Fix parser bug where "type" was misinterpreted as a keyword inside a match (#3950)

Fixes #3790

Slightly hacky, but I think this is correct and it should also improve performance somewhat.

diff --git a/CHANGES.md b/CHANGES.md
index 610a9de0e431af4bfff2a4fd39f2bfabeb31a4c9..f89b1b9df0a6acb4a31c39eb8791f5f0d76fc41e 100644 (file)
--- a/CHANGES.md
+++ b/CHANGES.md
@@ -37,6 +37,8 @@
 
 <!-- Changes to the parser or to version autodetection -->
 
 
 <!-- Changes to the parser or to version autodetection -->
 

+- Fix bug where attributes named `type` were not acccepted inside `match` statements
+  (#3950)

 - Add support for PEP 695 type aliases containing lambdas and other unusual expressions
   (#3949)
 
 - Add support for PEP 695 type aliases containing lambdas and other unusual expressions
   (#3949)
 

diff --git a/src/blib2to3/pgen2/parse.py b/src/blib2to3/pgen2/parse.py
index 299cc24a15f4bd353d155b62ef645d8d6a9ef148..ad51a3dad08dfbbe1fee7cc9b7845eeb7cb263fd 100644 (file)
--- a/src/blib2to3/pgen2/parse.py
+++ b/src/blib2to3/pgen2/parse.py
@@ -211,6 +211,7 @@ class Parser:
         # See note in docstring above. TL;DR this is ignored.
         self.convert = convert or lam_sub
         self.is_backtracking = False
         # 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.


@@ -236,6 +237,7 @@ class Parser:
         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: str, context: Context) -> bool:
         """Add a token; return True iff this is the end of the program."""
 
     def addtoken(self, type: int, value: str, context: Context) -> bool:
         """Add a token; return True iff this is the end of the program."""


@@ -317,6 +319,7 @@ class Parser:
                         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:


@@ -343,9 +346,23 @@ class Parser:
                 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)

diff --git a/tests/data/cases/pattern_matching_complex.py b/tests/data/cases/pattern_matching_complex.py
index b4355c7333a54021f41999637a7bed7adb1c0eeb..10b4d26e28969defb2b5c22daae726d61a1ff64e 100644 (file)
--- a/tests/data/cases/pattern_matching_complex.py
+++ b/tests/data/cases/pattern_matching_complex.py
@@ -143,3 +143,7 @@ match x:
         y = 1
     case []:
         y = 2
         y = 1
     case []:
         y = 2

+# issue 3790
+match (X.type, Y):
+    case _:
+        pass

diff --git a/tests/data/cases/type_aliases.py b/tests/data/cases/type_aliases.py
index 9631bfd5ccc3c3cf815ab2421ac531f37810ff18..7c2009e8202cf9a2820c69d485c36dbd2d1d4551 100644 (file)
--- a/tests/data/cases/type_aliases.py
+++ b/tests/data/cases/type_aliases.py
@@ -5,6 +5,8 @@ type Gen[T]=list[T]
 type Alias[T]=lambda: T
 type And[T]=T and T
 type IfElse[T]=T if T else T
 type Alias[T]=lambda: T
 type And[T]=T and T
 type IfElse[T]=T if T else T

+type One = int; type Another = str
+class X: type InClass = int

 
 type = aliased
 print(type(42))
 
 type = aliased
 print(type(42))


@@ -16,6 +18,13 @@ type Gen[T] = list[T]
 type Alias[T] = lambda: T
 type And[T] = T and T
 type IfElse[T] = T if T else T
 type Alias[T] = lambda: T
 type And[T] = T and T
 type IfElse[T] = T if T else T

+type One = int
+type Another = str
+
+
+class X:
+    type InClass = int
+

 
 type = aliased
 print(type(42))
 
 type = aliased
 print(type(42))