Infinidat/infi.clickhouse_orm
1
1
Fork 0
mirror of https://github.com/Infinidat/infi.clickhouse_orm.git synced 2024-11-13 04:46:33 +03:00
Code Issues Packages Projects Releases Wiki Activity

Finished Release v1.1.0

Browse Source
This commit is contained in:
Itai Shirav 2019-06-13 17:20:49 +03:00
commit 40725372fe
32 changed files with 469 additions and 236 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
CHANGELOG.md
View File

@ -1,6 +1,20 @@
Change Log
==========
v1.1.0
------
- Add PREWHERE support to querysets (M1hacka)
- Add WITH TOTALS support to querysets (M1hacka)
- Extend date field range (trthhrtz)
- Fix parsing of server errors in ClickHouse v19.3.3+
- Fix pagination when asking for the last page on a query that matches no records
- Use HTTP Basic Authentication instead of passing the credentials in the URL
- Support default/alias/materialized for nullable fields
- Add UUIDField (kpotehin)
- Add `log_statements` parameter to database initializer
- Fix test_merge which fails on ClickHouse v19.8.3
- Fix querysets using the SystemPart model
v1.0.4
------
- Added `timeout` parameter to database initializer (SUHAR1K)

2
buildout.cfg
View File

@ -49,7 +49,7 @@ eggs = ${project:name}
ipython<6
nose
coverage
enum34
enum-compat
infi.unittest
infi.traceback
memory_profiler

49
docs/class_reference.md
View File

@ -701,6 +701,13 @@ Extends BaseIntField
#### UInt8Field(default=None, alias=None, materialized=None, readonly=None)
### UUIDField
Extends Field
#### UUIDField(default=None, alias=None, materialized=None, readonly=None)
infi.clickhouse_orm.engines
---------------------------
@ -835,10 +842,10 @@ is equivalent to:
Returns the whole query as a SQL string.
#### conditions_as_sql()
#### conditions_as_sql(prewhere=False)
Returns the contents of the query's `WHERE` clause as a string.
Returns the contents of the query's `WHERE` or `PREWHERE` clause as a string.
#### count()
@ -854,17 +861,18 @@ Adds a DISTINCT clause to the query, meaning that any duplicate rows
in the results will be omitted.
#### exclude(**filter_fields)
#### exclude(*q, **kwargs)
Returns a copy of this queryset that excludes all rows matching the conditions.
Pass `prewhere=True` to apply the conditions as PREWHERE instead of WHERE.
#### filter(*q, **filter_fields)
#### filter(*q, **kwargs)
Returns a copy of this queryset that includes only rows matching the conditions.
Add q object to query if it specified.
Pass `prewhere=True` to apply the conditions as PREWHERE instead of WHERE.
#### final()
@ -908,6 +916,12 @@ The result is a namedtuple containing `objects` (list), `number_of_objects`,
`pages_total`, `number` (of the current page), and `page_size`.
#### select_fields_as_sql()
Returns the selected fields or expressions as a SQL string.
### AggregateQuerySet
Extends QuerySet
@ -943,10 +957,10 @@ This method is not supported on `AggregateQuerySet`.
Returns the whole query as a SQL string.
#### conditions_as_sql()
#### conditions_as_sql(prewhere=False)
Returns the contents of the query's `WHERE` clause as a string.
Returns the contents of the query's `WHERE` or `PREWHERE` clause as a string.
#### count()
@ -962,17 +976,18 @@ Adds a DISTINCT clause to the query, meaning that any duplicate rows
in the results will be omitted.
#### exclude(**filter_fields)
#### exclude(*q, **kwargs)
Returns a copy of this queryset that excludes all rows matching the conditions.
Pass `prewhere=True` to apply the conditions as PREWHERE instead of WHERE.
#### filter(*q, **filter_fields)
#### filter(*q, **kwargs)
Returns a copy of this queryset that includes only rows matching the conditions.
Add q object to query if it specified.
Pass `prewhere=True` to apply the conditions as PREWHERE instead of WHERE.
#### final()
@ -1022,3 +1037,17 @@ The result is a namedtuple containing `objects` (list), `number_of_objects`,
`pages_total`, `number` (of the current page), and `page_size`.
#### select_fields_as_sql()
Returns the selected fields or expressions as a SQL string.
#### with_totals()
Adds WITH TOTALS modifier ot GROUP BY, making query return extra row
with aggregate function calculated across all the rows. More information:
https://clickhouse.yandex/docs/en/query_language/select/#with-totals-modifier

43
docs/field_types.md
View File

@ -9,7 +9,7 @@ Currently the following field types are supported:
| ------------------ | ---------- | ------------------- | -----------------------------------------------------
| StringField | String | unicode | Encoded as UTF-8 when written to ClickHouse
| FixedStringField | String | unicode | Encoded as UTF-8 when written to ClickHouse
| DateField | Date | datetime.date | Range 1970-01-01 to 2038-01-19
| DateField | Date | datetime.date | Range 1970-01-01 to 2105-12-31
| DateTimeField | DateTime | datetime.datetime | Minimal value is 1970-01-01 00:00:00; Always in UTC
| Int8Field | Int8 | int | Range -128 to 127
| Int16Field | Int16 | int | Range -32768 to 32767
@ -25,6 +25,7 @@ Currently the following field types are supported:
| Decimal32Field | Decimal32 | Decimal | Ditto
| Decimal64Field | Decimal64 | Decimal | Ditto
| Decimal128Field | Decimal128 | Decimal | Ditto
| UUIDField | UUID | Decimal |
| Enum8Field | Enum8 | Enum | See below
| Enum16Field | Enum16 | Enum | See below
| ArrayField | Array | list | See below
@ -183,46 +184,6 @@ class BooleanField(Field):
return '1' if value else '0'
```
Here's another example - a field for storing UUIDs in the database as 16-byte strings. We'll use Python's built-in `UUID` class to handle the conversion from strings, ints and tuples into UUID instances. So in our Python code we'll have the convenience of working with UUID objects, but they will be stored in the database as efficiently as possible:
```python
from infi.clickhouse_orm.fields import Field
from infi.clickhouse_orm.utils import escape
from uuid import UUID
import six
class UUIDField(Field):
# The ClickHouse column type to use
db_type = 'FixedString(16)'
# The default value if empty
class_default = UUID(int=0)
def to_python(self, value, timezone_in_use):
# Convert valid values to UUID instance
if isinstance(value, UUID):
return value
elif isinstance(value, six.string_types):
return UUID(bytes=value.encode('latin1')) if len(value) == 16 else UUID(value)
elif isinstance(value, six.integer_types):
return UUID(int=value)
elif isinstance(value, tuple):
return UUID(fields=value)
else:
raise ValueError('Invalid value for UUIDField: %r' % value)
def to_db_string(self, value, quote=True):
# The value was already converted by to_python, so it's a UUID instance
val = value.bytes
if six.PY3:
val = str(val, 'latin1')
return escape(val, quote)
```
Note that the latin-1 encoding is used as an identity encoding for converting between raw bytes and strings. This is required in Python 3, where `str` and `bytes` are different types.
---
[<< Querysets](querysets.md) | [Table of Contents](toc.md) | [Table Engines >>](table_engines.md)

2
docs/models_and_databases.md
View File

@ -89,7 +89,7 @@ When values are assigned to model fields, they are immediately converted to thei
>>> suzy.birthday = 0.5
ValueError: Invalid value for DateField - 0.5
>>> suzy.birthday = '1922-05-31'
ValueError: DateField out of range - 1922-05-31 is not between 1970-01-01 and 2038-01-19
ValueError: DateField out of range - 1922-05-31 is not between 1970-01-01 and 2105-12-31
Inserting to the Database
-------------------------

21
docs/querysets.md
View File

@ -32,6 +32,14 @@ For filters with compound conditions you can use `Q` objects inside `filter` wit
>>> qs.conditions_as_sql()
u"((first_name = 'Ciaran' AND last_name = 'Carver') OR height <= 1.8) AND (NOT (first_name = 'David'))"
By default conditions from `filter` and `exclude` methods are add to `WHERE` clause.
For better aggregation performance you can add them to `PREWHERE` section using `prewhere=True` parameter
>>> qs = Person.objects_in(database)
>>> qs = qs.filter(first_name__startswith='V', prewhere=True)
>>> qs.conditions_as_sql(prewhere=True)
u"first_name LIKE 'V%'"
There are different operators that can be used, by passing `<fieldname>__<operator>=<value>` (two underscores separate the field name from the operator). In case no operator is given, `eq` is used by default. Below are all the supported operators.
| Operator | Equivalent SQL | Comments |
@ -202,6 +210,19 @@ This queryset is translated to:
After calling `aggregate` you can still use most of the regular queryset methods, such as `count`, `order_by` and `paginate`. It is not possible, however, to call `only` or `aggregate`. It is also not possible to filter the queryset on calculated fields, only on fields that exist in the model.
If you limit aggregation results, it might be useful to get total aggregation values for all rows.
To achieve this, you can use `with_totals` method. It will return extra row (last) with
values aggregated for all rows suitable for filters.
qs = Person.objects_in(database).aggregate('first_name', num='count()').with_totals().order_by('-count')[:3]
>>> print qs.count()
4
>>> for row in qs:
>>> print("'{}': {}".format(row.first_name, row.count))
'Cassandra': 2
'Alexandra': 2
'': 100
---
[<< Models and Databases](models_and_databases.md) | [Table of Contents](toc.md) | [Field Types >>](field_types.md)

4
docs/ref.md
View File

@ -482,9 +482,9 @@ infi.clickhouse_orm.query
#### QuerySet(model_cls, database)
#### conditions_as_sql()
#### conditions_as_sql(prewhere=True)
Return the contents of the queryset's WHERE clause.
Return the contents of the queryset's WHERE or `PREWHERE` clause.
#### count()

1
docs/toc.md
View File

@ -92,6 +92,7 @@
* [UInt32Field](class_reference.md#uint32field)
* [UInt64Field](class_reference.md#uint64field)
* [UInt8Field](class_reference.md#uint8field)
* [UUIDField](class_reference.md#uuidfield)
* [infi.clickhouse_orm.engines](class_reference.md#infi.clickhouse_orm.engines)
* [Engine](class_reference.md#engine)
* [TinyLog](class_reference.md#tinylog)

38
src/infi/clickhouse_orm/database.py
View File

@ -40,11 +40,19 @@ class ServerError(DatabaseException):
self.message = message
super(ServerError, self).__init__(message)
ERROR_PATTERN = re.compile(r'''
ERROR_PATTERNS = (
# ClickHouse prior to v19.3.3
re.compile(r'''
Code:\ (?P<code>\d+),
\ e\.displayText\(\)\ =\ (?P<type1>[^ \n]+):\ (?P<msg>.+?),
\ e.what\(\)\ =\ (?P<type2>[^ \n]+)
''', re.VERBOSE | re.DOTALL)
''', re.VERBOSE | re.DOTALL),
# ClickHouse v19.3.3+
re.compile(r'''
Code:\ (?P<code>\d+),
\ e\.displayText\(\)\ =\ (?P<type1>[^ \n]+):\ (?P<msg>.+)
''', re.VERBOSE | re.DOTALL),
)
@classmethod
def get_error_code_msg(cls, full_error_message):
@ -54,10 +62,11 @@ class ServerError(DatabaseException):
See the list of error codes here:
https://github.com/yandex/ClickHouse/blob/master/dbms/src/Common/ErrorCodes.cpp
"""
match = cls.ERROR_PATTERN.match(full_error_message)
for pattern in cls.ERROR_PATTERNS:
match = pattern.match(full_error_message)
if match:
# assert match.group('type1') == match.group('type2')
return int(match.group('code')), match.group('msg')
return int(match.group('code')), match.group('msg').strip()
return 0, full_error_message
@ -74,7 +83,7 @@ class Database(object):
def __init__(self, db_name, db_url='http://localhost:8123/',
username=None, password=None, readonly=False, autocreate=True,
timeout=60, verify_ssl_cert=True):
timeout=60, verify_ssl_cert=True, log_statements=False):
'''
Initializes a database instance. Unless it's readonly, the database will be
created on the ClickHouse server if it does not already exist.
@ -87,15 +96,17 @@ class Database(object):
- `autocreate`: automatically create the database if it does not exist (unless in readonly mode).
- `timeout`: the connection timeout in seconds.
- `verify_ssl_cert`: whether to verify the server's certificate when connecting via HTTPS.
- `log_statements`: when True, all database statements are logged.
'''
self.db_name = db_name
self.db_url = db_url
self.username = username
self.password = password
self.readonly = False
self.timeout = timeout
self.request_session = requests.Session()
self.request_session.verify = verify_ssl_cert
if username:
self.request_session.auth = (username, password or '')
self.log_statements = log_statements
self.settings = {}
self.db_exists = False # this is required before running _is_existing_database
self.db_exists = self._is_existing_database()
@ -276,7 +287,7 @@ class Database(object):
count = self.count(model_class, conditions)
pages_total = int(ceil(count / float(page_size)))
if page_num == -1:
page_num = pages_total
page_num = max(pages_total, 1)
elif page_num < 1:
raise ValueError('Invalid page number: %d' % page_num)
offset = (page_num - 1) * page_size
@ -287,7 +298,7 @@ class Database(object):
query += ' LIMIT %d, %d' % (offset, page_size)
query = self._substitute(query, model_class)
return Page(
objects=list(self.select(query, model_class, settings)),
objects=list(self.select(query, model_class, settings)) if count else [],
number_of_objects=count,
pages_total=pages_total,
number=page_num,
@ -325,6 +336,8 @@ class Database(object):
def _send(self, data, settings=None, stream=False):
if isinstance(data, string_types):
data = data.encode('utf-8')
if self.log_statements:
logger.info(data)
params = self._build_params(settings)
r = self.request_session.post(self.db_url, params=params, data=data, stream=stream, timeout=self.timeout)
if r.status_code != 200:
@ -336,10 +349,6 @@ class Database(object):
params.update(self.settings)
if self.db_exists:
params['database'] = self.db_name
if self.username:
params['user'] = self.username
if self.password:
params['password'] = self.password
# Send the readonly flag, unless the connection is already readonly (to prevent db error)
if self.readonly and not self.connection_readonly:
params['readonly'] = '1'
@ -352,6 +361,9 @@ class Database(object):
if '$' in query:
mapping = dict(db="`%s`" % self.db_name)
if model_class:
if model_class.is_system_model():
mapping['table'] = model_class.table_name()
else:
mapping['table'] = "`%s`.`%s`" % (self.db_name, model_class.table_name())
query = Template(query).safe_substitute(mapping)
return query

39
src/infi/clickhouse_orm/fields.py
View File

@ -1,11 +1,12 @@
from __future__ import unicode_literals
from six import string_types, text_type, binary_type
from six import string_types, text_type, binary_type, integer_types
import datetime
import iso8601
import pytz
import time
from calendar import timegm
from decimal import Decimal, localcontext
from uuid import UUID
from .utils import escape, parse_array, comma_join
@ -131,7 +132,7 @@ class FixedStringField(StringField):
class DateField(Field):
min_value = datetime.date(1970, 1, 1)
max_value = datetime.date(2038, 1, 19)
max_value = datetime.date(2105, 12, 31)
class_default = min_value
db_type = 'Date'
@ -452,6 +453,29 @@ class ArrayField(Field):
return 'Array(%s)' % self.inner_field.get_sql(with_default_expression=False)
class UUIDField(Field):
class_default = UUID(int=0)
db_type = 'UUID'
def to_python(self, value, timezone_in_use):
if isinstance(value, UUID):
return value
elif isinstance(value, binary_type):
return UUID(bytes=value)
elif isinstance(value, string_types):
return UUID(value)
elif isinstance(value, integer_types):
return UUID(int=value)
elif isinstance(value, tuple):
return UUID(fields=value)
else:
raise ValueError('Invalid value for UUIDField: %r' % value)
def to_db_string(self, value, quote=True):
return escape(str(value), quote)
class NullableField(Field):
class_default = None
@ -478,4 +502,13 @@ class NullableField(Field):
return self.inner_field.to_db_string(value, quote=quote)
def get_sql(self, with_default_expression=True):
return 'Nullable(%s)' % self.inner_field.get_sql(with_default_expression=False)
s = 'Nullable(%s)' % self.inner_field.get_sql(with_default_expression=False)
if with_default_expression:
if self.alias:
s = '%s ALIAS %s' % (s, self.alias)
elif self.materialized:
s = '%s MATERIALIZED %s' % (s, self.materialized)
elif self.default:
default = self.to_db_string(self.default)
s = '%s DEFAULT %s' % (s, default)
return s

15
src/infi/clickhouse_orm/models.py
View File

@ -311,9 +311,16 @@ class MergeModel(Model):
@classmethod
def create_table_sql(cls, db):
assert isinstance(cls.engine, Merge), "engine must be engines.Merge instance"
return super(MergeModel, cls).create_table_sql(db)
assert isinstance(cls.engine, Merge), "engine must be an instance of engines.Merge"
parts = ['CREATE TABLE IF NOT EXISTS `%s`.`%s` (' % (db.db_name, cls.table_name())]
cols = []
for name, field in iteritems(cls.fields()):
if name != '_table':
cols.append(' %s %s' % (name, field.get_sql()))
parts.append(',\n'.join(cols))
parts.append(')')
parts.append('ENGINE = ' + cls.engine.create_table_sql(db))
return '\n'.join(parts)
# TODO: base class for models that require specific engine
@ -324,7 +331,7 @@ class DistributedModel(Model):
"""
def set_database(self, db):
assert isinstance(self.engine, Distributed), "engine must be engines.Distributed instance"
assert isinstance(self.engine, Distributed), "engine must be an instance of engines.Distributed"
res = super(DistributedModel, self).set_database(db)
return res

202
src/infi/clickhouse_orm/query.py
View File

@ -2,7 +2,7 @@ from __future__ import unicode_literals
import six
import pytz
from copy import copy
from copy import copy, deepcopy
from math import ceil
from .engines import CollapsingMergeTree
@ -173,6 +173,11 @@ class FOV(object):
def to_sql(self, model_cls):
return self._operator.to_sql(model_cls, self._field_name, self._value)
def __deepcopy__(self, memodict={}):
res = copy(self)
res._value = deepcopy(self._value)
return res
class Q(object):
@ -181,17 +186,32 @@ class Q(object):
def __init__(self, **filter_fields):
self._fovs = [self._build_fov(k, v) for k, v in six.iteritems(filter_fields)]
self._l_child = None
self._r_child = None
self._children = []
self._negate = False
self._mode = self.AND_MODE
@property
def is_empty(self):
"""
Checks if there are any conditions in Q object
:return: Boolean
"""
return not bool(self._fovs or self._children)
@classmethod
def _construct_from(cls, l_child, r_child, mode):
if mode == l_child._mode:
q = deepcopy(l_child)
q._children.append(deepcopy(r_child))
elif mode == r_child._mode:
q = deepcopy(r_child)
q._children.append(deepcopy(l_child))
else:
# Different modes
q = Q()
q._l_child = l_child
q._r_child = r_child
q._children = [l_child, r_child]
q._mode = mode # AND/OR
return q
def _build_fov(self, key, value):
@ -202,16 +222,27 @@ class Q(object):
return FOV(field_name, operator, value)
def to_sql(self, model_cls):
condition_sql = []
if self._fovs:
sql = ' {} '.format(self._mode).join(fov.to_sql(model_cls) for fov in self._fovs)
condition_sql.extend([fov.to_sql(model_cls) for fov in self._fovs])
if self._children:
condition_sql.extend([child.to_sql(model_cls) for child in self._children if child])
if not condition_sql:
# Empty Q() object returns everything
sql = '1'
elif len(condition_sql) == 1:
# Skip not needed brackets over single condition
sql = condition_sql[0]
else:
if self._l_child and self._r_child:
sql = '({} {} {})'.format(
self._l_child.to_sql(model_cls), self._mode, self._r_child.to_sql(model_cls))
else:
return '1'
# Each condition must be enclosed in brackets, or order of operations may be wrong
sql = '(%s)' % ') {} ('.format(self._mode).join(condition_sql)
if self._negate:
sql = 'NOT (%s)' % sql
return sql
def __or__(self, other):
@ -225,6 +256,20 @@ class Q(object):
q._negate = True
return q
def __bool__(self):
return not self.is_empty
def __deepcopy__(self, memodict={}):
q = Q()
q._fovs = [deepcopy(fov) for fov in self._fovs]
q._negate = self._negate
q._mode = self._mode
if self._children:
q._children = [deepcopy(child) for child in self._children]
return q
@six.python_2_unicode_compatible
class QuerySet(object):
@ -242,7 +287,10 @@ class QuerySet(object):
self._model_cls = model_cls
self._database = database
self._order_by = []
self._q = []
self._where_q = Q()
self._prewhere_q = Q()
self._grouping_fields = []
self._grouping_with_totals = False
self._fields = model_cls.fields().keys()
self._limits = None
self._distinct = False
@ -284,20 +332,44 @@ class QuerySet(object):
qs._limits = (start, stop - start)
return qs
def select_fields_as_sql(self):
"""
Returns the selected fields or expressions as a SQL string.
"""
return comma_join('`%s`' % field for field in self._fields) if self._fields else '*'
def as_sql(self):
"""
Returns the whole query as a SQL string.
"""
distinct = 'DISTINCT ' if self._distinct else ''
fields = '*'
if self._fields:
fields = comma_join('`%s`' % field for field in self._fields)
ordering = '\nORDER BY ' + self.order_by_as_sql() if self._order_by else ''
limit = '\nLIMIT %d, %d' % self._limits if self._limits else ''
final = ' FINAL' if self._final else ''
params = (distinct, fields, self._model_cls.table_name(), final,
self.conditions_as_sql(), ordering, limit)
return u'SELECT %s%s\nFROM `%s`%s\nWHERE %s%s%s' % params
table_name = self._model_cls.table_name()
if not self._model_cls.is_system_model():
table_name = '`%s`' % table_name
params = (distinct, self.select_fields_as_sql(), table_name, final)
sql = u'SELECT %s%s\nFROM %s%s' % params
if self._prewhere_q and not self._prewhere_q.is_empty:
sql += '\nPREWHERE ' + self.conditions_as_sql(prewhere=True)
if self._where_q and not self._where_q.is_empty:
sql += '\nWHERE ' + self.conditions_as_sql(prewhere=False)
if self._grouping_fields:
sql += '\nGROUP BY %s' % comma_join('`%s`' % field for field in self._grouping_fields)
if self._grouping_with_totals:
sql += ' WITH TOTALS'
if self._order_by:
sql += '\nORDER BY ' + self.order_by_as_sql()
if self._limits:
sql += '\nLIMIT %d, %d' % self._limits
return sql
def order_by_as_sql(self):
"""
@ -308,14 +380,12 @@ class QuerySet(object):
for field in self._order_by
])
def conditions_as_sql(self):
def conditions_as_sql(self, prewhere=False):
"""
Returns the contents of the query's `WHERE` clause as a string.
Returns the contents of the query's `WHERE` or `PREWHERE` clause as a string.
"""
if self._q:
return u' AND '.join([q.to_sql(self._model_cls) for q in self._q])
else:
return u'1'
q_object = self._prewhere_q if prewhere else self._where_q
return q_object.to_sql(self._model_cls)
def count(self):
"""
@ -326,8 +396,10 @@ class QuerySet(object):
sql = u'SELECT count() FROM (%s)' % self.as_sql()
raw = self._database.raw(sql)
return int(raw) if raw else 0
# Simple case
return self._database.count(self._model_cls, self.conditions_as_sql())
conditions = (self._where_q & self._prewhere_q).to_sql(self._model_cls)
return self._database.count(self._model_cls, conditions)
def order_by(self, *field_names):
"""
@ -347,25 +419,43 @@ class QuerySet(object):
qs._fields = field_names
return qs
def filter(self, *q, **filter_fields):
"""
Returns a copy of this queryset that includes only rows matching the conditions.
Add q object to query if it specified.
"""
def _filter_or_exclude(self, *q, **kwargs):
inverse = kwargs.pop('_inverse', False)
prewhere = kwargs.pop('prewhere', False)
qs = copy(self)
if q:
qs._q = list(self._q) + list(q)
condition = Q()
for q_obj in q:
condition &= q_obj
if kwargs:
condition &= Q(**kwargs)
if inverse:
condition = ~condition
condition = copy(self._prewhere_q if prewhere else self._where_q) & condition
if prewhere:
qs._prewhere_q = condition
else:
qs._q = list(self._q) + [Q(**filter_fields)]
qs._where_q = condition
return qs
def exclude(self, **filter_fields):
def filter(self, *q, **kwargs):
"""
Returns a copy of this queryset that includes only rows matching the conditions.
Pass `prewhere=True` to apply the conditions as PREWHERE instead of WHERE.
"""
return self._filter_or_exclude(*q, **kwargs)
def exclude(self, *q, **kwargs):
"""
Returns a copy of this queryset that excludes all rows matching the conditions.
Pass `prewhere=True` to apply the conditions as PREWHERE instead of WHERE.
"""
qs = copy(self)
qs._q = list(self._q) + [~Q(**filter_fields)]
return qs
return self._filter_or_exclude(*q, _inverse=True, **kwargs)
def paginate(self, page_num=1, page_size=100):
"""
@ -459,7 +549,8 @@ class AggregateQuerySet(QuerySet):
self._grouping_fields = grouping_fields
self._calculated_fields = calculated_fields
self._order_by = list(base_qs._order_by)
self._q = list(base_qs._q)
self._where_q = base_qs._where_q
self._prewhere_q = base_qs._prewhere_q
self._limits = base_qs._limits
self._distinct = base_qs._distinct
@ -488,26 +579,11 @@ class AggregateQuerySet(QuerySet):
"""
raise NotImplementedError('Cannot re-aggregate an AggregateQuerySet')
def as_sql(self):
def select_fields_as_sql(self):
"""
Returns the whole query as a SQL string.
Returns the selected fields or expressions as a SQL string.
"""
distinct = 'DISTINCT ' if self._distinct else ''
grouping = comma_join('`%s`' % field for field in self._grouping_fields)
fields = comma_join(list(self._fields) + ['%s AS %s' % (v, k) for k, v in self._calculated_fields.items()])
params = dict(
distinct=distinct,
grouping=grouping or "''",
fields=fields,
table=self._model_cls.table_name(),
conds=self.conditions_as_sql()
)
sql = u'SELECT %(distinct)s%(fields)s\nFROM `%(table)s`\nWHERE %(conds)s\nGROUP BY %(grouping)s' % params
if self._order_by:
sql += '\nORDER BY ' + self.order_by_as_sql()
if self._limits:
sql += '\nLIMIT %d, %d' % self._limits
return sql
return comma_join(list(self._fields) + ['%s AS %s' % (v, k) for k, v in self._calculated_fields.items()])
def __iter__(self):
return self._database.select(self.as_sql()) # using an ad-hoc model
@ -519,3 +595,13 @@ class AggregateQuerySet(QuerySet):
sql = u'SELECT count() FROM (%s)' % self.as_sql()
raw = self._database.raw(sql)
return int(raw) if raw else 0
def with_totals(self):
"""
Adds WITH TOTALS modifier ot GROUP BY, making query return extra row
with aggregate function calculated across all the rows. More information:
https://clickhouse.yandex/docs/en/query_language/select/#with-totals-modifier
"""
qs = copy(self)
qs._grouping_with_totals = True
return qs

4
src/infi/clickhouse_orm/system_models.py
View File

@ -19,8 +19,8 @@ class SystemPart(Model):
"""
OPERATIONS = frozenset({'DETACH', 'DROP', 'ATTACH', 'FREEZE', 'FETCH'})
readonly = True
system = True
_readonly = True
_system = True
database = StringField() # Name of the database where the table that this part belongs to is located.
table = StringField() # Name of the table that this part belongs to.

2
tests/base_test_with_data.py
View File

@ -14,7 +14,7 @@ logging.getLogger("requests").setLevel(logging.WARNING)
class TestCaseWithData(unittest.TestCase):
def setUp(self):
self.database = Database('test-db')
self.database = Database('test-db', log_statements=True)
self.database.create_table(Person)
def tearDown(self):

2
tests/test_alias_fields.py
View File

@ -11,7 +11,7 @@ from infi.clickhouse_orm.engines import *
class MaterializedFieldsTest(unittest.TestCase):
def setUp(self):
self.database = Database('test-db')
self.database = Database('test-db', log_statements=True)
self.database.create_table(ModelWithAliasFields)
def tearDown(self):

2
tests/test_array_fields.py
View File

@ -11,7 +11,7 @@ from infi.clickhouse_orm.engines import *
class ArrayFieldsTest(unittest.TestCase):
def setUp(self):
self.database = Database('test-db')
self.database = Database('test-db', log_statements=True)
self.database.create_table(ModelWithArrays)
def tearDown(self):

65
tests/test_custom_fields.py
View File

@ -1,18 +1,15 @@
from __future__ import unicode_literals
import unittest
import six
from uuid import UUID
from infi.clickhouse_orm.database import Database
from infi.clickhouse_orm.fields import Field, Int16Field
from infi.clickhouse_orm.models import Model
from infi.clickhouse_orm.engines import Memory
from infi.clickhouse_orm.utils import escape
class CustomFieldsTest(unittest.TestCase):
def setUp(self):
self.database = Database('test-db')
self.database = Database('test-db', log_statements=True)
def tearDown(self):
self.database.drop_database()
@ -35,37 +32,6 @@ class CustomFieldsTest(unittest.TestCase):
with self.assertRaises(ValueError):
TestModel(i=1, f=value)
def test_uuid_field(self):
# Create a model
class TestModel(Model):
i = Int16Field()
f = UUIDField()
engine = Memory()
self.database.create_table(TestModel)
# Check valid values (all values are the same UUID)
values = [
'{12345678-1234-5678-1234-567812345678}',
'12345678123456781234567812345678',
'urn:uuid:12345678-1234-5678-1234-567812345678',
'\x12\x34\x56\x78'*4,
(0x12345678, 0x1234, 0x5678, 0x12, 0x34, 0x567812345678),
0x12345678123456781234567812345678,
]
for index, value in enumerate(values):
rec = TestModel(i=index, f=value)
self.database.insert([rec])
for rec in TestModel.objects_in(self.database):
self.assertEqual(rec.f, UUID(values[0]))
# Check that ClickHouse encoding functions are supported
for rec in self.database.select("SELECT i, UUIDNumToString(f) AS f FROM testmodel", TestModel):
self.assertEqual(rec.f, UUID(values[0]))
for rec in self.database.select("SELECT 1 as i, UUIDStringToNum('12345678-1234-5678-1234-567812345678') AS f", TestModel):
self.assertEqual(rec.f, UUID(values[0]))
# Check invalid values
for value in [None, 'zzz', -1, '123']:
with self.assertRaises(ValueError):
TestModel(i=1, f=value)
class BooleanField(Field):
@ -88,32 +54,3 @@ class BooleanField(Field):
# The value was already converted by to_python, so it's a bool
return '1' if value else '0'
class UUIDField(Field):
# The ClickHouse column type to use
db_type = 'FixedString(16)'
# The default value if empty
class_default = UUID(int=0)
def to_python(self, value, timezone_in_use):
# Convert valid values to UUID instance
if isinstance(value, UUID):
return value
elif isinstance(value, six.string_types):
return UUID(bytes=value.encode('latin1')) if len(value) == 16 else UUID(value)
elif isinstance(value, six.integer_types):
return UUID(int=value)
elif isinstance(value, tuple):
return UUID(fields=value)
else:
raise ValueError('Invalid value for UUIDField: %r' % value)
def to_db_string(self, value, quote=True):
# The value was already converted by to_python, so it's a UUID instance
val = value.bytes
if six.PY3:
val = str(val, 'latin1')
return escape(val, quote)

12
tests/test_database.py
View File

@ -112,6 +112,14 @@ class DatabaseTestCase(TestCaseWithData):
self.assertEqual([obj.to_tsv() for obj in page_a.objects],
[obj.to_tsv() for obj in page_b.objects])
def test_pagination_empty_page(self):
for page_num in (-1, 1, 2):
page = self.database.paginate(Person, 'first_name, last_name', page_num, 10, conditions="first_name = 'Ziggy'")
self.assertEqual(page.number_of_objects, 0)
self.assertEqual(page.objects, [])
self.assertEqual(page.pages_total, 0)
self.assertEqual(page.number, max(page_num, 1))
def test_pagination_invalid_page(self):
self._insert_and_check(self._sample_data(), len(data))
for page_num in (0, -2, -100):
@ -142,7 +150,7 @@ class DatabaseTestCase(TestCaseWithData):
exc = cm.exception
self.assertEqual(exc.code, 193)
self.assertEqual(exc.message, 'Wrong password for user default')
self.assertTrue(exc.message.startswith('Wrong password for user default'))
def test_nonexisting_db(self):
db = Database('db_not_here', autocreate=False)
@ -150,7 +158,7 @@ class DatabaseTestCase(TestCaseWithData):
db.create_table(Person)
exc = cm.exception
self.assertEqual(exc.code, 81)
self.assertEqual(exc.message, "Database db_not_here doesn't exist")
self.assertTrue(exc.message.startswith("Database db_not_here doesn't exist"))
# Create and delete the db twice, to ensure db_exists gets updated
for i in range(2):
# Now create the database - should succeed

2
tests/test_datetime_fields.py
View File

@ -10,7 +10,7 @@ from infi.clickhouse_orm.engines import *
class DateFieldsTest(unittest.TestCase):
def setUp(self):
self.database = Database('test-db')
self.database = Database('test-db', log_statements=True)
self.database.create_table(ModelWithDate)
def tearDown(self):

2
tests/test_decimal_fields.py
View File

@ -12,7 +12,7 @@ from infi.clickhouse_orm.engines import *
class DecimalFieldsTest(unittest.TestCase):
def setUp(self):
self.database = Database('test-db')
self.database = Database('test-db', log_statements=True)
self.database.add_setting('allow_experimental_decimal_type', 1)
try:
self.database.create_table(DecimalModel)

8
tests/test_engines.py
View File

@ -14,7 +14,7 @@ logging.getLogger("requests").setLevel(logging.WARNING)
class _EnginesHelperTestCase(unittest.TestCase):
def setUp(self):
self.database = Database('test-db')
self.database = Database('test-db', log_statements=True)
def tearDown(self):
self.database.drop_database()
@ -115,8 +115,8 @@ class EnginesTestCase(_EnginesHelperTestCase):
TestModel2(date='2017-01-02', event_id=2, event_group=2, event_count=2, event_version=2)
])
# event_uversion is materialized field. So * won't select it and it will be zero
res = self.database.select('SELECT *, event_uversion FROM $table ORDER BY event_id', model_class=TestMergeModel)
res = [row for row in res]
res = self.database.select('SELECT *, _table, event_uversion FROM $table ORDER BY event_id', model_class=TestMergeModel)
res = list(res)
self.assertEqual(2, len(res))
self.assertDictEqual({
'_table': 'testmodel1',
@ -209,7 +209,7 @@ class DistributedTestCase(_EnginesHelperTestCase):
exc = cm.exception
self.assertEqual(exc.code, 170)
self.assertEqual(exc.message, "Requested cluster 'cluster_name' not found")
self.assertTrue(exc.message.startswith("Requested cluster 'cluster_name' not found"))
def test_verbose_engine_two_superclasses(self):
class TestModel2(SampleModel):

2
tests/test_enum_fields.py
View File

@ -15,7 +15,7 @@ except NameError:
class EnumFieldsTest(unittest.TestCase):
def setUp(self):
self.database = Database('test-db')
self.database = Database('test-db', log_statements=True)
self.database.create_table(ModelWithEnum)
self.database.create_table(ModelWithEnumArray)

2
tests/test_fixed_string_fields.py
View File

@ -11,7 +11,7 @@ from infi.clickhouse_orm.engines import *
class FixedStringFieldsTest(unittest.TestCase):
def setUp(self):
self.database = Database('test-db')
self.database = Database('test-db', log_statements=True)
self.database.create_table(FixedStringModel)
def tearDown(self):

2
tests/test_join.py
View File

@ -9,7 +9,7 @@ from infi.clickhouse_orm import database, engines, fields, models
class JoinTest(unittest.TestCase):
def setUp(self):
self.database = database.Database('test-db')
self.database = database.Database('test-db', log_statements=True)
self.database.create_table(Foo)
self.database.create_table(Bar)
self.database.insert([Foo(id=i) for i in range(3)])

2
tests/test_materialized_fields.py
View File

@ -11,7 +11,7 @@ from infi.clickhouse_orm.engines import *
class MaterializedFieldsTest(unittest.TestCase):
def setUp(self):
self.database = Database('test-db')
self.database = Database('test-db', log_statements=True)
self.database.create_table(ModelWithMaterializedFields)
def tearDown(self):

2
tests/test_migrations.py
View File

@ -24,7 +24,7 @@ logging.getLogger("requests").setLevel(logging.WARNING)
class MigrationsTestCase(unittest.TestCase):
def setUp(self):
self.database = Database('test-db')
self.database = Database('test-db', log_statements=True)
self.database.drop_table(MigrationHistory)
def tearDown(self):

25
tests/test_nullable_fields.py
View File

@ -6,6 +6,7 @@ from infi.clickhouse_orm.database import Database
from infi.clickhouse_orm.models import Model
from infi.clickhouse_orm.fields import *
from infi.clickhouse_orm.engines import *
from infi.clickhouse_orm.utils import comma_join
from datetime import date, datetime
@ -13,7 +14,7 @@ from datetime import date, datetime
class NullableFieldsTest(unittest.TestCase):
def setUp(self):
self.database = Database('test-db')
self.database = Database('test-db', log_statements=True)
self.database.create_table(ModelWithNullable)
def tearDown(self):
@ -95,14 +96,19 @@ class NullableFieldsTest(unittest.TestCase):
ModelWithNullable(date_field='2016-08-30', null_str='', null_int=42, null_date=dt),
ModelWithNullable(date_field='2016-08-30', null_str='nothing', null_int=None, null_date=None),
ModelWithNullable(date_field='2016-08-31', null_str=None, null_int=42, null_date=dt),
ModelWithNullable(date_field='2016-08-31', null_str=None, null_int=None, null_date=None)
ModelWithNullable(date_field='2016-08-31', null_str=None, null_int=None, null_date=None, null_default=None)
])
def _assert_sample_data(self, results):
for r in results:
print(r.to_dict())
dt = date(1970, 1, 1)
self.assertEqual(len(results), 4)
self.assertIsNone(results[0].null_str)
self.assertEqual(results[0].null_int, 42)
self.assertEqual(results[0].null_default, 7)
self.assertEqual(results[0].null_alias, 21)
self.assertEqual(results[0].null_materialized, 420)
self.assertEqual(results[0].null_date, dt)
self.assertIsNone(results[1].null_date)
self.assertEqual(results[1].null_str, 'nothing')
@ -110,19 +116,27 @@ class NullableFieldsTest(unittest.TestCase):
self.assertIsNone(results[2].null_str)
self.assertEqual(results[2].null_date, dt)
self.assertEqual(results[2].null_int, 42)
self.assertEqual(results[2].null_default, 7)
self.assertEqual(results[2].null_alias, 21)
self.assertEqual(results[0].null_materialized, 420)
self.assertIsNone(results[3].null_int)
self.assertIsNone(results[3].null_default)
self.assertIsNone(results[3].null_alias)
self.assertIsNone(results[3].null_materialized)
self.assertIsNone(results[3].null_str)
self.assertIsNone(results[3].null_date)
def test_insert_and_select(self):
self._insert_sample_data()
query = 'SELECT * from $table ORDER BY date_field'
fields = comma_join(ModelWithNullable.fields().keys())
query = 'SELECT %s from $table ORDER BY date_field' % fields
results = list(self.database.select(query, ModelWithNullable))
self._assert_sample_data(results)
def test_ad_hoc_model(self):
self._insert_sample_data()
query = 'SELECT * from $db.modelwithnullable ORDER BY date_field'
fields = comma_join(ModelWithNullable.fields().keys())
query = 'SELECT %s from $db.modelwithnullable ORDER BY date_field' % fields
results = list(self.database.select(query))
self._assert_sample_data(results)
@ -133,5 +147,8 @@ class ModelWithNullable(Model):
null_str = NullableField(StringField(), extra_null_values={''})
null_int = NullableField(Int32Field())
null_date = NullableField(DateField())
null_default = NullableField(Int32Field(), default=7)
null_alias = NullableField(Int32Field(), alias='null_int/2')
null_materialized = NullableField(Int32Field(), alias='null_int*10')
engine = MergeTree('date_field', ('date_field',))

18
tests/test_querysets.py
View File

@ -29,6 +29,13 @@ class QuerySetTestCase(TestCaseWithData):
self.assertEqual(count, expected_count)
self.assertEqual(qs.count(), expected_count)
def test_prewhere(self):
# We can't distinguish prewhere and where results, it affects performance only.
# So let's control prewhere acts like where does
qs = Person.objects_in(self.database)
self.assertTrue(qs.filter(first_name='Connor', prewhere=True))
self.assertFalse(qs.filter(first_name='Willy', prewhere=True))
def test_no_filtering(self):
qs = Person.objects_in(self.database)
self._test_qs(qs, len(data))
@ -404,6 +411,17 @@ class AggregateTestCase(TestCaseWithData):
print(qs.as_sql())
self.assertEqual(qs.count(), 1)
def test_aggregate_with_totals(self):
qs = Person.objects_in(self.database).aggregate('first_name', count='count()').\
with_totals().order_by('-count')[:5]
print(qs.as_sql())
result = list(qs)
self.assertEqual(len(result), 6)
for row in result[:-1]:
self.assertEqual(2, row.count)
self.assertEqual(100, result[-1].count)
def test_double_underscore_field(self):
class Mdl(Model):
the__number = Int32Field()

4
tests/test_readonly.py
View File

@ -36,9 +36,9 @@ class ReadonlyTestCase(TestCaseWithData):
def _check_db_readonly_err(self, exc, drop_table=None):
self.assertEqual(exc.code, 164)
if drop_table:
self.assertEqual(exc.message, 'Cannot drop table in readonly mode')
self.assertTrue(exc.message.startswith('Cannot drop table in readonly mode'))
else:
self.assertEqual(exc.message, 'Cannot insert into table in readonly mode')
self.assertTrue(exc.message.startswith('Cannot insert into table in readonly mode'))
def test_readonly_db_with_default_user(self):
self._test_readonly_db('default')

32
tests/test_server_errors.py Normal file
View File

@ -0,0 +1,32 @@
from __future__ import unicode_literals
import unittest
from infi.clickhouse_orm.database import ServerError
class ServerErrorTest(unittest.TestCase):
def test_old_format(self):
code, msg = ServerError.get_error_code_msg("Code: 81, e.displayText() = DB::Exception: Database db_not_here doesn't exist, e.what() = DB::Exception (from [::1]:33458)")
self.assertEqual(code, 81)
self.assertEqual(msg, "Database db_not_here doesn't exist")
code, msg = ServerError.get_error_code_msg("Code: 161, e.displayText() = DB::Exception: Limit for number of columns to read exceeded. Requested: 11, maximum: 1, e.what() = DB::Exception\n")
self.assertEqual(code, 161)
self.assertEqual(msg, "Limit for number of columns to read exceeded. Requested: 11, maximum: 1")
def test_new_format(self):
code, msg = ServerError.get_error_code_msg("Code: 164, e.displayText() = DB::Exception: Cannot drop table in readonly mode")
self.assertEqual(code, 164)
self.assertEqual(msg, "Cannot drop table in readonly mode")
code, msg = ServerError.get_error_code_msg("Code: 48, e.displayText() = DB::Exception: Method write is not supported by storage Merge")
self.assertEqual(code, 48)
self.assertEqual(msg, "Method write is not supported by storage Merge")
code, msg = ServerError.get_error_code_msg("Code: 60, e.displayText() = DB::Exception: Table default.zuzu doesn't exist.\n")
self.assertEqual(code, 60)
self.assertEqual(msg, "Table default.zuzu doesn't exist.")

18
tests/test_system_models.py
View File

@ -13,8 +13,9 @@ from infi.clickhouse_orm.system_models import SystemPart
class SystemTest(unittest.TestCase):
def setUp(self):
self.database = Database('test-db')
self.database = Database('test-db', log_statements=True)
def tearDown(self):
self.database.drop_database()
@ -38,7 +39,7 @@ class SystemPartTest(unittest.TestCase):
BACKUP_DIRS = ['/var/lib/clickhouse/shadow', '/opt/clickhouse/shadow/']
def setUp(self):
self.database = Database('test-db')
self.database = Database('test-db', log_statements=True)
self.database.create_table(TestTable)
self.database.create_table(CustomPartitionedTable)
self.database.insert([TestTable(date_field=date.today())])
@ -54,6 +55,12 @@ class SystemPartTest(unittest.TestCase):
return dirnames
raise unittest.SkipTest('Cannot find backups dir')
def test_is_read_only(self):
self.assertTrue(SystemPart.is_read_only())
def test_is_system_model(self):
self.assertTrue(SystemPart.is_system_model())
def test_get_all(self):
parts = SystemPart.get(self.database)
self.assertEqual(len(list(parts)), 2)
@ -62,7 +69,8 @@ class SystemPartTest(unittest.TestCase):
parts = list(SystemPart.get_active(self.database))
self.assertEqual(len(parts), 2)
parts[0].detach()
self.assertEqual(len(list(SystemPart.get_active(self.database))), 1)
parts = list(SystemPart.get_active(self.database))
self.assertEqual(len(parts), 1)
def test_get_conditions(self):
parts = list(SystemPart.get(self.database, conditions="table='testtable'"))
@ -101,6 +109,10 @@ class SystemPartTest(unittest.TestCase):
# TODO Not tested, as I have no replication set
pass
def test_query(self):
SystemPart.objects_in(self.database).count()
list(SystemPart.objects_in(self.database).filter(table='testtable'))
class TestTable(Model):
date_field = DateField()

45
tests/test_uuid_fields.py Normal file
View File

@ -0,0 +1,45 @@
from __future__ import unicode_literals
import unittest
from uuid import UUID
from infi.clickhouse_orm.database import Database
from infi.clickhouse_orm.fields import Int16Field, UUIDField
from infi.clickhouse_orm.models import Model
from infi.clickhouse_orm.engines import Memory
class UUIDFieldsTest(unittest.TestCase):
def setUp(self):
self.database = Database('test-db', log_statements=True)
def tearDown(self):
self.database.drop_database()
def test_uuid_field(self):
# Create a model
class TestModel(Model):
i = Int16Field()
f = UUIDField()
engine = Memory()
self.database.create_table(TestModel)
# Check valid values (all values are the same UUID)
values = [
'12345678-1234-5678-1234-567812345678',
'{12345678-1234-5678-1234-567812345678}',
'12345678123456781234567812345678',
'urn:uuid:12345678-1234-5678-1234-567812345678',
b'\x12\x34\x56\x78'*4,
(0x12345678, 0x1234, 0x5678, 0x12, 0x34, 0x567812345678),
0x12345678123456781234567812345678,
UUID(int=0x12345678123456781234567812345678),
]
for index, value in enumerate(values):
rec = TestModel(i=index, f=value)
self.database.insert([rec])
for rec in TestModel.objects_in(self.database):
self.assertEqual(rec.f, UUID(values[0]))
# Check invalid values
for value in [None, 'zzz', -1, '123']:
with self.assertRaises(ValueError):
TestModel(i=1, f=value)