This repo is really old, and it's the first Go I ever wrote. It's probably pretty vile (I'm sure if I looked back at it now, I'd grimace a lot). Proceed with caution...
An Apache Solr library written in Go, which is my first Go project! Functionality includes:
- Select queries
- Raw Select tqueries - useful for more complex queries such as Function Queries
- Update queries (add/replace/delete)
- Faceting
For more information on Solr itself, please refer to Solr's wiki.
This library is released under the "do whatever you like" license.
Example programs can be found in the examples
folder here.
Import the solr
package (it is assumed you know how to build/install it, if not, see here) and create a "connection" to your solr server by calling the solr.Init(hostname, port int)
function supplying a hostname and port.
// connect to server running on localhost port 8983
s, err := solr.Init("localhost", 8983)
Select queries are performed using the solr.Select(q *Query)
method, passing it a pointer to a Query
type.
Here's an example:
q := solr.Query{
Params: solr.URLParamMap{
"q": []string{"id:31"},
"facet.field": []string{"some_field", "some_other_field"},
"facet": "true",
},
Rows: 10,
Sort: "title ASC",
}
Here we have defined a set of URL parameters - q
, facet.field
, facet
, rows
and sort
using the solr.Query{}
struct. Under the hood this would work out as the following Solr query string:
GET http://localhost:8983/solr/select?q=id:31&facet.field=some_field&facet.field=some_other_field&facet=true
Notice that facet_field
, like q
, is an array of strings and appears multiple times in the resulting query string (shown above)
Performing a query using our solr.Query()
is simple and shown below
res, err := s.Select(&q)
if err != nil {
fmt.Println(err)
}
// ...
A pointer to q
is passed to s.Select()
, and returned is a pointer to a SelectResponse
(res
) and an error
(err
) if an error occurred.
Iterating over the results is shown later in this document.
rsty/solr
supports raw queries where you can specify your exact query in string form. This is useful for specifying complex queries where a Query
type would be cumbersome. Raw queries are performed as follows:
q := "q={!func}add($v1,$v2)&v1=sqrt(popularity)&v2=100.0" // a solr query
res, err := s.SelectRaw(q)
if err != nil {
// handle error here
}
// ...
In other words, under the hood the following query will have been performed:
GET http://localhost:8983/solr/select?q={!func}add($v1,$v2)&v1=sqrt(popularity)&v2=100.0
As with solr.Select()
, solr.SelectRaw()
returns a pointer to a SelectResponse
and an error, err
.
Responses to select queries (solr.Select()
and solr.RawSelect()
) come in the form of pointers to SelectResponse
types. A SelectResponse
wraps a Solr response with a convenient interface. The following few paragraphs and sections describe the various parts of a SelectResponse
object
A pointer to a SelectResponse
and an error are returned from calls to solr.Select()
and solr.SelectRaw()
. A SelectResponse
mimics a Solr response and therefore has the following attributes:
Results
- a pointer to aDocumentCollection
(more on this later) which contains the documents returned by SolrStatus
- querystatus
indicator as returned by SolrQTime
-QTime
value as returned by Solr
More information on Status
and QTime
can be found here.
A DocumentCollection
wraps up a set of Document
s providing a convenient interface to them.
DocumentCollection
supports the following methods:
Len() int
- returns the length (int) of theDocument
s returnedGet(i int) *Document
- returns a pointer to the document at positioni
within the Collection
DocumentCollection
has the following properties:
NumFound
- the total number of results solr matched to your query (irrespective of the amount returned)Facets
- an array ofFacet
objectsNumFacets
- the number of facet fields returned (if any)
Document
s implement the following methods:
Field(field_name string) interface{}
- returns the value of the field, specified byfield_name
If your select query specifies facets, facets will be found under Response.Results.Facets
which is an array of Facet
s. A Facet
has the following attributes
Name
- the name of the facet (field) as returned by SolrCounts
- an array ofFacetCount
s, the corresponding value counts for the field.
A FacetCount
has the following attributes
Value
- the facet field valueCount
- the count (int) for the field value
Below is an example showing an iteration over a collection of Facet
s
// q is assumed to have been set up
// perform the query
res, err := s.Query(&q)
// handle error, err, here
results := res.Results
for i := 0; i < results.NumFacets; i++ {
facet := results.Facets[i]
fmt.Println("Facet:", facet.Name)
k := len(facet.Counts)
for j := 0; j < k; j++ {
fmt.Println(facet.Counts[j].Value, "=>", facet.Counts[j].Count)
}
fmt.Println("")
}
This might output the following:
Facets
------
Facet: category
cameras => 1
Facet: type
digital_slr => 10
compact => 2
Update queries are used to add, replace or delete documents in Solr's index. Please see the Solr Wiki for more information. Go-Solr uses JSON for update queries, not XML. Solr3.1 will need to be configured to support JSON for update messages, Solr 4.0+ supports JSON natively via /update
.
solr.Update(document map[string]interface{}, commit bool)
takes two arguments, an "update document" and a commit flag (boolean) which specifies whether or not a commit should be performed at the same time as the update is performed. An example may look like the following
q, err := solr.Update(document, true);
if err != nil {
// ...
}
An update document must be of type map[string]interface{}
, and may look like the following:
doc := map[string]interface {}{
"add":[]interface {}{
map[string]interface {}{"id": 22, "title": "abc"},
map[string]interface {}{"id": 23, "title": "def"},
map[string]interface {}{"id": 24, "title": "def"},
},
}
... which is equivalent to the following JSON:
{"add": [{"id": 22, "title": "abc"}, {"id": 23, "title": "def"}, {"id": 24, "title": "def"}]}
... which is an Update which adds (or replaces) 3 documents in a fictional Solr index.
You can define any type of document to send off to Solr in an update. Support will be added later to allow raw JSON strings to be used in Updates.
solr.Update()
returns an UpdateResponse
and an error
. UpdateResponse
has a Success
(bool) property.