I have two questions about grammar

[Pete-mathpresso]

What should I do if I don't want to ignore some of the grammar in grammar and some of it? In the Lark grammar, %ignore " " is used to control the spaces in all rules and terminal definitions, but I want to control only some of them.
For example, when you create a grammar that takes and parses a multiplication with latex-form of alphabet, integer and decimal numbers do not want spaces on both sides of the point, but I want to ignore them otherwise. All I can think of is using regular expressions. I wonder if there is a module that can be provided by the Lark library, such as the 'Combine' module provided by Pyparsing. (It's 1st Question)
[2nd Question]
I want to implement a grammar that parses integer or decimal characters, but I am having difficulty using or grammar.How do we deal with this?
I wrote like this,
grammar =
r"""
expr : integer | decimal_default

integer : INT
decimal_default : /\d+\.\d+/

%import common.INT
%import common.WS
%ignore WS

"""

parser = Lark(grammar, parser='lalr', start='expr')
test_text = r'3.5'
test_result = parser.parse(test_text)
result = parser.parse(test_text)
print('result : ', result)

But there is a bug report like that,

[erezsh]

1. You can simply add spaces to the terminal. Like so:

from lark import Lark

p = Lark("""
start: "a" "b" " c"
_WS: " "
%ignore _WS
""", parser="lalr")

p.parse("a b c")	# OK
p.parse("a bc")		# Error

If you were using Earley, you could even write _WS "c" instead.

2. The lexer is matching INT before the decimal. Simply define decimal with a high priority:

DECIMAL.2: /\d+\.\d+/

[erezsh]

1. Accomplishing this doesn't require any special whitespace control. Just define decimal as a terminal, like /\d+\.\d+/, which won't allow any whitespace in between. Don't care a rule like INT "." INT and it should be fine.

2. Notice that rules are lowercase and terminals are uppercase. It needs to be DECIMAL_DEFAULT to work.

[erezsh]

Oh btw, your decimal is badly defined. You're using . instead of \. which matches any character (including a digit) instead of just a dot.

[Pete-mathpresso]

Oh, It is my mistake to show . instead of \. in the code. In fact, it was the same when I used \. , so I changed it to . , but I showed it without knowing it. lalr parser tokenizes 3.5 to 3 and writes errors lark.exceptions.UnexpectedCharacters: No terminal defined for '.' at line 1 col 2

Surprisingly, earley parser works well for the same grammar. Perhaps there is a mathematical weakness in Lalr's own logic.

The reason for specifying variables as lower case instead of uppercase, in other words, the reason I designated the variables as rules instead of terminals is that I thought only the tree structure could be changed when transforming the parsed tree.

Q. When parsing, the rule is converted to Tree structure and Terminal is converted to Token, but I wonder if there is a way to transform Terminal as i want.

[Pete-mathpresso]

Using DECIMAL_DEFAULT instead of decimal_default seems work! surprise.
Then, Could I transform token to what I want? (like float)

[erezsh]

Yes, you can transform tokens.