Update specific field on SOLR index

问题:

I want to using solr for search on articles

I have 3 table:

1. Group (id , group name)
2. ArticleBase (id, groupId, some other field)
3. Article(id, articleBaseId, title, date, ...)

in solr schema.xml file i just define all article field that mixed with ArticleBase table (for use one index on solr) like this: (id, articleBaseId, groupId, ...)

problem: Admin want to change group (ArticleBase), therefore i must update (or replace) all indexed article in solr. right ?
can i update groupId only in solr index ?

have any solution ?

Note:Article table contains more than 200 million article, and i using solr for index only (not store any field data except article id)

回答1:

Solr does not support updating individual fields yet, but there is a JIRA issue about this (almost 3 years old as of this writing).

Until this is implemented, you have to update the whole document.

UPDATE: as of Solr 4+ this is implemented, here's the documentation.

回答2:

Please refer to this document about the "Partial Documents Update" Feature in Solr 4.0

Solr 4.0 is now final and production-ready.

This feature makes it possible to update fields and even adding values to multiValued fields.

Mauricio was right with his answer back in 2010, but this is the way things are today.

回答3:

SolrPHP doesn't provide any method to update a specific field in Solr.

However, you can make a Curl call in PHP to update a specific field:

<?php
// Update array
$update = array(
    'id' => $docId,
    $solrFieldName => array(
        'set' => $solrFieldValue
    )
);
$update = json_encode(array($update));

// Create curl resource and URL
$ch = curl_init('http://'.SOLR_HOSTNAME.':'.SOLR_PORT.'/'.SOLR_COLLECTION.'/update?commit=true');

// Set Login/Password auth (if required)
curl_setopt($ch, CURLOPT_USERPWD, SOLR_LOGIN.':'.SOLR_PASSWORD);

// Set POST fields
curl_setopt($ch, CURLOPT_POST,true);
curl_setopt($ch, CURLOPT_POSTFIELDS, $update);

// Return transfert
curl_setopt($ch, CURLOPT_RETURNTRANSFER, 1);

// Set type of data sent
curl_setopt($ch, CURLOPT_HTTPHEADER, array('Content-Type:application/json'));

// Get response result
$output = json_decode(curl_exec($ch));

// Get response code
$responseCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);

// Close Curl resource
curl_close($ch);

if ($responseCode == 200)
{
    echo 'SOLR: Updated successfully field '.$solrFieldName.' for id:'.$docId.' (query time: '.$output->responseHeader->QTime.'ms).';
}
else
{
    echo ('SOLR: Can\'t update field '.$solrFieldName.' for id:'.$docId.', response ('.$responseCode.') is: '.print_r($output,true));
}

I use this code to update in JSON, you can also provide data in XML.

回答4:

My Solution was something as below:

$client = new SolrClient($options);
$query = new SolrQuery();
// Find old Document
$query->setQuery('id:5458');
$query->setStart(0);
$query->setRows(1);
$query_response = $client->query($query);
// I had to set the parsemode to PARSE_SOLR_DOC
$query_response->setParseMode(SolrQueryResponse::PARSE_SOLR_DOC);
$response = $query_response->getResponse();
$doc = new SolrInputDocument();
// used the getInputDocument() to get the old document from the query
$doc = $response->response->docs[0]->getInputDocument();
if ($response->response->numFound) {
    $second_doc = new SolrInputDocument();
    $second_doc->addField('cat', "category123");
// Notice I removed the second parameter from the merge()
    $second_doc->merge($doc);
    $updateResponse = $client->addDocument($second_doc);
    $client->commit();
}