带有XPathEntityProcessor的SOLR DataImportHandler

Question

当我使用DIH XPathEntityProcessor索引Resource XML文件时，应该有2个solr文件

1. ID为1000的链接，带有2个标签ABC和DEF
2. id为2000的链接有3个标签GHI，JKL和MNO

Solr版本：4.10.2

问题： 我无法正确索引数据。

预期产出：

{
    "id": "1000",
    "field_name": "val1",
    "ABC": "ABC_VALUE",
    "DEF": "DEF_VALUE"
},
{
    "id": "2000",
    "field_name": "val2",
    "GHI": "GHI_VALUE",
    "JKL": "JKL_VALUE",
    "MNO": "MNO_VALUE"
}

=============================================== ==========================

资源XML：

<RESOURCE>
    <LINK ID="1000">
        <FIELD>val1</FIELD>
        <TAG>
            <TAG_CODE>ABC</TAG_CODE>
            <TAG_VALUE>ABC_VALUE</TAG_VALUE>
        </TAG>
        <TAG>
            <TAG_CODE>DEF</TAG_CODE>
            <TAG_VALUE>DEF_VALUE</TAG_VALUE>
        </TAG>
    </LINK>
    <LINK ID="2000">
        <FIELD>val2</FIELD>
        <TAG>
            <TAG_CODE>GHI</TAG_CODE>
            <TAG_VALUE>GHI_VALUE</TAG_VALUE>
        </TAG>
        <TAG>
            <TAG_CODE>JKL</TAG_CODE>
            <TAG_VALUE>JKL_VALUE</TAG_VALUE>
        </TAG>
        <TAG>
            <TAG_CODE>MNO</TAG_CODE>
            <TAG_VALUE>MNO_VALUE</TAG_VALUE>
        </TAG>
    </LINK> 
</RESOURCE>

=============================================== ======================

DataConfig XML（TRY 1）：

<dataConfig>
    <script><![CDATA[
        function f1(row) {
            var code = row.get("TAG_CODE");
            var val = row.get("TAG_VALUE");

            row.put(code, val);
            row.remove("TAG_CODE");
            row.remove("TAG_VALUE");
            return row;
        }
    ]]></script>
    <dataSource type="URLDataSource" />
    <document>
        <entity name="testdata" url="http://host:port/uri"
                processor="XPathEntityProcessor" forEach="/RESOURCE/LINK">
            <field column="id" xpath="/RESOURCE/LINK/@ID" />    
            <field column="field_name" xpath="/RESOURCE/LINK/FIELD" />
            <entity name="testdata" url="http://host:port/uri"
                processor="XPathEntityProcessor" forEach="/RESOURCE/LINK/TAG" transformer="script:f1">
                <field column="TAG_CODE" xpath="/RESOURCE/LINK/TAG/TAG_CODE" />
                <field column="TAG_VALUE" xpath="/RESOURCE/LINK/TAG/TAG_VALUE" />
            </entity>
        </entity>
    </document>
</dataConfig>

Output:
    {
        "id": "1000",
        "field_name": "val1",
        "ABC": "ABC_VALUE",
        "DEF": "DEF_VALUE",
        "GHI": "GHI_VALUE",
        "JKL": "JKL_VALUE",
        "MNO": "MNO_VALUE"
    },
    {
        "id": "2000",
        "field_name": "val2",
        "ABC": "ABC_VALUE",
        "DEF": "DEF_VALUE",
        "GHI": "GHI_VALUE",
        "JKL": "JKL_VALUE",
        "MNO": "MNO_VALUE"
    }

=============================================== =================

DataConfig XML（TRY 2）：

<dataConfig>
    <script><![CDATA[
        function f1(row) {
            var code = row.get("TAG_CODE");
            var val = row.get("TAG_VALUE");

            row.put(code, val);
            row.remove("TAG_CODE");
            row.remove("TAG_VALUE");
            return row;
        }
    ]]></script>
    <dataSource type="URLDataSource" />
    <document>
        <entity name="testdata" url="http://host:port/uri"
                processor="XPathEntityProcessor" forEach="/RESOURCE/LINK">
            <field column="id" xpath="/RESOURCE/LINK/@ID" />    
            <field column="field_name" xpath="/RESOURCE/LINK/FIELD" />
            <entity name="testdata" url="http://host:port/uri"
                processor="XPathEntityProcessor" forEach="/RESOURCE/LINK[@ID=${testdata.id}]/TAG" transformer="script:f1">
                <field column="TAG_CODE" xpath="/RESOURCE/LINK/TAG/TAG_CODE" />
                <field column="TAG_VALUE" xpath="/RESOURCE/LINK/TAG/TAG_VALUE" />
            </entity>
        </entity>
    </document>
</dataConfig>

Output:
    {
        "id": "1000",
        "field_name": "val1"        
    },
    {
        "id": "2000",
        "field_name": "val2"        
    }

=============================================== =================

DataConfig XML（TRY 3）：

<dataConfig>
    <script><![CDATA[
        function f1(row) {
            var code = row.get("TAG_CODE");
            var val = row.get("TAG_VALUE");

            row.put(code, val);
            row.remove("TAG_CODE");
            row.remove("TAG_VALUE");
            return row;
        }
    ]]></script>
    <dataSource type="URLDataSource" />
    <document>
        <entity name="testdata" url="http://host:port/uri"
                processor="XPathEntityProcessor" forEach="/RESOURCE/LINK">
            <field column="id" xpath="/RESOURCE/LINK/@ID" />    
            <field column="field_name" xpath="/RESOURCE/LINK/FIELD" />
            <entity name="testdata" url="http://host:port/uri"
                processor="XPathEntityProcessor" forEach="/RESOURCE/LINK[@ID=${testdata.id}]/TAG" transformer="script:f1">
                <field column="TAG_CODE" xpath="/RESOURCE/LINK[@ID=${testdata.id}]/TAG/TAG_CODE" />
                <field column="TAG_VALUE" xpath="/RESOURCE/LINK[@ID=${testdata.id}]/TAG/TAG_VALUE" />
            </entity>
        </entity>
    </document>
</dataConfig>

Output:
    {
        "id": "1000",
        "field_name": "val1"        
    },
    {
        "id": "2000",
        "field_name": "val2"        
    }