def test_replace_views():
dataset = DataSet()
adapter = dataset.add_adapter(MockAdapter())
no_managers = SelectionOp(
LoadOp('bogus'),
IsOp(Var('manager_id'), NullConst())
)
dataset.create_view(
'no_managers',
no_managers
)
view = AliasOp('no_managers', no_managers)
compare(
replace_views(LoadOp('no_managers'), dataset),
view
)
compare(
replace_views(
JoinOp(
LoadOp('no_managers'),
LoadOp('no_managers')
),
dataset
),
JoinOp(
view,
view
)
)
def test_relpace_views():
dataset = DataSet()
dataset.add_adapter(EmployeeAdapter())
no_managers = SelectionOp(
LoadOp('employees'),
IsOp(Var('manager_id'), NullConst())
)
dataset.create_view(
'no_managers',
no_managers
)
eq_(
replace_views(LoadOp('no_managers'), dataset),
no_managers
)
eq_(
replace_views(
JoinOp(
LoadOp('no_managers'),
LoadOp('no_managers')
),
dataset
),
JoinOp(
no_managers,
no_managers
)
)
def test_replace_views():
dataset = DataSet()
adapter = dataset.add_adapter(MockAdapter())
no_managers = SelectionOp(LoadOp('bogus'),
IsOp(Var('manager_id'), NullConst()))
dataset.create_view('no_managers', no_managers)
view = AliasOp('no_managers', no_managers)
compare(replace_views(LoadOp('no_managers'), dataset), view)
compare(
replace_views(JoinOp(LoadOp('no_managers'), LoadOp('no_managers')),
dataset), JoinOp(view, view))
def test_replace_view_within_a_view():
dataset = DataSet()
adapter = dataset.add_adapter(MockAdapter())
dataset.create_view(
'view1',
LoadOp('bogus')
)
dataset.create_view(
'view2',
LoadOp('view1')
)
dataset.create_view(
'view3',
SelectionOp(LoadOp('view2'), IsOp(Var('x'), NullConst()))
)
v1 = replace_views(
LoadOp('view3'),
dataset
)
compare(
v1,
AliasOp('view3',
SelectionOp(
AliasOp('view2', AliasOp('view1',LoadOp('bogus'))),
IsOp(Var('x'), NullConst())
)
)
)
def init(**tables):
"""
Returns a dataset to work with the discodb specified by path
"""
dataset = DataSet()
dataset.add_server(DiscoDBServer(**tables))
dataset.add_server(FileServer(
common_crawl=dict(
root_dir=join(dirname(__file__), 'data'),
pattern="sample.arc.gz",
decode="application/x-arc",
#description="Raw documents from http://commoncrawl.org"
)
))
dataset.add_server(FileServer(
top_sites=dict(
#description="Top Sites as reported by Alexa",
root_dir=join(dirname(__file__), 'data'),
pattern="alexa-top1m-{date}.csv",
decode="auto",
schema=dict(
fields=[
dict(name="date", type="DATE"),
dict(name="rank", type="STRING"),
dict(name="site", type="STRING")
]
)
)
))
dataset.frm('top_sites').limit(10).create_view('top_10')
dataset.create_view(
'outbound_links',
"select link_to, count() "
"from flatten(docs, 'link_to') "
"group by link_to order by count desc"
)
dataset.create_view(
'scripts',
"select scripts as script, count() "
"from flatten(docs, 'scripts') "
"group by script order by count desc"
)
dataset.create_view(
'servers',
"select headers_value as server_name, count() "
"from flatten(docs, 'headers') "
"where headers_name = 'Server' "
"group by server_name order by count desc"
)
return dataset
def test_replace_view_within_a_view():
dataset = DataSet()
adapter = dataset.add_adapter(MockAdapter())
dataset.create_view('view1', LoadOp('bogus'))
dataset.create_view('view2', LoadOp('view1'))
dataset.create_view(
'view3', SelectionOp(LoadOp('view2'), IsOp(Var('x'), NullConst())))
v1 = replace_views(LoadOp('view3'), dataset)
compare(
v1,
AliasOp(
'view3',
SelectionOp(AliasOp('view2', AliasOp('view1', LoadOp('bogus'))),
IsOp(Var('x'), NullConst()))))