Ggram

• Ggram
  • What is it ?
  • Installation or manual compilation
    • Compilation
  • What is a .gg syntax file ?
  • Options
    • Basic Usage
    • Command line options
      • Restype
  • Syntax
    • Lexer
    • Parser
  • Example

What is it ?

Ggram is a lexer-parser generator. It translates a .gg file describing a language syntax using something very close to the Backus–Naur form (BNF) into a lexer-parser written in cpp.

Installation or manual compilation

Compilation

Download the source code here.

Then, in the source directory, run :

make

What is a .gg syntax file ?

As said earlier, a .gg file describes a language syntax (so that ggram can generate a lexer-parser out of it). We'll look at a syntax example for simple calculations :

number "[1-9][0-9]*|0"

---

<program> ::= 

Options

Basic Usage

./ggram -f syntax_file_name.gg -o output_directory_name
./ggram syntax_file_name.gg -o output_directory_name

Command line options

Option	Description	Possible values
-f	Specify the syntax file to use.	A file path
-o	Specify the output directory.	A directory path
-h / --help	Display the help.
-v / --version	Display the version name
-r / --restype	Controls how the parser works interally	ORS, TRY_CATCH, ERROR_TOKEN

Restype

TODO

Syntax

Lexer

terminal "regexp_expression"

A terminal can be ignored by the parser by adding .ignore before it.

The lexer return a vector of Token. A Token is an object containing the terminal name (type), the value of the terminal, line and column.

Parser

<rule_name> ::= <rule> ;

Rule can contain parenthesis for priority, brackets for optionality, braces for repetition and the | operator for alternatives. Rule can also be a terminal. Rule can be recursive (but not left recursive). If you want check the value of a token, you can use "token_value" as a rule if the token is a terminal and if the token is lexed by the lexer.

The main rule is the one named program. Without it, the parser won't be able to parse anything.

The parser return a tree of Token (yes, the Token object contains a vector of Token, the "children").

Example

For simple arithmetic expressions :

int "[1-9][0-9]*|0"
float "[1-9][0-9]*.[0-9]*|[0-9]*.[0-9]*"
operator "[+-*/]"
parenthesis "[()]"
fun "[fgh]"
var "[x-z]"
const "[a-e]"
.ignore "[\n\t\r ]"

---

<program> ::= { <expression> ";" } <expression> | <expression> ;

<expression> ::= <fun> "(" <expression> ")"
    | <term> <operator> <expression>
    | <term> ;

<term> ::= <int> | <float> | <var> | <const> | "(" <expression> ")" ;