udf support for monetdbe backend

This commit is contained in:
2022-06-29 17:26:33 +08:00
parent 3b659b99f8
commit ece2a6cc9f
111 changed files with 2540 additions and 780 deletions
+30 -5
View File
@@ -1,12 +1,37 @@
from engine.ast import Context, ast_node
import engine.ddl, engine.projection
saved_cxt = None
def initialize():
return Context()
def initialize(cxt = None, keep = False):
global saved_cxt
if cxt is None or not keep or type(cxt) is not Context:
if saved_cxt is None or not keep:
cxt = Context()
saved_cxt = cxt
else:
cxt = saved_cxt
cxt.new()
return cxt
def generate(ast, cxt):
for k in ast.keys():
if k in ast_node.types.keys():
root = ast_node.types[k](None, ast, cxt)
__all__ = ["initialize", "generate"]
def exec(stmts, cxt = None, keep = None):
cxt = initialize(cxt, keep)
stmts_stmts = stmts['stmts']
if type(stmts_stmts) is list:
for s in stmts_stmts:
generate(s, cxt)
else:
generate(stmts_stmts, cxt)
cxt.Info(cxt.ccode)
with open('out.cpp', 'wb') as outfile:
outfile.write((cxt.finalize()).encode('utf-8'))
return cxt
__all__ = ["initialize", "generate", "exec", "saved_cxt"]
+9 -5
View File
@@ -1,5 +1,6 @@
from engine.utils import base62uuid
from copy import copy
from typing import *
# replace column info with this later.
class ColRef:
def __init__(self, cname, _ty, cobj, cnt, table:'TableInfo', name, id, compound = False):
@@ -65,7 +66,7 @@ class TableInfo:
self.views = set()
#keep track of temp vars
self.local_vars = dict()
self.rec = None
self.rec = None # a hook on get_col_d to record tables being referenced in the process
self.groupinfo = None
self.add_cols(cols)
# runtime
@@ -207,14 +208,17 @@ class Context:
LOG_INFO = 'INFO'
LOG_ERROR = 'ERROR'
LOG_SILENT = 'SILENT'
from engine.types import Types
type_table : Dict[str, Types] = dict()
def new(self):
self.tmp_names = set()
self.udf_map = dict()
self.headers = set(['\"./server/libaquery.h\"'])
self.finalized = False
# read header
self.ccode = ''
self.ccodelet = ''
self.ccode = str()
self.ccodelet = str()
with open('header.cxx', 'r') as outfile:
self.ccode = outfile.read()
# datasource will be availible after `from' clause is parsed
@@ -236,8 +240,8 @@ class Context:
self.log_level = Context.LOG_SILENT
self.print = print
# read header
self.ccode = ''
self.ccodelet = ''
self.ccode = str()
self.ccodelet = str()
self.columns_in_context = dict()
self.tables_in_context = dict()
with open('header.cxx', 'r') as outfile:
+5 -2
View File
@@ -28,11 +28,14 @@ class expr(ast_node):
'xor' : '^',
'gt':'>',
'lt':'<',
'le':'<=',
'gt':'>='
'lte':'<=',
'gte':'>=',
'neq':'!=',
'eq':'=='
}
compound_ops = {
'missing' : ['missing', lambda x: f'{x[0]} == nullval<decays<decltype({x[0]})>>'],
}
unary_ops = {
+222 -26
View File
@@ -1,34 +1,230 @@
from engine.ast import Context
from engine.utils import defval
from typing import Dict, List
type_table: Dict[str, "Types"] = {}
class Types:
name = 'Any'
cname = 'void*'
ctype_name = "types::NONE"
def __init__(self, context:Context):
self.cxt = context
def cast_to(self, *_):
return self
def cast_from(self, *_):
return self
def init_any(self):
self.name : str = 'Any'
self.sqlname : str = 'Int'
self.cname : str = 'void*'
self.ctype_name : str = "types::NONE"
self.null_value = 0
self.priority : int= 0
self.cast_to_dict = dict()
self.cast_from_dict = dict()
def __init__(self, priority = 0, *,
name = None, cname = None, sqlname = None,
ctype_name = None, null_value = None,
fp_type = None, long_type = None, is_fp = False,
cast_to = None, cast_from = None
):
self.is_fp = is_fp
if name is None:
self.init_any()
else:
self.name = name
self.cname = defval(cname, name.lower() + '_t')
self.sqlname = defval(sqlname, name.upper())
self.ctype_name = defval(ctype_name, f'types::{name.upper()}')
self.null_value = defval(null_value, 0)
self.cast_to_dict = defval(cast_to, dict())
self.cast_from_dict = defval(cast_from, dict())
self.priority = priority
self.long_type = defval(long_type, self)
self.fp_type = defval(fp_type, self)
global type_table
type_table[name] = self
def cast_to(self, ty : "Types"):
if ty in self.cast_to_dict:
return self.cast_to_dict[ty.name](ty)
else:
raise Exception(f'Illeagal cast: from {self.name} to {ty.name}.')
def cast_from(self, ty : "Types"):
if ty in self.cast_from_dict:
return self.cast_from_dict[ty.name](ty)
else:
raise Exception(f'Illeagal cast: from {ty.name} to {self.name}.')
def __repr__(self) -> str:
return self.cname
return self.sqlname
def __str__(self) -> str:
return self.sqlname
class TypeCollection:
def __init__(self, sz, deftype, fptype = None, utype = None, *, collection = None) -> None:
self.size = sz
self.type = deftype
self.fptype = fptype
self.utype = utype
self.all_types = [deftype]
if fptype is not None:
self.all_types.append(fptype)
if utype is not None:
self.all_types.append(utype)
if collection is not None:
for ty in collection:
self.all_types.append(ty)
type_table = dict()
AnyT = Types(0)
LazyT = Types(240, name = 'Lazy', cname = '', sqlname = '', ctype_name = '')
DoubleT = Types(17, name = 'double', cname='double', sqlname = 'DOUBLE', is_fp = True)
FloatT = Types(16, name = 'float', cname = 'float', sqlname = 'REAL',
long_type = DoubleT, is_fp = True)
LongT = Types(4, name = 'int64', sqlname = 'BIGINT', fp_type = DoubleT)
ByteT = Types(1, name = 'int8', sqlname = 'TINYINT', long_type=LongT, fp_type=FloatT)
ShortT = Types(2, name = 'int16', sqlname='SMALLINT', long_type=LongT, fp_type=FloatT)
IntT = Types(3, name = 'int', cname = 'int', long_type=LongT, fp_type=FloatT)
StrT = Types(200, name = 'str', cname = 'const char*', sqlname='VARCHAR', ctype_name = 'types::STRING')
def _ty_make_dict(fn : str, *ty : Types):
return {eval(fn):t for t in ty}
class String(Types):
name = 'String'
cname = 'const char*'
ctype_name = "types::ASTR"
def cast_from(self, ty, val, container = None):
if type(ty) is Int:
self.cxt.emit()
int_types : Dict[str, Types] = _ty_make_dict('t.sqlname.lower()', LongT, ByteT, ShortT, IntT)
fp_types : Dict[str, Types] = _ty_make_dict('t.sqlname.lower()', FloatT, DoubleT)
builtin_types : Dict[str, Types] = {**_ty_make_dict('t.sqlname.lower()', AnyT, StrT), **int_types, **fp_types}
class Int(Types):
name = "Int"
cname = "int"
ctype_name = "types::AINT"
type_bylength : Dict[int, TypeCollection] = {}
type_bylength[1] = TypeCollection(1, ByteT)
type_bylength[2] = TypeCollection(2, ShortT)
type_bylength[4] = TypeCollection(4, IntT, FloatT)
type_bylength[8] = TypeCollection(8, LongT, DoubleT, collection=[AnyT])
class OperatorBase:
def extending_type(ops:Types):
return ops.long_type
def fraction_type (ops:Types):
return ops.fp_type
def __init__(self, opname, n_ops, return_fx, * ,
optypes = None, cname = None, sqlname = None,
call = None):
self.name = opname
self.cname = defval(cname, opname)
self.sqlname = defval(sqlname, opname.upper())
self.n_ops = n_ops
self.optypes = optypes
self.return_type = defval(return_fx, lambda: self.optypes[0])
self.call = defval(call, lambda _, c_code = False, *args:
f'{self.cname if c_code else self.sqlname}({", ". join(args)})')
def __call__(self, c_code = False, *args) -> str:
return self.call(self, c_code, *args)
def get_return_type(self, inputs):
return self.return_type(inputs)
class Float(Types):
name = "Float"
cname = "float"
ctype_name = "types::AFLOAT"
def __repr__(self) -> str:
return self.name
def __str__(self) -> str:
return self.name
# TODO: Type checks, Type catagories, e.g.: value type, etc.
# return type deduction
def auto_extension(*args : Types) -> Types:
final_type = AnyT
is_fp = False
for a in args:
if not is_fp and a.is_fp:
is_fp = True
final_type = final_type.fp_type
elif is_fp:
a = a.fp_type
final_type = a if a.priority > final_type.priority else final_type
return final_type
def auto_extension_int(*args : Types) -> Types:
final_type = AnyT
for a in args:
final_type = a if a.priority > final_type.priority else final_type
return final_type
def ty_clamp(fn, l:int = None, r:int = None):
return lambda *args : fn(*args[l: r])
def logical(*_ : Types) -> Types:
return ByteT
def int_return(*_ : Types) -> Types:
return IntT
def as_is (t: Types) -> Types:
return t
def fp (fx):
return lambda *args : fx(*args).fp_type
def ext (fx):
return lambda *args : fx(*args).long_type
# operator call behavior
def binary_op_behavior(op:OperatorBase, c_code, x, y):
name = op.cname if c_code else op.sqlname
return f'({x} {name} {y})'
def unary_op_behavior(op:OperatorBase, c_code, x):
name = op.cname if c_code else op.sqlname
return f'({x} {name})'
def fn_behavior(op:OperatorBase, c_code, *x):
name = op.cname if c_code else op.sqlname
return f'{name}({", ".join([f"{xx}" for xx in x])})'
def windowed_fn_behavor(op: OperatorBase, c_code, *x):
if not c_code:
return f'{op.sqlname}({", ".join([f"{xx}" for xx in x])})'
else:
name = op.cname if len(x) == 1 else op.cname[:-1] + 'w'
return f'{name}({", ".join([f"{xx}" for xx in x])})'
# arithmetic
opadd = OperatorBase('add', 2, auto_extension, cname = '+', sqlname = '+', call = binary_op_behavior)
opdiv = OperatorBase('div', 2, fp(auto_extension), cname = '/', sqlname = '/', call = binary_op_behavior)
opmul = OperatorBase('mul', 2, fp(auto_extension), cname = '*', sqlname = '*', call = binary_op_behavior)
opsub = OperatorBase('sub', 2, auto_extension, cname = '-', sqlname = '-', call = binary_op_behavior)
opmod = OperatorBase('mod', 2, auto_extension_int, cname = '%', sqlname = '%', call = binary_op_behavior)
opneg = OperatorBase('neg', 1, as_is, cname = '-', sqlname = '-', call = unary_op_behavior)
# logical
opand = OperatorBase('and', 2, logical, cname = '&&', sqlname = ' AND ', call = binary_op_behavior)
opor = OperatorBase('or', 2, logical, cname = '||', sqlname = ' OR ', call = binary_op_behavior)
opxor = OperatorBase('xor', 2, logical, cname = '^', sqlname = ' XOR ', call = binary_op_behavior)
opgt = OperatorBase('gt', 2, logical, cname = '>', sqlname = '>', call = binary_op_behavior)
oplt = OperatorBase('lt', 2, logical, cname = '<', sqlname = '<', call = binary_op_behavior)
opge = OperatorBase('gte', 2, logical, cname = '>=', sqlname = '>=', call = binary_op_behavior)
oplte = OperatorBase('lte', 2, logical, cname = '<=', sqlname = '<=', call = binary_op_behavior)
opneq = OperatorBase('neq', 2, logical, cname = '!=', sqlname = '!=', call = binary_op_behavior)
opeq = OperatorBase('eq', 2, logical, cname = '==', sqlname = '=', call = binary_op_behavior)
opnot = OperatorBase('not', 1, logical, cname = '!', sqlname = 'NOT', call = unary_op_behavior)
# functional
fnmax = OperatorBase('max', 1, as_is, cname = 'max', sqlname = 'MAX', call = fn_behavior)
fnmin = OperatorBase('min', 1, as_is, cname = 'min', sqlname = 'MIN', call = fn_behavior)
fnsum = OperatorBase('sum', 1, ext(auto_extension), cname = 'sum', sqlname = 'SUM', call = fn_behavior)
fnavg = OperatorBase('avg', 1, fp(ext(auto_extension)), cname = 'avg', sqlname = 'AVG', call = fn_behavior)
fnmaxs = OperatorBase('maxs', [1, 2], ty_clamp(as_is, -1), cname = 'maxs', sqlname = 'MAXS', call = windowed_fn_behavor)
fnmins = OperatorBase('mins', [1, 2], ty_clamp(as_is, -1), cname = 'mins', sqlname = 'MINS', call = windowed_fn_behavor)
fnsums = OperatorBase('sums', [1, 2], ext(ty_clamp(auto_extension, -1)), cname = 'sums', sqlname = 'SUMS', call = windowed_fn_behavor)
fnavgs = OperatorBase('avgs', [1, 2], fp(ext(ty_clamp(auto_extension, -1))), cname = 'avgs', sqlname = 'AVGS', call = windowed_fn_behavor)
fncnt = OperatorBase('count', 1, int_return, cname = 'count', sqlname = 'COUNT', call = fn_behavior)
# special
def is_null_call_behavior(op:OperatorBase, c_code : bool, x : str):
if c_code :
return f'{x} == nullval<decays<decltype({x})>>'
else :
return f'{x} IS NULL'
spnull = OperatorBase('missing', 1, logical, cname = "", sqlname = "", call = is_null_call_behavior)
# cstdlib
fnsqrt = OperatorBase('sqrt', 1, lambda *_ : DoubleT, cname = 'sqrt', sqlname = 'SQRT', call = fn_behavior)
# type collections
def _op_make_dict(*items : OperatorBase):
return { i.name: i for i in items}
builtin_binary_arith = _op_make_dict(opadd, opdiv, opmul, opsub, opmod)
builtin_binary_logical = _op_make_dict(opand, opor, opxor, opgt, oplt, opge, oplte, opneq, opeq)
builtin_unary_logical = _op_make_dict(opnot)
builtin_unary_arith = _op_make_dict(opneg)
builtin_unary_special = _op_make_dict(spnull)
builtin_cstdlib = _op_make_dict(fnsqrt)
builtin_func = _op_make_dict(fnmax, fnmin, fnsum, fnavg, fnmaxs, fnmins, fnsums, fnavgs, fncnt)
builtin_operators : dict[str, OperatorBase] = {**builtin_binary_arith, **builtin_binary_logical,
**builtin_unary_arith, **builtin_unary_logical, **builtin_unary_special, **builtin_func, **builtin_cstdlib}
+33 -2
View File
@@ -1,7 +1,8 @@
import uuid
base62alp = '0123456789abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ'
nums = '0123456789'
reserved_monet = ['month']
def base62uuid(crop=8):
id = uuid.uuid4().int
ret = ''
@@ -12,6 +13,33 @@ def base62uuid(crop=8):
return ret[:crop] if len(ret) else '0'
def get_leagl_name(name, lower = True):
if name is not None:
if lower:
name = name.lower()
name = ''.join([n for n in name if n in base62alp or n == '_'])
if name is None or len(name) == 0 or set(name) == set('_'):
name = base62uuid(8)
if(name[0] in nums):
name = '_' + name
return name
def check_leagl_name(name):
all_underscores = True
for c in name:
if c not in base62alp and c != '_':
return False
if c != '_':
all_underscores = False
if all_underscores:
return False
if name[0] in nums:
return False
return True
def enlist(l):
return l if type(l) is list else [l]
@@ -22,4 +50,7 @@ def has_other(a, b):
for ai in a:
if ai not in b:
return True
return False
return False
def defval(val, default):
return default if val is None else val