add all page documents to my master document with addChildDocument()..
现在我的问题是，如何在所有文档的所有页面中搜索给定的单词，例如：
Now my question is, how do I perform a search for a given word inside all pages of all documents for example like this:
Document1.pdf 'this is my doc1 title' [2 matches]
[Page 14] 'Example phrase for special a <em>word</em> given by....
[Page 55] 'another <em>word</em> for this test
Document2.pdf 'doc2 title' [X matches]
[Page 1] 'given <em>word</em> in this text
[Page 2] '<em>words</em> hit more than fists
[Page 99] 'some <em>words</em> of wisdom
我的第一个想法是使用' text：word~ '然后按父文档分组，但我找不到一个好方法： - （
My first idea was to simple search with 'text:word~' and then group by the parent document but I didn't find a good way :-(
看来嵌套文档对solr来说有点新，我没有找不到SolrJ的简单解决方案。
It seems that nested documents are a little new to solr and I didn't find an easy solution with SolrJ.
提前谢谢
推荐答案
我已经用父子关系创建了以下格式的solr数据。其中一个insurance_accounts有多个人的车辆保险。一个人可以有多辆车，如汽车，自行车等。我把人作为父母和车辆作为儿童文件。
I have created data on solr in below format with parent child relation. Where one insurance_accounts have multiple person's vehicle insurance. One person can have multiple vehicles like car, bike etc. I have taken person as parent and vehicle as child document.
<insurance_accounts>
<person>
<firstname>John</firstname>
<lastname>Jones</lastname>
...other details...
<cars>
<car>
<make>honda</make>
<model>accord</model>
</car>
<car>
<make>Nissan</make>
<model>Maxima</model>
</car>
</cars>
<bikes>
<bike>
<make>yamaha</make>
<model>passion</model>
</bike>
<bike>
<make>Peugeot</make>
<model>Vivacity</model>
</bike>
</bikes>
</person>
<person>
...
</person>
</insurance_accounts>
在下面的java代码中，我使用solrj 4.9创建文档并在SOLR上运行搜索查询。我已经处理了QueryResponse以显示所需的结果，并且还给出了solr查询URL。
In below java code, I have used solrj 4.9 to create documents and run search query on SOLR. I have processed QueryResponse to show the required result and also given solr query URL.
您可以从给定的代码片段获取帮助，并告诉我它是否有效。
You can take help from the given code snippet and let me know if it works or not.
import java.util.ArrayList;
import java.util.Collection;
import java.util.List;
import java.util.Map;
import java.util.Map.Entry;
import java.util.LinkedHashMap;
import org.apache.solr.client.solrj.SolrQuery;
import org.apache.solr.client.solrj.impl.HttpSolrServer;
import org.apache.solr.client.solrj.response.QueryResponse;
import org.apache.solr.common.SolrDocument;
import org.apache.solr.common.SolrDocumentList;
import org.apache.solr.common.SolrInputDocument;
public class SolrNestedSearch {
static final String SOLR_URL = "http://localhost:8983/solr/collection1";
public static void main ( String[] args ) throws Exception {
HttpSolrServer solr = new HttpSolrServer( SOLR_URL );
// Docs to submit
Collection<SolrInputDocument> batch = new ArrayList<SolrInputDocument>();
// Parent Doc 1, a person mamed John Jones
SolrInputDocument person1 = new SolrInputDocument();
person1.addField( "id", "john_jones" );
person1.addField( "content_type", "person" );
// "_t" suffix tells Solr that it's text
person1.addField( "first_name_t", "John" );
person1.addField( "last_name_t", "Jones" );
// states and history used in edismax examples
person1.addField( "states_t", "California Nevada Idaho Maine" );
person1.addField( "history_t", "safe accident accident accident accident accident" );
// child docs, the vehicles he owns
SolrInputDocument p1_car1 = new SolrInputDocument();
p1_car1.addField( "id", "jj_car1" );
p1_car1.addField( "content_type", "car" );
// For cars "make" is an alias for "manufacturer"
p1_car1.addField( "make_t", "Honda" );
p1_car1.addField( "model_t", "Accord" );
SolrInputDocument p1_car2 = new SolrInputDocument();
p1_car2.addField( "id", "jj_car2" );
p1_car2.addField( "content_type", "car" );
p1_car2.addField( "make_t", "Nissan" );
p1_car2.addField( "model_t", "Maxima" );
SolrInputDocument p1_bike1 = new SolrInputDocument();
p1_bike1.addField( "id", "jj_bike1" );
p1_bike1.addField( "content_type", "bike" );
p1_bike1.addField( "make_t", "Yamaha" );
p1_bike1.addField( "model_t", "Passion" );
SolrInputDocument p1_bike2 = new SolrInputDocument();
p1_bike2.addField( "id", "jj_bike2" );
p1_bike2.addField( "content_type", "bike" );
p1_bike2.addField( "make_t", "Peugeot" );
p1_bike2.addField( "model_t", "Vivacity" );
// Add children to parent
person1.addChildDocument( p1_car1 );
person1.addChildDocument( p1_car2 );
person1.addChildDocument( p1_bike1 );
person1.addChildDocument( p1_bike2 );
// Add parent to batch
batch.add( person1 );
// Parent Doc 2, person mamed Satish Smith
SolrInputDocument person2 = new SolrInputDocument();
person2.addField( "id", "satish_smith" );
person2.addField( "content_type", "person" );
person2.addField( "first_name_t", "Satish" );
person2.addField( "last_name_t", "Smith" );
person2.addField( "states_t", "California Texas California Maine Vermont Connecticut" );
person2.addField( "history_t", "safe safe safe safe safe safe safe safe accident" );
// Vehicles (child docs)
SolrInputDocument p2_car1 = new SolrInputDocument();
p2_car1.addField( "id", "ss_car1" );
p2_car1.addField( "content_type", "car" );
p2_car1.addField( "make_t", "Peugeot" );
p2_car1.addField( "model_t", "iOn" );
SolrInputDocument p2_bike1 = new SolrInputDocument();
p2_bike1.addField( "id", "ss_bike1" );
p2_bike1.addField( "content_type", "bike" );
p2_bike1.addField( "make_t", "Honda" );
p2_bike1.addField( "model_t", "Spree" );
// link objects and add to batch
person2.addChildDocument( p2_car1 );
person2.addChildDocument( p2_bike1 );
batch.add( person2 );
System.out.println( "Adding batch of " + batch.size() + " parent docs" );
// Submit as a group
solr.add( batch );
solr.commit();
Map<String,String> params = new LinkedHashMap<>();
params.put( "parent_filter", "content_type:person" );
params.put( "defType", "edismax" );
params.put( "qf", "history_t states_t^100" );
doQuery(solr,
"eDismax3: Drivers matching accidents and/or California, and all of their vehicles, boost on State",
"California accident report",
"{!parent which=$parent_filter}",
"*,[child parentFilter=$parent_filter]",
params );
}
static void doQuery( HttpSolrServer solr, String description, String queryStr, String optFilter,
String optFields, Map<String,String>extraParams ) throws Exception
{
SolrQuery q = new SolrQuery( queryStr );
if ( null!=optFilter ) {
q.addFilterQuery( optFilter );
}
if ( null!=optFields ) {
q.setParam( "fl", optFields );
}
else {
q.addField( "*" );
}
if ( null!=extraParams ) {
for ( Entry<String,String> param : extraParams.entrySet() ) {
q.set( param.getKey(), param.getValue() );
}
}
// Run and show results
QueryResponse rsp = solr.query( q );
SolrDocumentList docs = rsp.getResults();
long numFound = docs.getNumFound();
System.out.println( "Matched: " + numFound );
int docCounter = 0;
for (SolrDocument doc : docs) {
docCounter++;
System.out.println( "Doc # " + docCounter );
for ( Entry<String, Object> field : doc.entrySet() ) {
String name = field.getKey();
Object value = field.getValue();
System.out.println( "\t" + name + "=" + value );
}
List<SolrDocument> childDocs = doc.getChildDocuments();
if ( null!=childDocs ) {
for ( SolrDocument child : childDocs ) {
System.out.println( "\tChild doc:" );
for ( Entry<String, Object> field : child.entrySet() ) {
String name = field.getKey();
Object value = field.getValue();
System.out.println( "\t\t" + name + "=" + value );
}
}
}
}
System.out.println( "Query URL:" );
System.out.println( SOLR_URL + "/select?" + q );
}
}
这篇关于在Solr（J）中搜索嵌套文档的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！
