(function () {
var $P = Date.Parsing;
var _ = $P.Operators = {
//
// Tokenizers
//
rtoken: function (r) { // regex token
return function (s) {
var mx = s.match(r);
if (mx) {
return ([ mx[0], s.substring(mx[0].length) ]);
} else {
throw new $P.Exception(s);
}
};
},
token: function () { // whitespace-eating token
return function (s) {
return _.rtoken(new RegExp("^\\s*" + s + "\\s*"))(s);
};
},
stoken: function (s) { // string token
return _.rtoken(new RegExp("^" + s));
},
// Atomic Operators
until: function (p) {
return function (s) {
var qx = [], rx = null;
while (s.length) {
try {
rx = p.call(this, s);
} catch (e) {
qx.push(rx[0]);
s = rx[1];
continue;
}
break;
}
return [ qx, s ];
};
},
many: function (p) {
return function (s) {
var rx = [], r = null;
while (s.length) {
try {
r = p.call(this, s);
} catch (e) {
return [ rx, s ];
}
rx.push(r[0]);
s = r[1];
}
return [ rx, s ];
};
},
// generator operators -- see below
optional: function (p) {
return function (s) {
var r = null;
try {
r = p.call(this, s);
} catch (e) {
return [ null, s ];
}
return [ r[0], r[1] ];
};
},
not: function (p) {
return function (s) {
try {
p.call(this, s);
} catch (e) {
return [null, s];
}
throw new $P.Exception(s);
};
},
ignore: function (p) {
return p ?
function (s) {
var r = null;
r = p.call(this, s);
return [null, r[1]];
} : null;
},
product: function () {
var px = arguments[0],
qx = Array.prototype.slice.call(arguments, 1), rx = [];
for (var i = 0 ; i < px.length ; i++) {
rx.push(_.each(px[i], qx));
}
return rx;
},
cache: function (rule) {
var cache = {}, cache_length = 0, cache_keys = [], CACHE_MAX = Date.Config.CACHE_MAX || 100000, r = null;
var cacheCheck = function () {
if (cache_length === CACHE_MAX) {
// kill several keys, don't want to have to do this all the time...
for (var i=0; i < 10; i++) {
var key = cache_keys.shift();
if (key) {
delete cache[key];
cache_length--;
}
}
}
};
return function (s) {
cacheCheck();
try {
r = cache[s] = (cache[s] || rule.call(this, s));
} catch (e) {
r = cache[s] = e;
}
cache_length++;
cache_keys.push(s);
if (r instanceof $P.Exception) {
throw r;
} else {
return r;
}
};
},
// vector operators -- see below
any: function () {
var px = arguments;
return function (s) {
var r = null;
for (var i = 0; i < px.length; i++) {
if (px[i] == null) {
continue;
}
try {
r = (px[i].call(this, s));
} catch (e) {
r = null;
}
if (r) {
return r;
}
}
throw new $P.Exception(s);
};
},
each: function () {
var px = arguments;
return function (s) {
var rx = [], r = null;
for (var i = 0; i < px.length ; i++) {
if (px[i] == null) {
continue;
}
try {
r = (px[i].call(this, s));
} catch (e) {
throw new $P.Exception(s);
}
rx.push(r[0]);
s = r[1];
}
return [ rx, s];
};
},
all: function () {
var px = arguments, _ = _;
return _.each(_.optional(px));
},
// delimited operators
sequence: function (px, d, c) {
d = d || _.rtoken(/^\s*/);
c = c || null;
if (px.length === 1) {
return px[0];
}
return function (s) {
var r = null, q = null;
var rx = [];
for (var i = 0; i < px.length ; i++) {
try {
r = px[i].call(this, s);
} catch (e) {
break;
}
rx.push(r[0]);
try {
q = d.call(this, r[1]);
} catch (ex) {
q = null;
break;
}
s = q[1];
}
if (!r) {
throw new $P.Exception(s);
}
if (q) {
throw new $P.Exception(q[1]);
}
if (c) {
try {
r = c.call(this, r[1]);
} catch (ey) {
throw new $P.Exception(r[1]);
}
}
return [ rx, (r?r[1]:s) ];
};
},
//
// Composite Operators
//
between: function (d1, p, d2) {
d2 = d2 || d1;
var _fn = _.each(_.ignore(d1), p, _.ignore(d2));
return function (s) {
var rx = _fn.call(this, s);
return [[rx[0][0], r[0][2]], rx[1]];
};
},
list: function (p, d, c) {
d = d || _.rtoken(/^\s*/);
c = c || null;
return (p instanceof Array ?
_.each(_.product(p.slice(0, -1), _.ignore(d)), p.slice(-1), _.ignore(c)) :
_.each(_.many(_.each(p, _.ignore(d))), px, _.ignore(c)));
},
set: function (px, d, c) {
d = d || _.rtoken(/^\s*/);
c = c || null;
return function (s) {
// r is the current match, best the current 'best' match
// which means it parsed the most amount of input
var r = null, p = null, q = null, rx = null, best = [[], s], last = false;
// go through the rules in the given set
for (var i = 0; i < px.length ; i++) {
// last is a flag indicating whether this must be the last element
// if there is only 1 element, then it MUST be the last one
q = null;
p = null;
r = null;
last = (px.length === 1);
// first, we try simply to match the current pattern
// if not, try the next pattern
try {
r = px[i].call(this, s);
} catch (e) {
continue;
}
// since we are matching against a set of elements, the first
// thing to do is to add r[0] to matched elements
rx = [[r[0]], r[1]];
// if we matched and there is still input to parse and
// we don't already know this is the last element,
// we're going to next check for the delimiter ...
// if there's none, or if there's no input left to parse
// than this must be the last element after all ...
if (r[1].length > 0 && ! last) {
try {
q = d.call(this, r[1]);
} catch (ex) {
last = true;
}
} else {
last = true;
}
// if we parsed the delimiter and now there's no more input,
// that means we shouldn't have parsed the delimiter at all
// so don't update r and mark this as the last element ...
if (!last && q[1].length === 0) {
last = true;
}
// so, if this isn't the last element, we're going to see if
// we can get any more matches from the remaining (unmatched)
// elements ...
if (!last) {
// build a list of the remaining rules we can match against,
// i.e., all but the one we just matched against
var qx = [];
for (var j = 0; j < px.length ; j++) {
if (i !== j) {
qx.push(px[j]);
}
}
// now invoke recursively set with the remaining input
// note that we don't include the closing delimiter ...
// we'll check for that ourselves at the end
p = _.set(qx, d).call(this, q[1]);
// if we got a non-empty set as a result ...
// (otw rx already contains everything we want to match)
if (p[0].length > 0) {
// update current result, which is stored in rx ...
// basically, pick up the remaining text from p[1]
// and concat the result from p[0] so that we don't
// get endless nesting ...
rx[0] = rx[0].concat(p[0]);
rx[1] = p[1];
}
}
// at this point, rx either contains the last matched element
// or the entire matched set that starts with this element.
// now we just check to see if this variation is better than
// our best so far, in terms of how much of the input is parsed
if (rx[1].length < best[1].length) {
best = rx;
}
// if we've parsed all the input, then we're finished
if (best[1].length === 0) {
break;
}
}
// so now we've either gone through all the patterns trying them
// as the initial match; or we found one that parsed the entire
// input string ...
// if best has no matches, just return empty set ...
if (best[0].length === 0) {
return best;
}
// if a closing delimiter is provided, then we have to check it also
if (c) {
// we try this even if there is no remaining input because the pattern
// may well be optional or match empty input ...
try {
q = c.call(this, best[1]);
} catch (ey) {
throw new $P.Exception(best[1]);
}
// it parsed ... be sure to update the best match remaining input
best[1] = q[1];
}
// if we're here, either there was no closing delimiter or we parsed it
// so now we have the best match; just return it!
return best;
};
},
forward: function (gr, fname) {
return function (s) {
return gr[fname].call(this, s);
};
},
//
// Translation Operators
//
replace: function (rule, repl) {
return function (s) {
var r = rule.call(this, s);
return [repl, r[1]];
};
},
process: function (rule, fn) {
return function (s) {
var r = rule.call(this, s);
return [fn.call(this, r[0]), r[1]];
};
},
min: function (min, rule) {
return function (s) {
var rx = rule.call(this, s);
if (rx[0].length < min) {
throw new $P.Exception(s);
}
return rx;
};
}
};
// Generator Operators And Vector Operators
// Generators are operators that have a signature of F(R) => R,
// taking a given rule and returning another rule, such as
// ignore, which parses a given rule and throws away the result.
// Vector operators are those that have a signature of F(R1,R2,...) => R,
// take a list of rules and returning a new rule, such as each.
// Generator operators are converted (via the following _generator
// function) into functions that can also take a list or array of rules
// and return an array of new rules as though the function had been
// called on each rule in turn (which is what actually happens).
// This allows generators to be used with vector operators more easily.
// Example:
// each(ignore(foo, bar)) instead of each(ignore(foo), ignore(bar))
// This also turns generators into vector operators, which allows
// constructs like:
// not(cache(foo, bar))
var _generator = function (op) {
function gen() {
var args = null, rx = [], px, i;
if (arguments.length > 1) {
args = Array.prototype.slice.call(arguments);
} else if (arguments[0] instanceof Array) {
args = arguments[0];
}
if (args) {
px = args.shift();
if (px.length > 0) {
args.unshift(px[i]);
rx.push(op.apply(null, args));
args.shift();
return rx;
}
} else {
return op.apply(null, arguments);
}
}
return gen;
};
var gx = "optional not ignore cache".split(/\s/);
for (var i = 0 ; i < gx.length ; i++) {
_[gx[i]] = _generator(_[gx[i]]);
}
var _vector = function (op) {
return function () {
if (arguments[0] instanceof Array) {
return op.apply(null, arguments[0]);
} else {
return op.apply(null, arguments);
}
};
};
var vx = "each any all".split(/\s/);
for (var j = 0 ; j < vx.length ; j++) {
_[vx[j]] = _vector(_[vx[j]]);
}
}()); |