def name_usage(self, name, start_year=MAX_YEAR, stop_year=MAX_YEAR):
year_gender = dict()
for i in xrange(start_year, stop_year + 1):
year_gender[i] = {'M': 0, 'F': 0}
for gender in ['M', 'F']:
q = Query(and_args(eq('name', name),
eq('gender', gender),
gte('year', start_year),
lte('year', stop_year)))
categorize = CategorizeSpec()
categorize.stats = StatsSpec('year', 'count')
q.add_categorization(categorize)
q.limit(0)
result = self.names.find(q.build())
try:
for year, stat in result.categories['year'].items():
year_gender[int(float(year))][gender] = stat['sum']
except KeyError: # No stats exists for year meaning no resuls for that name
pass
return year_gender
def test_categorize_by_query(self):
q = Query(regex('name', '.*'))
cat_spec = CategorizeSpec()
cat_spec.add_categorize_query(eq('name', 'Dan'))
q.add_categorization(cat_spec)
q.limit(0)
result = self.test_collection.find(q.build())
self.assertTrue(len(result.categories) > 0)
def test_categorize_by_query(self):
q = Query(regex('name', '.*'))
cat_spec = CategorizeSpec()
cat_spec.add_categorize_query(eq('name', 'Dan'))
q.add_categorization(cat_spec)
q.limit(0)
result = self.test_collection.find(q.build())
self.assertTrue(len(result.categories) > 0)
def test_categorize_with_stats(self):
print '\nTest categorize with stats\n'
q = Query(regex('name', '.*'))
cat_spec = CategorizeSpec()
cat_spec.stats = StatsSpec('name', 'age')
q.add_categorization(cat_spec)
q.limit(0)
result = self.test_collection.find(q.build())
self.assertTrue(len(result.categories) > 0)
def test_categorize_by_field(self):
print '\nTest categorize by field\n'
q = Query(regex('name', '.*'))
cat_spec = CategorizeSpec()
cat_spec.field = 'name'
q.add_categorization(cat_spec)
q.limit(0)
result = self.test_collection.find(q.build())
self.assertTrue(len(result.categories) > 0)
def test_categorize_with_stats(self):
print '\nTest categorize with stats\n'
q = Query(regex('name', '.*'))
cat_spec = CategorizeSpec()
cat_spec.stats = StatsSpec('name', 'age')
q.add_categorization(cat_spec)
q.limit(0)
result = self.test_collection.find(q.build())
self.assertTrue(len(result.categories) > 0)
def test_categorize_by_field(self):
print '\nTest categorize by field\n'
q = Query(regex('name', '.*'))
cat_spec = CategorizeSpec()
cat_spec.field = 'name'
q.add_categorization(cat_spec)
q.limit(0)
result = self.test_collection.find(q.build())
self.assertTrue(len(result.categories) > 0)
def test_categorize_by_range(self):
print '\nTest categorize by range'
q = Query(regex('name', '.*'))
range_spec = RangeSpec('age', 1, 10, 1)
cat_spec = CategorizeSpec()
cat_spec.range_spec = range_spec
q.add_categorization(cat_spec)
q.limit(0)
result = self.test_collection.find(q.build())
self.assertTrue(len(result.categories) > 0)
def test_categorize_by_range(self):
print '\nTest categorize by range'
q = Query(regex('name', '.*'))
range_spec = RangeSpec('age', 1, 10, 1)
cat_spec = CategorizeSpec()
cat_spec.range_spec = range_spec
q.add_categorization(cat_spec)
q.limit(0)
result = self.test_collection.find(q.build())
self.assertTrue(len(result.categories) > 0)
def categorize_by_field(query_field, field):
print '\nCategorize by value for field, {field}'.format(field=field)
q = Query(regex(query_field, '.*'))
q.limit(0)
category = CategorizeSpec()
category.field = field
q.add_categorization(category)
query = json.dumps(q.build())
query_pprint(q.build())
print '\n' + curl('PUT', 'query/all', data=query, accept_header=True)
result = demo_collection.find(q.build())
json_pprint(result.response_doc)
def categorize_with_stats(field, stat_field):
print '\nCategorize with stats on field, {stat_field}'.format(stat_field=stat_field)
q = Query(regex(field, '.*'))
q.limit(0)
category = CategorizeSpec()
category.stats = StatsSpec(field, stat_field)
q.add_categorization(category)
query = json.dumps(q.build())
query_pprint(q.build())
print '\n' + curl('PUT', 'query/all', data=query, accept_header=True)
result = demo_collection.find(q.build())
json_pprint(result.response_doc)
def categorize_by_field(query_field, field):
print '\nCategorize by value for field, {field}'.format(field=field)
q = Query(regex(query_field, '.*'))
q.limit(0)
category = CategorizeSpec()
category.field = field
q.add_categorization(category)
query = json.dumps(q.build())
query_pprint(q.build())
print '\n' + curl('PUT', 'query/all', data=query, accept_header=True)
result = demo_collection.find(q.build())
json_pprint(result.response_doc)
def partial_name_search(self, name_prefix, start_year=MAX_YEAR, stop_year=MAX_YEAR):
q = Query(and_args(regex('name', name_prefix + '.*'),
gte('year', start_year),
lte('year', stop_year)))
q.limit(0)
categorize = CategorizeSpec()
categorize.stats = StatsSpec('name', 'count')
q.add_categorization(categorize)
#print q.build()
result = self.names.find(q.build())
name_count = dict()
for field, stat in result.categories['name'].items():
name_count[field] = stat['sum']
return name_count
def categorize_with_stats(field, stat_field):
print '\nCategorize with stats on field, {stat_field}'.format(
stat_field=stat_field)
q = Query(regex(field, '.*'))
q.limit(0)
category = CategorizeSpec()
category.stats = StatsSpec(field, stat_field)
q.add_categorization(category)
query = json.dumps(q.build())
query_pprint(q.build())
print '\n' + curl('PUT', 'query/all', data=query, accept_header=True)
result = demo_collection.find(q.build())
json_pprint(result.response_doc)
def categorize_by_range(query_field, field, start, end, increment):
print '\nCategorize by range for field, {field}'.format(field=field)
q = Query(regex(query_field, '.*'))
q.limit(0)
category = CategorizeSpec()
_range = RangeSpec(field, start, end, increment)
category.range_spec = _range
q.add_categorization(category)
query = json.dumps(q.build())
query_pprint(q.build())
print '\n' + curl('PUT', 'query/all', data=query, accept_header=True)
result = demo_collection.find(q.build())
json_pprint(result.response_doc)
def categorize_by_range(query_field, field, start, end, increment):
print '\nCategorize by range for field, {field}'.format(field=field)
q = Query(regex(query_field, '.*'))
q.limit(0)
category = CategorizeSpec()
_range = RangeSpec(field, start, end, increment)
category.range_spec = _range
q.add_categorization(category)
query = json.dumps(q.build())
query_pprint(q.build())
print '\n' + curl('PUT', 'query/all', data=query, accept_header=True)
result = demo_collection.find(q.build())
json_pprint(result.response_doc)