Awesome
esniff
Low footprint JavaScript source code parser
Low footprint, fast source code parser, which allows you to find all code fragment occurrences with respect to all syntax rules that cannot be handled with plain regular expression search.
It aims at use cases where we don't need full AST tree, but instead we're interested in finding usages of given function, property etc. in syntactically valid code.
Installation
npm
$ npm install esniff
Usage
Using main module you can configure sophisticated parser on your own. However, first, see preprared API utilities that may already address use cases you have.
esniff(code, executor)
code
- Code to parseexecutor
- A function to be executed immediately by the constructor, It receives anemitter
parameter.
emitter
emits following events:
trigger:<char>
- When char is a code character approached in code, that is not a whitespace, is not in a middle of identificator, is not part of a comment, string, template string or regular expression.
Emitter passes to listener the accessor
object, which provides access to current parser state and allows to manipulate parsing process. accessor
exposes following methods:
skipCodePart(codePart)
- Skips forward through input codePart assuming parser index points start of given part. Returns true if givencodePart
was foundskipIdentifier
- Skips approached identifier (can be function name or property name), returns{ name, start, end }
meta object that describes location and identifier nameskipWhitespace
- Skips any whitespace and comments found at current parsing indexcollectScope
- If at current index(
character is found, it registers given paranthesis scope for registrations (its content will be returned as one of the results after finished parsing)stop
- Stops parsing processindex
- Returns currently parsed indexpreviousToken
- Previous non-whitespace characterscopeDepth
- Current scope depthshouldCollectComments
- Whether data about code comments should be collected in the result
Example
Parse all require(..)
calls:
var esniff = require("esniff");
var parseRequires = function (code) {
return esniff(code, function (emitter) {
emitter.on("trigger:r", function (accessor) {
if (accessor.previousToken === ".") return;
if (!accessor.skipCodePart("require")) return;
accessor.skipWhitespace();
accessor.collectScope();
});
});
};
console.log(parseRequires("var x = require('foo/bar')"));
[{ type: "scope", point: 17, column: 17, line: 1, raw: "'foo/bar'" }];
Predefined utils for common use cases
accessedProperties(objName) (esniff/accessed-properties)
Returns function which allows us to find all accessed property names on given object name
var findProperties = require("esniff/accessed-properties");
var findContextProperties = findProperties("this");
var result = findContextProperties(
"var foo = \"0\"; this.bar = foo; this.someMethod(); otherFunction()"
);
console.log(result); // [ { name: 'bar', start: 20, end: 23 }, { name: 'someMethod', start: 36, end: 46 } ]
function(name[, options]) (esniff/function)
Returns function which allows us to find all occurrences of given function (or method) being invoked
Through options we can restrict cases which we're after:
asProperty
(default:false
), on true will allowx.name()
when we search forname
callsasPlain
(default:true
), on true it allows plain calls e.g.name()
when we search forname
. Should be set tofalse
if we're strictly about method calls.
Setting both asProperty
and asPlain
to false, will always produce empty result
var findRequires = require("esniff/function")("require");
findRequires("var x = require('foo/bar')");
// [{ point: 17, column: 17, line: 1, raw: '\'foo/bar\'' }]
resolveArguments(code[, limit]) (esniff/resolve-arguments)
Resolves expressions separated with commas, with additional limit
you can specify after which number of arguments resolver should stop
var resolveArgs = require("esniff/resolve-arguments");
var result = resolveArgs("'raz', 'dwa', ['raz', 'dwa'], 'trzy'", 3);
console.log(result); // ['"raz"', ' "dwa"', ' [\'raz\', \'dwa\']']
Limitations
- esniff assumes code that you pass is syntactically correct, it won't inform you about any syntax errors and may produce unexpected and nonsense results when such code is used.
- There's single case of syntactically correct code, which will make esniff produce incorrect results, it's division made directly on object literal (e.g.
x = { foo: 'bar' } / 14
, esniff in that case will assume that/
starts regular expression). Still there's not known use case where such code may make any sense, and many popular JS source code parsers share very same vulnerability.
Tests
$ npm test
Security contact information
To report a security vulnerability, please use the Tidelift security contact. Tidelift will coordinate the fix and disclosure.