Can I improve this GOLD Parser Grammar?

问题

I have to parse a file that looks like this:

versioninfo
{
    "editorversion" "400"
    "editorbuild" "4715"
}
visgroups
{
}
world
{
    "id" "1"
    "mapversion" "525"
    "classname" "worldspawn"
    solid
    {
        "id" "2"
        side
        {
            "id" "1"
            "plane" "(-544 -400 0) (-544 -240 0) (-272 -240 0)"
        }
        side
        {
            "id" "2"
            "plane" "(-544 -240 -16) (-544 -400 -16) (-272 -400 -16)"
        }
    }
}

I have a parser written from scratch, but it has a few bugs that I can't track down and I imagine it'll be difficult to maintain if the format changes in the future. I decided to use the GOLD Parsing System to generate a parser, instead. My grammar looks like this:

"Start Symbol" = <SectionList>

! SETS

{Section Chars} = {AlphaNumeric} + [_]
{Property Chars} = {Printable} - ["]

! TERMINALS

SectionName = {Section Chars}+ 
PropertyPart = '"' {Property Chars}* '"'

! RULES

<SectionList> ::= <Section>
               |  <Section> <SectionList>

<SectionBody> ::= <PropertyList>
               |  <SectionList>
               |  <PropertyList> <SectionList>

<Section> ::= SectionName '{' '}'
           |  SectionName '{' <SectionBody> '}'

<PropertyList> ::= <Property>
                |  <Property> <PropertyList>

<Property> ::= PropertyPart PropertyPart

There are no errors and it parses my 2000-line test file just fine. However, this is my first time writing a custom grammar, so I'm not sure if I'm doing it correctly.

Are there any improvements I could make to the grammar above?

回答1:

below are some changes i would request to change for better performance

1) make the grammar left recursive rules. this is better in terms of making shift reduce operations as gold parser is a shift reduce LR parser.

SectionList ::= Section

           |   SectionList Section

PropertyList ::= Property

            | PropertyList Property

2) third rule in below section forces you to have propertylist only before sectionlist but not between different 's. make sure its as per requirement

SectionBody ::= PropertyList

           |  SectionList

           |  PropertyList SectionList

i can help you better if required and if you let me know the language saying " it should accept this , shouldn't accept this" rather than a sample input which will not give 100% picture of your language. or let me know the bugs you felt from which we can define the language description also.

Regards, V M Rakesh (rakesh.vm@gmail.com)

来源：https://stackoverflow.com/questions/3327951/can-i-improve-this-gold-parser-grammar

标签

parsing

bnf

gold-parser