jinja2.optimizer
The optimizer tries to constant fold expressions and modify the AST in place so that it should be faster to evaluate.
Because the AST does not contain all the scoping information and the compiler has to find that out, we cannot do all the optimizations we want. For example, loop unrolling doesn't work because unrolled loops would have a different scope. The solution would be a second syntax tree that stored the scoping rules.
1"""The optimizer tries to constant fold expressions and modify the AST 2in place so that it should be faster to evaluate. 3 4Because the AST does not contain all the scoping information and the 5compiler has to find that out, we cannot do all the optimizations we 6want. For example, loop unrolling doesn't work because unrolled loops 7would have a different scope. The solution would be a second syntax tree 8that stored the scoping rules. 9""" 10 11import typing as t 12 13from . import nodes 14from .visitor import NodeTransformer 15 16if t.TYPE_CHECKING: 17 from .environment import Environment 18 19 20def optimize(node: nodes.Node, environment: "Environment") -> nodes.Node: 21 """The context hint can be used to perform an static optimization 22 based on the context given.""" 23 optimizer = Optimizer(environment) 24 return t.cast(nodes.Node, optimizer.visit(node)) 25 26 27class Optimizer(NodeTransformer): 28 def __init__(self, environment: "t.Optional[Environment]") -> None: 29 self.environment = environment 30 31 def generic_visit( 32 self, node: nodes.Node, *args: t.Any, **kwargs: t.Any 33 ) -> nodes.Node: 34 node = super().generic_visit(node, *args, **kwargs) 35 36 # Do constant folding. Some other nodes besides Expr have 37 # as_const, but folding them causes errors later on. 38 if isinstance(node, nodes.Expr): 39 try: 40 return nodes.Const.from_untrusted( 41 node.as_const(args[0] if args else None), 42 lineno=node.lineno, 43 environment=self.environment, 44 ) 45 except nodes.Impossible: 46 pass 47 48 return node
21def optimize(node: nodes.Node, environment: "Environment") -> nodes.Node: 22 """The context hint can be used to perform an static optimization 23 based on the context given.""" 24 optimizer = Optimizer(environment) 25 return t.cast(nodes.Node, optimizer.visit(node))
The context hint can be used to perform an static optimization based on the context given.
28class Optimizer(NodeTransformer): 29 def __init__(self, environment: "t.Optional[Environment]") -> None: 30 self.environment = environment 31 32 def generic_visit( 33 self, node: nodes.Node, *args: t.Any, **kwargs: t.Any 34 ) -> nodes.Node: 35 node = super().generic_visit(node, *args, **kwargs) 36 37 # Do constant folding. Some other nodes besides Expr have 38 # as_const, but folding them causes errors later on. 39 if isinstance(node, nodes.Expr): 40 try: 41 return nodes.Const.from_untrusted( 42 node.as_const(args[0] if args else None), 43 lineno=node.lineno, 44 environment=self.environment, 45 ) 46 except nodes.Impossible: 47 pass 48 49 return node
Walks the abstract syntax tree and allows modifications of nodes.
The NodeTransformer will walk the AST and use the return value of the
visitor functions to replace or remove the old node. If the return
value of the visitor function is None the node will be removed
from the previous location otherwise it's replaced with the return
value. The return value may be the original node in which case no
replacement takes place.
32 def generic_visit( 33 self, node: nodes.Node, *args: t.Any, **kwargs: t.Any 34 ) -> nodes.Node: 35 node = super().generic_visit(node, *args, **kwargs) 36 37 # Do constant folding. Some other nodes besides Expr have 38 # as_const, but folding them causes errors later on. 39 if isinstance(node, nodes.Expr): 40 try: 41 return nodes.Const.from_untrusted( 42 node.as_const(args[0] if args else None), 43 lineno=node.lineno, 44 environment=self.environment, 45 ) 46 except nodes.Impossible: 47 pass 48 49 return node
Called if no explicit visitor function exists for a node.