A parser class for simple formulae, like those of algebra and propositional logic.
Produces ASTs in JSON format.
The algorithm is a fully-immutable JavaScript adaptation of precedence climbing.
npm install formula-parserES module:
import FormulaParser from 'formula-parser';Node:
const FormulaParser = require('formula-parser');Browser:
<script src="node_modules/formula-parser/dist/formula-parser.js"></script>FormulaParser is a parser class for operator-precedence languages, i.e.,
context-free languages
which have only variables, (prefix) unary operators, and (infix) binary operators.
This restriction means that the grammar for a parser instance is wholly specified by the operator definitions (and a key with which to label variable nodes).
As the algebraParser example demonstrates,
an operator definition is an object like the following:
{ symbol: '+', key: 'plus', precedence: 1, associativity: 'left' }It specifies a symbol, a key for its AST node,
a precedence level, and (for binaries) an associativity direction.
Once the definitions are assembled, creating a parser instance is straightforward:
const algebraParser = new FormulaParser(variableKey, unaries, binaries);After creating a FormulaParser instance, calling its parse method will produce an AST for a formula:
algebraParser.parse('(a + b * c) ^ -d');→
{ "exp": [
{ "plus": [
{ "var": "a" },
{ "mult": [
{ "var": "b" },
{ "var": "c" }
]}
]},
{ "neg": { "var": "d" } }
]}Technically, constants aren't supported—the leaves of the formula are all treated as variables, the values of which are to be evaluated at some post-parse stage.
That said, since a "variable" for present purposes is any alphanumeric string (including underscores),
'true', 'PI', and even '3' will all be happily parsed as such.
(Of course, numbers in decimal notation will fail.)
Function symbols aren't explicitly supported either, but they can be simulated by operator symbols.
Specifying sin as a unary symbol will accept sin x or sin(x),
while specifying mod as a binary symbol will accept x mod y.
Unfortunately, this one's hard and fast:
Unary symbols must be in prefix notation and binary symbols must be in infix notation.