more comprehensive table scanning and querying

This commit is contained in:
Steve Pulec 2013-03-13 22:44:54 -04:00
parent ad4b6c4ee2
commit 3e6f802a20
5 changed files with 78 additions and 16 deletions

View File

@ -6,13 +6,13 @@ COMPARISON_FUNCS = {
'LT': lambda item_value, test_value: item_value < test_value,
'GE': lambda item_value, test_value: item_value >= test_value,
'GT': lambda item_value, test_value: item_value > test_value,
'NULL': lambda item_value: item_value is None,
'NOT_NULL': lambda item_value: item_value is not None,
'NULL': lambda item_value, test_value: item_value is None,
'NOT_NULL': lambda item_value, test_value: item_value is not None,
'CONTAINS': lambda item_value, test_value: test_value in item_value,
'NOT_CONTAINS': lambda item_value, test_value: test_value not in item_value,
'BEGINS_WITH': lambda item_value, test_value: item_value.startswith(test_value),
'IN': lambda item_value, test_value: item_value in test_value,
'BETWEEN': lambda item_value, test_values: test_values[0] <= item_value <= test_values[1],
'BETWEEN': lambda item_value, lower_test_value, upper_test_value: lower_test_value <= item_value <= upper_test_value,
}

View File

@ -97,14 +97,14 @@ class Table(object):
except KeyError:
return None
def query(self, hash_key, range_comparison, range_value):
def query(self, hash_key, range_comparison, range_values):
results = []
last_page = True # Once pagination is implemented, change this
possible_results = self.items.get(hash_key, [])
comparison_func = get_comparison_func(range_comparison)
for result in possible_results.values():
if comparison_func(result.range_key, range_value):
if comparison_func(result.range_key, *range_values):
results.append(result)
return results, last_page
@ -121,12 +121,24 @@ class Table(object):
for result in self.all_items():
scanned_count += 1
passes_all_conditions = True
for attribute_name, (comparison_operator, comparison_value) in filters.iteritems():
for attribute_name, (comparison_operator, comparison_values) in filters.iteritems():
comparison_func = get_comparison_func(comparison_operator)
attribute_value = result.attrs[attribute_name].values()[0]
if not comparison_func(attribute_value, comparison_value):
attribute = result.attrs.get(attribute_name)
if attribute:
# Attribute found
attribute_value = attribute.values()[0]
if not comparison_func(attribute_value, *comparison_values):
passes_all_conditions = False
break
elif comparison_operator == 'NULL':
# Comparison is NULL and we don't have the attribute
continue
else:
# No attribute found and comparison is no NULL. This item fails
passes_all_conditions = False
break
if passes_all_conditions:
results.append(result)
@ -172,12 +184,12 @@ class DynamoDBBackend(BaseBackend):
return table.get_item(hash_key, range_key)
def query(self, table_name, hash_key, range_comparison, range_value):
def query(self, table_name, hash_key, range_comparison, range_values):
table = self.tables.get(table_name)
if not table:
return None
return table.query(hash_key, range_comparison, range_value)
return table.query(hash_key, range_comparison, range_values)
def scan(self, table_name, filters):
table = self.tables.get(table_name)

View File

@ -2,6 +2,7 @@ import json
from moto.core.utils import headers_to_dict
from .models import dynamodb_backend
from .utils import values_from_dynamo_types
class DynamoHandler(object):
@ -128,8 +129,9 @@ class DynamoHandler(object):
hash_key = body['HashKeyValue'].values()[0]
range_condition = body['RangeKeyCondition']
range_comparison = range_condition['ComparisonOperator']
range_value = range_condition['AttributeValueList'][0].values()[0]
items, last_page = dynamodb_backend.query(name, hash_key, range_comparison, range_value)
range_values = values_from_dynamo_types(range_condition['AttributeValueList'])
items, last_page = dynamodb_backend.query(name, hash_key, range_comparison, range_values)
result = {
"Count": len(items),
@ -152,8 +154,11 @@ class DynamoHandler(object):
for attribute_name, scan_filter in scan_filters.iteritems():
# Keys are attribute names. Values are tuples of (comparison, comparison_value)
comparison_operator = scan_filter["ComparisonOperator"]
comparison_value = scan_filter["AttributeValueList"][0].values()[0]
filters[attribute_name] = (comparison_operator, comparison_value)
if scan_filter.get("AttributeValueList"):
comparison_values = values_from_dynamo_types(scan_filter.get("AttributeValueList"))
else:
comparison_values = [None]
filters[attribute_name] = (comparison_operator, comparison_values)
items, scanned_count, last_page = dynamodb_backend.scan(name, filters)

View File

@ -5,3 +5,18 @@ def unix_time(dt):
epoch = datetime.datetime.utcfromtimestamp(0)
delta = dt - epoch
return delta.total_seconds()
def value_from_dynamo_type(dynamo_type):
"""
Dynamo return attributes like {"S": "AttributeValue1"}.
This function takes that value and returns "AttributeValue1".
# TODO eventually this should be smarted to actually read the type of
the attribute
"""
return dynamo_type.values()[0]
def values_from_dynamo_types(dynamo_types):
return [value_from_dynamo_type(dynamo_type) for dynamo_type in dynamo_types]

View File

@ -224,6 +224,15 @@ def test_query():
results = table.query(hash_key='the-key', range_key_condition=condition.GT('9999'))
results.response['Items'].should.have.length_of(0)
results = table.query(hash_key='the-key', range_key_condition=condition.CONTAINS('12'))
results.response['Items'].should.have.length_of(1)
results = table.query(hash_key='the-key', range_key_condition=condition.BEGINS_WITH('7'))
results.response['Items'].should.have.length_of(1)
results = table.query(hash_key='the-key', range_key_condition=condition.BETWEEN('567', '890'))
results.response['Items'].should.have.length_of(1)
@mock_dynamodb
def test_scan():
@ -249,6 +258,13 @@ def test_scan():
)
item.put()
item_data = {
'Body': 'http://url_to_lolcat.gif',
'SentBy': 'User B',
'ReceivedTime': '12/9/2011 11:36:03 PM',
'Ids': {1, 2, 3},
'PK': 7,
}
item = table.new_item(
hash_key='the-key',
range_key='789',
@ -257,12 +273,26 @@ def test_scan():
item.put()
results = table.scan(scan_filter={'SentBy': condition.EQ('User B')})
results.response['Items'].should.have.length_of(0)
results.response['Items'].should.have.length_of(1)
results = table.scan(scan_filter={'Body': condition.BEGINS_WITH('http')})
results.response['Items'].should.have.length_of(3)
results = table.scan(scan_filter={'Ids': condition.CONTAINS(2)})
results.response['Items'].should.have.length_of(1)
results = table.scan(scan_filter={'Ids': condition.NOT_NULL()})
results.response['Items'].should.have.length_of(1)
results = table.scan(scan_filter={'Ids': condition.NULL()})
results.response['Items'].should.have.length_of(2)
results = table.scan(scan_filter={'PK': condition.BETWEEN(8, 9)})
results.response['Items'].should.have.length_of(0)
results = table.scan(scan_filter={'PK': condition.BETWEEN(5, 8)})
results.response['Items'].should.have.length_of(1)
# Batch read
# Batch write
# scan