jsonselect.js

/*! Copyright (c) 2011, Lloyd Hilaiel, ISC License */
/*
 * This is the JSONSelect reference implementation, in javascript.
 */
(function() {
    var w = window;
    var jp = (w.JSON ? w.JSON.parse : w.eval);
    function jsonParse(s) { try { return jp(s); } catch(e) { te("ijs"); }; }

    // emitted error codes.  Strip this table for an, uh, "optimized build"
    var _es = {}; // overshadow any globals when the table is stripped
    /** --->>> **/
    var _es = {
        "ijs": "invalid json string",
        "mpc": "multiple pseudo classes (:xxx) not allowed",
        "mepf": "malformed expression in pseudo-function",
        "nmi": "multiple ids not allowed",
        "se": "selector expected",
        "sra": "string required after '.'",
        "uc": "unrecognized char",
        "ujs": "unclosed json string",
        "upc": "unrecognized pseudo class"
    };
    /** <<<--- **/
    // throw a full or abbreviated error message depending on the existence of the
    // _es table
    var te = function(ec) { throw (_es[ec] ? _es[ec] : "jsonselect error: "+ec) }

    // THE LEXER
    var toks = {
        psc: 1, // pseudo class
        psf: 2, // pseudo class function
        typ: 3, // type
        str: 4, // string
    };

    var pat = /^(?:([\r\n\t\ ]+)|([*.,>])|(string|boolean|null|array|object|number)|(:(?:root|first-child|last-child|only-child))|(:(?:nth-child|nth-last-child))|(:\w+)|(\"(?:[^\\]|\\[^\"])*\")|(\")|((?:[_a-zA-Z]|[^\0-\0177]|\\[^\r\n\f0-9a-fA-F])(?:[_a-zA-Z0-9-]|[^\u0000-\u0177]|(?:\\[^\r\n\f0-9a-fA-F]))*))/;
    var exprPat = /^\s*\(\s*(?:([+-]?)([0-9]*)n\s*(?:([+-])\s*([0-9]))?|(odd|even)|([+-]?[0-9]+))\s*\)/;
    var lex = function (str, off) {
        if (!off) off = 0;
        var m = pat.exec(str.substr(off));
        if (!m) return undefined;
        off+=m[0].length;
        var a;
        if (m[1]) a = [off, " "];
        else if (m[2]) a = [off, m[0]];
        else if (m[3]) a = [off, toks.typ, m[0]];
        else if (m[4]) a = [off, toks.psc, m[0]];
        else if (m[5]) a = [off, toks.psf, m[0]];
        else if (m[6]) te("upc");
        else if (m[7]) a = [off, toks.str, jsonParse(m[0])];
        else if (m[8]) te("ujs");
        else if (m[9]) a = [off, toks.str, m[0].replace(/\\([^\r\n\f0-9a-fA-F])/g,"$1")];
        return a;
    };

    // THE PARSER

    var parse = function (str) {
        var am = undefined, a = [], off = 0;
        while (true) {
            var s = parse_selector(str, off);
            a.push(s[1]);
            s = lex(str, off = s[0]);
            if (s && s[1] === ' ') s = lex(str, off = s[0]);
            if (!s) break;
            // now we've parsed a selector, and have something else...
            if (s[1] === ">") {
                a.push(">");
                off = s[0];
            } else if (s[1] === ",") {
                if (am == undefined) am = [ ",", a ];
                else am.push(a);
                a = [];
                off = s[0];
            }
        }
        if (am) am.push(a);
        return am ? am : a;
    };

    var parse_selector = function(str, off) {
        var soff = off;
        var s = { };
        var l = lex(str, off);
        // skip space
        if (l && l[1] === " ") { soff = off = l[0]; l = lex(str, off); }
        if (l && l[1] === toks.typ) {
            s.type = l[2];
            l = lex(str, (off = l[0]));
        } else if (l && l[1] === "*") {
            // don't bother representing the universal sel, '*' in the
            // parse tree, cause it's the default
            l = lex(str, (off = l[0]));
        }

        // now support either an id or a pc
        while (true) {
            if (l === undefined) {
                break;
            } else if (l[1] === '.') {
                l = lex(str, (off = l[0]));
                if (!l || l[1] !== toks.str) te("sra");
                if (s.id) te("nmi");
                s.id = l[2];
            } else if (l[1] === toks.psc) {
                if (s.pc || s.pf) te("mpc");
                // collapse first-child and last-child into nth-child expressions
                if (l[2] === ':first-child') {
                    s.pf = ":nth-child";
                    s.a = 0;
                    s.b = 1;
                } else if (l[2] === ':last-child') {
                    s.pf = ":nth-last-child";
                    s.a = 0;
                    s.b = 1;
                } else {
                    s.pc = l[2];
                }
            } else if (l[1] === toks.psf) {
                if (s.pc || s.pf ) te("mpc");
                s.pf = l[2];
                var m = exprPat.exec(str.substr(l[0]));
                if (!m) te("mepf");
                if (m[5]) {
                    s.a = 2;
                    s.b = (m[5] === 'odd') ? 1 : 0;
                } else if (m[6]) {
                    s.a = 0;
                    s.b = parseInt(m[6], 10);
                } else {
                    s.a = parseInt((m[1] ? m[1] : "+") + (m[2] ? m[2] : "1"),10)
                    s.b = m[3] ? parseInt(m[3] + m[4],10) : 0;
                }
                l[0] += m[0].length;
            } else {
                break;
            }
            l = lex(str, (off = l[0]));
        }

        // now if we didn't actually parse anything it's an error
        if (soff === off) te("se");

        return [off, s];
    };

    // THE EVALUATOR

    function isArray(o) {
        return Object.prototype.toString.call(o) === '[object Array]';
    }

    function mytypeof(o) {
        if (o === null) return 'null';
        var to = typeof o;
        if (to === 'object' && isArray(o)) to = 'array';
        return to;
    }

    function mn(node, sel, id, num, tot) {
        var sels = [];
        var cs = (sel[0] === '>') ? sel[1] : sel[0];
        var m = true;
        if (cs.type) m = m && (cs.type === mytypeof(node));
        if (cs.id)   m = m && (cs.id === id);
        if (m && cs.pf) {
            if (cs.pf === ":nth-last-child") num = tot - num;
            else num++;
            if (cs.a === 0) {
                m = cs.b === num;
            } else {
                m = (!((num - cs.b) % cs.a) && ((num*cs.a + cs.b) >= 0));
            }
        }

        // should we repeat this selector for descendants?
        if (sel[0] !== '>' && sel[0].pc !== ":root") sels.push(sel);

        if (m) {
            // is there a fragment that we should pass down?
            if (sel[0] === '>') { if (sel.length > 2) { m = false; sels.push(sel.slice(2)); } }
            else if (sel.length > 1) { m = false; sels.push(sel.slice(1)); }
        }

        return [m, sels];
    }

    function forEach(sel, obj, fun, id, num, tot) {
        var a = (sel[0] === ',') ? sel.slice(1) : [sel];
        var a0 = [];
        var call = false;
        for (var i = 0; i < a.length; i++) {
            var x = mn(obj, a[i], id, num, tot);
            if (x[0]) call = true;
            for (var j = 0; j < x[1].length; j++) a0.push(x[1][j]);
        }
        if (a0.length && typeof obj === 'object') {
            if (a0.length >= 1) a0.unshift(",");
            if (isArray(obj)) {
                for (var i = 0; i < obj.length; i++) forEach(a0, obj[i], fun, undefined, i, obj.length);
            } else {
                // it's a shame to do this for :last-child and other
                // properties which count from the end when we don't
                // even know if they're present.  Also, the stream
                // parser is going to be pissed.
                var l = 0;
                for (var k in obj) if (obj.hasOwnProperty(k)) l++;
                var i = 0;
                for (var k in obj) if (obj.hasOwnProperty(k)) forEach(a0, obj[k], fun, k, i++, l);
            }
        }
        if (call && fun) fun(obj);
    };

    function match(sel, obj) {
        var a = [];
        forEach(sel, obj, function(x) { a.push(x); });
        return a;
    };

    function compile(sel) {
        return {
            sel: parse(sel),
            match: function(obj){return match(this.sel, obj)},
            forEach: function(obj, fun) { return forEach(this.sel, obj, fun) }
        };
    }

    w.JSONSelect = {
        // expose private API for testing
        _lex: lex,
        _parse: parse,
        // public API
        match: function (sel, obj) { return compile(sel).match(obj) },
        forEach: function(sel, obj, fun) { return compile(sel).forEach(obj, fun) },
        compile: compile
    };
})();