You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
AQuery/aquery_parser/sql_parser.py

730 lines
24 KiB

# encoding: utf-8
#
# This Source Code Form is subject to the terms of the Mozilla Public
# License, v. 2.0. If a copy of the MPL was not distributed with this file,
# You can obtain one at http://mozilla.org/MPL/2.0/.
#
# Contact: Kyle Lahnakoski (kyle@lahnakoski.com)
#
from sre_parse import WHITESPACE
from mo_parsing.helpers import restOfLine
from mo_parsing.infix import delimited_list
from mo_parsing.whitespaces import NO_WHITESPACE, Whitespace
from aquery_parser.keywords import *
from aquery_parser.types import get_column_type, time_functions
from aquery_parser.utils import *
from aquery_parser.windows import window
digit = Char("0123456789")
simple_ident = (
Char(FIRST_IDENT_CHAR)
+ Char(IDENT_CHAR)[...] # let's not support dashes in var_names.
)
simple_ident = Regex(simple_ident.__regex__()[1])
def common_parser():
combined_ident = Combine(delimited_list(
ansi_ident | mysql_backtick_ident | simple_ident, separator=".", combine=True,
)).set_parser_name("identifier")
return parser(ansi_string | mysql_doublequote_string, combined_ident)
def mysql_parser():
mysql_string = ansi_string | mysql_doublequote_string
mysql_ident = Combine(delimited_list(
mysql_backtick_ident | sqlserver_ident | simple_ident,
separator=".",
combine=True,
)).set_parser_name("mysql identifier")
return parser(mysql_string, mysql_ident)
def sqlserver_parser():
combined_ident = Combine(delimited_list(
ansi_ident
| mysql_backtick_ident
| sqlserver_ident
| Word(FIRST_IDENT_CHAR, IDENT_CHAR),
separator=".",
combine=True,
)).set_parser_name("identifier")
return parser(ansi_string, combined_ident, sqlserver=True)
def parser(literal_string, ident, sqlserver=False):
with Whitespace() as engine:
engine.add_ignore(Literal("--") + restOfLine)
engine.add_ignore(Literal("#") + restOfLine)
engine.add_ignore(Literal("/*") + SkipTo("*/", include=True))
var_name = ~RESERVED + ident
inline_kblock = (L_INLINE + SkipTo(R_INLINE, include=True))("c")
# EXPRESSIONS
expr = Forward()
column_type, column_definition, column_def_references = get_column_type(
expr, var_name, literal_string
)
# CASE
case = (
CASE
+ Group(ZeroOrMore(
(WHEN + expr("when") + THEN + expr("then")) / to_when_call
))("case")
+ Optional(ELSE + expr("else"))
+ END
) / to_case_call
switch = (
CASE
+ expr("value")
+ Group(ZeroOrMore(
(WHEN + expr("when") + THEN + expr("then")) / to_when_call
))("case")
+ Optional(ELSE + expr("else"))
+ END
) / to_switch_call
cast = (
Group(CAST("op") + LB + expr("params") + AS + column_type("params") + RB)
/ to_json_call
)
trim = (
Group(
keyword("trim").suppress()
+ LB
+ Optional(
(keyword("both") | keyword("trailing") | keyword("leading"))
/ (lambda t: t[0].lower())
)("direction")
+ (
assign("from", expr)
| expr("chars") + Optional(assign("from", expr))
)
+ RB
).set_parser_name("trim")
/ to_trim_call
)
_standard_time_intervals = MatchFirst([
keyword(d) / (lambda t: durations[t[0].lower()]) for d in durations.keys()
]).set_parser_name("duration")("params")
duration = (
real_num | int_num | literal_string
)("params") + _standard_time_intervals
interval = (
INTERVAL + ("'" + delimited_list(duration) + "'" | duration)
) / to_interval_call
timestamp = (
time_functions("op")
+ (
literal_string("params")
| MatchFirst([
keyword(t) / (lambda t: t.lower()) for t in times
])("params")
)
) / to_json_call
extract = (
keyword("extract")("op")
+ LB
+ (_standard_time_intervals | expr("params"))
+ FROM
+ expr("params")
+ RB
) / to_json_call
alias = Optional((
(
AS
+ (var_name("name") + Optional(LB + delimited_list(ident("col")) + RB))
| (
var_name("name")
+ Optional(
(LB + delimited_list(ident("col")) + RB)
| (AS + delimited_list(var_name("col")))
)
)
)
/ to_alias
)("name"))
named_column = Group(Group(expr)("value") + alias)
stack = (
keyword("stack")("op")
+ LB
+ int_num("width")
+ ","
+ delimited_list(expr)("args")
+ RB
) / to_stack
# ARRAY[foo],
# ARRAY < STRING > [foo, bar], INVALID
# ARRAY < STRING > [foo, bar],
create_array = (
keyword("array")("op")
+ Optional(LT.suppress() + column_type("type") + GT.suppress())
+ (
LB + delimited_list(Group(expr))("args") + RB
| (Literal("[") + delimited_list(Group(expr))("args") + Literal("]"))
)
)
if not sqlserver:
# SQL SERVER DOES NOT SUPPORT [] FOR ARRAY CONSTRUCTION (USED FOR IDENTIFIERS)
create_array = (
Literal("[") + delimited_list(Group(expr))("args") + Literal("]")
| create_array
)
create_array = create_array / to_array
create_map = (
keyword("map")
+ Literal("[")
+ expr("keys")
+ ","
+ expr("values")
+ Literal("]")
) / to_map
create_struct = (
keyword("struct")("op")
+ Optional(
LT.suppress() + delimited_list(column_type)("types") + GT.suppress()
)
+ LB
+ delimited_list(Group((expr("value") + alias) / to_select_call))("args")
+ RB
).set_parser_name("create struct") / to_struct
distinct = (
DISTINCT("op") + delimited_list(named_column)("params")
) / to_json_call
query = Forward().set_parser_name("query")
call_function = (
ident("op")
+ LB
+ Optional(Group(query) | delimited_list(Group(expr)))("params")
+ Optional(
(keyword("respect") | keyword("ignore"))("nulls")
+ keyword("nulls").suppress()
)
+ RB
).set_parser_name("call function") / to_json_call
with NO_WHITESPACE:
def scale(tokens):
return {"mul": [tokens[0], tokens[1]]}
scale_function = ((real_num | int_num) + call_function) / scale
scale_ident = ((real_num | int_num) + ident) / scale
compound = (
NULL
| TRUE
| FALSE
| NOCASE
| interval
| timestamp
| extract
| case
| switch
| cast
| distinct
| trim
| stack
| create_array
| create_map
| create_struct
| (LB + Group(query) + RB)
| (LB + Group(delimited_list(expr)) / to_tuple_call + RB)
| literal_string.set_parser_name("string")
| hex_num.set_parser_name("hex")
| scale_function
| scale_ident
| real_num.set_parser_name("float")
| int_num.set_parser_name("int")
| call_function
| Combine(var_name + Optional(".*"))
)
sort_column = (
expr("value").set_parser_name("sort1")
+ Optional(DESC("sort") | ASC("sort"))
+ Optional(assign("nulls", keyword("first") | keyword("last")))
)
window_clause, over_clause = window(expr, var_name, sort_column)
expr << (
(
Literal("*")
| infix_notation(
compound,
[
(
Literal("[").suppress() + expr + Literal("]").suppress(),
1,
LEFT_ASSOC,
to_offset,
),
(
Literal(".").suppress() + simple_ident,
1,
LEFT_ASSOC,
to_offset,
),
(window_clause, 1, LEFT_ASSOC, to_window_mod),
(
assign("filter", LB + WHERE + expr + RB),
1,
LEFT_ASSOC,
to_window_mod,
),
]
+ [
(
o,
1 if o in unary_ops else (3 if isinstance(o, tuple) else 2),
unary_ops.get(o, LEFT_ASSOC),
to_lambda if o is LAMBDA else to_json_operator,
)
for o in KNOWN_OPS
],
)
)("value").set_parser_name("expression")
)
select_column = (
Group(
expr("value") + alias | Literal("*")("value")
).set_parser_name("column")
/ to_select_call
)
table_source = Forward()
assumption = Group((ASC|DESC) ("sort") + var_name("value"))
assumptions = Optional(ASSUMING.suppress() + Group(delimited_list(assumption)))
join = (
Group(joins)("op")
+ (table_source )("join")
+ Optional((ON + expr("on")) | (USING + expr("using")))
| (
Group(WINDOW)("op")
+ Group(var_name("name") + AS + over_clause("value"))("join")
)
) / to_join_call
definable_name = Forward()
dindex = definable_name("l") + LSB + expr("idx") + RSB
definable_name << var_name | dindex
# lname = Forward()
# ptr = (lname("l") + LAMBDA + var_name("r"))
# member = (lname("l") + DOT + var_name("r"))
# idx = (expr | COLON)
# index = (lname("l") + LSB + expr("lidx") + "," + idx("ridx") + RSB)
# lname << var_name | ptr | member | index
assignment = expr("var") + (FASSIGN|PASSIGN|MASSIGN|MULASSIGN|DASSIGN)("op") + expr("expr")
declaration = definable_name("var") + Optional(Suppress(FASSIGN) + expr("expr"))
fassign = Group(assignment + Suppress(";"))("assignment")
static_decl = Group(STATIC + delimited_list(declaration))("static_decl")
stmt = Forward()
elifstmt = Group(ELIF + LB + expr("cond") + RB + stmt)("elif")
elsestmt = Group(ELSE + stmt)("else")
ifstmt = Group(IF + LB + expr("cond") + RB + stmt +
ZeroOrMore(elifstmt) + Optional(elsestmt))("if")
forstmt = Group(FOR + LB + ( delimited_list(assignment)("defs")
+ Suppress(";") + expr("cond") +
Suppress(";") + delimited_list(assignment)("tail"))
+ RB + stmt)("for")
block = Forward()
stmt << (fassign|ifstmt|forstmt|block|Suppress(";"))
stmts = (ZeroOrMore(stmt("stmt"), Whitespace()))
block << (LBRACE + Optional(stmts) + RBRACE)("code_block")
fbody = (Optional(static_decl) + Optional(stmts) + expr("ret"))
udf = (
Optional(AGGREGATION("Agg")) +
FUNCTION
+ var_name("fname")
+ LB
+ Optional(delimited_list(var_name)("params"))
+ RB
+ LBRACE
+ fbody
+ RBRACE
)("udf")
selection = (
(SELECT + DISTINCT + ON + LB)
+ delimited_list(select_column)("distinct_on")
+ RB
+ delimited_list(select_column)("select")
| SELECT + DISTINCT + delimited_list(select_column)("select_distinct")
| (
SELECT
+ Optional(
TOP
+ expr("value")
+ Optional(keyword("percent"))("percent")
+ Optional(WITH + keyword("ties"))("ties")
)("top")
/ to_top_clause
+ delimited_list(select_column)("select")
)
)
row = (LB + delimited_list(Group(expr)) + RB) / to_row
values = VALUES + delimited_list(row) / to_values
unordered_sql = Group(
values
| selection
+ Optional(INTO + table_source("into"))
+ Optional(
(
FROM
+ (delimited_list(table_source)
+ ZeroOrMore(join))("table_source")
+ Optional(assumptions) ("assumptions")
)("from")
+ Optional(WHERE + expr("where"))
+ Optional(GROUP_BY + delimited_list(Group(named_column))("groupby"))
+ Optional(HAVING + expr("having"))
)
).set_parser_name("unordered sql")
with NO_WHITESPACE:
def mult(tokens):
amount = tokens["bytes"]
scale = tokens["scale"].lower()
return {
"bytes": amount
* {"b": 1, "k": 1_000, "m": 1_000_000, "g": 1_000_000_000}[scale]
}
ts_bytes = (
(real_num | int_num)("bytes") + Char("bBkKmMgG")("scale")
) / mult
tablesample = assign(
"tablesample",
LB
+ (
(
keyword("bucket")("op")
+ int_num("params")
+ keyword("out of")
+ int_num("params")
+ Optional(ON + expr("on"))
)
/ to_json_call
| (real_num | int_num)("percent") + keyword("percent")
| int_num("rows") + keyword("rows")
| ts_bytes
)
+ RB,
)
table_source << Group(
((LB + query + RB) | stack | call_function | var_name)("value")
+ Optional(flag("with ordinality"))
+ Optional(tablesample)
+ alias
).set_parser_name("table_source") / to_table
rows = Optional(keyword("row") | keyword("rows"))
limit = (
Optional(assign("offset", expr) + rows)
& Optional(
FETCH
+ Optional(keyword("first") | keyword("next"))
+ expr("fetch")
+ rows
+ Optional(keyword("only"))
)
& Optional(assign("limit", expr))
)
outfile = Optional(
(
INTO
+ keyword("outfile").suppress()
+ literal_string ("loc")
+ Optional (
keyword("fields")
+ keyword("terminated")
+ keyword("by")
+ literal_string ("term")
)
)("outfile")
)
ordered_sql = (
(
(unordered_sql | (LB + query + RB))
+ ZeroOrMore(
Group(
(UNION | INTERSECT | EXCEPT | MINUS) + Optional(ALL | DISTINCT)
)("op")
+ (unordered_sql | (LB + query + RB))
)
)("union")
+ Optional(ORDER_BY + delimited_list(Group(sort_column))("orderby"))
+ limit
+ outfile
).set_parser_name("ordered sql") / to_union_call
with_expr = delimited_list(Group(
(
(var_name("name") + Optional(LB + delimited_list(ident("col")) + RB))
/ to_alias
)("name")
+ (AS + LB + (query | expr)("value") + RB)
))
query << (
Optional(assign("with recursive", with_expr) | assign("with", with_expr))
+ Group(ordered_sql)("query")
) / to_query
#####################################################################
# DML STATEMENTS
#####################################################################
# MySQL's index_type := Using + ( "BTREE" | "HASH" )
index_type = Optional(assign("using", ident("index_type")))
index_column_names = LB + delimited_list(var_name("columns")) + RB
column_def_delete = assign(
"on delete",
(keyword("cascade") | keyword("set null") | keyword("set default")),
)
table_def_foreign_key = FOREIGN_KEY + Optional(
Optional(var_name("index_name"))
+ index_column_names
+ column_def_references
+ Optional(column_def_delete)
)
index_options = ZeroOrMore(var_name)("table_constraint_options")
table_constraint_definition = Optional(CONSTRAINT + var_name("name")) + (
assign("primary key", index_type + index_column_names + index_options)
| (
Optional(flag("unique"))
+ Optional(INDEX | KEY)
+ Optional(var_name("name"))
+ index_type
+ index_column_names
+ index_options
)("index")
| assign("check", LB + expr + RB)
| table_def_foreign_key("foreign_key")
)
table_element = (
column_definition("columns") | table_constraint_definition("constraint")
)
create_table = (
keyword("create")
+ Optional(keyword("or") + flag("replace"))
+ Optional(flag("temporary"))
+ TABLE
+ Optional((keyword("if not exists") / (lambda: False))("replace"))
+ var_name("name")
+ Optional(LB + delimited_list(table_element) + RB)
+ ZeroOrMore(
assign("engine", EQ + var_name)
| assign("collate", EQ + var_name)
| assign("auto_increment", EQ + int_num)
| assign("comment", EQ + literal_string)
| assign("default character set", EQ + var_name)
| assign("default charset", EQ + var_name)
)
+ Optional(AS.suppress() + infix_notation(query, [])("query"))
)("create_table")
create_view = (
keyword("create")
+ Optional(keyword("or") + flag("replace"))
+ Optional(flag("temporary"))
+ VIEW.suppress()
+ Optional((keyword("if not exists") / (lambda: False))("replace"))
+ var_name("name")
+ AS
+ query("query")
)("create_view")
# CREATE INDEX a ON u USING btree (e);
create_index = (
keyword("create index")
+ Optional(keyword("or") + flag("replace"))(INDEX | KEY)
+ Optional((keyword("if not exists") / (lambda: False))("replace"))
+ var_name("name")
+ ON
+ var_name("table")
+ index_type
+ index_column_names
+ index_options
)("create index")
cache_options = Optional((
keyword("options").suppress()
+ LB
+ Dict(delimited_list(Group(
literal_string / (lambda tokens: tokens[0]["literal"])
+ Optional(EQ)
+ var_name
)))
+ RB
)("options"))
create_cache = (
keyword("cache").suppress()
+ Optional(flag("lazy"))
+ TABLE
+ var_name("name")
+ cache_options
+ Optional(AS + query("query"))
)("cache")
drop_table = (
keyword("drop table") + Optional(flag("if exists")) + var_name("table")
)("drop")
drop_view = (
keyword("drop view") + Optional(flag("if exists")) + var_name("view")
)("drop")
drop_index = (
keyword("drop index") + Optional(flag("if exists")) + var_name("index")
)("drop")
insert = (
keyword("insert").suppress()
+ (
flag("overwrite") + keyword("table").suppress()
| keyword("into").suppress() + Optional(keyword("table").suppress())
)
+ var_name("table")
+ Optional(LB + delimited_list(var_name)("columns") + RB)
+ Optional(flag("if exists"))
+ (values | query)("query")
) / to_insert_call
update = (
keyword("update")("op")
+ var_name("params")
+ assign("set", Dict(delimited_list(Group(var_name + EQ + expr))))
+ Optional(assign("where", expr))
) / to_json_call
delete = (
keyword("delete")("op")
+ keyword("from").suppress()
+ var_name("params")
+ Optional(assign("where", expr))
) / to_json_call
load_data = (
keyword("data") ("file_type")
+ keyword("infile")("loc")
+ literal_string ("file")
+ INTO
+ keyword("table").suppress()
+ var_name ("table")
+ Optional(
keyword("fields").suppress()
+ keyword("terminated").suppress()
+ keyword("by").suppress()
+ literal_string ("term")
)
)
module_func_def = (
var_name("fname")
+ LB
+ Optional(
delimited_list(
(
var_name("arg")
+ COLON
+ var_name("type")
)("vars")
)
)
+ RB
+ LAMBDA
+ var_name("ret_type")
)
load_module = (
keyword("module") ("file_type")
+ FROM
+ literal_string ("file")
+ Optional(
keyword("FUNCTIONS").suppress()
+ LB
+ module_func_def("funcs")
+ ZeroOrMore(Suppress(',')
+ module_func_def("funcs"),
Whitespace()
)
+ RB
)
)
load = (
keyword("load")
+ (load_data | load_module)
) ("load")
sql_stmts = delimited_list( (
query
| (insert | update | delete | load)
| (create_table | create_view | create_cache | create_index)
| (drop_table | drop_view | drop_index)
)("stmts"), ";")
other_stmt = (
inline_kblock
| udf
) ("stmts")
stmts = ZeroOrMore(
sql_stmts
|other_stmt
| keyword(";").suppress() # empty stmt
)
return stmts.finalize()