mirror of https://github.com/morpheus65535/bazarr
469 lines
13 KiB
Python
469 lines
13 KiB
Python
|
"""
|
||
|
Collection of postgres-specific extensions, currently including:
|
||
|
|
||
|
* Support for hstore, a key/value type storage
|
||
|
"""
|
||
|
import logging
|
||
|
import uuid
|
||
|
|
||
|
from peewee import *
|
||
|
from peewee import ColumnBase
|
||
|
from peewee import Expression
|
||
|
from peewee import Node
|
||
|
from peewee import NodeList
|
||
|
from peewee import SENTINEL
|
||
|
from peewee import __exception_wrapper__
|
||
|
|
||
|
try:
|
||
|
from psycopg2cffi import compat
|
||
|
compat.register()
|
||
|
except ImportError:
|
||
|
pass
|
||
|
|
||
|
from psycopg2.extras import register_hstore
|
||
|
try:
|
||
|
from psycopg2.extras import Json
|
||
|
except:
|
||
|
Json = None
|
||
|
|
||
|
|
||
|
logger = logging.getLogger('peewee')
|
||
|
|
||
|
|
||
|
HCONTAINS_DICT = '@>'
|
||
|
HCONTAINS_KEYS = '?&'
|
||
|
HCONTAINS_KEY = '?'
|
||
|
HCONTAINS_ANY_KEY = '?|'
|
||
|
HKEY = '->'
|
||
|
HUPDATE = '||'
|
||
|
ACONTAINS = '@>'
|
||
|
ACONTAINS_ANY = '&&'
|
||
|
TS_MATCH = '@@'
|
||
|
JSONB_CONTAINS = '@>'
|
||
|
JSONB_CONTAINED_BY = '<@'
|
||
|
JSONB_CONTAINS_KEY = '?'
|
||
|
JSONB_CONTAINS_ANY_KEY = '?|'
|
||
|
JSONB_CONTAINS_ALL_KEYS = '?&'
|
||
|
JSONB_EXISTS = '?'
|
||
|
JSONB_REMOVE = '-'
|
||
|
|
||
|
|
||
|
class _LookupNode(ColumnBase):
|
||
|
def __init__(self, node, parts):
|
||
|
self.node = node
|
||
|
self.parts = parts
|
||
|
super(_LookupNode, self).__init__()
|
||
|
|
||
|
def clone(self):
|
||
|
return type(self)(self.node, list(self.parts))
|
||
|
|
||
|
|
||
|
class _JsonLookupBase(_LookupNode):
|
||
|
def __init__(self, node, parts, as_json=False):
|
||
|
super(_JsonLookupBase, self).__init__(node, parts)
|
||
|
self._as_json = as_json
|
||
|
|
||
|
def clone(self):
|
||
|
return type(self)(self.node, list(self.parts), self._as_json)
|
||
|
|
||
|
@Node.copy
|
||
|
def as_json(self, as_json=True):
|
||
|
self._as_json = as_json
|
||
|
|
||
|
def concat(self, rhs):
|
||
|
return Expression(self.as_json(True), OP.CONCAT, Json(rhs))
|
||
|
|
||
|
def contains(self, other):
|
||
|
clone = self.as_json(True)
|
||
|
if isinstance(other, (list, dict)):
|
||
|
return Expression(clone, JSONB_CONTAINS, Json(other))
|
||
|
return Expression(clone, JSONB_EXISTS, other)
|
||
|
|
||
|
def contains_any(self, *keys):
|
||
|
return Expression(
|
||
|
self.as_json(True),
|
||
|
JSONB_CONTAINS_ANY_KEY,
|
||
|
Value(list(keys), unpack=False))
|
||
|
|
||
|
def contains_all(self, *keys):
|
||
|
return Expression(
|
||
|
self.as_json(True),
|
||
|
JSONB_CONTAINS_ALL_KEYS,
|
||
|
Value(list(keys), unpack=False))
|
||
|
|
||
|
def has_key(self, key):
|
||
|
return Expression(self.as_json(True), JSONB_CONTAINS_KEY, key)
|
||
|
|
||
|
|
||
|
class JsonLookup(_JsonLookupBase):
|
||
|
def __getitem__(self, value):
|
||
|
return JsonLookup(self.node, self.parts + [value], self._as_json)
|
||
|
|
||
|
def __sql__(self, ctx):
|
||
|
ctx.sql(self.node)
|
||
|
for part in self.parts[:-1]:
|
||
|
ctx.literal('->').sql(part)
|
||
|
if self.parts:
|
||
|
(ctx
|
||
|
.literal('->' if self._as_json else '->>')
|
||
|
.sql(self.parts[-1]))
|
||
|
|
||
|
return ctx
|
||
|
|
||
|
|
||
|
class JsonPath(_JsonLookupBase):
|
||
|
def __sql__(self, ctx):
|
||
|
return (ctx
|
||
|
.sql(self.node)
|
||
|
.literal('#>' if self._as_json else '#>>')
|
||
|
.sql(Value('{%s}' % ','.join(map(str, self.parts)))))
|
||
|
|
||
|
|
||
|
class ObjectSlice(_LookupNode):
|
||
|
@classmethod
|
||
|
def create(cls, node, value):
|
||
|
if isinstance(value, slice):
|
||
|
parts = [value.start or 0, value.stop or 0]
|
||
|
elif isinstance(value, int):
|
||
|
parts = [value]
|
||
|
else:
|
||
|
parts = map(int, value.split(':'))
|
||
|
return cls(node, parts)
|
||
|
|
||
|
def __sql__(self, ctx):
|
||
|
return (ctx
|
||
|
.sql(self.node)
|
||
|
.literal('[%s]' % ':'.join(str(p + 1) for p in self.parts)))
|
||
|
|
||
|
def __getitem__(self, value):
|
||
|
return ObjectSlice.create(self, value)
|
||
|
|
||
|
|
||
|
class IndexedFieldMixin(object):
|
||
|
default_index_type = 'GIN'
|
||
|
|
||
|
def __init__(self, *args, **kwargs):
|
||
|
kwargs.setdefault('index', True) # By default, use an index.
|
||
|
super(IndexedFieldMixin, self).__init__(*args, **kwargs)
|
||
|
|
||
|
|
||
|
class ArrayField(IndexedFieldMixin, Field):
|
||
|
passthrough = True
|
||
|
|
||
|
def __init__(self, field_class=IntegerField, field_kwargs=None,
|
||
|
dimensions=1, convert_values=False, *args, **kwargs):
|
||
|
self.__field = field_class(**(field_kwargs or {}))
|
||
|
self.dimensions = dimensions
|
||
|
self.convert_values = convert_values
|
||
|
self.field_type = self.__field.field_type
|
||
|
super(ArrayField, self).__init__(*args, **kwargs)
|
||
|
|
||
|
def bind(self, model, name, set_attribute=True):
|
||
|
ret = super(ArrayField, self).bind(model, name, set_attribute)
|
||
|
self.__field.bind(model, '__array_%s' % name, False)
|
||
|
return ret
|
||
|
|
||
|
def ddl_datatype(self, ctx):
|
||
|
data_type = self.__field.ddl_datatype(ctx)
|
||
|
return NodeList((data_type, SQL('[]' * self.dimensions)), glue='')
|
||
|
|
||
|
def db_value(self, value):
|
||
|
if value is None or isinstance(value, Node):
|
||
|
return value
|
||
|
elif self.convert_values:
|
||
|
return self._process(self.__field.db_value, value, self.dimensions)
|
||
|
else:
|
||
|
return value if isinstance(value, list) else list(value)
|
||
|
|
||
|
def python_value(self, value):
|
||
|
if self.convert_values and value is not None:
|
||
|
conv = self.__field.python_value
|
||
|
if isinstance(value, list):
|
||
|
return self._process(conv, value, self.dimensions)
|
||
|
else:
|
||
|
return conv(value)
|
||
|
else:
|
||
|
return value
|
||
|
|
||
|
def _process(self, conv, value, dimensions):
|
||
|
dimensions -= 1
|
||
|
if dimensions == 0:
|
||
|
return [conv(v) for v in value]
|
||
|
else:
|
||
|
return [self._process(conv, v, dimensions) for v in value]
|
||
|
|
||
|
def __getitem__(self, value):
|
||
|
return ObjectSlice.create(self, value)
|
||
|
|
||
|
def _e(op):
|
||
|
def inner(self, rhs):
|
||
|
return Expression(self, op, ArrayValue(self, rhs))
|
||
|
return inner
|
||
|
__eq__ = _e(OP.EQ)
|
||
|
__ne__ = _e(OP.NE)
|
||
|
__gt__ = _e(OP.GT)
|
||
|
__ge__ = _e(OP.GTE)
|
||
|
__lt__ = _e(OP.LT)
|
||
|
__le__ = _e(OP.LTE)
|
||
|
__hash__ = Field.__hash__
|
||
|
|
||
|
def contains(self, *items):
|
||
|
return Expression(self, ACONTAINS, ArrayValue(self, items))
|
||
|
|
||
|
def contains_any(self, *items):
|
||
|
return Expression(self, ACONTAINS_ANY, ArrayValue(self, items))
|
||
|
|
||
|
|
||
|
class ArrayValue(Node):
|
||
|
def __init__(self, field, value):
|
||
|
self.field = field
|
||
|
self.value = value
|
||
|
|
||
|
def __sql__(self, ctx):
|
||
|
return (ctx
|
||
|
.sql(Value(self.value, unpack=False))
|
||
|
.literal('::')
|
||
|
.sql(self.field.ddl_datatype(ctx)))
|
||
|
|
||
|
|
||
|
class DateTimeTZField(DateTimeField):
|
||
|
field_type = 'TIMESTAMPTZ'
|
||
|
|
||
|
|
||
|
class HStoreField(IndexedFieldMixin, Field):
|
||
|
field_type = 'HSTORE'
|
||
|
__hash__ = Field.__hash__
|
||
|
|
||
|
def __getitem__(self, key):
|
||
|
return Expression(self, HKEY, Value(key))
|
||
|
|
||
|
def keys(self):
|
||
|
return fn.akeys(self)
|
||
|
|
||
|
def values(self):
|
||
|
return fn.avals(self)
|
||
|
|
||
|
def items(self):
|
||
|
return fn.hstore_to_matrix(self)
|
||
|
|
||
|
def slice(self, *args):
|
||
|
return fn.slice(self, Value(list(args), unpack=False))
|
||
|
|
||
|
def exists(self, key):
|
||
|
return fn.exist(self, key)
|
||
|
|
||
|
def defined(self, key):
|
||
|
return fn.defined(self, key)
|
||
|
|
||
|
def update(self, **data):
|
||
|
return Expression(self, HUPDATE, data)
|
||
|
|
||
|
def delete(self, *keys):
|
||
|
return fn.delete(self, Value(list(keys), unpack=False))
|
||
|
|
||
|
def contains(self, value):
|
||
|
if isinstance(value, dict):
|
||
|
rhs = Value(value, unpack=False)
|
||
|
return Expression(self, HCONTAINS_DICT, rhs)
|
||
|
elif isinstance(value, (list, tuple)):
|
||
|
rhs = Value(value, unpack=False)
|
||
|
return Expression(self, HCONTAINS_KEYS, rhs)
|
||
|
return Expression(self, HCONTAINS_KEY, value)
|
||
|
|
||
|
def contains_any(self, *keys):
|
||
|
return Expression(self, HCONTAINS_ANY_KEY, Value(list(keys),
|
||
|
unpack=False))
|
||
|
|
||
|
|
||
|
class JSONField(Field):
|
||
|
field_type = 'JSON'
|
||
|
|
||
|
def __init__(self, dumps=None, *args, **kwargs):
|
||
|
if Json is None:
|
||
|
raise Exception('Your version of psycopg2 does not support JSON.')
|
||
|
self.dumps = dumps
|
||
|
super(JSONField, self).__init__(*args, **kwargs)
|
||
|
|
||
|
def db_value(self, value):
|
||
|
if value is None:
|
||
|
return value
|
||
|
if not isinstance(value, Json):
|
||
|
return Json(value, dumps=self.dumps)
|
||
|
return value
|
||
|
|
||
|
def __getitem__(self, value):
|
||
|
return JsonLookup(self, [value])
|
||
|
|
||
|
def path(self, *keys):
|
||
|
return JsonPath(self, keys)
|
||
|
|
||
|
def concat(self, value):
|
||
|
return super(JSONField, self).concat(Json(value))
|
||
|
|
||
|
|
||
|
def cast_jsonb(node):
|
||
|
return NodeList((node, SQL('::jsonb')), glue='')
|
||
|
|
||
|
|
||
|
class BinaryJSONField(IndexedFieldMixin, JSONField):
|
||
|
field_type = 'JSONB'
|
||
|
__hash__ = Field.__hash__
|
||
|
|
||
|
def contains(self, other):
|
||
|
if isinstance(other, (list, dict)):
|
||
|
return Expression(self, JSONB_CONTAINS, Json(other))
|
||
|
return Expression(cast_jsonb(self), JSONB_EXISTS, other)
|
||
|
|
||
|
def contained_by(self, other):
|
||
|
return Expression(cast_jsonb(self), JSONB_CONTAINED_BY, Json(other))
|
||
|
|
||
|
def contains_any(self, *items):
|
||
|
return Expression(
|
||
|
cast_jsonb(self),
|
||
|
JSONB_CONTAINS_ANY_KEY,
|
||
|
Value(list(items), unpack=False))
|
||
|
|
||
|
def contains_all(self, *items):
|
||
|
return Expression(
|
||
|
cast_jsonb(self),
|
||
|
JSONB_CONTAINS_ALL_KEYS,
|
||
|
Value(list(items), unpack=False))
|
||
|
|
||
|
def has_key(self, key):
|
||
|
return Expression(cast_jsonb(self), JSONB_CONTAINS_KEY, key)
|
||
|
|
||
|
def remove(self, *items):
|
||
|
return Expression(
|
||
|
cast_jsonb(self),
|
||
|
JSONB_REMOVE,
|
||
|
Value(list(items), unpack=False))
|
||
|
|
||
|
|
||
|
class TSVectorField(IndexedFieldMixin, TextField):
|
||
|
field_type = 'TSVECTOR'
|
||
|
__hash__ = Field.__hash__
|
||
|
|
||
|
def match(self, query, language=None, plain=False):
|
||
|
params = (language, query) if language is not None else (query,)
|
||
|
func = fn.plainto_tsquery if plain else fn.to_tsquery
|
||
|
return Expression(self, TS_MATCH, func(*params))
|
||
|
|
||
|
|
||
|
def Match(field, query, language=None):
|
||
|
params = (language, query) if language is not None else (query,)
|
||
|
field_params = (language, field) if language is not None else (field,)
|
||
|
return Expression(
|
||
|
fn.to_tsvector(*field_params),
|
||
|
TS_MATCH,
|
||
|
fn.to_tsquery(*params))
|
||
|
|
||
|
|
||
|
class IntervalField(Field):
|
||
|
field_type = 'INTERVAL'
|
||
|
|
||
|
|
||
|
class FetchManyCursor(object):
|
||
|
__slots__ = ('cursor', 'array_size', 'exhausted', 'iterable')
|
||
|
|
||
|
def __init__(self, cursor, array_size=None):
|
||
|
self.cursor = cursor
|
||
|
self.array_size = array_size or cursor.itersize
|
||
|
self.exhausted = False
|
||
|
self.iterable = self.row_gen()
|
||
|
|
||
|
@property
|
||
|
def description(self):
|
||
|
return self.cursor.description
|
||
|
|
||
|
def close(self):
|
||
|
self.cursor.close()
|
||
|
|
||
|
def row_gen(self):
|
||
|
while True:
|
||
|
rows = self.cursor.fetchmany(self.array_size)
|
||
|
if not rows:
|
||
|
return
|
||
|
for row in rows:
|
||
|
yield row
|
||
|
|
||
|
def fetchone(self):
|
||
|
if self.exhausted:
|
||
|
return
|
||
|
try:
|
||
|
return next(self.iterable)
|
||
|
except StopIteration:
|
||
|
self.exhausted = True
|
||
|
|
||
|
|
||
|
class ServerSideQuery(Node):
|
||
|
def __init__(self, query, array_size=None):
|
||
|
self.query = query
|
||
|
self.array_size = array_size
|
||
|
self._cursor_wrapper = None
|
||
|
|
||
|
def __sql__(self, ctx):
|
||
|
return self.query.__sql__(ctx)
|
||
|
|
||
|
def __iter__(self):
|
||
|
if self._cursor_wrapper is None:
|
||
|
self._execute(self.query._database)
|
||
|
return iter(self._cursor_wrapper.iterator())
|
||
|
|
||
|
def _execute(self, database):
|
||
|
if self._cursor_wrapper is None:
|
||
|
cursor = database.execute(self.query, named_cursor=True,
|
||
|
array_size=self.array_size)
|
||
|
self._cursor_wrapper = self.query._get_cursor_wrapper(cursor)
|
||
|
return self._cursor_wrapper
|
||
|
|
||
|
|
||
|
def ServerSide(query, database=None, array_size=None):
|
||
|
if database is None:
|
||
|
database = query._database
|
||
|
with database.transaction():
|
||
|
server_side_query = ServerSideQuery(query, array_size=array_size)
|
||
|
for row in server_side_query:
|
||
|
yield row
|
||
|
|
||
|
|
||
|
class _empty_object(object):
|
||
|
__slots__ = ()
|
||
|
def __nonzero__(self):
|
||
|
return False
|
||
|
__bool__ = __nonzero__
|
||
|
|
||
|
__named_cursor__ = _empty_object()
|
||
|
|
||
|
|
||
|
class PostgresqlExtDatabase(PostgresqlDatabase):
|
||
|
def __init__(self, *args, **kwargs):
|
||
|
self._register_hstore = kwargs.pop('register_hstore', False)
|
||
|
self._server_side_cursors = kwargs.pop('server_side_cursors', False)
|
||
|
super(PostgresqlExtDatabase, self).__init__(*args, **kwargs)
|
||
|
|
||
|
def _connect(self):
|
||
|
conn = super(PostgresqlExtDatabase, self)._connect()
|
||
|
if self._register_hstore:
|
||
|
register_hstore(conn, globally=True)
|
||
|
return conn
|
||
|
|
||
|
def cursor(self, commit=None):
|
||
|
if self.is_closed():
|
||
|
self.connect()
|
||
|
if commit is __named_cursor__:
|
||
|
return self._state.conn.cursor(name=str(uuid.uuid1()))
|
||
|
return self._state.conn.cursor()
|
||
|
|
||
|
def execute(self, query, commit=SENTINEL, named_cursor=False,
|
||
|
array_size=None, **context_options):
|
||
|
ctx = self.get_sql_context(**context_options)
|
||
|
sql, params = ctx.sql(query).query()
|
||
|
named_cursor = named_cursor or (self._server_side_cursors and
|
||
|
sql[:6].lower() == 'select')
|
||
|
if named_cursor:
|
||
|
commit = __named_cursor__
|
||
|
cursor = self.execute_sql(sql, params, commit=commit)
|
||
|
if named_cursor:
|
||
|
cursor = FetchManyCursor(cursor, array_size)
|
||
|
return cursor
|