why does my rails generator template method results in 'file_clash' - ruby-on-rails

So I'm writing a rails generator to do the most simple of things: copy some model files from a gem's lib/generators/pathways/templates directory into a project in the app/models directory. My possibly mistaken understanding is that the template method is basically a file copy with source/target.
(note that "pathways" is the name of my gem that installs this generator)
Here's the guts of the copy code in the generator:
def copy_models
project_models_location = "#{Rails.root}/app/models/"
[
"pathways_experiment.rb",
...
].each do |filename|
puts "copying #{filename} to #{project_models_location}"
template filename, "#{project_models_location}"
end
end
The puts displays what I expected:
copying pathways_experiment.rb to
/Users/meuser/Projects/testing_gem/exp_gem_test/app/models/
however, the call to the templates method dumps this output:
file_clash app/models
I checked the target directory and there are no files in it, so it doesn't seem to be because the code is trying to overwrite the file.
here's the source of the entire generator in case I simply haven't included or extended the correct classes/modules:
require 'rails/generators'
require 'rails/generators/active_record'
module Pathways
class InstallGenerator < ActiveRecord::Generators::Base
include Rails::Generators::Migration
source_root File.expand_path("../templates", __FILE__)
def code_that_runs
puts "PATHWAYS: installing models"
copy_models
end
private
def copy_models
project_models_location = "#{Rails.root}/app/models/"
[
"pathways_experiment.rb",
...
].each do |filename|
puts "copying #{filename} to #{project_models_location}"
template filename, "#{project_models_location}"
end
end
end
end

Related

undefined local variable or method for method located in lib directory file

I have some code i've inherited and am in the process of upgrading it to Rails 3.1. I'm suuuuper close to done but I got a bug.
In Rails Console I run User.first and I get this error
undefined local variable or method `acts_as_userstamp' for #<Class:0x000000046bef50>
Now acts_as_userstamp is a method located on line two inside my User model
class User < ActiveRecord::Base
#TODO /lib is not loading??? or is it??? why this method not work in browser?
acts_as_userstamp
And is defined in a file called app/lib/model_modifications.rb.
Now I recently discovered that my app/lib folder was not being autoloaded in my application.rb file and I think that's been fixed...or has it? Is this file correct? Or no?
require File.expand_path('../boot', __FILE__)
require 'rails/all'
# evil outdated soap middleware, TODO: kill it with fire
# Does this have to be loaded BEFORE the first line???
$LOAD_PATH.unshift(File.join(File.dirname(__FILE__), '..', "vendor", "soap4r"))
$LOAD_PATH.unshift(File.join(File.dirname(__FILE__), '..', "vendor", "plugins", "soap4r-middleware", "lib"))
# evil outdated soap middleware, TODO: kill it with fire
require 'soap4r-middleware'
require File.join(File.dirname(__FILE__), '..', 'app', 'lib', 'soap.rb')
if defined?(Bundler)
# If you precompile assets before deploying to production, use this line
Bundler.require *Rails.groups(:assets => %w(development test))
# If you want your assets lazily compiled in production, use this line
# Bundler.require(:default, :assets, Rails.env)
end
module MyappDev
class Application < Rails::Application
# startup the lib directory goodies <-- IS THIS CORRECT???
# config.autoload_paths << "#{Rails.root}/lib"
# config.autoload_paths += %W( lib/ )
config.autoload_paths += %W(#{config.root}/lib)
config.autoload_paths += Dir["#{config.root}/lib/**/"]
# Configure the default encoding used in templates for Ruby 1.9.
config.encoding = "utf-8"
# Configure sensitive parameters which will be filtered from the log file.
config.filter_parameters += [:password]
config.middleware.use MyAPIMiddleware
end
end
I'm trying to debug this file as I post this now. Here is a peak at it's internal structure...(i've just included the overall structure for the sake of brevity)
app/lib/model_modificatons.rb
class Bignum
...
end
class Fixnum
...
end
class ProcessorDaemon
...
end
module ActiveRecord
module UserMonitor
...
end
module MyLogger
...
end
end
class Object
...
end
class Struct
...
end
class String
...
end
class Fixnum
...
end
class OpenStruct
...
end
class ActiveRecord::Base
def self.visible_columns
...
end
...
def self.acts_as_userstamp
logger.info "HI fonso - acts_as_userstamp is called"
include ActiveRecord::UserMonitor
end
...
protected
def self.range_math(*ranges)
...
end
end
class Array
...
end
class DB
...
end
If you can spot a problem with the overall structure or anywhere else please let me know.
So why is this method not found? I'm trying to debug it as I'm posting this and I'm getting nothing.
I suspect the file app/lib/model_modifications.rb is not being loading. That nothing in the /lib directory is being loaded..but how do I confirm this?
Thank you for reading this far, I hope I've not rambled on too much.
autoload_path configuration does not load all the given files on the boot but defines folders where rails will be searching for defined constants.
When your application is loaded, most of the constants in your application are not there. Rails have a "clever" way of delaying loading the files by using a constant_missing method on Module. Basically, when Ruby encounters a constant in the code and fails to resolve it, it executes said method. THe sntandard implementation of this method is to raise UndefinedConstant exception, but rails overrides it to search all of its autoload_paths for a file with a name matching the missing constant, require it and then check again if the missing constant is now present.
So, in your code everything works as expected and you need to load this extension file manually. If you want to have some code that executes on the application boot, put your file within config/initializers folder.
Aside: Try avoiding monkey patching whenever possible. It might be looking clever, but adding more methods to already overpopulated classes will not make them easier to use.

Ruby Gem Project -- Thor Generator Causing Read-only File System Error

As a personal project, I decided to write a minified-version of Ruby on Rails and upload it as a gem using Bundle called railz_lite.
Inside of my project, I was hoping to implement a Generator similar to rails new, which would create the necessary folders for a web app i.e. controllers/, views/, models/, etc.
To do so, I included Thor as a dependency, then created the following files:
require 'thor/group'
module RailzLite
module Generators
class Project < Thor::Group
include Thor::Actions
def self.source_root
File.dirname(__FILE__) + "/templates"
end
def add_controllers
empty_directory("controllers")
end
def add_models
empty_directory("models")
end
def add_server
template("server.rb", "config/server.rb")
end
def add_views
empty_directory("views")
end
def add_public
empty_directory("public")
end
end
end
end
Inside the gem project's root folder, when I run bundle exec railz_lite new, the generator works just fine and the necessary files are created.
However, if I create a new project, puts my gem (railz_lite) in the Gemfile, run bundle install, then execute bundle exec rails_lite new, I am greeted with the following error:
.rbenv/versions/2.5.1/lib/ruby/2.5.0/fileutils.rb:232:in `mkdir':
: Read-only file system # dir_s_mkdir - /controllers (Errno::EROFS)
I suspect the error is because the empty_directory command is not referring to the root directory of the project I just created. I am hoping that there is a simple way to fix this.
For further reference, the CLI script and class look as follows:
railz_lite
#!/usr/bin/env ruby
require 'railz_lite/cli'
RailzLite::CLI.start
cli.rb
require 'thor'
require 'railz_lite'
require 'railz_lite/generators/project'
module RailzLite
class CLI < Thor
desc 'new', 'Generates a new RailzLite project'
def new
RailzLite::Generators::Project.start([])
end
end
end
Any solutions would be greatly appreciated!
Note: I am running this on macOS Catalina.
So I found a solution after searching extensively through gem forums and looking at the Rails source code.
Inside of the generation I have to manually set the destination_root to the working directory of the project. The working directory can be found with Dir.pwd
require 'thor/group'
module RailzLite
module Generators
class Project < Thor::Group
include Thor::Actions
def self.source_root
File.dirname(__FILE__) + "/templates"
end
def self.destination_root # this method fixes the problem!
Dir.pwd # get the current project directory
end
def add_controllers
empty_directory("controllers")
end
def add_models
empty_directory("models")
end
def add_server
template("server.rb", "config/server.rb")
end
def add_views
empty_directory("views")
end
def add_public
empty_directory("public")
end
end
end
end

Add Migrations to a Rails Engine's Install Generator

In my rails 5 engine I want to include the installation of the engine's migrations with the custom install generator I have at:
myengine/lib/generators/myengine/install_generator.rb
This generator currently looks like this:
module Myengine
module Generators
class InstallGenerator < Rails::Generators::Base
source_root File.expand_path("../../templates", __FILE__)
desc "Creates a Myengine initializer and copys template files to your application."
def copy_initializer
template "myengine.rb", "config/initializers/myengine.rb"
end
end
end
end
When I add the engine to a rails app, instead of having to call:
rails g myengine:install
then
rails myengine:install:migrations
How can I add the creation of those migrations to the custom generator?
Fun question! And one that I hope I can answer. Let's say you have an engine named "Buttafly" and a generator living at:
#lib/generators/buttafly/install/install_generator.rb
At the top of your generator, require the 'date' library like so:
require 'date'
Then in the body of your generator, add a method definition like:
def copy_migrations
# get an array of the migrations in your engine's db/migrate/
# folder:
migrations = Dir[Buttafly::Engine.root.join("db/migrate/*.rb")]
migrations.each_with_index do |migration, i|
# The migrations will be created with the same timestamp if you
# create them all at once. So if you have more than one migration
# in your engine, add one second to the second migration's file
# timestamp and a third second to the third migration's timestamp
# and so on:
seconds = (DateTime.now.strftime("%S").to_i + i).to_s
seconds = seconds.to_s.length == 2 ? seconds : "0#{seconds}"
timestamp = (DateTime.now.strftime "%Y%m%d%H%M") + seconds
# get the filename from the engine migration minus the timestamp:
name = migration.split("/").split("_").last
# See if a the name of your engine migration is already in your
# host app's db/migrate folder:
if Rails.root.join("db/migrate/*#{name}").exist?
# do nothing:
puts "Migration #{name} has already been copied to your app"
else
# copy your engine migration over to the host app with a new
# timestamp:
copy_file m, "db/migrate/#{timestamp}_buttafly_#{name}"
end
end
end

Ruby on Rails docsplit file path

I have a pdftotext.rb file in /lib and the code is
module Pdftotext
require 'rubygems'
require 'docsplit'
class << self
def convert
Docsplit.extract_text("hello.pdf")
end
end
end
I have the hello.pdf file in the /assets folder and I tried "assets/hello.pdf" but it keeps telling me Error: Couldn't open file '/assets/hello.pdf': No such file or directory.
How can I get the right path to get the file to be converted?
By the way I am using rails 3.2.1, thanks.
Do you mean it is in RAILS_ROOT/assets/hello.pdf?
You should use File.join to get at the file. Like this:
module Pdftotext
require 'rubygems'
require 'docsplit'
class << self
def convert
Docsplit.extract_text(File.join(Rails.root, "assets", "hello.pdf"))
end
end
end
Using "/assets/hello.pdf" will try to get it from the file system root.

What is the best way to get an empty temporary directory in Ruby on Rails?

What is the best way to get a temporary directory with nothing in it using Ruby on Rails? I need the API to be cross-platform compatible. The stdlib tmpdir won't work.
The Dir object has a method mktmpdir which creates a temporary directory:
require 'tmpdir' # Not needed if you are using rails.
Dir.mktmpdir do |dir|
puts "My new temp dir: #{dir}"
end
The temporary directory will be removed after execution of the block.
The Dir#tmpdir function in the Ruby core (not stdlib that you linked to) should be cross-platform.
To use this function you need to require 'tmpdir'.
A general aprox I'm using now:
def in_tmpdir
path = File.expand_path "#{Dir.tmpdir}/#{Time.now.to_i}#{rand(1000)}/"
FileUtils.mkdir_p path
yield path
ensure
FileUtils.rm_rf( path ) if File.exists?( path )
end
So in your code you can:
in_tmpdir do |tmpdir|
puts "My tmp dir: #{tmpdir}"
# work with files in the dir
end
The temporary dir will be removed automatically when your method will finish.
Ruby has Dir#mktmpdir, so just use that.
require 'tempfile'
Dir.mktmpdir('prefix_unique_to_your_program') do |dir|
### your work here ###
end
See http://www.ruby-doc.org/stdlib-1.9.3/libdoc/tmpdir/rdoc/Dir.html
Or build your own using Tempfile tempfile that is process and thread unique, so just use that to build a quick Tempdir.
require 'tempfile'
Tempfile.open('prefix_unique_to_your_program') do |tmp|
tmp_dir = tmp.path + "_dir"
begin
FileUtils.mkdir_p(tmp_dir)
### your work here ###
ensure
FileUtils.rm_rf(tmp_dir)
end
end
See http://www.ruby-doc.org/stdlib-1.9.3/libdoc/tempfile/rdoc/Tempfile.html for optional suffix/prefix options.
require 'tmpdir' # not needed if you are loading Rails
tmp_dir = File.join(Dir::tmpdir, "my_app_#{Time.now.to_i}_#{rand(100)}")
Dir.mkdir(tmp_dir)
Works for me.
You can use Dir.mktmpdir.
Using a block will get rid of the temporary directory when it closes.
Dir.mktmpdir do |dir|
File.open("#{dir}/foo", 'w') { |f| f.write('foo') }
end
Or if you need multiple temp directories to exist at the same time, for example
context 'when there are duplicate tasks' do
it 'raises an DuplicateTask error' do
begin
tmp_dir1 = Dir.mktmpdir('foo')
tmp_dir2 = Dir.mktmpdir('bar')
File.new("#{tmp_dir1}/task_name", 'w+')
File.new("#{tmp_dir2}/task_name", 'w+')
expect { subject.filepath('task_name') }.to raise_error(TaskFinder::DuplicateTask)
ensure
FileUtils.remove_entry tmp_dir1
FileUtils.remove_entry tmp_dir2
end
end
end
Dir.mktmpdir creates a temporary directory under Dir.tmpdir (you'll need to require 'tmpdir' to see what that evaluates to).
If you want to use your own path, Dir.mktmpdir takes an optional second argument tmpdir if non-nil value is given. E.g.
Dir.mktmpdir(nil, "/var/tmp") { |dir| "dir is '/var/tmp/d...'" }
I started to tackle this by hijacking Tempfile, see below.
It should clean itself up as Tempfile does, but doesn't always yet..
It's yet to delete files in the tempdir.
Anyway I share this here, might be useful as a starting point.
require 'tempfile'
class Tempdir < Tempfile
require 'tmpdir'
def initialize(basename, tmpdir = Dir::tmpdir)
super
p = self.path
File.delete(p)
Dir.mkdir(p)
end
def unlink # copied from tempfile.rb
# keep this order for thread safeness
begin
Dir.unlink(#tmpname) if File.exist?(#tmpname)
##cleanlist.delete(#tmpname)
#data = #tmpname = nil
ObjectSpace.undefine_finalizer(self)
rescue Errno::EACCES
# may not be able to unlink on Windows; just ignore
end
end
end
This can be used the same way as Tempfile, eg:
Tempdir.new('foo')
All methods on Tempfile , and in turn, File should work.
Just briefly tested it, so no guarantees.
Update: gem install files, then
require "files"
dir = Files do
file "hello.txt", "stuff"
end
See below for more examples.
Here's another solution, inspired by a few other answers. This one is suitable for inclusion in a test (e.g. rspec or spec_helper.rb). It makes a temporary dir based on the name of the including file, stores it in an instance variable so it persists for the duration of the test (but is not shared between tests), and deletes it on exit (or optionally doesn't, if you want to check its contents after the test run).
def temp_dir options = {:remove => true}
#temp_dir ||= begin
require 'tmpdir'
require 'fileutils'
called_from = File.basename caller.first.split(':').first, ".rb"
path = File.join(Dir::tmpdir, "#{called_from}_#{Time.now.to_i}_#{rand(1000)}")
Dir.mkdir(path)
at_exit {FileUtils.rm_rf(path) if File.exists?(path)} if options[:remove]
File.new path
end
end
(You could also use Dir.mktmpdir (which has been around since Ruby 1.8.7) instead of Dir.mkdir but I find the API of that method confusing, not to mention the naming algorithm.)
Usage example (and another useful test method):
def write name, contents = "contents of #{name}"
path = "#{temp_dir}/#{name}"
File.open(path, "w") do |f|
f.write contents
end
File.new path
end
describe "#write" do
before do
#hello = write "hello.txt"
#goodbye = write "goodbye.txt", "farewell"
end
it "uses temp_dir" do
File.dirname(#hello).should == temp_dir
File.dirname(#goodbye).should == temp_dir
end
it "writes a default value" do
File.read(#hello).should == "contents of hello.txt"
end
it "writes a given value" do
# since write returns a File instance, we can call read on it
#goodbye.read.should == "farewell"
end
end
Update: I've used this code to kickstart a gem I'm calling files which intends to make it super-easy to create directories and files for temporary (e.g. unit test) use. See https://github.com/alexch/files and https://rubygems.org/gems/files . For example:
require "files"
files = Files do # creates a temporary directory inside Dir.tmpdir
file "hello.txt" # creates file "hello.txt" containing "contents of hello.txt"
dir "web" do # creates directory "web"
file "snippet.html", # creates file "web/snippet.html"...
"<h1>Fix this!</h1>" # ...containing "<h1>Fix this!</h1>"
dir "img" do # creates directory "web/img"
file File.new("data/hello.png") # containing a copy of hello.png
file "hi.png", File.new("data/hello.png") # and a copy of hello.png named hi.png
end
end
end # returns a string with the path to the directory
Check out the Ruby STemp library: http://ruby-stemp.rubyforge.org/rdoc/
If you do something like this:
dirname = STemp.mkdtemp("#{Dir.tmpdir}/directory-name-template-XXXXXXXX")
dirname will be a string that points to a directory that's guaranteed not to exist previously. You get to define what you want the directory name to start with. The X's get replaced with random characters.
EDIT: someone mentioned this didn't work for them on 1.9, so YMMV.

Resources