A lightly-typescriptified version of jison-lex
npm install @ts-jison/lexer-generator@ts-jison/lexer-generator:
====
A lightly-typescriptified API for creating parsers in JavaScript
-----------------------------------------
A lexical analyzer generator used by @ts-jison/parser-generator. It takes a lexical grammar definition (either in JSON or Bison's lexical grammar format) and outputs a JavaScript lexer.
This is a fork of Zach Carter parse(text, yyobject) methods.
Installation
------------@ts-jison/lexer-generator can be installed for Node using npm
Using npm:
npm install @ts-jison/lexer-generator
add a package.json script for a javascript file for e.g. an LR parser for "ShEx":
"build-parser": "ts-jison -n ShExJison -t javascript -p lr -o ./lib/ShExJison.js ./lib/ShExJison.jison"
or e.g. an LALR typescript parser:
"build-parser": "ts-jison -n ShExJison -t typescript -p lalr -o ./lib/ShExJison.ts ./lib/ShExJison.jison"
Status:
=====
This works (I'm using it in a few javascript and typescritp projects) and runs the original tests. If you want to geek about this, ping ericP on discord or ericprud on gitter.
* issues
Changes from jison
---
Most changes are in the output. Specifically, the constructed regular expressions (mostly) elide unnecessary () and eliminate capture groups on semanticly-necessary (?:), e.g.:
``
/^(?:\s+|(#[^\u000a\u000d]|\/\([^]|\([^/]|\\\/))\\/))/
is now:
/^(?:\s+|#[^\u000a\u000d]|\/\(?:[^]|\(?:[^/]|\\\/))\\/)/
This resulted in a ~300X speedup for the ShExC parser and kept the stack from blowing up on large input like the FHIR ShEx.
When invoking the lexer through the API, the old convetion for rules was an optional array of start conditions, followed by a string for the pattern and a string for the action:
` JSON`
[ "enter-test", "this.begin('test');" ],
[ ["INITIIAL", "test"], "x", "return 'T';" },
Those are now distinguised by properties in an object: JSON`
{ pattern: "enter-test", action: "this.begin('test');" },
{ start: ["INITIAL", "test"], pattern: "x", action: "return 'T';" },
Components:
=====
* parser-generator - A lightly-typescriptified version of jison
* lexer-generator - A lightly-typescriptified version of jison-lex
* parser - runtime library for parsers
* lexer - runtime library for lexers
* common - functions needed by parser and lexer
Usage: ts-lex [file] [options]file file containing a lexical grammar
Options:
-o FILE, --outfile FILE Filename and base module name of the generated parser
-t TYPE, --module-type TYPE The type of module to generate (commonjs, js)
--version print version and exit
programatic usage
var grammar = {
rules: [
["x", "return 'X';" ],
["y", "return 'Y';" ],
["$", "return 'EOF';" ]
]
};
// or load from a file
// var grammar = fs.readFileSync('mylexer.l', 'utf8');
// generate source
var lexerSource = JisonLex.generate(grammar);
// or create a parser in memory
var lexer = new JisonLex(grammar);
lexer.setInput('xyxxy');
lexer.lex();
// => 'X'
lexer.lex();
// => 'Y'