aiosolr
AsyncIO Python client for Apache Solr
Requirements
This project requires Python 3.7+
Installation
To install with pip
pip install aiosolr
Usage
The connection to the Solr backend is defined during object initialization. The accepted kwargs to
init are scheme
, host
, port
, and collection
.
collection
may optionally be passed at query time
import aiosolr
client = aiosolr.Client(host=localhost, collection="example", port=8983)
Alternatively you may instantiate via passing connection_url
like:
import aiosolr
client = aiosolr.Client(connection_url="http://host:1234/path/to/solr/collection")
Once you have your aiosolr.Client
instance, set up the session:
await client.setup()
There are methods available for querying. You can use Solr's built-in get handler with the get
method to retrieve a single document:
await client.get(document_id)
You can use a pre-defined suggestions handler by using the suggestions
method:
await client.suggestions("suggest_handler", query="asdf")
You can also use the suggestions
method to build your suggestions:
await client.suggestions("suggest_handler", build=True)
handler
is a required argument for suggestions unlike for get or query
You can use the query
method to query your search handler. The default handler
used is select
.
If you would like spellcheck suggestion turned on, pass spellcheck=True
(default is False
).
await client.query(handler="my_handler", query="asdf", spellcheck=True)
If spellcheck
is True
the query method returns a tuple with the first element being an array of
documents and the 2nd element being an array of spellcheck suggestions. Otherwise, the query method
returns a simple array of documents.
You can use the update
method to access Solr's built-in update handler like:
await client.update(my_data)
At any point that you need to commit data to your collection you can use the commit
method.
Arguments should be the handler
(update
by default) and soft
as a boolean indicating whether
it should be a hard or soft commit (defaults to False
).
There is one more method you might want to use before querying Solr especially if the query is
coming from an untrusted end user. There is a clean_query
method which can be used to strip out
unwanted characters. Use it like:
trusted_query = aiosolr.clean_query(users_query)
Once you are finished with the Solr instance, you should call the method close
to cleanup sessions
like:
await client.close()
Timeouts
You can initialize the client with read_timeout
and write_timeout
to limit how long to wait for
requests to complete. read_timeout
applies to get
and query
whereas write_timeout
applies to
update
:
import aiosolr
client = aiosolr.Client(connection_url=connection_url, read_timeout=5, write_timeout=30)
You can override the timeouts for a specific request:
await client.get(document_id, read_timeout=1)
await client.update(doc, write_timeout=60)
aiosolr
uses
asyncio.wait_for
internally, so if a timeout occurs the exception raised is asyncio.TimeoutError
.
Debugging
To get more information from the Client you can initialize with debug=True
:
import aiosolr
client = aiosolr.Client(host=localhost, collection="example", port=8983, debug=True)
This sets the aiosolr
logger to DEBUG
level, and also sets the internally used HTTP session
(provided by aiohttp) to the DEBUG
level. This
makes it easier to see the actual network requests going to Solr.