Langshot

A strongly-typed language extension framework. We found a slick way to make strongly typed parsers in TypeScript, and are building a language extension framework around it. Very much a work in progress.

Quick Start

A Simple Example

Let's build a super simple parser to parse the following JavaScript snippet:

constmyString="Hello, world!";letmyNumber=42;myNumber=0;

We can define a parser for this like so:

import{LangNode,token,and,or,optional,many}from"./Langshot";classFileextendsLangNode.from(()=>({statements: many(Statement),})){}classStatementextendsLangNode.from(()=>({statement: or(VarDef,VarAssign),separator: optional(token(`Separator`,/;/)),})){}classVarDefextendsLangNode.from(()=>({modifier: token(`Modifier`,/var|let|const/),ident: Ident,defOp: AssignOp,initValue: or(Ident,LitNum,LitStr),})){getisConst(){returnthis.modifier.text==="const";}}classVarAssignextendsLangNode.from(()=>({ident: Ident,assignOp: AssignOp,value: or(Ident,LitNum,LitStr),})){}classAssignOpextendsLangNode.from(()=>/=/){}classIdentextendsLangNode.from(()=>/_*[A-Za-z][A-Za-z0-9_]*/){}classLitNumextendsLangNode.from(()=>/-?[0-9]+(\.[0-9]+)?/){getasFloat(){returnparseFloat(this.text);}}classLitStrextendsLangNode.from(()=>/"([^"\\]*(\\.[^"\\]*)*)"/){getasString(){returnthis.text.slice(1,-1);}}

If we parse the snippet above using File.parse(testFile), which yields the following AST:

{
"nodeType": "File",
"startIndex": 0,
"endIndex": 66,
"statements": [
{
"nodeType": "Statement",
"startIndex": 0,
"endIndex": 33,
"statement": {
"nodeType": "VarDef",
"startIndex": 0,
"endIndex": 32,
"modifier": {
"nodeType": "Modifier",
"startIndex": 0,
"endIndex": 5,
"text": "const"
},
"ident": {
"nodeType": "Ident",
"startIndex": 6,
"endIndex": 14,
"text": "myString"
},
"defOp": {
"nodeType": "AssignOp",
"startIndex": 15,
"endIndex": 16,
"text": "="
},
"initValue": {
"nodeType": "LitStr",
"startIndex": 17,
"endIndex": 32,
"text": "\"Hello, world!\""
}
},
"separator": {
"nodeType": "Separator",
"startIndex": 32,
"endIndex": 33,
"text": ";"
}
},
{
"nodeType": "Statement",
"startIndex": 34,
"endIndex": 52,
"statement": {
"nodeType": "VarDef",
"startIndex": 34,
"endIndex": 51,
"modifier": {
"nodeType": "Modifier",
"startIndex": 34,
"endIndex": 37,
"text": "let"
},
"ident": {
"nodeType": "Ident",
"startIndex": 38,
"endIndex": 46,
"text": "myNumber"
},
"defOp": {
"nodeType": "AssignOp",
"startIndex": 47,
"endIndex": 48,
"text": "="
},
"initValue": {
"nodeType": "LitNum",
"startIndex": 49,
"endIndex": 51,
"text": "42"
}
},
"separator": {
"nodeType": "Separator",
"startIndex": 51,
"endIndex": 52,
"text": ";"
}
},
{
"nodeType": "Statement",
"startIndex": 53,
"endIndex": 66,
"statement": {
"nodeType": "VarAssign",
"startIndex": 53,
"endIndex": 65,
"ident": {
"nodeType": "Ident",
"startIndex": 53,
"endIndex": 61,
"text": "myNumber"
},
"assignOp": {
"nodeType": "AssignOp",
"startIndex": 62,
"endIndex": 63,
"text": "="
},
"value": {
"nodeType": "LitNum",
"startIndex": 64,
"endIndex": 65,
"text": "0"
}
},
"separator": {
"nodeType": "Separator",
"startIndex": 65,
"endIndex": 66,
"text": ";"
}
}
]
}

Caveats

We have big dreams for Langshot, but right now it's really small and has a lot of limitations.

• Only parsing works right now.
• There is no elegant system for comments yet.
• There is no elegant parser logging system yet.
• There is no elegant system to handle code actively being typed.