bill/AQuery
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

regression: nested aggregation support

Browse Source
This commit is contained in:
bill
2022-09-02 07:01:51 +08:00
parent 3a56c19e2e
commit 44ccc0b835
14 changed files with 4843 additions and 54 deletions
Download Patch File Download Diff File Expand all files Collapse all files
reconstruct/ast.py
+24 -20
View File
@@ -70,14 +70,14 @@ class projection(ast_node):
self.context.postproc_begin(self.postproc_fname)
def spawn(self, node):
self.datasource = None # datasource is Join instead of TableInfo
self.datasource = join(self, [], self.context) # datasource is Join instead of TableInfo
self.assumptions = []
if 'from' in node:
from_clause = node['from']
self.datasource = join(self, from_clause)
if 'assumptions' in from_clause:
self.assumptions = enlist(from_clause['assumptions'])
else:
self.assumptions = []
if self.datasource is not None:
self.datasource_changed = True
self.prev_datasource = self.context.datasource
@@ -157,7 +157,6 @@ class projection(ast_node):
def finialize(astnode:ast_node):
if(astnode is not None):
self.add(astnode.sql)
self.add('FROM')
finialize(self.datasource)
finialize(self.where)
if self.group_node and not self.group_node.use_sp_gb:
@@ -469,6 +468,7 @@ class join(ast_node):
self.tables : List[TableInfo] = []
self.tables_dir = dict()
self.rec = None
self.top_level = self.parent and type(self.parent) is projection
# self.tmp_name = 'join_' + base62uuid(4)
# self.datasource = TableInfo(self.tmp_name, [], self.context)
def append(self, tbls, __alias = ''):
@@ -547,9 +547,12 @@ class join(ast_node):
@property
def all_cols(self):
return set([c for t in self.tables for c in t.columns])
def consume(self, _):
def consume(self, node):
self.sql = ', '.join(self.joins)
return super().consume(_)
if node and self.sql and self.top_level:
self.sql = ' FROM ' + self.sql
return super().consume(node)
def __str__(self):
return ', '.join(self.joins)
def __repr__(self):
@@ -644,18 +647,18 @@ class load(ast_node):
ret_type = Types.decode(f['ret_type'])
nargs = 0
arglist = ''
if 'var' in f:
if 'vars' in f:
arglist = []
for v in enlist(f['var']):
for v in enlist(f['vars']):
arglist.append(f'{Types.decode(v["type"]).cname} {v["arg"]}')
nargs = len(arglist)
arglist = ', '.join(arglist)
# create c++ stub
cpp_stub = f'{ret_type.cname} (*{fname})({arglist});'
cpp_stub = f'{ret_type.cname} (*{fname})({arglist}) = nullptr;'
self.context.module_stubs += cpp_stub + '\n'
self.context.module_map[fname] = cpp_stub
#registration for parser
self.functions[fname] = user_module_function(fname, nargs, ret_type)
self.functions[fname] = user_module_function(fname, nargs, ret_type, self.context)
def produce_aq(self, node):
node = node['load']
@@ -723,6 +726,12 @@ class outfile(ast_node):
class udf(ast_node):
name = 'udf'
first_order = name
@staticmethod
def try_init_udf(context : Context):
if context.udf is None:
context.udf = '/*UDF Start*/\n'
context.headers.add('\"./udf.hpp\"')
@dataclass
class builtin_var:
enabled : bool = False
@@ -754,13 +763,7 @@ class udf(ast_node):
}
self.var_table = {}
self.args = []
if self.context.udf is None:
self.context.udf = (
Context.udf_head
+ self.context.module_stubs
+ self.context.get_init_func()
)
self.context.headers.add('\"./udf.hpp\"')
udf.try_init_udf(self.context)
self.vecs = set()
self.code_list = []
self.builtin_used = None
@@ -983,10 +986,11 @@ class udf(ast_node):
return udf.ReturnPattern.bulk_return
class user_module_function(OperatorBase):
def __init__(self, name, nargs, ret_type):
super().__init__(name, nargs, lambda: ret_type, call=fn_behavior)
def __init__(self, name, nargs, ret_type, context : Context):
super().__init__(name, nargs, lambda *_: ret_type, call=fn_behavior)
user_module_func[name] = self
builtin_operators[name] = self
# builtin_operators[name] = self
udf.try_init_udf(context)
def include(objs):
import inspect
reconstruct/expr.py
+14 -4
View File
@@ -42,10 +42,12 @@ class expr(ast_node):
if(type(parent) is expr):
self.inside_agg = parent.inside_agg
self.is_udfexpr = parent.is_udfexpr
self.is_agg_func = parent.is_agg_func
self.root : expr = parent.root
self.c_code = parent.c_code
self.builtin_vars = parent.builtin_vars
else:
self.is_agg_func = False
self.is_udfexpr = type(parent) is udf
self.root : expr = self
self.c_code = self.is_udfexpr or type(parent) is projection
@@ -71,8 +73,8 @@ class expr(ast_node):
else:
self.datasource = self.context.datasource
self.udf_map = parent.context.udf_map
self.func_maps = {**builtin_func, **self.udf_map}
self.operators = {**builtin_operators, **self.udf_map}
self.func_maps = {**builtin_func, **self.udf_map, **user_module_func}
self.operators = {**builtin_operators, **self.udf_map, **user_module_func}
def produce(self, node):
from engine.utils import enlist
@@ -81,6 +83,12 @@ class expr(ast_node):
if type(node) is dict:
for key, val in node.items():
if key in self.operators:
if key in builtin_func:
if self.is_agg_func:
self.root.is_special = True # Nested Aggregation
else:
self.is_agg_func = True
op = self.operators[key]
val = enlist(val)
@@ -95,13 +103,15 @@ class expr(ast_node):
self.type = AnyT
self.sql = op(self.c_code, *str_vals)
special_func = [*self.context.udf_map.keys(), "maxs", "mins", "avgs", "sums"]
special_func = [*self.context.udf_map.keys(), *self.context.module_map.keys(), "maxs", "mins", "avgs", "sums"]
if key in special_func and not self.is_special:
self.is_special = True
if key in self.context.udf_map:
self.root.udf_called = self.context.udf_map[key]
if self.is_udfexpr and key == self.root.udf.name:
self.root.is_recursive_call_inudf = True
elif key in user_module_func.keys():
udf.try_init_udf(self.context)
# TODO: make udf_called a set!
p = self.parent
while type(p) is expr and not p.udf_called:
@@ -201,7 +211,7 @@ class expr(ast_node):
if self.c_code and self.datasource is not None:
self.sql = f'{{y(\"{self.sql}\")}}'
elif type(node) is bool:
self.type = ByteT
self.type = BoolT
if self.c_code:
self.sql = '1' if node else '0'
else:
reconstruct/storage.py
+17 -3
View File
@@ -136,20 +136,24 @@ class Context:
'#include \"./server/libaquery.h\"\n'
'#include \"./server/aggregations.h\"\n\n'
)
def get_init_func(self):
if self.module_map:
if not self.module_map:
return ''
ret = 'void init(Context* cxt){\n'
ret = '__AQEXPORT__(void) __builtin_init_user_module(Context* cxt){\n'
for fname in self.module_map.keys():
ret += f'{fname} = (decltype({fname}))(cxt->get_module_function("{fname}"));\n'
self.queries.insert(0, f'P__builtin_init_user_module')
return ret + '}\n'
def sql_begin(self):
self.sql = ''
def sql_end(self):
self.queries.append('Q' + self.sql)
if self.sql.strip():
self.queries.append('Q' + self.sql)
self.sql = ''
def postproc_begin(self, proc_name: str):
self.ccode = self.function_deco + proc_name + self.function_head
@@ -158,6 +162,16 @@ class Context:
self.ccode = ''
self.queries.append('P' + proc_name)
def finalize_udf(self):
if self.udf is not None:
return (Context.udf_head
+ self.module_stubs
+ self.get_init_func()
+ self.udf
)
else:
return None
def finalize(self):
if not self.finalized:
headers = ''