Skip to content

Commit

Permalink
Refine the generated Python3 code
Browse files Browse the repository at this point in the history
Signed-off-by: Phil Elson <[email protected]>
  • Loading branch information
pelson committed Oct 1, 2024
1 parent 2a7904a commit 7a89e15
Show file tree
Hide file tree
Showing 2 changed files with 205 additions and 120 deletions.
124 changes: 82 additions & 42 deletions runtime/Python3/src/antlr4/xpath/XPathLexer.py
Original file line number Diff line number Diff line change
@@ -1,40 +1,46 @@
# Generated from XPathLexer.g4 by ANTLR 4.13.1
from antlr4 import *
from io import StringIO
import sys
if sys.version_info[1] > 5:
from typing import TextIO
else:
from typing.io import TextIO
import typing

from antlr4 import *


def serializedATN():
return [
4,0,8,50,6,-1,2,0,7,0,2,1,7,1,2,2,7,2,2,3,7,3,2,4,7,4,2,5,7,5,2,
6,7,6,2,7,7,7,1,0,1,0,1,0,1,1,1,1,1,2,1,2,1,3,1,3,1,4,1,4,5,4,29,
8,4,10,4,12,4,32,9,4,1,4,1,4,1,5,1,5,3,5,38,8,5,1,6,1,6,1,7,1,7,
5,7,44,8,7,10,7,12,7,47,9,7,1,7,1,7,1,45,0,8,1,3,3,4,5,5,7,6,9,7,
11,0,13,0,15,8,1,0,2,5,0,48,57,95,95,183,183,768,879,8255,8256,13,
0,65,90,97,122,192,214,216,246,248,767,880,893,895,8191,8204,8205,
8304,8591,11264,12271,12289,55295,63744,64975,65008,65533,50,0,1,
1,0,0,0,0,3,1,0,0,0,0,5,1,0,0,0,0,7,1,0,0,0,0,9,1,0,0,0,0,15,1,0,
0,0,1,17,1,0,0,0,3,20,1,0,0,0,5,22,1,0,0,0,7,24,1,0,0,0,9,26,1,0,
0,0,11,37,1,0,0,0,13,39,1,0,0,0,15,41,1,0,0,0,17,18,5,47,0,0,18,
19,5,47,0,0,19,2,1,0,0,0,20,21,5,47,0,0,21,4,1,0,0,0,22,23,5,42,
0,0,23,6,1,0,0,0,24,25,5,33,0,0,25,8,1,0,0,0,26,30,3,13,6,0,27,29,
3,11,5,0,28,27,1,0,0,0,29,32,1,0,0,0,30,28,1,0,0,0,30,31,1,0,0,0,
31,33,1,0,0,0,32,30,1,0,0,0,33,34,6,4,0,0,34,10,1,0,0,0,35,38,3,
13,6,0,36,38,7,0,0,0,37,35,1,0,0,0,37,36,1,0,0,0,38,12,1,0,0,0,39,
40,7,1,0,0,40,14,1,0,0,0,41,45,5,39,0,0,42,44,9,0,0,0,43,42,1,0,
0,0,44,47,1,0,0,0,45,46,1,0,0,0,45,43,1,0,0,0,46,48,1,0,0,0,47,45,
1,0,0,0,48,49,5,39,0,0,49,16,1,0,0,0,4,0,30,37,45,1,1,4,0
4, 0, 8, 50, 6, -1, 2, 0, 7, 0, 2, 1, 7, 1, 2, 2, 7, 2, 2, 3, 7,
3, 2, 4, 7, 4, 2, 5, 7, 5, 2, 6, 7, 6, 2, 7, 7, 7, 1, 0, 1, 0, 1,
0, 1, 1, 1, 1, 1, 2, 1, 2, 1, 3, 1, 3, 1, 4, 1, 4, 5, 4, 29, 8,
4, 10, 4, 12, 4, 32, 9, 4, 1, 4, 1, 4, 1, 5, 1, 5, 3, 5, 38, 8,
5, 1, 6, 1, 6, 1, 7, 1, 7, 5, 7, 44, 8, 7, 10, 7, 12, 7, 47, 9,
7, 1, 7, 1, 7, 1, 45, 0, 8, 1, 3, 3, 4, 5, 5, 7, 6, 9, 7, 11, 0,
13, 0, 15, 8, 1, 0, 2, 5, 0, 48, 57, 95, 95, 183, 183, 768, 879,
8255, 8256, 13, 0, 65, 90, 97, 122, 192, 214, 216, 246, 248, 767,
880, 893, 895, 8191, 8204, 8205, 8304, 8591, 11264, 12271, 12289,
55295, 63744, 64975, 65008, 65533, 50, 0, 1, 1, 0, 0, 0, 0, 3, 1,
0, 0, 0, 0, 5, 1, 0, 0, 0, 0, 7, 1, 0, 0, 0, 0, 9, 1, 0, 0, 0, 0,
15, 1, 0, 0, 0, 1, 17, 1, 0, 0, 0, 3, 20, 1, 0, 0, 0, 5, 22, 1,
0, 0, 0, 7, 24, 1, 0, 0, 0, 9, 26, 1, 0, 0, 0, 11, 37, 1, 0, 0,
0, 13, 39, 1, 0, 0, 0, 15, 41, 1, 0, 0, 0, 17, 18, 5, 47, 0, 0,
18, 19, 5, 47, 0, 0, 19, 2, 1, 0, 0, 0, 20, 21, 5, 47, 0, 0, 21,
4, 1, 0, 0, 0, 22, 23, 5, 42, 0, 0, 23, 6, 1, 0, 0, 0, 24, 25, 5,
33, 0, 0, 25, 8, 1, 0, 0, 0, 26, 30, 3, 13, 6, 0, 27, 29, 3, 11,
5, 0, 28, 27, 1, 0, 0, 0, 29, 32, 1, 0, 0, 0, 30, 28, 1, 0, 0, 0,
30, 31, 1, 0, 0, 0, 31, 33, 1, 0, 0, 0, 32, 30, 1, 0, 0, 0, 33,
34, 6, 4, 0, 0, 34, 10, 1, 0, 0, 0, 35, 38, 3, 13, 6, 0, 36, 38,
7, 0, 0, 0, 37, 35, 1, 0, 0, 0, 37, 36, 1, 0, 0, 0, 38, 12, 1, 0,
0, 0, 39, 40, 7, 1, 0, 0, 40, 14, 1, 0, 0, 0, 41, 45, 5, 39, 0,
0, 42, 44, 9, 0, 0, 0, 43, 42, 1, 0, 0, 0, 44, 47, 1, 0, 0, 0, 45,
46, 1, 0, 0, 0, 45, 43, 1, 0, 0, 0, 46, 48, 1, 0, 0, 0, 47, 45,
1, 0, 0, 0, 48, 49, 5, 39, 0, 0, 49, 16, 1, 0, 0, 0, 4, 0, 30, 37,
45, 1, 1, 4, 0
]


class XPathLexer(Lexer):

atn = ATNDeserializer().deserialize(serializedATN())

decisionsToDFA = [ DFA(ds, i) for i, ds in enumerate(atn.decisionToState) ]
decisionsToDFA = [DFA(ds, i) for i, ds in enumerate(atn.decisionToState)]

TOKEN_REF = 1
RULE_REF = 2
Expand All @@ -45,43 +51,77 @@ class XPathLexer(Lexer):
ID = 7
STRING = 8

channelNames = [ u"DEFAULT_TOKEN_CHANNEL", u"HIDDEN" ]
channelNames = [
"DEFAULT_TOKEN_CHANNEL",
"HIDDEN",
]

modeNames = [ "DEFAULT_MODE" ]
modeNames = [
"DEFAULT_MODE",
]

literalNames = [ "<INVALID>",
"'//'", "'/'", "'*'", "'!'" ]
literalNames = [
"<INVALID>",
"'//'",
"'/'",
"'*'",
"'!'",
]

symbolicNames = [ "<INVALID>",
"TOKEN_REF", "RULE_REF", "ANYWHERE", "ROOT", "WILDCARD", "BANG",
"ID", "STRING" ]
symbolicNames = [
"<INVALID>",
"TOKEN_REF",
"RULE_REF",
"ANYWHERE",
"ROOT",
"WILDCARD",
"BANG",
"ID",
"STRING",
]

ruleNames = [ "ANYWHERE", "ROOT", "WILDCARD", "BANG", "ID", "NameChar",
"NameStartChar", "STRING" ]
ruleNames = [
"ANYWHERE",
"ROOT",
"WILDCARD",
"BANG",
"ID",
"NameChar",
"NameStartChar",
"STRING",
]

grammarFileName = "XPathLexer.g4"

def __init__(self, input=None, output:TextIO = sys.stdout):
def __init__(
self,
input = None,
output: typing.TextIO = sys.stdout,
) -> None:
super().__init__(input, output)
self.checkVersion("4.13.1")
self._interp = LexerATNSimulator(self, self.atn, self.decisionsToDFA, PredictionContextCache())
self._interp = LexerATNSimulator(
self,
self.atn,
self.decisionsToDFA,
PredictionContextCache(),
)
self._actions = None
self._predicates = None


def action(self, localctx:RuleContext, ruleIndex:int, actionIndex:int):
def action(self, localctx: RuleContext, ruleIndex: int, actionIndex: int) -> None:
if self._actions is None:
actions = dict()
actions[4] = self.ID_action
actions = {}
actions[4] = self.ID_action
self._actions = actions
action = self._actions.get(ruleIndex, None)
if action is not None:
action(localctx, actionIndex)
else:
raise Exception("No registered action for:" + str(ruleIndex))

raise Exception(f"No registered action for: {ruleIndex}")

def ID_action(self, localctx:RuleContext , actionIndex:int):
def ID_action(self, localctx: RuleContext, actionIndex: int) -> None:
if actionIndex == 0:

char = self.text[0]
Expand Down
Loading

0 comments on commit 7a89e15

Please sign in to comment.