RFC: Number lexer lookahead restriction

Implements and adds the tests described by graphql/graphql-spec#601

Replicates graphql/graphql-js@ca1c1df

Author: Cito

README.md

@@ -16,7 +16,7 @@ The current stable version 3.0.1 of GraphQL-core is up-to-date
 with GraphQL.js version 14.5.8.
 
 All parts of the API are covered by an extensive test suite
-of currently 1992 unit tests.
+of currently 1993 unit tests.
 
 
 ## Documentation

src/graphql/language/lexer.py

@@ -215,8 +215,8 @@ def read_number(
             position = self.read_digits(position, char)
             char = body[position : position + 1]
 
-        # Numbers cannot be followed by . or e
-        if char and char in ".eE":
+        # Numbers cannot be followed by . or NameStart
+        if char and (char == "." or is_name_start(char)):
             raise GraphQLSyntaxError(
                 source,
                 position,
@@ -436,3 +436,8 @@ def char2hex(a: str):
     elif "a" <= a <= "f":  # a-f
         return ord(a) - 87
     return -1
+
+
+def is_name_start(char: str) -> bool:
+    """Check whether char is an underscore or a plain ASCII letter"""
+    return char == "_" or "A" <= char <= "Z" or "a" <= char <= "z"

tests/language/test_lexer.py

@@ -19,9 +19,9 @@ def lex_second(s: str) -> Token:
     return lexer.advance()
 
 
-def assert_syntax_error(text, message, location):
+def assert_syntax_error(text, message, location, second=False):
     with raises(GraphQLSyntaxError) as exc_info:
-        lex_one(text)
+        lex_second(text) if second else lex_one(text)
     error = exc_info.value
     assert error.message == f"Syntax Error: {message}"
     assert error.locations == [location]
@@ -330,6 +330,32 @@ def lex_reports_useful_number_errors():
             "1.23.4", "Invalid number, expected digit but got: '.'.", (1, 5)
         )
 
+    def lex_does_not_allow_name_start_after_a_number():
+        assert_syntax_error(
+            "0xF1", "Invalid number, expected digit but got: 'x'.", (1, 2)
+        )
+        assert_syntax_error(
+            "0b10", "Invalid number, expected digit but got: 'b'.", (1, 2)
+        )
+        assert_syntax_error(
+            "123abc", "Invalid number, expected digit but got: 'a'.", (1, 4)
+        )
+        assert_syntax_error(
+            "1_1234", "Invalid number, expected digit but got: '_'.", (1, 2)
+        )
+        assert_syntax_error(
+            "1_1234", "Invalid number, expected digit but got: '_'.", (1, 2)
+        )
+        assert_syntax_error(
+            "1ß", "Cannot parse the unexpected character 'ß'.", (1, 2), second=True,
+        )
+        assert_syntax_error(
+            "1.23f", "Invalid number, expected digit but got: 'f'.", (1, 5)
+        )
+        assert_syntax_error(
+            "12ß", "Cannot parse the unexpected character 'ß'.", (1, 3), second=True,
+        )
+
     # noinspection PyArgumentEqualDefault
     def lexes_punctuation():
         assert lex_one("!") == Token(TokenKind.BANG, 0, 1, 1, 1, None, None)