pyroonga

What's this?

Python interface for groonga fulltext search engine.

Requirements

• Python 2.6 or 3.x and later
• groonga

Installation

from pypi:

% pip install pyroonga

from source:

% python setup.py install

Usage

First, Please run groonga by server mode or daemon mode. see following:

# server mode
% groonga -s DB_PATH_NAME

# daemon mode
% groonga -d DB_PATH_NAME

See groonga --help for more options.

Create Table

from pyroonga import tablebase, Column, Groonga

# create the base class for table definition.
Table = tablebase()

# define the table
class Site(Table):
   title = Column()
   name = Column()

class Blog(Table):
   entry = Column()

# create and bind the groonga connection object
grn = Groonga()
Table.bind(grn)

# create the all table on groonga's database
Table.create_all()

Data loading

data = [Site(_key='key1', title='foo', name='hoge'),
        Site(_key='key2', title='bar', name='fuga'),
        Site(_key='key3', title='baz', name='piyo'),
        Site(_key='key4', title='qux', name='xyzzy')]

Site.load(data)

The example above is load the data to groonga immediately. Also pseudo commit and rollback are supported:

data1 = [Site(_key='key5', title='Constellation', name='Sagittarius'),
         Site(_key='key6', title='Constellation', name='Pisces')]

# first load, but not load to groonga actually
data = Site.load(data1, immediate=False)

data2 = [Site(_key='key7', title='Constellation', name='Aquarius')]
data.load(data2)  # same as previous

# load data to groonga actually
data.commit()

Or reset loaded data:

data.rollback()

Note that loaded data reset is only if set immediate=False.

Query and get data as a mapped object

Get the all data from Site table:

data = Site.select().all()

And print the data:

for row in data:
    print(row._id, row._key, row.title)

Fulltext search queries:

Site.select().match_columns(Site.title).query('foo').all()
Site.select().match_columns(Site.title, Site.name).query('bar').all()

The above example is same as following queries:

select --table Site --match_columns 'title' --query "foo"
select --table Site --match_columns 'title OR name' --query "bar"

For more complex queries using pyroonga.odm.GE:

from pyroonga.odm import GE

Site.select().match_columns(Site.title).query(GE('foo') | GE('bar')).all()

The above example is same as following query:

select --table Site --match_columns 'title' --query "(foo OR bar)"

And also not use match_columns:

Site.select(title='foo').all()
Site.select(title='foo', name='bar').all()  # "or" search

The above example is same as following queries:

select --table Site --query "(title:@\"foo\")"
select --table Site --query "(title:@\"foo\" OR name:@\"bar\")"

Conditional search query:

Site.select(Site.title == 'bar').all()

Conbination for a condition:

Site.select((Site._id > 3) & (Site.title == 'baz')).all()

Limit and offset:

Site.select().limit(3).offset(2).all()

Sortby:

Site.select().sortby(Site._id).all()   # asc
Site.select().sortby(-Site._id).all()  # desc

Select the output columns:

# get the title and name columns
Site.select().output_columns(Site.title, Site.name).all()

# get the all columns
Site.select().output_columns(Site.ALL).all()

Drilldown

Switch to the drilldown query after the call of drilldown() from select() method chain:

data = Site.select().sortby(Site._key).drilldown(Site.title).all()

Result of drilldown will be stored to the drilldown attribute of the return value from all() method:

for drilldown in data.drilldown:
    print(drilldown._key, drilldown._nsubrecs)

A sortby() method in example above, It is query option of --sortby. For sortby of drilldown, Please call of sortby() method after the call of drilldown() method:

Site.select().drilldown(Site.title).sortby(Site._key).all()

A sortby() method in example above, It is query option of --drilldown_sortby. Of course, As well as limit() , offset() and output_columns() methods.

Suggest

N.B. The Groonga's suggest feature is still in draft.

First, Create table if still not created:

from pyroonga import SuggestTable

grn = Groonga()
SuggestTable.bind(grn)
SuggestTable.create_all()

Second, Data loading:

import time
from pyroonga import event_query

data = [event_query(time=time.time(), sequence=1, item='e'),
        event_query(time=time.time(), sequence=1, item='en'),
        event_query(time=time.time(), sequence=1, item='eng'),
        event_query(time=time.time(), sequence=1, item='engi'),
        event_query(time=time.time(), sequence=1, item='engin'),
        event_query(time=time.time(), sequence=1, item='engine', type='submit')]
event_query.load(data)

Finally, Querying:

from pyroonga import item_query, SuggestType

query = 'en'
result = item_query.suggest(query).types(SuggestType.complete). \
        frequency_threshold(1).all()
for r in result.complete:
    print("key is '%s', score is %s" % (r._key, r._score))

See also http://groonga.org/docs/suggest.html

More information

Still not written.

See also

http://groonga.org/ (Japanese: http://groonga.org/ja/ )

LICENSE

pyroonga is licensed under the MIT license.