= WiceGrid Version:: 3.3.1 Author:: Yuri Leikind Sources:: https://github.com/leikind/wice_grid/ Examples online:: http://wicegrid.herokuapp.com News:: http://leikind.org/pages/wicegrid/ Email:: "Yuri Leikind" FOR RAILS 3 USE VERSION 3.x FOR RAILS 2 USE VERSION 0.6 (https://github.com/leikind/wice_grid/tree/master). == Intro WiceGrid is a Rails grid plugin. One of the goals of this plugin was to allow the programmer to define the contents of the cell by himself, just like one does when rendering a collection via a simple table (and this is what differentiates WiceGrid from various scaffolding solutions), but automate implementation of filters, ordering, paginations, CSV export, and so on. Ruby blocks provide an elegant means for this. WiceGrid builds the call to the ActiveRecord layer for you and creates a table view with the results of the call including: * paging * sortable columns * filtering by multiple columns * CSV export * saving queries All working nicely together. Filters are added automatically according to the type of the underlying DB column. Filtering by more than one column at the same time is possible. More than one such grid can appear on a page, and manipulations with one grid do not have any impact on the other. WiceGrid does not take a collection as an input, it works directly with ActiveRecord. WiceGrid does not use AJAX calls to reload itself, instead simple GET requests are used for this, nevertheless, all other page parameters are respected and preserved. WiceGrid views do not contain forms so you can include it in your own forms. There are two major branches of WiceGrid. WiceGrid is known to work with MySQL and Postgres. === Examples This tutorial is accompanied by a sample application with WiceGrid examples which you can browse online ( http://wicegrid.herokuapp.com ), or just view the code ( https://github.com/leikind/wice_grid_testbed ). == Requirements Rails version 3.2.x or newer, jQuery, jQuery Datepicker. For Rails 3.0.x and 3.1.x versions use version 3.0.4. For Rails 2 use version 0.6 (https://github.com/leikind/wice_grid/tree/master). WARNING: Since 3.2.pre2 WiceGrid is not compatible with will_paginate because internally it uses kaminari for pagination, and kaminari is not compatible with will_paginate! == How-To === Support for javascript frameworks WiceGrid started as a plugin using the Prototype javascript framework. Support for jQuery was added for version 0.6. Beginning from version 3.2 only jQuery is supported. === Installation Add the following to your Gemfile: gem "wice_grid", '3.2.1' and run the bundle command. Run the generator: rails g wice_grid:install This adds the config file wice_grid_config.rb to config/initializers/, the locale file wice_grid.yml to config/locales/, and the styles file wice_grid.css.scss to app/assets/stylesheets/. Require WiceGrid javascript in your js index file: //= require wice_grid Make sure that jQuery is loaded. If the application uses Date and DateTime filters, you have to install jQuery Datepicker by yourself. Here is an example of application.js with everything WiceGrid needs: //= require jquery //= require jquery_ujs //= require jquery-ui //= require wice_grid //= require jquery.ui.datepicker //= require_tree . WiceGrid provides some very basic styles, not specifying exactly how the table should look like, but if the application uses Twitter Bootstrap, the markup generated by WiceGrid will have correct classes and will fit nicely. Generally it is advised to modify WiceGrid css to match the application style. === Basics The simplest example of a WiceGrid for one simple DB table called ApplicationAccount is the following: Controller: @tasks_grid = initialize_grid(Task) It is also possible to use an ActiveRecord::Relation instance as the first argument: @tasks_grid = initialize_grid(Task.where(:active => true)) View: <%= grid(@tasks_grid) do |g| g.column do |task| task.id end g.column do |task| task.title end g.column do |task| task.description end g.column do |task| task.archived? ? 'Yes' : 'No' end g.column do |task| link_to('Edit', edit_task_path(task)) end end -%> Code g.column do |task| ... end defines everything related to a column in the resulting view table including column names, sorting, filtering, the content of the column cells, etc. The return value of the block is the table cell content. Column names are defined with parameter :name: <%= grid(@tasks_grid) do |g| g.column :name => 'ID' do |task| task.id end g.column :name => 'Title' do |task| task.title end g.column :name => 'Description' do |task| task.description end g.column :name => 'Archived' do |task| task.archived? ? 'Yes' : 'No' end g.column do |task| link_to('Edit', edit_task_path(task)) end end -%> To add filtering and ordering, declare to which column in the underlying database table(s) the view column corresponds using parameter :attribute : <%= grid(@tasks_grid) do |g| g.column :name => 'ID', :attribute => 'id' do |task| task.id end g.column :name => 'Title', :attribute => 'title' do |task| task.title end g.column :name => 'Description', :attribute => 'description' do |task| task.description end g.column :name => 'Archived', :attribute => 'archived' do |task| task.archived? ? 'Yes' : 'No' end g.column do |task| link_to('Edit', edit_task_path(task)) end end -%> This will add sorting links and filters for columns +Username+ and +Active+. The plugin automatically creates filters according to the type of the database column. In the above example a text field will be created for column Title (title is a string), for column +Archived+ a dropdown filter will be created with options 'Yes', 'No', and '--', and for the integer ID two short text fields are added which can contain the numeric range (more than, less than). It is important to remember that :attribute is the name of the database column, not a model attribute. Of course, all database columns have corresponding model attributes, but not all model attributes map to columns in the same table with the same name. Read more about available filters in the documentation for the column method. Read the section about custom dropdown filters for more advanced filters. For columns like g.column :name => 'Title', :attribute => 'title' do |task| task.title end where the block contains just a call to the same attribute declared by :attribute, the block can be omitted: <%= grid(@tasks_grid) do |g| g.column :name => 'ID', :attribute => 'id' g.column :name => 'Title', :attribute => 'title' g.column :name => 'Description', :attribute => 'description' g.column :name => 'Archived', :attribute => 'archived' do |task| task.archived? ? 'Yes' : 'No' end g.column do |task| link_to('Edit', edit_task_path(task)) end end -%> In this case +name+ will be used as the method name to send to the ActiveRecord instance. If only ordering is needed, and no filter, we can turn off filters using :filter : g.column :name => 'ID', :attribute => 'id', :filter => false If no ordering links are needed, use :ordering => false: g.column :name => 'Added', :attribute => 'created_at', :ordering => false It is important to understand that it is up to the developer to make sure that the value returned by a column block (the content of a cell) corresponds to the underlying database column specified by :attribute (and :model discussed below). === Rendering filter panel The filter panel can be shown and hidden clicking the icon with binoculars. The way the filter panel is shown after the page is loaded is controlled via parameter :show_filters of the grid helper. Possible values are: * :when_filtered - the filter is shown when the current table is the result of filtering * :always - always show the filter * :no - never show the filter Example: <%= grid(@tasks_grid, :show_filters => :always) do |g| ...... end -%> Filter related icons (filter icon, reset icon, show/hide icon) are placed in the header of the last column if it doesn't have any filter or a column name, otherwise an additional table column is added. To always place the icons in the additional column, set Wice::Defaults::REUSE_LAST_COLUMN_FOR_FILTER_ICONS to +false+ in the configuration file. === Initial Ordering Initializing the grid we can also define the column by which the record will be ordered on the first rendering of the grid, when the user has not set their ordering setting by clicking the column label, and the order direction: @tasks_grid = initialize_grid(Task, :order => 'tasks.title', :order_direction => 'desc' ) === Records Per Page The number of rows per page is set with :per_page: @tasks_grid = initialize_grid(Task, :per_page => 40) === Conditions The +initialize_grid+ method supports a :conditions parameter which is passed on to the underlying ActiveRecord, so it can be in any format processable by ActiveRecord: @tasks_grid = initialize_grid(Task, :conditions => ["archived = false and estimated_time > ?", 100] ) @tasks_grid = initialize_grid(Task, :include => :project, :conditions => {:archived => false, :project => {:active => true}} ) A good example is substituting a common pattern like @user_groups = @portal_application.user_groups with WiceGrid code: @user_groups_grid = initialize_grid(UserGroup, :conditions => ['portal_application_id = ?', @portal_application]) Alternatively, instead of a Class object as the first parameter, you can use ActiveRecord::Relation: @tasks_grid = initialize_grid(Task. where(:archived => false, :projects => {:active => true}). joins(:project) ) Please note that though all queries inside of WiceGrid are run without the default scope, if you use an ActiveRecord::Relation instance to initialize grid, it will already include the default scope. Thus you might consider using unscoped: @tasks_grid = initialize_grid(Task.unscoped. where(:archived => false, :projects => {:active => true}). joins(:project) ) === Queries with join tables WiceGrid also supports ActiveRecord's :joins and :include. @products_grid = initialize_grid(Product, :include => :category, :order => 'products.name', :per_page => 20) Note that if we want to order initially by a column from a joined table we have to specify the table and the column name with the sql dot notation, that is, products.name To show columns of joined tables in the view table, the ActiveRecord model class name has to be specified, that corresponds to the joined table: <%= grid(@products_grid) do |g| g.column :name => 'Product Name', :attribute => 'name' do |product| # primary table link_to(product.name, product_path(product)) end g.column :name => 'Category', :attribute => 'name', :model => Category |product| # joined table product.category.name end %> Please note that the blockless definition of the column only works with columns from the main table and it won't work with columns with :model === Joined associations referring to the same table In case there are two joined associations both referring to the same table, ActiveRecord constructs a query where the second join provides an alias for the joined table. To enable WiceGrid to order and filter by columns belonging to different associations but originating from the same table, set :table_alias to this alias: Model: class Project < ActiveRecord::Base belongs_to :customer, :class_name => 'Company' belongs_to :supplier, :class_name => 'Company' end Controller: @projects_grid = initialize_grid(Project, :include => [:customer, :supplier] ) View: <%= grid(@projects_grid, :show_filters => :always) do |g| g.column :name => 'Project Name', :attribute => 'name' g.column :name => 'Customer company', :model => 'Company', :attribute => 'name' do |task| task.customer.name if task.customer end g.column :name => 'Supplier company', :model => 'Company', :attribute => 'name', :table_alias => ' suppliers_projects' do |task| task.supplier.name if task.supplier end end -%> === More than one grid on a page It is possible to use more that one grid on a page, each with its own state. To do so, you must specify the name of the grid in +initialize_grid+ using parameter :name The name serves as the base name for HTTP parameters, DOM IDs, etc, so it is important that all grids on a page have different names. The default name is 'grid'. The name can only contain alphanumeric characters. @projects_grid = initialize_grid(Project, :name => 'g1' ) @tasks_grid = initialize_grid(Task, :name => 'g2') === Custom Ordering It is possible to change the way results are ordered injecting a chunk of SQL code, for example, use ORDER BY INET_ATON(ip_address) instead of ORDER BY ip_address. To do so, provide parameter :custom_order in the initialization of the grid with a hash where keys are fully qualified names of database columns, and values the required chunks of SQL to use in the ORDER BY clause. For example: @hosts_grid = initialize_grid(Host, :custom_order => { 'hosts.ip_address' => 'INET_ATON(hosts.ip_address)' }) It is possible to use the '?' character instead of the name of the column in the hash value: @hosts_grid = initialize_grid(Host, :custom_order => { 'hosts.ip_address' => 'INET_ATON( ? )' }) Values can also be Proc objects. The parameter supplied to such a Proc object is the name of the column: @hosts_grid = initialize_grid(Host, :custom_order => { 'hosts.ip_address' => lambda{|f| "INET_ATON( #{f} )"} }) === Custom dropdown filters It is possible to construct custom dropdown filters. Depending on the value of column parameter:custom_filter different modes are available: ==== Array of two-element arrays or a hash An array of two-element arrays or a hash are semantically identical ways of creating a custom filter. Every first item of the two-element array is used for the label of the select option while the second element is the value of the select option. In case of a hash the keys become the labels of the generated dropdown list, while the values will be values of options of the dropdown list: g.column :name => 'Status', :attribute => 'status', :custom_filter => {'Development' => 'development', 'Testing' => 'testing', 'Production' => 'production'} g.column :name => 'Status', :attribute => 'status', :custom_filter => [['Development', 'development'], ['Testing', 'testing'], ['Production', 'production']] It is also possible to submit a array of strings or numbers, in this case every item will be used both as the value of the select option and as its label: g.column :name => 'Status', :attribute => 'status', :custom_filter => ['development', 'testing', 'production'] ==== :auto :auto - a powerful option which populates the dropdown list with all unique values of the column specified by :attribute and :model. g.column :name => 'Status', :attribute => 'status', :custom_filter => ['development', 'testing', 'production'] Note that in the above example all names of all possible categories will appear even if they don't appear in the current resultset. To only show those which do appear in the resutset, use an array of symbol messages (see section 'An array of symbols'). ==== Custom filters and associations (joined tables) In most cases custom fields are needed for one-to-many and many-to-many associations. To correctly build a filter condition foreign keys have to be used, not the actual values rendered in the column. For example, if there is a column: g.column :name => 'Project Name', :attribute => 'name', :model => 'Project' do |task| task.project.name if task.project end adding :custom_filter like this: g.column :name => 'Project Name', :attribute => 'name', :model => 'Project', :custom_filter => Project.find(:all).map{|pr| [pr.name, pr.name]} do |task| task.project.name if task.project end is bad style and can fail, because the resulting condition will compare the name of the project, projects.name to a string, and in some databases it is possible that different records (projects in our example) have the same name. To use filter with foreign keys, we have to change the declaration of the column from projects.name, to tasks.project_id, and build the dropdown with foreign keys as values: g.column :name => 'Project Name', :attribute => 'tasks.project_id', :custom_filter => Project.find(:all).map{|pr| [pr.id, pr.name]} do |task| task.project.name if task.project end However, this will break the ordering of the column - the column will be ordered by the integer foreign key. To fix this, we can override the ordering using :custom_order: @tasks_grid = initialize_grid(Task, :include => :project, :custom_order => { 'tasks.project_id' => 'projects.name' } ) ==== Any other symbol (method name) or an array of symbols (method names) For one symbol (different from :auto) the dropdown list is populated by all unique values returned by the method with this name sent to all ActiveRecord objects throughout all pages. The conditions set up by the user are ignored, that is, the records used are all those found on all pages without any filters active. For an array of symbols, the first method name is sent to the ActiveRecord object if it responds to this method, the second method name is sent to the returned value unless it is +nil+, and so on. In other words, a single symbol mode is the same as an array of symbols where the array contains just one element. g.column :name => 'Version', :attribute => 'expected_version_id', :custom_filter => [:expected_version, :to_option] do |task| task.expected_version.name if task.expected_version end There are two important differences from :auto: 1. The method does not have to be a field in the result set, it is just some value computed in the method after the database call and ActiveRecord instantiation. 2. Filtering by any option of such a custom filter will bring a non-empty list, unlike with :auto. This mode has one major drawback - this mode requires an additional query without +offset+ and +limit+ clauses to instantiate _all_ ActiveRecord objects, and performance-wise it brings all the advantages of pagination to nothing. Thus, memory- and performance-wise this can be really bad for some queries and tables and should be used with care. If the final method returns a atomic value like a string or an integer, it is used for both the value and the label of the select option element: However, if the retuned value is a two element array, the first element is used for the option label and the second - for the value. Typically, a model method like the following: def to_option [name, id] end together with :custom_filter => :to_option would do the trick: Alternatively, a hash with the single key-value pair can be used, where the key will be used for the label, and the key - for the value: def to_option {name => id} end ==== Special treatment of values 'null' and 'not null' Values 'null' and 'not null' in a generated custom filter are treated specially, as SQL +null+ statement and not as strings. Value 'null' is transformed into SQL condition IS NULL, and 'not null' into IS NOT NULL Thus, if in a filter defined by :custom_filter => {'No' => 'null', 'Yes' => 'not null', '1' => 1, '2' => '2', '3' => '3'} values '1', '2' and 'No' are selected (in a multi-select mode), this will result in the following SQL: ( table.field IN ( '1', '2' ) OR table.field IS NULL ) ==== Multiple selection By default it is possible for any dropdown list to switch between single and multiple selection modes. To only allow single selection use :allow_multiple_selection: g.column :name => 'Expected in version', :attribute => 'expected_version_id', :custom_filter => [:expected_version, :to_option], :allow_multiple_selection => false do |task| ... end ==== Numeric Filters Before version 3.2.1 the filter used for numeric columns was a range filter with two limits. Beginning with version 3.2.1 the default is a direct comparison filter with one input field. The old range filter can still be loaded using parameter :filter_type with value :range: g.column :filter_type => :range do |task| ... end === Defaults Default values like can be changed in config/initializers/wice_grid_config.rb. === Submit/Reset buttons Instead of using default Submit and Reset icons you can use external HTML elements to trigger these actions. Add a button or any other clickable HTML element with class +wg-external-submit-button+ or +wg-external-reset-button+, and attribute +data-grid-name+ whose value is the name of the grid: To hide the default icons use :hide_submit_button => true and :hide_reset_button => true in the +grid+ helper. === Auto-reloading filters It is possible to configure a grid to reload itself once a filter has been changed. It works with all filter types including the JS calendar, the only exception is the standard Rails date/datetime filters. Use option :auto_reload in the column definiton: <%= grid(@tasks_grid, :show_filters => :always, :hide_submit_button => true) do |g| # String g.column :name => 'Title', :attribute => 'title', :auto_reload => true # Boolean g.column :name => 'Archived', :attribute => 'archived', :auto_reload => true # Custom (dropdown) g.column :name => 'Status', :attribute => 'status_id', :custom_filter => Status.to_dropdown, :auto_reload => true do |task| task.status.name if task.status end # Datetime g.column :name => 'Added', :attribute => 'created_at', :auto_reload => true, :helper_style => :calendar do |task| task.created_at.to_s(:short) end end -%> To make this behavior default change constant +AUTO_RELOAD+ in the configuration file. === Styling the grid ==== Adding classes and styles The +grid+ helper accepts parameter :html which is a hash of HTML attributes for the table tag. Another +grid+ parameter is header_tr_html which is a hash of HTML attributes to be added to the first +tr+ tag (or two first +tr+'s if the filter row is present). :html is a parameter for the +column+ method setting HTML attributes of +td+ tags for a certain column. ==== Adding classes and styles dynamically WiceGrid offers ways to dynamically add classes and styles to +TR+ and +TD+ based on the current ActiveRecord instance. For , let the +column+ return an array where the first item is the usual string output whole the second is a hash of HTML attributes to be added for the tag of the current cell: g.column do |portal_application| css_class = portal_application.public? ? 'public' : 'private' [portal_application.name, {:class => css_class}] end For adding classes/styles to use special clause +row_attributes+ , similar to +column+, only returning a hash: <%= grid(@versions_grid) do |g| g.row_attributes do |version| if version.in_production? {:style => 'background-color: rgb(255, 255, 204);'} end end g.column{|version| ... } g.column{|version| ... } end -%> Naturally, there can be only one +row_attributes+ definition for a WiceGrid instance. Various classes do not overwrite each other, instead, they are concatenated. === Adding rows to the grid It is possible to add your own handcrafted HTML after and/or before each grid row. This works similar to +row_attributes+, by adding blocks +after_row+, +before_row+, and +last_row+: <%= grid(@tasks_grid) do |g| g.before_row do |task, number_of_columns| if task.active? "Custom line for #{t.name}" # this would add a row # before every active task row else nil end end g.last_row do |number_of_columns| # This row will always be added to the bottom of the grid content_tag(:tr, content_tag(:td, 'Last row', :colspan => 10), :class => 'last_row') end ....... end %> It is up for the developer to return the correct HTML code, or return +nil+ if no row is needed for this record. Naturally, there is only one +before_row+ definition and one +after_row+ definition for a WiceGrid instance. The second variable injected into to before_row and after_row block, and the first parameter injected into the last_row is the number of columns in the current grid. === Rendering a grid without records If the grid does not contain any records to show, it is possible show some alternative view instead of an empty grid. Bear in mind that if the user sets up the filters in such a way that the selection of records is empty, this will still render the grid and it will be possible to reset the grid clicking on the Reset button. Thus, this only works if the initial number of records is 0. <%= grid(@grid) do |g| g.blank_slate do "There are no records" end g.column do |product| ... end end -%> There are two alternative ways to do the same, submitting a string to +blank_slate+: g.blank_slate "some text to be rendered" Or a partial: g.blank_slate :partial => "partial_name" === Action Column It is possible to add a column with checkboxes for each record. This is useful for actions with multiple records, for example, deleting selected records. Please note that +action_column+ only creates the checkboxes and the 'Select All' and 'Deselect All' buttons, and the form itself as well as processing the parameters should be taken care of by the application code. <%= grid(@tasks_grid, :show_filters => :always) do |g| ... g.action_column ... end -%> By default the name of the HTTP parameter follows pattern "#{grid_name}[#{param_name}][]", thus params[grid_name][param_name] will contain an array of object IDs. You can hide a certain action checkbox if you add the usual block to +g.action_column+, just like with the +g.column+ definition. If the block returns +nil+ or +false+ no checkbox will be rendered. <%= grid(@tasks_grid, :show_filters => :always) do |g| ... g.action_column do |task| task.finished? end ... end -%> WiceGrid is form-friendly: submitting grid in a form retains the state of the form. === Integration of the grid with other forms on page Imagine that the user should be able to change the behavior of the grid using some other control on the page, and not a grid filter. For example, on a page showing tasks, change between 'Show active tasks' to 'Show archived tasks' using a dropdown box. WiceGrid allows to keep the status of the grid with all the filtering and sorting using helper +dump_filter_parameters_as_hidden_fields+ which takes a grid object and dumps all current sorting and filtering parameters as hidden fields. Just include dump_filter_parameters_as_hidden_fields(@grid) inside your form, and the newly rendered grid will keep ordering and filtering. <% form_tag('', :method => :get) do %> <%= dump_filter_parameters_as_hidden_fields(@tasks_grid) %> <%= select_tag 'archived', options_for_select([['View active tasks', 0], ['View archived tasks', 1]], @archived ? 1 : 0), :onchange => 'this.form.submit()' %> <% end -%> === Javascript Calendar for Date and DateTime Filters. WiceGrid uses jQuery datepicker[http://jqueryui.com/demos/datepicker/] for Date and DateTime filters by default. Because this is part of the standard jQuery codebase, it is not bundled together with the plugin, and it is the responsibility of the programmer to include all necessary assets including localization files if the application is multilingual. jQuery datepicker does not have any time related controls, and when dealing with DateTime filters, the time value is ignored. Another option is standard Rails helpers. It's possible to change between the two styles of Date/DateTime filters in config/initializers/wice_grid_config.rb using constant HELPER_STYLE, or set it on per-column basis: g.column :name => 'Due Date', :attribute => 'due_date', :helper_style => :calendar do |task| task.due_date.to_s(:short) if task.due_date end g.column :name => 'Updated', :attribute => 'updated_at', :helper_style => :standard do |task| task.created_at.to_s(:short) end Constants +DATE_FORMAT+ and +DATETIME_FORMAT+ in the configuration file define the format of dates the user will see, as well as the format of the string sent in a HTTP parameter. If you change the formats, make sure that lamdbas defined in +DATETIME_PARSER+ and +DATE_PARSER+ return valid DateTime and Date objects. jQuery +datepicker+ uses a different format flavor, therefore there is an additional constant +DATE_FORMAT_JQUERY+. While +DATE_FORMAT_JQUERY+ is fed to +datepicker+, +DATE_FORMAT+ is still used for presenting initial date values in filters, so make sure that +DATE_FORMAT_JQUERY+ and +DATE_FORMAT+ result in an identical date representation. Constant +DATEPICKER_YEAR_RANGE+ defined the range of years in the Datepicker year dropdown. Alternatively, you can always change this range dynamically with the following javascript: $( ".hasDatepicker" ).datepicker( "option", "yearRange", "2000:2042" ); === Show All Records It is possible to switch to the All Records mode clicking on link "show all" in the bottom right corner. This functionality should be used with care. To turn this mode off for all grid instances, change constant +ALLOW_SHOWING_ALL_QUERIES+ in config/initializers/wice_grid_config.rb to +false+. To do so for a specific grid, use initializer parameter :allow_showing_all_records. Configuration constant +START_SHOWING_WARNING_FROM+ sets the threshold number of all records after which clicking on the link results in a javascript confirmation dialog. === CSV Export It is possible to export the data displayed on a grid to a CSV file. The dumped data is the current resultset with all the current filters and sorting applied, only without the pagination constraint (i.e. all pages). To enable CSV export add parameters +enable_export_to_csv+ and +csv_file_name+ to the initialization of the grid: @projects_grid = initialize_grid(Project, :include => [:customer, :supplier], :name => 'g2', :enable_export_to_csv => true, :csv_file_name => 'projects' ) +csv_file_name+ is the name of the downloaded file. This parameter is optional, if it is missing, the name of the grid is used instead. The export icon will appear at the bottom right corner of the grid. Next, each grid view helper should be placed in a partial of its own, requiring it from the master template for the usual flow. There must be no HTML or ERB code in this partial except for the grid helper. By convention the name of such a partial follows the following pattern: _GRID_NAME_grid.html.erb In other words, a grid named +tasks+ is expected to be found in a template called _tasks_grid.html.erb (remember that the default name of grids is '+grid+'.) Next, method +export_grid_if_requested+ should be added to the end of each action containing grids with enabled CSV export. +export_grid_if_requested+ intercepts CSV export requests and evaluates the partial with the required grid helper. The naming convention for grid partials can be easily overridden by supplying a hash parameter to +export_grid_if_requested+ where each key is the name of a grid, and the value is the name of the template (like it is specified for +render+, i.e. without '_' and extensions): export_grid_if_requested('g1' => 'tasks_grid', 'g2' => 'projects_grid') If the request is not a CSV export request, +export_grid_if_requested+ does nothing and returns +false+, if it is a CSV export request, the method returns +true+. If the action has no explicit +render+ call, it's OK to just place +export_grid_if_requested+ as the last line of the action: def index @tasks_grid = initialize_grid(Task, :name => 'g1', :enable_export_to_csv => true, :csv_file_name => 'tasks' ) @projects_grid = initialize_grid(Project, :name => 'g2', :enable_export_to_csv => true, :csv_file_name => 'projects' ) export_grid_if_requested end Otherwise, to avoid double rendering, use the return value of the method to conditionally call your +render+ : def index ........... export_grid_if_requested || render(:action => 'my_template') end It's also possible to supply a block which will be called if no CSV export is requested: def index ........... export_grid_if_requested do render(:action => 'my_template') end end If a column has to be excluded from the CSV export, set +column+ parameter +in_csv+ to +false+: g.column :in_csv => false do |task| link_to('Edit', edit_task_path(task)) end If a column must appear both in HTML and CSV, but with different output, duplicate the column and use parameters +in_csv+ and +in_html+ to include one of them to html output only, the other to CSV only: # html version g.column :name => 'Title', :attribute => 'title', :in_csv => false do |task| link_to('Edit', edit_task_path(task.title)) end # plain text version g.column :name => 'Title', :in_html => false do |task| task.title end The default field separator in generated CSV is a comma, but it's possible to override it: @products_grid = initialize_grid(Product, :enable_export_to_csv => true, :csv_field_separator => ';', :csv_file_name => 'products') If you need an external CSV export button , add class +wg-external-csv-export-button+ to any clickable element on page and set its attribute +data-grid-name+ to the name of the grid: If you need to disable the default export icon in the grid, add :hide_csv_button => true to the +grid+ helper. === Detached Filters Filters can also be detached from the grid table and placed anywhere on page. This is a 3-step process. First, define the grid with helper +define_grid+ instead of +grid+. Everything should be done the same way as with +grid+, but every column which will have an external filter, add :detach_with_id => :some_filter_name+ in the column definition. The value of +:detach_with_id+ is an arbitrary string or a symbol value which will be used later to identify the filter. <%= define_grid(@tasks_grid, :show_filters => :always) do |g| g.column :name => 'Title', :attribute => 'title', :detach_with_id => :title_filter do |task| link_to('Edit', edit_task_path(task.title)) end g.column :name => 'Archived', :attribute => 'archived', :detach_with_id => :archived_filter do |task| task.archived? ? 'Yes' : 'No' end g.column :name => 'Added', :attribute => 'created_at', :detach_with_id => :created_at_filter do |task| task.created_at.to_s(:short) end end -%> Then, use grid_filter(grid, :some_filter_name) to render filters: <% # rendering filter with key :title_filter %> <%= grid_filter @tasks_grid, :title_filter %> <% # rendering filter with key :archived_filter %> <%= grid_filter @tasks_grid, :archived_filter %> <% # rendering filter with key :created_at_filter %> <%= grid_filter @tasks_grid, :created_at_filter %> <% # Rendering the grid body %> <%= grid(@tasks_grid) %> Finally, use render_grid(@grid) to actually output the grid table . Using custom submit and reset buttons together with :hide_submit_button => true and :hide_reset_button => true allows to completely get rid of the default filter row and the default icons (see section 'Submit/Reset Buttons'). If a column was declared with :detach_with_id, but never output with +grid_filter+, filtering the grid in development mode will result in an warning javascript message and the missing filter will be ignored. There is no such message in production. === Access to Records From Outside The Grid There are two ways you can access the records outside the grid - using methods of the WiceGrid object and using callbacks. ==== Accessing Records Via The WiceGrid Object Method +current_page_records+ returns exactly the same list of objects displayed on page: <%= grid(@tasks_grid) do |g| ... end -%>

IDs of records on the current page: <%= @tasks_grid.current_page_records.map(&:id).to_sentence %>

Method +all_pages_records+ returns a list of objects browsable through all pages with the current filters: <%= grid(@tasks_grid) do |g| ... end -%>

IDs of all records: <%= @tasks_grid.all_pages_records.map(&:id).to_sentence %>

Mind that this helper results in an additional SQL query. Because of the current implementation of WiceGrid these helpers work only after the declaration of the grid in the view. This is due to the lazy nature of WiceGrid - the actual call to the database is made during the execution of the +grid+ helper, because to build the correct query columns declarations are required. ==== Accessing Records Via Callbacks It is possible to set up callbacks which are executed from within the plugin just after the call to the database. The callbacks are called before rendering the grid cells, so the results of this processing can be used in the grid. There are 3 ways you can set up such callbacks: Via a lambda object: def index @tasks_grid = initialize_grid(Task, :with_paginated_resultset => lambda do |records| ... end ) end Via a symbol which is the name of a controller method: def index @tasks_grid = initialize_grid(Task, :with_paginated_resultset => :process_selection ) end def process_selection(records) ... end Via a separate block: def index @tasks_grid = initialize_grid(Task) @tasks_grid.with_paginated_resultset do |records| ... end end There are two callbacks: * :with_paginated_resultset - used to process records of the current page * :with_resultset - used to process all records browsable through all pages with the current filters While the :with_paginated_resultset callback just receives the list of records, :with_resultset receives an ActiveRelation object which can be used to obtain the list of all records: def index @tasks_grid = initialize_grid(Task) @tasks_grid.with_resultset do |active_relation| all_records = active_relation.find(:first) ... end end This lazy nature exists for performance reasons. Reading all records leads to an additional call, and there can be cases when processing all records should be triggered only under certain circumstances: def index @tasks_grid = initialize_grid(Task) @tasks_grid.with_resultset do |wrapper| if params[:process_all_records] all_records = wrapper.call ... end end end == Icons Icons used by the plugin are courtesy of Mark James, the creator of the SILK icon set - http://www.famfamfam.com/lab/icons/silk/.