What am I doing wrong here? I'm using the fat arrows => for my callbacks, however when the code reaches cb.onEndElement and calls @returner I get an null object exception. So why doesn't @returner exist?
class Parser
constructor: () ->
@returner = (data) ->
searchParser: new xml.SaxParser (cb) =>
cb.onStartElementNS (elem, attrs, prefix, url, ns) =>
if elem is "results" then @results = []
else if elem is "title" then @curr = "title"
else @curr = "none"
cb.onCdata (cdata) =>
if @curr is "title" then @book.title = cdata
cb.onEndElementNS (elem, prefix, url) =>
@results.push @book if elem is "book"
cb.onEndDocument =>
@returner @results
search: (str, callback) ->
@returner = callback
@searchParser.parseString str
p = new Parser
p.search "somexml", (data) ->
console.log JSON.stringify data
Your method search
needs a fat-arrow =>
in order to bind it to instances of Parser
.
Additionally, although the line searchParser: new xml.SaxParser (cb) =>
compiles, it's probably not doing what you want, because the fat arrow is binding the callback to Parser
and not this
. You have two options:
- you should probably put
@searchParser = new xml.SaxParser (cb) => ...
in your constructor instead, given the way you are calling it.
- otherwise you could use
searchParser: () => new xml.SaxParser (cb) =>
and call it with parens lower down: @searchParser().parseString str
, which would create a searchParser
method bound to this
As an example, here are my two solutions, as well as your original line, slightly simplified, as well as the compiled code, for compare and contrast purposes:
Simplified example in CoffeeScript:
class Parser
constructor: () -> @searchParser1 = new xml.SaxParser (x) => console.log(x)
searchParser2: () => new xml.SaxParser (x) => console.log(x)
searchParser: new xml.SaxParser (x) => console.log(x)
Compiled JavaScript:
var Parser;
var __bind = function(fn, me){ return function(){ return fn.apply(me, arguments); }; };
Parser = (function() {
function Parser() {
this.searchParser2 = __bind(this.searchParser2, this);
this.searchParser1 = new xml.SaxParser(__bind(function(x) {
return console.log(x);
}, this));
}
Parser.prototype.searchParser2 = function() {
return new xml.SaxParser(__bind(function(x) {
return console.log(x);
}, this));
};
Parser.prototype.searchParser = new xml.SaxParser(__bind(function(x) {
return console.log(x);
}, Parser));
return Parser;
}).call(this);
Note how searchParser1
and searchParser2
have their callbacks bound to this
and searchParser
's is bound to Parser
.
As always, the "Try CoffeeScript" button at the CoffeeScript homepage (http://jashkenas.github.com/coffee-script/) is your friend!
First off, the concept you're talking about isn't "scope"—it's this
, also informally called the "context." And it's certainly one of the trickiest concepts in JavaScript (and thus CoffeeScript), even though the rules are fairly simple. Perhaps it's because the word this
itself, which doesn't seem like its meaning should be so easily changed depending on how a function is called...
Nicholas' answer is dead-on, but I recommend that you also read up on this
and try to really understand it, rather than just using =>
all the time (it's a great tool, but not always the right one). A few resources I recommend:
- JavaScript Garden's section on
this
- How to Node's article What is
this
?
- My book, CoffeeScript: Accelerated JavaScript Development (a shameless plug, yes, but it does explain
this
, as well as scope, in a fair amount of depth)