It's Probably Interactive

Wax talks to Twitter

2009-10-20T18:33:54-07:00

One of my favorite parts about Wax is how easy HTTP calls are, especially to REST APIs. Basically, to make an HTTP request you just need this code...

function getSomeTwitterTrends(self)
  local yourCallback = function(json, response)
    -- Do whatever you want here... it is called when the response is received.
    -- This is where you can refresh a view, update a tableview, or whatevs
  end

  local url = "http://search.twitter.com/trends.json"
  HTTPotluck.request{url, callback = yourCallback}
end

That's it! You've got yourself an asynchronous HTTP call that parses a JSON response into a Lua Table!

Below I will step you through creating a simple Twitter app that lists the recent trends in a UITableView. You can get the source here.

Setup Wax

Open up Xcode and create a new project using the Wax template. You can name it whatever you want, I'm going to call mine TwitterApp. If you haven't setup Wax yet you should start with this article.

Since we are going to be talking to the Twitter's API we need to add the HTTPotluck and JSON extensions. It's easy though because they are bundled with Wax.

Open up the TwitterAppAppDelegate.m file.
Replace wax_start(); with wax_startWithExtensions(luaopen_HTTPotluck, luaopen_json, nil);
Make sure you import the extension header files json.h && HTTPotluck.h as well.

// The other imports are omitted from this sample code.

#import "json.h"
#import "HTTPotluck.h"

@implementation TwitterAppAppDelegate

@synthesize window;

- (void)dealloc {
    [window release];
    [super dealloc];
}

- (void)applicationDidFinishLaunching:(UIApplication *)application {    
    [window makeKeyAndVisible];
    wax_startWithExtensions(luaopen_HTTPotluck, luaopen_json, nil);
}


@end

Let's write some Lua!

We are going to list the current Twitter trends in a UITableView. First create a new file called APP_ROOT/data/scripts/TwitterTableViewController.lua

Create a UITableViewController subclass with this call

waxClass{"TwitterTableViewController", UI.TableViewController}

Create the init function

function init(self)
  self.super:initWithStyle(UITableViewStylePlain)
  self.trends = {}

  return self
end

Implement UIDataSource as shown below, then run the app! you should have a real simple(boring) UITableView up and running!

function numberOfSectionsInTableView(self, tableView)
  return 1
end

function tableView_numberOfRowsInSection(self, tableView, section)
  return #self.trends
end

function tableView_cellForRowAtIndexPath(self, tableView, indexPath)
  local identifier = "TwitterTableViewControllerCell"
  local cell = tableView:dequeueReusableCellWithIdentifier(identifier) or
               UI.TableViewCell:initWithStyle_reuseIdentifier(UITableViewCellStyleDefault, identifier)

  local object = self.trends[indexPath:row() + 1] -- Must +1 because lua arrays are 1 based
  cell:textLabel():setText(object)

  return cell
end

Twitter API

Now we are going to use HTTPotluck to get info from the Twitter api. HTTPotluck is a wrapper around the iPhone SDK's NSURLConnection class. Lua makes the entire processes SO MUCH easier. Here is the code we need to add to our TwitterTableViewController

function viewDidLoad(self)
  local trendsDownloaded = function(json, response)
    if response:statusCode() == 200 then
      self.trends = json["trends"]
    else
      -- Ignore errors for now...
    end
    self:tableView():reloadData()
  end

  HTTPotluck.request{"http://search.twitter.com/trends.json", callback = trendsDownloaded}
end

It's not terribly exciting, but that's the best part! Since Lua has scoped closures we can create the function trendsDownloaded and pass it into the HTTPotluck request. When the request is finished trendsDownloaded is called with the json results (as a Lua table) and the NSHTTPURLResponse arguments. All thats left is to fill up our trends variable with the returned results!

Since Twitter returns Content-type=application/json HTTPotluck transforms the response body's JSON text into Lua table. You can check out what's inside a Lua table with a call like this puts(json).

You'll also need to update the tableView_cellForRowAtIndexPath and the init functions (check out the source)
Run it! You've got Twitter trends in a table view!

Improvements

There are obvious improvements we can make but for now that seems like a good jumping off point for making HTTP requests with Wax!

How does iPhone Wax work?

2009-10-19T09:15:16-07:00

Here is a quick tour of the Wax framework. By the end you should have a better understanding of how Wax interacts with Objective-C.

Wax's Lua Syntax

In Wax, Cocoa classes are namespaced to keep the global scope clean. To access a class you put a dot . between the framework prefix and the class name.
```
NS.String -- Returns the NSString class
UI.View -- Returns the UIView class

UI.TableView -- This is just syntactic sugar for the line below.
wax.class["UITableView"] -- This is the raw Wax call
```
To see how this works check out the cocoaNamespace calls in APP_ROOT/wax/wax-scripts/init.lua

To write your own obj-c object, create a new file with a call to waxClass(CLASS_NAME, PARENT_CLASS) at the top.

waxClass{"MyClass", NS.Object}

-- If you need to add protocols, add them like this
waxClass{"MyClass", NS.Object, protocols = {"UITableViewDelegate", "UITableViewDataSource"}}

-- REMEMBER! Protocols need to be added as strings since they are not classes.

Calling a Lua function with brackets {} instead of parenthesis () (like in the above call to waxClass) is just syntactic sugar for calling a function with a single table as the argument.
```
waxClass{"MyClass", NS.Object}
-- is equivalent to...
waxClass({"MyClass", NS.Object})
-- ...omitting the parenthesis is much more pretty, so that's how we roll.
```
The waxClass function creates an obj-c class via the obj-c runtime. All subsequent functions (within the same file) you add to that Lua file will be automatically added to your new class as instance methods.

For waxClass functions, the first argument must be self. This is how Wax mimics obj-c's object orientated model.

waxClass{"MyClass", NS.Object}

-- Notice the first argument is self... this will hold an instance of MyClass 
function storeWords(self, words)
  self.words = words
  
  -- Wax creates a super variable on the self object. 
  -- Use that to call a method on the object's super class.
  self.super:words(self)
end

Use a colon : instead of a dot . when you call any obj-c method. This secretly passes the calling object in as the first argument to the method.

-- A method called with a colon : in Lua ...
UI.Application:sharedApplication()

-- ... is equivalent to this
UI.Application.sharedApplication(UI.Application)

-- So we just use the colon : to keep this DRY and readable!

You can dynamically create member variables for any obj-c object using the dot . operator. Unlike the colon : operator (used to call methods on an obj-c class/instance) the dot . operator dynamically creates member variables on the Lua side of the object (The obj-c side of the object has no knowledge of these variables). The member variables are available throughout the lifetime of the object.
```
-- A method called with a colon : in Lua ...
local view = UI.View:init()

view.someRandomVariable = "YEAH!"
-- You can assign anything to an instance and it will stick around
```

If an obj-c method takes multiple arguments, the method call uses underscores _ instead of colons like in obj-c.

In obj-c...

[UIAlertView initWithTitle:@"title" message:@"message" delegate:nil];

With Wax...

UI.AlertView:initWithTitle_message_delegate("title", "message", nil)

Wax will try and coerce obj-c objects into Lua. It also does the reverse, if an obj-c methods takes an NSString argument, you can just pass it a lua string.

local fileManager = NS.FileManager:defaultManager()
local contents = fileManager:directoryContentsAtPath(".")
-- directoryContentsAtPath returns an NSArray, but Wax turns this into a Lua
-- table, which is totally cooler.

print(contents[1]) --> "info.plist"

-- NSDictionaries become Lua tables
-- NSArrays and NSSets become Lua tables
-- NSStrings become Lua strings
-- NSNumbers become Lua numbers

Sometimes you don't want an obj-c object to be coerced, for that use the toobjc method.

-- if you are trying to call an NSString method, this will fail ...
local string = "a string"
string:capitalizedString()
-- ... because string is coerced into a Lua string

-- This will force string to behave like an NSString
toobjc(string):capitalizedString()

Enums, like UITableViewStylePlain, are a bit tricky. For now the most commonly used ones are declared in APP_ROOT/wax/wax-scripts/enums.lua. In the future I hope to get BridgeSupport working for the iPhone so all enums & structs can be created automatically.

Structs are also tricky, the most common ones are decalred in APP_ROOT/wax/wax-scripts/structs.lua. It is easy to add structs by following the pattern found in the structs.lua file.

wax.struct.create("CGRect", "ffff", "x", "y", "width", "height")

-- The creates a global function called CGRect that takes 4 float arguments,
-- the second argument "ffff" is what defines the argument types.

local rect = CGRect(1, 2, 3, 4)
print(rect.x) --> 1
rect.x = 200
print(rect.x) --> 200

Setting up iPhone Wax

2009-10-18T09:14:32-07:00

iPhone Wax is is a Lua framework to make iPhone app development faster! What is the advantage of writing apps in Lua? Lua is a modern dynamic language, like Ruby or Python. Because of that you can be more expressive in less lines of code, take advantage of modern language techniques and extend Lua any way you see fit!

Install Wax

Go to http://github.com/probablycorey/wax and do what the README says.
Don't know Lua? If you are familiar with Javascript, Python, Ruby or any dynamic language you can learn the basics of Lua in 15 minutes! You can check out these free tutorials or buy a copy of Programing in Lua.

Create a new project

Open up Xcode and create a new project using the Wax template (The wax template is installed by following these instructions). You can name the project whatever you want, I'm going to be super generic and call mine TableView.
From now on APP_ROOT references the path of this Xcode project.
Run the app! You'll get a simple "Hello Lua" program running in your iPhone Simulator!

Let's write some Lua!

You don't need to edit the Lua code from Xcode, it's all located in the APP_ROOT/date/scripts directory, so open up your favorite text editor and go at it! You should also look at Wax's support files at APP_ROOT/wax/wax-scripts.
- If you want to use TextMate type rake tm from APP_ROOT to setup a project. Also install the wax-bundle for some helpful keyboard shortcuts.
- If Xcode strikes your fancy, you could try capgo.com's syntax highlighting plugin.
To create the TableView, first create a new file called APP_ROOT/data/scripts/MyTableViewController.lua

Create a new obj-c controller class with this call

waxClass{"MyTableViewController", UI.TableViewController}

Now add the init function, it will setup the tableView's contents and call the super's init (just like in obj-c!)

function init(self)
  self.super:initWithStyle(UITableViewStylePlain)
  
   -- Here are the tableView's contents
  self.things = {"Planes", "Trains", "Automobiles"}

  return self
end

Now we need to implement methods from the UIDataSource protocol. The final MyTableViewController.lua file should look like this.

waxClass{"MyTableViewController", UI.TableViewController}
    
function init(self)
  self.super:initWithStyle(UITableViewStylePlain)

   -- Here are the tableView's contents
  self.things = {"Planes", "Trains", "Automobiles"}

  return self
end
    
function numberOfSectionsInTableView(self, tableView)
  return 1
end

function tableView_numberOfRowsInSection(self, tableView, section)
  return #self.things
end

function tableView_cellForRowAtIndexPath(self, tableView, indexPath)
  local identifier = "MyTableViewControllerCell"
  local cell = tableView:dequeueReusableCellWithIdentifier(identifier) or
               UI.TableViewCell:initWithStyle_reuseIdentifier(UITableViewCellStyleDefault, identifier)

  local thing = self.things[indexPath:row() + 1] -- Must +1 because Lua arrays are 1 based
  cell:textLabel():setText(thing)

  return cell
end

All that is left is to create an instance of MyTableViewController and add it to the main window. We do this by editing APP_ROOT/data/scripts/init.lua. We do this in init.lua because Wax runs this file when the app starts.
```
require "wax"
require "MyTableViewController"

window = UI.Application:sharedApplication():keyWindow()

controller = MyTableViewController:init()
window:addSubview(controller:view())
```
Run the app... You will have a really simple/boring UITableView that was created via Lua!

Learn More

That's just the beginning of what you can do with iPhone Wax! To learn more, check out the article about how iPhone Wax works!

Ruby (tinyrb) on iPhone

2009-05-03T17:54:52-07:00

Some caveats...

First, tinyrb isn't technically ruby. It's a subset of ruby intended to be fast, embedable, portable and have low memory usage. It's also still in development and as the author (macournoyer) says "not useable for real things" yet. And last, I was having trouble getting garbage collection working so I gave up and just turned it off.

But who cares! Tinyrb is awesomely tiny, and super fun to play around with! So I wanted to see if I could get it running on the iPhone – and I did. To get it running on your phone check out the source and sample project http://github.com/probablycorey/iphone-loves-tinyrb/tree/master

The sample app doesn't do much of anything. It just prints some words out to stdout, but it does so using ruby!

Tinyrb on the iPhone from scratch

Grab the original tinyrb source from github.
Remove the vendor dependencies we won't need for the iphone (freegetopt, gc, peg)
Compile a PCRE static lib for the iPhone. If you don't know how check out this post.
Delete tinyrb/vm/tr.c we don't need it since will will instantiate the vm directly from within the app.
tinyrb uses GC_malloc to allocate space, since I couldn't figure out how to get GC working on the iphone I punted and just have tinyrb never cleanup after itself. Below is the code I changed in tr.h.

// Need this function because tinyrb expects malloc's to return zeroed out memory
void *zeroed_malloc(size_t size) {
  void *value = malloc(size);
  if (value != 0) memset(value, 0, size);
  return value;
}

// Don't let the GC function do anything!
void _noop();

/* allocation macros */
#define TR_MALLOC            zeroed_malloc
#define TR_CALLOC(m,n)       calloc((m)*(n))
#define TR_REALLOC           realloc
#define TR_FREE(S)           free(s)

#define GC_INIT              _noop
#define GC_gcollect          _noop

Get rid of #include <gc.h> and change #include <pcre.h> to #include "pcre.h"
Create an xcode iPhone project. Add the "pcre", "vm" and "lib" folders to the project. When adding the "lib" folder make sure you choose "Create Folder References for added folders" instead of "Recursively create groups for added folders". This will add the lib files to the application bundle so the tinyrb VM can access them at runtime.
You have to change the app's current working directory to the bundle path. Tinyrb expects to find the "lib" folder to be in the cwd.
You need to name the VM variable "vm", a lot of the tinyrb macros rely on this.
Now just start up the VM and send it some ruby. It will output it all to stdout.

vm = TrVM_new();    

NSString *input = [NSString stringWithFormat:@"a = 'does this work?'; puts a + ' YEAH'"];
NSString *input = [NSString stringWithFormat:@"a = 4; puts a + 6"];
TR_FAILSAFE(TrVM_eval(vm, [input UTF8String], "<eval>"));

Building PCRE static lib for the iPhone

2009-05-02T22:09:52-07:00

Building a static lib for the iphone is kind of tricky. I couldn't find much documentation on the process, so here is the script I used to compile PCRE for the iphone. I'm using the 2.2 SDK in these examples... if you are using something different don't forget to change the path.

#!/bin/bash

# This script will compile a PCRE static lib for the device and simulator

cd pcre

# Compile a version for the device...

CC=/Developer/Platforms/iPhoneOS.platform/Developer/usr/bin/gcc \
CPP=/Developer/Platforms/iPhoneOS.platform/Developer/usr/bin/cpp \
AR=/Developer/Platforms/iPhoneOS.platform/Developer/usr/bin/ar \
./configure --disable-shared --host=arm-apple-darwin --disable-cpp \
CFLAGS="-arch armv6 -isysroot /Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS2.2.sdk"

make

mv .libs/libpcre.a .libs/libpcre_device.a

# Compile a version for the simulator...
CC=/Developer/Platforms/iPhoneSimulator.platform/Developer/usr/bin/gcc \
CPP=/Developer/Platforms/iPhoneSimulator.platform/Developer/usr/bin/cpp \
AR=/Developer/Platforms/iPhoneSimulator.platform/Developer/usr/bin/ar \
./configure --disable-shared --disable-cpp \
CFLAGS="-isysroot /Developer/Platforms/iPhoneSimulator.platform/Developer/SDKs/iPhoneSimulator2.2.sdk"

make

mv .libs/libpcre.a .libs/libpcre_simulator.a

# Combine the two libs into one lib
lipo -create .libs/libpcre_simulator.a .libs/libpcre_device.a -output  .libs/libpcre.a

cp .libs/libpcre.a YOU_PROJECT_DIR
cp pcre.h YOU_PROJECT_DIR

Then all you have to do is add libpcre.a and pcre.h to your project and you'll be set for both simulator and device builds.

Amazon EC2 + Chef = Mmmmm

2009-03-29T11:34:39-07:00

UPDATE

This is a little out of date now, but it still contains some helpful information on getting EC2 setup

Amazon's Elastic Computing Cloud (EC2) has always intrigued me. But now that you can get an EC2 instance for around $50 a month it's an amazing deal I couldn't pass up!

One of the perceived downsides of EC2 is that the instances aren't persistent. That's where Chef comes in! Chef is like a little system admin robot... you tell it how you want your system configured and it will do all the dirty work on the server.

Below I describe how to get started with EC2, use Chef to prepare the server for rails apps, and then use Capistrano to start up a sample rails app! It's really pretty easy!

Getting started with EC2

Read the Getting Started Guide
- Setup an account
- Setup the tools
Generate an SSH Key Pair
- $ ec2-add-keypair ec2-keypair > ~/.ssh/ec2-keypair # I call it ec2-keypair, you can call it whatever you want
- $ chmod 600 ~/.ssh/ec2-keypair # Keep it secret – keep it safe
Open up some ports on the default group
- $ ec2-authorize default -p 22 # open up ssh port
- $ ec2-authorize default -p 80 # open up web port
Get AMI instance number from alestic.com... I'm using ami-71fd1a18 (Ubuntu 8.04 LTS Hardy)
$ ec2-run-instances ami-71fd1a18 -k ec2-keypair --availability-zone 'us-east-1a'
$ ec2-describe-instances # Copy the public-dns, you'll need it later, it looks like ec2-XXX-XXX-XXX-XXX.compute-1.amazonaws.com
Create and attach a "db" Elastic Block Storage (EBS) volume. Read more about EBS Volumes here
- $ ec2-create-volume --availability-zone 'us-east-1a' --size '5' # Grab the volume id
- $ c2-attach-volume VOLUME_ID_GOES_HERE --device 'sdq2' --instance INSTANCE_ID_GOES_HERE # The instance has to be running for this to work
Create and attach a "data" Elastic Block Storage (EBS) volume
- $ ec2-create-volume --availability-zone 'us-east-1a' --size '5' # Grab the volume id
- $ c2-attach-volume VOLUME_ID_GOES_HERE --device 'sdq1' --instance INSTANCE_ID_GOES_HERE

You have an EC2 instance!

$ ssh -i ~/.ssh/ec2-keypair root@PUBLIC-DNS-OF-YOUR-EC2-INSTANCE # Don't get too excited yet, we need chef to make this tasty
$ logout
$ ssh-add ~/.ssh/ec2-keypair # Phew, Now you don't have to use the -i option again (until you restart you computer)

Setup chef

save this file on your local machine somewhere like /tmp/chef-bootstrap

#!/usr/bin/env ruby

def cmd(cmd)
  puts cmd; system(cmd)
end


puts "--------------"
puts "Chef Chef Chef"
puts

cmd "aptitude -y update"
cmd "aptitude -y install irb ri rdoc libshadow-ruby1.8 ruby1.8-dev gcc g++ curl"

cmd "curl -L 'http://rubyforge.org/frs/download.php/45905/rubygems-1.3.1.tgz' | tar xvzf -"
cmd "cd rubygems* && ruby setup.rb --no-ri --no-rdoc"
 
cmd "ln -sfv /usr/bin/gem1.8 /usr/bin/gem"

cmd "gem install rdoc chef ohai --no-ri --no-rdoc --source http://gems.opscode.com --source http://gems.rubyforge.org"

cmd "yes | mkfs -t ext3 /dev/sdq1"
cmd "yes | mkfs -t ext3 /dev/sdq2"

puts
puts "It seems to have worked!"
puts "------------------------"

$ scp /tmp/chef-bootstrap root@PUBLIC-DNS-OF-YOUR-EC2-INSTANCE:/tmp # Send this up to your EC2 Server
$ ssh "ruby /tmp/chef-bootstrap" # run it!
Chef and all it's dependencies are all setup!

Get cooking with chef

Grab my chef-sample cookbook/config from http://github.com/probablycorey/chef-sample
edit the file dna.rb (using the dna.sample.rb as a reference) to add apps, users, cronjobs, gems, packages, blah blah blah. Make sure you edit the authorized_keys parameter for each user so you can ssh in as them.
rake cook server=root@PUBLIC-DNS-OF-YOUR-EC2-INSTANCE

Put up a sample rails app

Grab my rails-sample app from http://github.com/probablycorey/rails-sample
edit RAILS_ROOT/config/deploy.rb to make it work with your dna.rb file
$ cap deploy:setup
$ cap deploy:start
$ cap deploy:migrations
Visit http://PUBLIC-DNS-OF-YOUR-EC2-INSTANCE You should see amazing things.

Helpful Links

Objective-c key paths

2009-02-13T15:02:35-08:00

Key paths are an insanely helpful concept in objective-c. What are they? Basically they are like key-value coding but nested... plus a touch of magic. With key-value coding you can get at the variables and methods of an object.

NSString *name = @"Corey";
NSLog(@"Output: %@", [name valueForKey:@"length"]);

// Output: 5

That example was pretty boring since we can get at the length property in simpler ways, but it's a simple introduction to key-value coding.

Hello Key Paths

What if you have an array of strings and you want to get an array of their lengths (this example is contrived, but just stick with me.) You could loop through the strings and store the lengths in an array or you could use key paths!

NSArray *animals = [NSArray arrayWithObjects:@"pig", 
  @"dog", 
  @"human", 
  @"bear", 
  @"frog", 
  nil];
  
NSLog(@"Output: %@", [animals valueForKeyPath:@"length"]);  

// Output: (3, 3, 5, 4, 4)

The same technique applies to dictionaries as well. It's super helpful if you have nested dictionaries. For example assume you have an NSDictionary called animals where the animal name is the key, and the value is an NSDictionary of that animals traits (i.e. legs, teeth, eyes, nipples). Then...

NSLog(@"Legs: %@", [animals valueForKeyPath:@"pig.legs"]

// Output: 4

But that's not all! key paths have magic keywords you can throw in like sum, distinctUnionOfObjects and avg.

NSArray *words = [NSArray arrayWithObjects:@"bob", 
  @"dog", 
  @"human", 
  @"bear", 
  @"frog", 
  nil];
  
NSLog(@"Sum: %@", [words valueForKeyPath:@"@sum.length"]); // Add up all the string lengths
NSLog(@"Avg: %@", [words valueForKeyPath:@"@avg.length"]); // Average length of all strings

// Sum: 19
// Avg: 3.8

Take a look at the Apple Docs for information about all the operators.

Real World Example

How would you use this in the real world? What if you had a shockingly descriptive object model of your friends, and you wanted to find everyone who had a dad named "Bob". No problem with key paths! Just throw in this NSArray category and you are set!

@implementation NSArray (Find)

- (NSArray *)findAllWhereKeyPath:(NSString *)keyPath equals:(id)value {
  NSMutableArray *matches = [NSMutableArray array];
    for (id object in self) {
      id objectValue = [object valueForKeyPath:keyPath];
      if ([objectValue isEqual:value] || objectValue == value) [matches addObject:object];         
    }
   
    return matches;
}

@end

// Implementation example
NSArray *friendsWithDadsNamedBob = [friends findAllWhereKeyPath:@"father.name" equals:@"Bob"]

POW!

2008-12-26T12:01:30-08:00

Ever get frustrated dealing with files or directories in ruby? Maybe it's just me, but using File, Dir, FileUtils, Pathname, etc makes me queasy. So I wrote POW! to make life easier. Git it here http://github.com/probablycorey/pow or sudo gem install pow Since path manipulation is kind of boring, I'll just give a few examples of how it's helped me. Consider this directory structure...

        /tmp
        /sub_dir
          /deeper_dir
        /extra_dir
        file.txt
        program
        README

To open a directory

path = Pow("tmp")

Open a file

Pow(:tmp, :sub_dir, :file.txt) do |file|
  # All the stuff you usually do with an open file!
  file.read
end

Check out what's inside a directory

path = Pow("tmp")
path.each {|child| puts "#{child} - #{child.class.name}!" }

# Output!
# -------
/tmp/README - Pow::File
/tmp/subdir - Pow::Directory
/tmp/sub_dir/deeper_dir - Pow::Directory
/tmp/sub_dir/file.txt - Pow::File
/tmp/sub_dir/program - Pow::File
/tmp/extra_dir - Pow::Directory

Access a file in that directory

# The / operator accepcts symbols, strings or other Pow objects
file = path / :subdir / "file.txt"

# You can also use brackets to access paths
file = path[:subdir, "file.txt"]

Create nested directories

# Great for code generation

Pow("MOAR").create do
Pow("sub_dir").create do
Pow("info.txt").create do |file|
  file.puts "Here is the info you desired!"
end

Pow("README").create_file do |file|
  file.puts "I'm so glad you read this."
end

# Creates directory structure
/MOAR
  /sub_dir
    info.txt
  README

Sort a bunch of files by the modified date

Pow("images").files.sort_by {|file| file.modified_at}

Abusing Ruby's question mark methods.

2008-11-28T06:25:24-08:00

I've always enjoyed the ruby convention of using question marks to denote boolean methods. empty?, exist?, any?, alive?, etc... They're concise and easy to read. I do have one gripe with it, sometimes testing the inverse makes my code read like yoda wrote it.

not sting.empty?
not path.file?
not param.blank?

In rails I use not something.blank? a lot, so creating a not_blank? method wrapper around blank? was a nice simple fix. But I'm lazy and easily give into the temptation of the dark side, so I created a Bizarro Object. Which lets me do this...

# Meaningless examples that prove my point

"".not.empty?       #=> false
"words".not.empty?  #=> true

0.not.zero?         #=> false
1.not.zero?         #=> true

And here is the bit of code that makes it work...

class BizarroObject
  # Since this is a proxy, get rid of every default method
  # Copied from Jim Weirich's BlankSlate class
  # http://onestepback.org/index.cgi/Tech/Ruby/BlankSlate.rdoc
  instance_methods.each { |m| undef_method m unless m =~ /^__/ }

  # Takes an object as a parameter and will invert the return values of all it's methods.
  def initialize(object)
    @object = object
  end

  def method_missing(symbol, *args)
    !@object.send(symbol, *args)
  end
end

class Object
  # This is where the proxy/bizarro object is created
  def not
    BizarroObject.new(self)
  end
end

It's not optimized, and it's a little ridiculous, but I love that ruby lets me abuse it like this.

Git hooks make me giddy

2008-11-07T08:16:29-08:00

I have a rails project that is too wily for autotest, but I still want a simple way to make sure my tests are passing. Using git hooks is a simple solution to this problem. Git Hooks? Hooks (more info here) are little scripts that git triggers after certain commands. They are located in your projects GIT_DIR/hooks directory and are disabled by default. To enable them just chmod +x the hook you want to run. How is this helpful for testing? You can run your specs from the GIT_DIR/hooks/pre-commit hook! The script is run every time you call git commit. If the script exits with a non-zero status your changes won't be committed and you can scurry to fix your specs. Here is a little example of how I'm using in one of my apps

#!/usr/bin/env ruby

html_path = "spec_results.html"

`spec -f h:#{html_path} -f p spec` # run the spec. send progress to screen. save html results to html_path

# find out how many errors were found
html = open(html_path).read
examples = html.match(/(\d+) examples/)[0].to_i rescue 0
failures = html.match(/(\d+) failures/)[0].to_i rescue 0
pending = html.match(/(\d+) pending/)[0].to_i rescue 0

if failures.zero?
  puts "0 failures! #{examples} run, #{pending} pending"
else
  puts "\aDID NOT COMMIT YOUR FILES!"
  puts "View spec results at #{File.expand_path(html_path)}"
  puts
  puts "#{failures} failures! #{examples} run, #{pending} pending"
  exit 1
end

Ruby Equality! equal? eql? == and ===

2008-10-26T18:37:11-07:00

The equality methods in ruby confuse me about once a month. So it's time to permanently embed this information into my brain.

The first confusing part is remembering which object is doing the comparing.

1 == 2
# It makes more sense if you think of it like this
1.==(2)

1 calls it's method == with the argument 2. I like to think of it as 1 is put in charge of checking if it equals 2. 2 is just along for the ride.The second confusing part is there are several methods that deal with equality but in a slightly different ways.

equal?

Returns true if the object_id's are equal. Rarely used, and should not be overridden. You can basically forget out it.

a = b = "bob"
a.equal? b # true

a.equal? "bob" # false since the object_id for the literal "bob" is different than will be different

==

The most common equality operator. It is generally used to test the value of objects instead of the object_id's.

string = "bob"
array = [1,2,3]

array == [1,2,3] # true
string == "bob" # true
1 == 1.0 # true

eql?

Just like == but more strict (i.e. returns false for objects are different types.) As far as I can tell, this is only used to compare hash keys. Like the equal? method, you can basically forget about this one.

string = "bob"
string.eql? "bob" # true
1.eql? 1.0 # false since 1 is a Fixnum and 1.0 is a Float

===

Used for case statements. It is full of magic and sprinkled with mystery. This case statement...

case 1
when Numeric then "Number"
when String then "String"
end

# It is equivalent to...
if Numeric === 1 then "Number"
elsif String === 1 then "String"
end

This seems kind of unintuitive since the arguments for === seem to be in the reversed in the case statement. But this technique has more power than greyskull. Check out Regex, Range, and Module to see clever === uses. Also remember a === b may be true, but it's converse b === a could be false!

Ruby, Rails and Google Sitemaps

2008-10-20T09:46:45-07:00

Setting up a google sitemap is an easy way to force google to notice your site. A sitemap is just a simple xml file that lists every url you want google to know about. They are especially useful if...

You have dynamic content.
Your site is new and google is unaware of it.
You use a lot of AJAX or Flash.

You also get the added benefit of seeing where the googlebot looked last, where it encountered errors, and your sites top search keywords.

So it's helpful, but is it easy to setup? If you're using Ruby on Rails (or any other ruby based framework) it's cake!

Step 1: Created a script (RAILS_ROOT/scripts/sitemap)

This script will collect all relevant urls and create a file at RAILS_ROOT/public/sitemap.xml that contains info about each url. For example, let's pretend we have a site devoted to hippo pictures, our script would look like this...

#!/usr/bin/env ruby

ENV['RAILS_ENV'] ||= "production"

Dir.chdir(File.expand_path(File.dirname(__FILE__) + "/..")) # Change current directory to RAILS_ROOT
require "config/environment" # Start up rails

# These two lines make life super easy... It allows you to call url_for/link_to outside of a controller or view
include ActionController::UrlWriter
default_url_options[:host] = 'www.hippos-are-awesome.com'

filename = "#{RAILS_ROOT}/public/sitemap.xml"

hippo_pics = HippoPic.find(:all) # Such a wonderful collection

File.open(filename, "w") do |file|
  xml = Builder::XmlMarkup.new(:target => file, :indent => 2)

  # This
  xml.instruct!
  xml.urlset "xmlns" => "http://www.sitemaps.org/schemas/sitemap/0.9" do
    for hippo_pic in hippo_pics
      xml.url do
        xml.loc url_for(:controller => "hippos", :id => hippo_pic.id)
        xml.lastmod hippo_pic.updated_at.xmlschema
        xml.changefreq "weekly"
        xml.priority 0.5
      end
    end
  end
end

For more info about what the lastmod, changefreq and priority mean in the sitemap, google explains it all here. Basically they tell google which urls are more important.

Step 2: Create a daily or weekly cronjob to run the sitemap script

Just switch to the user that runs your ruby apps and add this to its crontab.

20 2 * PATH_TO_RAILS_APP/script/sitemap # Runs the sitemap script every morning

Step 3: Let google know about your sitemap

Head over to google's webmaster tools and follow the instructions on how to point google to your sitemap

That's it. Some other additional things to consider are

gzip your sitemap. Google can read them just fine and you save on bandwidth.
If you have more than 50,000 links you need to split your sitemap into several files.
Other search engines (like yahoo) can take google style sitemaps too.