def test_attachment(self): from ambry.bundle import new_analysis_bundle ab = new_analysis_bundle(source='foo.com', dataset='crime', subset='attach', variation='test', revision=2) print "Bundle Dir", ab.bundle_dir with ab.config.about as a: a.title = 'This is an Example Analysis Bundle?' a.tags = ['example', 'another'] a.groups = ['examples'] ab.config.build.dependencies = { 'incidents': 'clarinova.com-crime-incidents-casnd-incidents', 'addresses': 'clarinova.com-crime-incidents-casnd-addresses' } incidents = ab.library.dep('incidents').partition addresses = ab.library.dep('addresses').partition incidents.attach(addresses, 'addr') df = incidents.select("SELECT * FROM incidents " "LEFT JOIN addr.addresses as addresses ON addresses.id = incidents.address_id " "LIMIT 1000", index_col='id').pandas print df.dtypes ab.post_build()
def testBasic(self): from ambry.bundle import new_analysis_bundle ab = new_analysis_bundle(source='foo.com', dataset='dataset', subset='subset', variation='test', revision=2) print "Bundle Dir", ab.bundle_dir with ab.config.about as a: a.title = 'This is an Example Analysis Bundle?' a.tags = ['example','another'] a.groups = ['examples'] ab.config.build.dependencies = { 'random': 'example.com-random-example1' } p = ab.library.dep('random').partition df = p.select("SELECT * FROM example1",index_col='id').pandas gt90 = df[df.int > 90] print gt90.head(10) out = ab.partitions.new_db_from_pandas(gt90,table = 'gt90') # Try attaching ab.post_build() print p._repr_html_()