Prepare release 23.10.0 (#3951)

[etc/vim.git] / src / blib2to3 / pgen2 / parse.py

diff --git a/src/blib2to3/pgen2/parse.py b/src/blib2to3/pgen2/parse.py
index 792e8e66698246f578fef7ce88d7bf69258af3e9..ad51a3dad08dfbbe1fee7cc9b7845eeb7cb263fd 100644 (file)
--- a/src/blib2to3/pgen2/parse.py
+++ b/src/blib2to3/pgen2/parse.py
@@ -9,33 +9,32 @@ See Parser/parser.c in the Python distribution for additional info on
 how this parsing engine works.
 
 """
 how this parsing engine works.
 
 """

-import copy

 from contextlib import contextmanager
 from contextlib import contextmanager

-
-# Local imports
-from . import grammar, token, tokenize

 from typing import (
 from typing import (

-    cast,
+    TYPE_CHECKING,

     Any,
     Any,

-    Optional,
-    Text,
-    Union,
-    Tuple,
+    Callable,

     Dict,
     Dict,

-    List,

     Iterator,
     Iterator,

-    Callable,
+    List,
+    Optional,

     Set,
     Set,

-    TYPE_CHECKING,
+    Tuple,
+    Union,
+    cast,

 )
 )

+

 from blib2to3.pgen2.grammar import Grammar
 from blib2to3.pgen2.grammar import Grammar

-from blib2to3.pytree import convert, NL, Context, RawNode, Leaf, Node
+from blib2to3.pytree import NL, Context, Leaf, Node, RawNode, convert
+
+# Local imports
+from . import grammar, token, tokenize

 
 if TYPE_CHECKING:
 
 if TYPE_CHECKING:

-    from blib2to3.driver import TokenProxy
+    from blib2to3.pgen2.driver import TokenProxy

 
 
 
 

-Results = Dict[Text, NL]
+Results = Dict[str, NL]

 Convert = Callable[[Grammar, RawNode], Union[Node, Leaf]]
 DFA = List[List[Tuple[int, int]]]
 DFAS = Tuple[DFA, Dict[int, int]]
 Convert = Callable[[Grammar, RawNode], Union[Node, Leaf]]
 DFA = List[List[Tuple[int, int]]]
 DFAS = Tuple[DFA, Dict[int, int]]


@@ -46,6 +45,17 @@ def lam_sub(grammar: Grammar, node: RawNode) -> NL:
     return Node(type=node[0], children=node[3], context=node[2])
 
 
     return Node(type=node[0], children=node[3], context=node[2])
 
 

+# A placeholder node, used when parser is backtracking.
+DUMMY_NODE = (-1, None, None, None)
+
+
+def stack_copy(
+    stack: List[Tuple[DFAS, int, RawNode]]
+) -> List[Tuple[DFAS, int, RawNode]]:
+    """Nodeless stack copy."""
+    return [(dfa, label, DUMMY_NODE) for dfa, label, _ in stack]
+
+

 class Recorder:
     def __init__(self, parser: "Parser", ilabels: List[int], context: Context) -> None:
         self.parser = parser
 class Recorder:
     def __init__(self, parser: "Parser", ilabels: List[int], context: Context) -> None:
         self.parser = parser


@@ -53,8 +63,8 @@ class Recorder:
         self.context = context  # not really matter
 
         self._dead_ilabels: Set[int] = set()
         self.context = context  # not really matter
 
         self._dead_ilabels: Set[int] = set()

-        self._start_point = copy.deepcopy(self.parser.stack)
-        self._points = {ilabel: copy.deepcopy(self._start_point) for ilabel in ilabels}
+        self._start_point = self.parser.stack
+        self._points = {ilabel: stack_copy(self._start_point) for ilabel in ilabels}

 
     @property
     def ilabels(self) -> Set[int]:
 
     @property
     def ilabels(self) -> Set[int]:


@@ -62,15 +72,34 @@ class Recorder:
 
     @contextmanager
     def switch_to(self, ilabel: int) -> Iterator[None]:
 
     @contextmanager
     def switch_to(self, ilabel: int) -> Iterator[None]:

-        self.parser.stack = self._points[ilabel]
+        with self.backtrack():
+            self.parser.stack = self._points[ilabel]
+            try:
+                yield
+            except ParseError:
+                self._dead_ilabels.add(ilabel)
+            finally:
+                self.parser.stack = self._start_point
+
+    @contextmanager
+    def backtrack(self) -> Iterator[None]:
+        """
+        Use the node-level invariant ones for basic parsing operations (push/pop/shift).
+        These still will operate on the stack; but they won't create any new nodes, or
+        modify the contents of any other existing nodes.
+
+        This saves us a ton of time when we are backtracking, since we
+        want to restore to the initial state as quick as possible, which
+        can only be done by having as little mutatations as possible.
+        """
+        is_backtracking = self.parser.is_backtracking

         try:
         try:

+            self.parser.is_backtracking = True

             yield
             yield

-        except ParseError:
-            self._dead_ilabels.add(ilabel)

         finally:
         finally:

-            self.parser.stack = self._start_point
+            self.parser.is_backtracking = is_backtracking

 
 

-    def add_token(self, tok_type: int, tok_val: Text, raw: bool = False) -> None:
+    def add_token(self, tok_type: int, tok_val: str, raw: bool = False) -> None:

         func: Callable[..., Any]
         if raw:
             func = self.parser._addtoken
         func: Callable[..., Any]
         if raw:
             func = self.parser._addtoken


@@ -84,7 +113,9 @@ class Recorder:
                     args.insert(0, ilabel)
                 func(*args)
 
                     args.insert(0, ilabel)
                 func(*args)
 

-    def determine_route(self, value: Text = None, force: bool = False) -> Optional[int]:
+    def determine_route(
+        self, value: Optional[str] = None, force: bool = False
+    ) -> Optional[int]:

         alive_ilabels = self.ilabels
         if len(alive_ilabels) == 0:
             *_, most_successful_ilabel = self._dead_ilabels
         alive_ilabels = self.ilabels
         if len(alive_ilabels) == 0:
             *_, most_successful_ilabel = self._dead_ilabels


@@ -101,10 +132,10 @@ class ParseError(Exception):
     """Exception to signal the parser is stuck."""
 
     def __init__(
     """Exception to signal the parser is stuck."""
 
     def __init__(

-        self, msg: Text, type: Optional[int], value: Optional[Text], context: Context
+        self, msg: str, type: Optional[int], value: Optional[str], context: Context

     ) -> None:
         Exception.__init__(
     ) -> None:
         Exception.__init__(

-            self, "%s: type=%r, value=%r, context=%r" % (msg, type, value, context)
+            self, f"{msg}: type={type!r}, value={value!r}, context={context!r}"

         )
         self.msg = msg
         self.type = type
         )
         self.msg = msg
         self.type = type


@@ -112,7 +143,7 @@ class ParseError(Exception):
         self.context = context
 
 
         self.context = context
 
 

-class Parser(object):
+class Parser:

     """Parser engine.
 
     The proper usage sequence is:
     """Parser engine.
 
     The proper usage sequence is:


@@ -179,6 +210,8 @@ class Parser(object):
         self.grammar = grammar
         # See note in docstring above. TL;DR this is ignored.
         self.convert = convert or lam_sub
         self.grammar = grammar
         # See note in docstring above. TL;DR this is ignored.
         self.convert = convert or lam_sub

+        self.is_backtracking = False
+        self.last_token: Optional[int] = None

 
     def setup(self, proxy: "TokenProxy", start: Optional[int] = None) -> None:
         """Prepare for parsing.
 
     def setup(self, proxy: "TokenProxy", start: Optional[int] = None) -> None:
         """Prepare for parsing.


@@ -204,8 +237,9 @@ class Parser(object):
         self.rootnode: Optional[NL] = None
         self.used_names: Set[str] = set()
         self.proxy = proxy
         self.rootnode: Optional[NL] = None
         self.used_names: Set[str] = set()
         self.proxy = proxy

+        self.last_token = None

 
 

-    def addtoken(self, type: int, value: Text, context: Context) -> bool:
+    def addtoken(self, type: int, value: str, context: Context) -> bool:

         """Add a token; return True iff this is the end of the program."""
         # Map from token to label
         ilabels = self.classify(type, value, context)
         """Add a token; return True iff this is the end of the program."""
         # Map from token to label
         ilabels = self.classify(type, value, context)


@@ -238,6 +272,10 @@ class Parser(object):
                     break
 
                 next_token_type, next_token_value, *_ = proxy.eat(counter)
                     break
 
                 next_token_type, next_token_value, *_ = proxy.eat(counter)

+                if next_token_type in (tokenize.COMMENT, tokenize.NL):
+                    counter += 1
+                    continue
+

                 if next_token_type == tokenize.OP:
                     next_token_type = grammar.opmap[next_token_value]
 
                 if next_token_type == tokenize.OP:
                     next_token_type = grammar.opmap[next_token_value]
 


@@ -249,7 +287,7 @@ class Parser(object):
 
         return self._addtoken(ilabel, type, value, context)
 
 
         return self._addtoken(ilabel, type, value, context)
 

-    def _addtoken(self, ilabel: int, type: int, value: Text, context: Context) -> bool:
+    def _addtoken(self, ilabel: int, type: int, value: str, context: Context) -> bool:

         # Loop until the token is shifted; may raise exceptions
         while True:
             dfa, state, node = self.stack[-1]
         # Loop until the token is shifted; may raise exceptions
         while True:
             dfa, state, node = self.stack[-1]


@@ -281,6 +319,7 @@ class Parser(object):
                         dfa, state, node = self.stack[-1]
                         states, first = dfa
                     # Done with this token
                         dfa, state, node = self.stack[-1]
                         states, first = dfa
                     # Done with this token

+                    self.last_token = type

                     return False
 
             else:
                     return False
 
             else:


@@ -294,7 +333,7 @@ class Parser(object):
                     # No success finding a transition
                     raise ParseError("bad input", type, value, context)
 
                     # No success finding a transition
                     raise ParseError("bad input", type, value, context)
 

-    def classify(self, type: int, value: Text, context: Context) -> List[int]:
+    def classify(self, type: int, value: str, context: Context) -> List[int]:

         """Turn a token into a label.  (Internal)
 
         Depending on whether the value is a soft-keyword or not,
         """Turn a token into a label.  (Internal)
 
         Depending on whether the value is a soft-keyword or not,


@@ -307,9 +346,23 @@ class Parser(object):
                 return [self.grammar.keywords[value]]
             elif value in self.grammar.soft_keywords:
                 assert type in self.grammar.tokens
                 return [self.grammar.keywords[value]]
             elif value in self.grammar.soft_keywords:
                 assert type in self.grammar.tokens

+                # Current soft keywords (match, case, type) can only appear at the
+                # beginning of a statement. So as a shortcut, don't try to treat them
+                # like keywords in any other context.
+                # ('_' is also a soft keyword in the real grammar, but for our grammar
+                # it's just an expression, so we don't need to treat it specially.)
+                if self.last_token not in (
+                    None,
+                    token.INDENT,
+                    token.DEDENT,
+                    token.NEWLINE,
+                    token.SEMI,
+                    token.COLON,
+                ):
+                    return [self.grammar.tokens[type]]

                 return [
                 return [

-                    self.grammar.soft_keywords[value],

                     self.grammar.tokens[type],
                     self.grammar.tokens[type],

+                    self.grammar.soft_keywords[value],

                 ]
 
         ilabel = self.grammar.tokens.get(type)
                 ]
 
         ilabel = self.grammar.tokens.get(type)


@@ -317,30 +370,42 @@ class Parser(object):
             raise ParseError("bad token", type, value, context)
         return [ilabel]
 
             raise ParseError("bad token", type, value, context)
         return [ilabel]
 

-    def shift(self, type: int, value: Text, newstate: int, context: Context) -> None:
+    def shift(self, type: int, value: str, newstate: int, context: Context) -> None:

         """Shift a token.  (Internal)"""
         """Shift a token.  (Internal)"""

-        dfa, state, node = self.stack[-1]
-        rawnode: RawNode = (type, value, context, None)
-        newnode = convert(self.grammar, rawnode)
-        assert node[-1] is not None
-        node[-1].append(newnode)
-        self.stack[-1] = (dfa, newstate, node)
+        if self.is_backtracking:
+            dfa, state, _ = self.stack[-1]
+            self.stack[-1] = (dfa, newstate, DUMMY_NODE)
+        else:
+            dfa, state, node = self.stack[-1]
+            rawnode: RawNode = (type, value, context, None)
+            newnode = convert(self.grammar, rawnode)
+            assert node[-1] is not None
+            node[-1].append(newnode)
+            self.stack[-1] = (dfa, newstate, node)

 
     def push(self, type: int, newdfa: DFAS, newstate: int, context: Context) -> None:
         """Push a nonterminal.  (Internal)"""
 
     def push(self, type: int, newdfa: DFAS, newstate: int, context: Context) -> None:
         """Push a nonterminal.  (Internal)"""

-        dfa, state, node = self.stack[-1]
-        newnode: RawNode = (type, None, context, [])
-        self.stack[-1] = (dfa, newstate, node)
-        self.stack.append((newdfa, 0, newnode))
+        if self.is_backtracking:
+            dfa, state, _ = self.stack[-1]
+            self.stack[-1] = (dfa, newstate, DUMMY_NODE)
+            self.stack.append((newdfa, 0, DUMMY_NODE))
+        else:
+            dfa, state, node = self.stack[-1]
+            newnode: RawNode = (type, None, context, [])
+            self.stack[-1] = (dfa, newstate, node)
+            self.stack.append((newdfa, 0, newnode))

 
     def pop(self) -> None:
         """Pop a nonterminal.  (Internal)"""
 
     def pop(self) -> None:
         """Pop a nonterminal.  (Internal)"""

-        popdfa, popstate, popnode = self.stack.pop()
-        newnode = convert(self.grammar, popnode)
-        if self.stack:
-            dfa, state, node = self.stack[-1]
-            assert node[-1] is not None
-            node[-1].append(newnode)
+        if self.is_backtracking:
+            self.stack.pop()

         else:
         else:

-            self.rootnode = newnode
-            self.rootnode.used_names = self.used_names
+            popdfa, popstate, popnode = self.stack.pop()
+            newnode = convert(self.grammar, popnode)
+            if self.stack:
+                dfa, state, node = self.stack[-1]
+                assert node[-1] is not None
+                node[-1].append(newnode)
+            else:
+                self.rootnode = newnode
+                self.rootnode.used_names = self.used_names