Cornucopia

This is a gem with a bunch of tools which I think are useful when testing. There are many parts to the gem and not all parts are useful in all projects. The gem is designed to work so that you can only use the parts that you need. If the gem ever grows large enough or complex enough, I might break it into pieces similar to rspec. For now, this is just the way it is.

Installation

Add this line to your application's Gemfile:

gem 'cornucopia'

And then execute:

$ bundle install

Or install it yourself as:

$ gem install cornucopia

Usage

Hooks

The primary usefulness of the system is the built in integration system. Using your favorite supported test system include the appropriate files.

RSPEC:

spec_helper.rb or rails_helper.rb depending on the version of RSpec you are using, or where you want to put it:

require "cornucopia/rspec_hooks"

Cucumber:

env.rb:

require "cornucopia/cucumber_hooks"

Spinach:

env.rb:

require "cornucopia/spinach_hooks"

Once required in the hooks are installed, the system will automatically generate a report detailing the error. The automatic report includes information about the failure including:

  • The exception
  • The callstack
  • Any instance variables for the test
  • Details about the failed test and/or step
  • Any Capybara windows and their details including a screen shot

Capybara

I have added the following functions to Capybara to help simplify using it.

  • synchronize_test(seconds=Capybara.default_wait_time, options = {}, &block)

This function yields to a block until the Capybara timeout occurs or the block returns true. I added it because I find that simple synchronized find functions are not always sufficient. This will allow the system to wait till any random block returns true, synchronizing to a wide variety of conditions.

# Wait until either #some_element or #another_element exists.
Capybara::current_session.synchronize_test do
  page.all("#some_element").length > 0 ||
      page.all("#another_element").length > 0
end
  • select_value

This function selects the option from a select box based on the value for the selected option instead of the value string.

If I have the a SELECT list with the following option:

<option value="AZ">Arizona</option>

The following line will select the "Arizona" option:

my_address.state_abbreviation = "AZ"
page.find("#state_selector").select_value(my_address.state_abbreviation)
  • value_text

This function returns the string value of the currently selected option(s).

page.find("#state_selector").value_text  # returns "Arizona" instead of "AZ" if "AZ" is selected.

SitePrism

I love using SitePrism. Unfortunately, I find it involves a lot of copy/paste. To simplify the use of SitePrism, I have created what are basically some macros:

  • patterned_elements(pattern, *element, options = {})

This allows you to define a list of multiple elements where the pattern for the finder includes the text for what you want the call the element.

You can specify the type of finder to be used with the finder_type parameter, and if necessary include additional parameters as well.

Examples:

class MySection < SitePrism::Section
  patterned_elements "td.column_%{element_name}",
                     :my_element_1,
                     :my_element_2,
                     :my_element_3

  # instead of:
  # element :my_element_1, "td.column_my_element_1"
  # element :my_element_2, "td.column_my_element_2"
  # element :my_element_3, "td.column_my_element_3"

  patterned_elements "//td[name = \"${element_name}\"]",
                     :my_element_1,
                     :my_element_2,
                     :my_element_3,
                     finder_type: :xpath,
                     visible: false

  # instead of:
  # element :my_element_1, :xpath, "//td[name = \"my_element_1\"]", visible: false
  # element :my_element_2, :xpath, "//td[name = \"my_element_2\"]", visible: false
  # element :my_element_3, :xpath, "//td[name = \"my_element_3\"]", visible: false
end
  • form_elements(form_type, *elements)

This provides a quick and easy way to define elements for the items in forms. The ids of the elements in a form follow the simple pattern of: <form_name><element_id>. Most of the time, you want the name of the element to match the <element_id>.

Example:

= form_for my_table_object do |form|
  = form.select :field_1
  = form.text_box :field_2
  = form.check_box :field_3

class MySection < SitePrism::Section
  form_elements :my_table,
                :field_1,
                :field_2,
                :field_3,

  # instead of:
  # element :field_1, "#my_table_field_1"
  # element :field_2, "#my_table_field_2"
  # element :field_3, "#my_table_field_3"
end
  • id_elements(*elements)

This is a quick and easy way to define elements where the name of the element is the same as the id for the element.

Example:

class MySection < SitePrism::Section
  id_elements :my_item_1,
              :my_item_2,
              :my_item_3

  # instead of:
  # element :my_item_1, "#my_item_1"
  # element :my_item_2, "#my_item_2"
  # element :my_item_3, "#my_item_3"
end
  • class_elements(*elements)

This is a quick and easy way to define elements where the name of the element is the same as the class name used to identify the element.

Example:

class MySection < SitePrism::Section
  id_elements :my_item_1,
              :my_item_2,
              :my_item_3,
              "my-class-name"

  # instead of:
  # element :my_item_1, ".my_item_1"
  # element :my_item_2, ".my_item_2"
  # element :my_item_3, ".my_item_3"
  # element :my_class_name, ".my-class-name"
end
  • indexed_elements(pattern, *element, options = {})

This allows you to define a list of multiple elements where the pattern for the finder is based on an index value. This is useful for things like table columns.

You can specify the type of finder to be used with the finder_type parameter, and if necessary include additional parameters as well with additional_options.

Examples:

class MySection < SitePrism::Section
  patterned_elements "td:nth-child(%{element_index})",
                     :my_element_1,
                     :my_element_2,
                     :my_element_3

  # instead of:
  # element :my_element_1, "td:nth-child(1)"
  # element :my_element_2, "td:nth-child(2)"
  # element :my_element_3, "td:nth-child(3)"

  patterned_elements "//td_%{element_index}",
                     :my_element_1,
                     :my_element_2,
                     :my_element_3,
                     finder_type: :xpath,
                     additional_options: { visible: false },
                     start_index: 3,
                     increment: 12

  # instead of:
  # element :my_element_1, :xpath, "//td_3", visible: false
  # element :my_element_2, :xpath, "//td_15", visible: false
  # element :my_element_3, :xpath, "//td_27", visible: false
end
  • PageApplication

SitePrism recommends memoizing the page objects. I found it annoying to create a class for the pages and define a new function for each page. To simplify my life, I created the PageApplication class. It will memoize pages for me automatically as long as I follow some simple rules.

Basically, just create all of my pages underneath a single module in a single folder, and the a class derived from PageApplication will find the pages and memoize them for you automatically.

class MyPageApplication < Cornucopia::SitePrism::PageApplication
  def pages_module
    MyPagesModule
  end
end

module MyPagesModule
  module MyModule
    class MyPage < SitePrism::Page
      # your SitePirsm page definition here...
    end
  end
end

module MyPagesModule
  class MyOtherPage < SitePrism::Page
    # your SitePirsm page definition here...
  end
end

a_memoized_page       = MyPageApplication.my_module__my_page
a_memoized_other_page = MyPageApplication.my_other_page

Capybara::Node::Simple integration

SitePrism is very useful, but it only works with the main page: Capybara::current_session.page. I have found that there are times where it is very useful to use the Capybara::Node::Simple object. The problem I have is that I can't use my SitePrism pages with the SimpleNode.

I therefore added a property owner_node to the SitePrism::Page and SitePrism::Section classes. You can assign to this property the Capybara::Node that you want to execute the finder functions against.

An example might be something like:

RSpec.describe MyController, type: :controller do
  get :index

  expect(response.status).to eq 200

  my_node = Capybara::Node::Simple.new(response.body)
  my_page = MyPageApplication.my_page

  my_page.owner_node = my_node

  expect(my_page.my_section.my_list[0].my_element.text).to eq(my_expectation)
end

Utilities

Configuration

Cornucopia::Util::Configuration

The configuration class contains the various configurations that are used by the system.

  • seed

    The seed value represents the seed value for rand. It is used by the testing hooks to allow tests with randomized values to still be repeatable. This value can be set or read.

  • context_seed

    The context_seed value represents the seed value for rand if called outside of a test within a group. It is used by the testing hooks to allow tests with randomized values to still be repeatable. This value can be set or read.

  • suite_seed

    The seed value represents the seed value for rand. It is used by the testing hooks to allow test runs to be re-run with the same seed for the entire run. It should make randomized values including the context and seed values to still be repeatable. This value can be set or read.

  • order_seed

    Experimental The order_seed value represents the seed value for the order that RSpec tests are run in if they are run randomly. This value in the configurations actually doesn't always work as it is very dependent on when it is set during the RSpec configuration process. The system tries, but it may be best just to do what this does and set RSpec.configuration.seed = <your value> youself in spec_helper.rb. Really I've just started playing around with it and seeing what I can do with it.

  • grab_logs

    Indicates if the Cornucopia::Util::LogCapture class will capture any log files or not.

  • backup_logs_on_failure

    Indicates if the Cornucopia::Util::LogCapture class will backup the full log file at the end of the test suite if there is an error.

  • user_log_files

    Returns a list of the log files which will be captures. Changing the returned value will not affect log capturing settings.

  • num_lines(log_file_name=nil)

    This gets the number of lines that are captured for the indicated log file. If no value is passed in for log_file_name, the default number of lines will be returned.

  • default_num_lines=

    Set the default number of lines to be captured when capturing log files.

  • add_log_file(log_file_name, options = {})

    Add a log file to be captured. The log path needs to be relative to the Rails log folder, or if Rails is not being used in the project, relative to the current working directory. If a file is added multiple times, the file will be captured only once. Subsequent calls will update the options for the file.

    Options:

    num_lines - The number of lines to capture for the file.

  • remove_log_file

    Removes a log file from the list of files to be captured.

  • report_configuration(report_name)

    Returns the Cornucopia::Util::ConfiguredReport for different reports.

    Supported Reports:

    rspec

    cucumber

    spinach

    capybara_page_diagnostics

  • print_timeout_min

    This is the amount of time (in seconds) the will be allowed for the rendering of a variable when printing a value using Cornucopia::Util::ReportBuilder.pretty_object.

    This value exists because in real testing I found a few vaules (which I now exclude automatically) which took literally hours to print out using pretty_inspect. (I suspect an infinite loop.) This value prevents that by interrupting a printout which takes too long.

  • auto_open_report_after_generation

    This allows you to tell the system to open a report with errors in it after the report is closed/finished. If you do not specify a type of report the value you specify will be the default value used for any report that is generated.

  • record_test_start_and_end_in_log

    When true, this causes Cornucopia to output a line in the log file indicating the start and end of a test.

    The default value is true.

  • record_test_start_and_end_format

    This parameter specifies the format of the message that is recorded in the log file. You can use the following variables in the format string to get the appropriate values:

    • %{start_end} - This will be the words "Start" or "End"
    • %{test_name} - This will be the name of the test.

    The default value is "******** %start_end: %test_name".

  • And more...

    There are more configurations. They are commented in the configuraiton.rb file. If I didn't include them here, then I probably thought that they were well commented in the file, or that they weren't important enough to put here, or that they shouldn't be used normally, or (hopefully not) I just forgot to update the readme.

    If I forgot one and you think it should be here, let me know.

ConfiguredReport

The Cornucopia::Util::ConfiguredReport class allows you to configure what information is exported to generated report files.

I've tried to create reasonable default configuration for what gets reported when there is an error in the test environments that I know enough to support: Cucumber, RSpec, and Spinach. I also provided a default configuration for what to report when an error occurs a Capybara page is open.

You can override these defaults if you find a need and specify the configured report that is used when an exception occurs in a particular environment.

The default configurations will output the following information that it can find/is available:

  • The exception that caused the problem and its call stack
  • Test details such as the test name, file path, etc.
  • Any instance variables for the test
  • Any values defined using let
  • The log for the current environment
  • Any additional log files specified by the user
  • Any Capybara details that can be determined like the HTML source, a screen shot, etc.

The ConfiguredReport class and the example configurations detail how the configured reports work if you feel the need to create your own.

ReportBuilder

The ReportBuilder is the tool which is used to create the reports that are generated when an exception is caught.
There are 3 basic objects to work with in a report:

Tests

A test is basically a new sub-report. When you create a new test, you will pass in a name for the test. This name will appear in the left portion of the report. When the user clicks on the test, the report for the test will appear on the right side. You will likely not need to create your own tests. To do so, you call within_test.

An example:

Cornucopia::ReportBuilder.current_report.within_test("This is the name of my test") do
  # build your test report here.
end
Sections

A section is a block within a test. The section has a header that describes the section and is the primary container for tables. Multiple sections are allowed within a test and are colored alternating colors to distinguish them.

An example:

Cornucopia::ReportBuilder.current_report.within_section("Section header") do |section|
  # Build the details for the section here.

  # NOTE:  Currently section is == Cornucopia::ReportBuilder.current_report  This may or may not be so in the 
  #        future.
end

Note that only one test is allowed to be active at all times and that if no other test is active a defaut "unknown" test will be used. As a result, you can call the within_section function directly from the report object and it will be within the currently active test.

Tables

A table is exactly what it sounds like it is a table of information. Tables have rows of information pairs - a label and the information to be shown. Unlike Sections and Tests, Tables can be nested inside each other. That is the information in a table row can be another table.

If the information for a particular cell is too large, that information will be partially hidden from view so that the table size doesn't get out of hand.

To write out a value, you simply use write_stats and pass in a label and a value.

An example:

Cornucopia::ReportBuilder.current_report.within_section("Section header") do |section|
  section.within_table do |table|
    table.write_stats "Statistic name", "Statistic value"

    ReportTable.new nested_table: table, nested_table_label: "Sub Table" do |sub_table|
      sub_table.write_stats "Sub statistic name", "Sub statistic value"
    end
  end
end

Tables have a lot of options:

  • table_prefix - This is the value to use to "open" the table.
  • table_postfix - This is the value to use to "close" the table.
  • report_table - If set, all table calls are passed through to this table object. The purpose of this value is to allow for an optional sub-table. That is the code acts as if it is working on a sub-table, but in reality it is working in the report_table.
  • nested_table - This is the table that the table being created will be a sub-table of. When the new table is completed, it till be output into a row of the specified table.
  • nested_table_label - This is the lable that will be used for this table when it is output in the nested_table.
  • nested_table_options - A hash of options that will be used to determine the look and feel of the nested table when it is output. These options are passed into write_stats when the table is output.
  • not_a_table - If set, then when write_stats is called, the label value is ignored and the value is simply appended to the table as-is.
  • suppress_blank_table - If set, the table will not be output if it is blank.

Contributing

  1. Fork it
  2. Create your feature branch (git checkout -b my-new-feature)
  3. Commit your changes (git commit -am 'Add some feature')
  4. Push to the branch (git push origin my-new-feature)
  5. Create new Pull Request

Todos...

ReportBuilder - reformat and styling of report?