Click here to Skip to main content
Click here to Skip to main content

Day 1: Spider Database Navigator Website

, 3 Nov 2013 CPOL
Rate this:
Please Sign up or sign in to vote.
Creating a website in Ruby on Rails for dynamically displaying and navigating a SQL Server database

(click here for a full screen image)

The latest versions of this code can be obtained from my GitHub account here.

In The Beginning...

"In the beginning of creation, when Marc made the website, the website was without form and void, with darkness over the face of the HTML, and a mighty script that swept over the surface of the IDE.  Marc said, 'Let there be a database spider UI', and there was a database spider UI; and Marc saw that the UI was good, and he separated primary keys from foreign keys.  He called the foreign keys parent/child relationships, and the darkness Ruby on Rails.  So evening came, and morning came, the first day." 

I've been wanting to create a database navigator, what a friend of mine termed a "Spider UI", for quite some time now.  I originally wrote an implementation in WinForms for an Oracle database but that never went as far as I wanted it to go.  Currently, I'm working a lot with Ruby on Rails and also encountered a couple layers called Slim ("a lightweight templating engine") and Sass ("Syntactically Awesome Style Sheets") in another project.  Wanting to learn more about how Slim and Sass work together, I came up with this project idea.  In its full glory, I'm wanting to add all sorts of interesting features, such as custom layout for editing, but first, I wanted to get some basic functionality in place first. 

Internally, the website must support:

  • Connect to SQL Server (yes, Ruby on Rails can connect to SQL Server.)
  • Use a separate database for session and other metadata store, leaving the database we're connecting to for navigation untouched.
  • Rather than physical Models for each table, implement a dynamic Model.
  • Dynamic discovery of user tables and foreign key relationships by inspecting the database schema rather than relying on the Rails application schema (which would be otherwise generated from concrete Models backed by physical tables.)

The user interface must support:

  • Selecting a table from a list
  • Viewing the data for that table
  • Selecting parent and child tables to navigate to
  • Selecting records to qualify navigation and display of parent/child records
  • Pagination
  • Breadcrumb trail of parent/child navigation

What I'm leaving for "Day 2" are several metadata features:

  • Replacing foreign key ID's with "display field" lookups.
  • Describing the display fields in a lookup table to use when resolving foreign keys.
  • Specifying fields that need not be displayed.
  • Aliasing field names
  • Aliasing table names

"Day 3" will consist of supporting:

  • type-based automated generation of UI's to edit table data.
  • custom record editing templates.

"Day 4" will consist of:

  • views (described in metadata as opposed to database-side views), as creating views with full schema information is required.
  • creating SQL statements to support transactions on views

What comes after that will probably involve the support of custom processing of data during transactions (both code and PL/SQL calls) using Ruby as well as server-side triggers on transactions.  We'll see!

If Your New To Ruby on Rails

Code Project is primarily (at the time of this writing!) a site for all things Microsoft, so if this is the first time you're encountering Ruby on Rails, I'd recommend reading through some of my other articles on the Ruby language and Ruby on Rails:

as these are more tutorial based for developing Ruby on Rails (RoR) applications.  This article assumes that you already have some familiarity with the project structure of a RoR application.

AdventureWorks2008

This article uses Microsoft's example database, Adventure Works, to demonstrate Ruby on Rails connectivity to SQL Server as well as example dataset for all the screenshots in this article.

About the Ruby Code

You'll notice that I tend to write very short Ruby functions - this is intended to promote the clarity of higher-level functions, which can otherwise detract from the purpose of the function when embedding idiomatic Ruby and arcane operations.  I also like to explicitly state what the return value of a function is, even when it's unnecessary.  This promotes further clarity to someone who is unfamiliar with the application code.

About the Slim and Sass Markup

I've tried to keep the markup concise and I've put comments where it seems appropriate to describe the intent of the markup.  This is especially salient in the Sass markup, where the intention behind the styling is not necessarily obvious.

Gems and What They Are Used For

Gems are Ruby on Rails' plug-in mechanism for adding functionality from the vast amount of free and open source components that people have contributed to over the years.  Besides the Rails gems, the ones I'm taking advantage of are:

gem 'tiny_tds'
gem 'activerecord-sqlserver-adapter'
gem 'sqlite3'
gem 'slim'
gem 'thin'
gem 'sass'
gem 'will_paginate'

TinyTDS

The gem tiny_tds necessary for connecting to SQL Server.  TinyTDS requires a SQL Server Authentication login, as opposed to a Windows Authentication login.  I have a short blog entry here on connecting to SQL Server Express from Ruby which covers configuring SQL Server Express and testing the TinyTDS connectivity.  In this application, TinyTDS is used to acquire the schema information from SQL Server -- see the section "Schema Class" below.

activerecord-sqlserver-adapter

This gem uses the "dblib" connection mode which in turn is dependent upon TinyTDS.  What this gem enables you to do is to work with the Rails ActiveRecord API for all transactions with the database.  It's important that we use ActiveRecord for table queries because the pagination system relies on our Model classes being derived from ActiveRecord::Base.  In future articles, we'll also be relying, in part, on ActiveRecord for other transactions on the table records.

Rails expects the connection information  to be specified in the config\database.yml file.  Here we set up our development connection, specifying the sqlserver adapter which the above gem provides us, along with the connection information required by TinyTDS.

development:
  adapter: sqlserver
  database: AdventureWorks2008
  dataserver: localhost\SQLEXPRESS
  username: ruby
  password: rubyist1

Sqlite3

One of the requirements of this application is to not change the schema of the database that we're "spidering."  Also, there's a lot of session information that is being preserved -- too much to place is session cookies on the client.  We're using Sqlite3 for storing session information independent of our SQL Server database and this gem provides the connectivity.  To see how this is done, read the section on "Storing Session Information in a Separate Database."

Slim

This gem eliminates the angle brackets and ending tags of the HTML script.  Here's a simple example of the lighter-weight syntax:

doctype
html
  head
    title Database Spider
    = stylesheet_link_tag "application", media: "all"
    = javascript_include_tag "application"
    = csrf_meta_tags
  body
    = yield

Notice how indentation is used to determine where the closing tags should go in the generated HTML.

Sass

This gem is also "lightens" the description of CSS and works very well in conjunction with Slim.  You'll see Slim/Sass used such that the structure of the Slim and Sass markup are identical.  Here's a short example taken from the user table list markup:

The Slim Markup

.table_list_region
  p Tables:
  .table_list
    ul
      - @tables.each do |table_name|
        li
          = link_to(table_viewer_path(table_name: table_name)) do
            t #{table_name}

Notice how Slim knows how to work with both HTML and Ruby script in the markup!

The Sass Markup

.table_list_region
  width: 240px
  height: 700px
  float: left
  p
    text-align: left
    margin-bottom: 2px
  .table_list
    text-align: left
    border: 1px solid
    border-radius: 3px
    width: 100%
    height: 100%
    line-height: 1.5em
    ul
      width: 190px
      height: 99%
      overflow: auto
      margin-top: 2px
      li
        list-style-type: none
        margin-left: -30px
        a:visited
          color: #000000
        a:link
          color: #000000
          text-decoration: none
        a:hover
          color: #0000FF

You can see how Sass markup can follow the same structure as the Slim markup.  The problem with this approach is that it obfuscates re-use of CSS.  For example, I might want to re-use the table_list styling but since it's a child of table_list_region, I can't (at least not without a table_list_region container).  However, I can certainly outdent the styling:

.table_list_region
  width: 240px
  height: 700px
  float: left
  p
    text-align: left
    margin-bottom: 2px

.table_list
  text-align: left
  ... etc ...

without affecting the structure of the Slim markup.  Now table_list is accessible inside the container as well as reusable by other containers.  Unfortunately, I have worked on at least one major open source project in which the developers chose not to promote styling re-use, resulting in complex Sass structures in which one has to follow the hiearchy of the Slim exactly to locate the styling that I wanted to change.  Don't do this unless your styling is truly specific to its container context!

Will_paginate

This is an amazing gem that paginates your data and provides a variety of pre-existing styling and also the ability to customize the styling of the pagination bar.  And of course, one of the advantages of using pagination is that for tables with large numbers of records, only the records for that page are returned from the database, greatly improving usability.

The Code

I'm not going to go into every detail of the code, but I will point out some of the more interesting features.

Schema Class

This class (schema.rb) encapsulates the static functions we need for connecting to SQL Server directly and obtaining schema information and thus relies on the TinyTDS gem for the direct SQL connection.  The main workhorse is this function:

def self.execute(sql)
  client = create_db_client
  result = client.execute(sql)
  records = result.each(as: :array, symbolize_keys: true)
  array = convert_to_array_of_hashes(result.fields, records)

  array
end

We rely on these helper methods (helpers\my_utils.rb) to create a client connection:

# create a client connection.
def create_db_client
  config = get_current_database_config
  config_as_symbol = symbolize_hash_key(config)
  client = TinyTds::Client.new(config_as_symbol)

  client
end

# Returns the current database config has a dictionary of string => string
def get_current_database_config
  Rails.configuration.database_configuration[Rails.env]
end

# Given a dictionary of string => string, returns :symbol => string
# Example: config_as_symbol = symbolize_hash_key(config)
def symbolize_hash_key(hash)
  hash.each_with_object({}){|(k,v), h| h[k.to_sym] = v}
end

Once the records are returned from TinyTDS, I want to package them into an array of hashes (field => value) so that there's a consistent representation of the resulting data.  This requires a couple post-processing functions:

# Convert the array of records from the TinyTDS query into an array of hashes, where
# the keys are the field names.
def self.convert_to_array_of_hashes(fields, records)
  array = []
  records.each { |record|
    dict = hash_from_key_value_arrays(fields, record)
    array << dict
  }

  array
end

# Given two arrays of equal length, 'keys' and 'values', returns a hash of key => value
def hash_from_key_value_arrays(keys, values)
  Hash[keys.zip values]
end

Schema Queries

We can now define the three functions that we need for our Spider UI:

  1. get_user_tables
  2. get_parent_table_info_for
  3. get_child_table_info_for

In the last two functions, we replace the string "[table]" with the table name before executing the query:

  sql.sub!('[table]', table_name.partition('.')[2])

What we're not dealing with at the moment is that the table name being passed in is fully qualified (it includes also the schema name) but the query doesn't pay attention to the schema name, hence we need to extract just the table name from the parameter value.  This is one of those "TODO" items for a later date.

These functions rely on our storing the actual queries in a "queries.yml" file, which is a hierarchical file similar to XML in concept but very different in implementation.  In this file, we store our schema queries:

Schema:
  user_tables: "
    select 
      s.name + '.' + o.name as table_name
    from 
      sys.objects o
        left join sys.schemas s on s.schema_id = o.schema_id
    where 
      type_desc = 'USER_TABLE'"

  get_parents: "
    SELECT
      f.parent_object_id as ChildObjectID,
      SCHEMA_NAME(f.schema_id) SchemaName,
      OBJECT_NAME(f.parent_object_id) TableName,
      COL_NAME(fc.parent_object_id,fc.parent_column_id) ColName,
      SCHEMA_NAME(ref.schema_id) ReferencedSchemaName,
      OBJECT_NAME(f.referenced_object_id) ReferencedTableName,
      COL_NAME(fc.referenced_object_id, fc.referenced_column_id) ReferencedColumnName
    FROM 
      sys.foreign_keys AS f
        INNER JOIN sys.foreign_key_columns AS fc ON f.OBJECT_ID = fc.constraint_object_id
        INNER JOIN sys.tables t ON t.object_id = fc.referenced_object_id
        INNER JOIN sys.tables ref ON ref.object_id = f.referenced_object_id
    WHERE
      OBJECT_NAME (f.parent_object_id) = '[table]'"

  get_children: "
    SELECT 
      f.parent_object_id as ChildObjectID,
      SCHEMA_NAME(f.schema_id) SchemaName,
      OBJECT_NAME(f.parent_object_id) TableName,
      COL_NAME(fc.parent_object_id,fc.parent_column_id) ColName,
      SCHEMA_NAME(ref.schema_id) ReferencedSchemaName,
      OBJECT_NAME(f.referenced_object_id) ReferencedTableName,
      COL_NAME(fc.referenced_object_id, fc.referenced_column_id) ReferencedColumnName
    FROM
      sys.foreign_keys AS f
        INNER JOIN sys.foreign_key_columns AS fc ON fc.constraint_object_id = f.OBJECT_ID
        INNER JOIN sys.tables t ON t.OBJECT_ID = fc.referenced_object_id
        INNER JOIN sys.tables ref ON ref.object_id = f.referenced_object_id
    WHERE
      OBJECT_NAME (f.referenced_object_id) = '[table]'"

and we again use a small helper function to retrieve the text:

# Gets the specified query from the config.yml file
# Example: sql = get_query("Schema", "user_tables")
# TODO: cache keys
def get_query(key1, key2)
  sql = YAML.load_file(File.expand_path('config/queries.yml'))
  sql[key1][key2]
end

Here's another "TODO" item for later: eventually this structure will be "refactored" to include the database context, because of course querying the schema of a database is very database specific!

Get User Tables

This function simply returns a collection of user table names:

# Returns an array of strings containing the user tables in the database to which we're connecting.
def self.get_user_tables
  client = create_db_client
  sql = get_query("Schema", "user_tables")
  result = client.execute(sql)
  records = result.each(as: :array, symbolize_keys: true)
  names = get_column(records, 0)

  names
end

Get Parent Table Info

This function returns the foreign key associations of the current table, resulting in an array of all parent tables along with schema information describing the columns of both child and parent tables that describe the foreign key relationship:

# Returns an array of hashes (column name => value) of the parent table schemas of the specified table.
def self.get_parent_table_info_for(table_name)
  sql = get_query("Schema", "get_parents")
  sql.sub!('[table]', table_name.partition('.')[2])
  execute(sql)
end

Get Child Table Info

This function returns the child tables and their foreign key column relationships to the specified parent table.  This query has a similar structure as what is returned in the previous function, courtesy of how the SQL is formatted (see earlier.)

# Returns an array of hashes (column name => value) of the child table schemas of the specified table.
def self.get_child_table_info_for(table_name)
  sql = get_query("Schema", "get_children")
  sql.sub!('[table]', table_name.partition('.')[2])
  execute(sql)
end

Storing Session Information in a Separate Database

We need to tell Rails in initializers\session_store.rb that we want to use a database rather than cookies for storing session information:

DatabaseSpider::Application.config.session_store :active_record_store

However, we want to tell Rails not to use the SQL Server database, so we also need:

ActiveRecord::SessionStore::Session.establish_connection(:sessions)

and finally, we need to put the session definition in the config\database.yml file:

# use sqlite3 as the DB for storing session information.
sessions:
  adapter: sqlite3
  database: db/session.sqlite3
  pool: 5
  timeout: 5000

Dynamic Active Records

Another requirement is to avoid creating a concrete Model class for every physical table in the database.  To accomplish this, we derive a class from ActiveRecord::Base but specify the table name to which it is associated:

class DynamicTable < ActiveRecord::Base

  # Returns an array of records for the specified table.
  def set_table_data(table_name)
    DynamicTable::table_name = table_name
  end

  # Returns the field names given at least one record.
  def get_record_fields(records)
    fields = []
    fields = records[0].attributes.keys if records.count > 0

    fields
  end
end

This enables us to interact with the table just as we would with any other ActiveRecord instance.

The function get_record_fields is used to return the field names for the record - we arbitrarily pick the first record for the field list, assuming any records exist at all.  This is a bit problematic because even if there are no records returned, we'd like the table to at least display the field names.  So one of the "TODO" items is to use a TinyTDS query with "where 1=0", which, though it returns zero rows, will populate the field names for us.

TableViewerController, the Index Function

The index function of the TableViewerController is the main workhorse of the website:

def index
  initialize_attributes
  update_model_page_numbers

  if self.table_name
    restore_page_number_on_nav # restores the page number when navigating back along the breadcrumbs
    self.last_page_num = self.model_page_nums[self.table_name+'_page'] # preserve the page number so selected navigation records are selected from the correct page.
    @data_table = load_DDO(self.table_name, self.last_page_num, self.qualifier, MAIN_TABLE_ROWS)
    add_hidden_index_values(@data_table)
    load_navigators(self.table_name)
    @parent_dataset = load_fk_tables(@parent_tables)
    @child_dataset = load_fk_tables(@child_tables)
    # Update the parent tab index based on the existence and value of the selected_parent_table_index parameter
    update_parent_child_tab_indices
  end
end

Various attributes are initialized, of most important are:

  • the list of user tables.
  • a DynamicTable instance.
  • the navigation breadcrumb trail is restored from the session.

Session Variables

There are several session variables defined in the controller:

attr_session_accessor :table_name         # the selected user table
attr_session_accessor :qualifier          # the qualifier currently being used to filter the selected user table
attr_session_accessor :breadcrumbs        # the breadcrumb trail
attr_session_accessor :last_page_num      # the last page number of the user table
attr_session_accessor :force_page_num     # if set, forces the pagination to a different page, used with breadcrumbs
attr_session_accessor :model_page_nums 	  # dictionary of page numbers for all the models being displayed.

which help preserve the state of the page between posts.

Custom Attribute Accessor

I get tired of having to write code like this:

@qualifier = session[:qualifier]
session[:table_name] = @table_name

so I wrote a custom attribute accessor that simplifies getting and setting session values:

# Adds an "attr_session_accessor" declarator that, in addition to setting/getting the value
# to the attribute, it also gets/sets the value from the session.
# Usage inside the class defining the attribute: self.foobar = 1
# Note that "self." must prefix the usage of the attribute.
class Class
  def attr_session_accessor(*args)
    args.each do |arg|
      self.class_eval("def #{arg}; @#{arg}=session['#{arg}']; end")
      self.class_eval("def #{arg}=(val); @#{arg}=val; session['#{arg}']=val; end")
    end
  end
end

For the getter, this code initializes the specified attribute from the session name, and is equivalent to writing (for the variable qualifier):

@qualifier = session[:qualifier]

The setter initializes both the attribute and the session key with the specified value, and is the equivalent of writing (for the variable table_name):

@table_name = val
session[:table_name] = @table_name

Thus, we can code like this (this is the handler for clicking on a breadcrumb):

# Navigate back to the selected table in the nav history and pop the stack to that point.
# Use the qualifier that was specified when navigating to this table.
# Restore the page number the user was previously on for this table.
def nav_back
  stack_idx = params[:index].to_i
  breadcrumb = self.breadcrumbs[stack_idx]          # get the current breadcrumb
  self.table_name = breadcrumb.table_name           # we want to go back to this table and its qualifier
  self.qualifier = breadcrumb.qualifier
  self.breadcrumbs = self.breadcrumbs[0..stack_idx] # remove all the other items on the stack
  self.force_page_num = breadcrumb.page_num

  redirect_to table_viewer_path+"/index"
end

I suppose it would be more readable if I named these variables with some prefix, perhaps "sess_" to make it clear to the reader that we're accessing session data, however, an more useful "TODO" would be to put together a Session class that has all the session attributes that the controller referenecs.  Internally, this class could still use my custom attribute accessor, but it would also allow intellisense to work within the IDE, improving the programmer's experience and providing clarity as to what is going on behind the scenes.

Byte Encoding

One of the issues I encountered was with SQL Server fields that were using byte encoding.  This gave Rails fits and required a helper method, applied to each field value:

# Fixes encoding to UTF-8 for certain field types that cause problems.
# http://stackoverflow.com/questions/13003287/encodingundefinedconversionerror
helper_method :fix_encoding
def fix_encoding(value)
  value.to_s.encode('UTF-8', {:invalid => :replace, :undef => :replace, :replace => '?'})
end

Hiding Fields

Certain fields that are found in the Adventure Works database can simply always be hidden.  Also, the paginator adds a column that it uses internally for keeping track of what page the user is viewing, and finally, I add a column that the checkbox on each row uses to identify the selected row.  None of these need to be displayed, therefore we have a couple helper methods:

# Return true if the field can be displayed.
# All sorts of interesting things can be done here:
# Hide primary keys, ModifiedDate, rowguid, etc.
helper_method :display_field?
def display_field?(table_name, field_name)
  # '__rn' is something that will_paginate adds.
  # '__idx' is my hidden column for creating a single column unique ID to identify selected rows, since
  # we can't guarantee single-field PK's and we need some way to identify a row uniquely other than the actual data.
  return false if ['__rn', '__idx', 'rowguid', 'ModifiedDate'].include?(field_name)
  true
end

# Returns only the visible fields
helper_method :get_visible_fields
def get_visible_fields(data_table)
  data_table.fields.keep_if {|f| display_field?(data_table.table_name, f)}
end

The User Interface

The user interface consists of six areas:

  1. The list of user tables in the database
  2. The selected table's data
  3. The parent tables of the selected table
  4. The child tables of the selected table
  5. Navigation options
  6. List of navigations (breadcrumbs)

I decided to put each of these sections into their own render block, thus the resulting index.html.slim file is simply:

=form_for @table_viewer do |f|
  = render 'breadcrumbs'
  = render 'user_tables'

  - # The selected table data
  - if !@table_name.nil?
    = render 'parent_tables'
    = render 'selected_table'
    = render 'navigation', f: f
    = render 'child_tables'

- # Restore current page selections
javascript:
  select_fk_tab('#parent_tab', '#parent_tab_content', #{@parent_tab_index}, #{@parent_tables.count})
  select_fk_tab('#child_tab', '#child_tab_content', #{@child_tab_index}, #{@child_tables.count})

Breadcrumbs

The breadcrumbs is a clickable list of tables that the user has already navigated to using the "navigate to parent" and/or "navigate to child" options.  When a breadcrumb is selected, any selection that was made to qualify the data for that table is restored as well as the table's data.

The Slim Markup

- # The navigation breadcrumbs
.navigation_history
  p = "Nav History:"
  br
  - if @breadcrumbs
    - @breadcrumbs.each_with_index do |breadcrumb, index|
      - table_name = breadcrumb.table_name
      = link_to content_tag(:span, "#{table_name.partition('.')[2]}"), navigate_back_path(index: index), {:class => "button"}, :onclick => 'this.blur();'

User Tables

This is simply an alphabetical list of user tables.

The Slim Markup

- # The list of all tables.
.table_list_region
  p Tables:
  .table_list
    ul
      - @tables.each do |table_name|
        li
          = link_to(table_viewer_path(table_name: table_name)) do
            t #{table_name}

Selected Table Data

The selected user table is displayed with a checkbox on each row so that the user can select a specific row or rows and navigate to a child or parent table whose resulting data is displayed in the same box but qualified by the selection.  If no selection is made, then all the parent or child data is displayed.  At the bottom of the table is the paginator (see the parent and child table data for visual examples.)

The Slim Markup

- # The selected user table data.
.table_data_region
  p #{@data_table.table_name}
  = render "table_data", data_table: @data_table
  .digg_pagination
    = will_paginate @data_table.records, param_name: @data_table.table_name+"_page"

Note that this code (and the code used for child and table navigation) all re-use another render called "table_data".

_table_data.html.slim

This is file used in common for rendering the table data in the selected table, parent tables and child tables:

- # render for table data. The paginator is separate because it requires additional parameters that are specific
- # to the data being paginated.
.table_data
  - visible_fields = get_visible_fields(data_table)
  table
    - # Display header
    tr
      - # Dummy header column for checkbox
      th

      - # Header of field names
      - visible_fields.each do |field_name|
        th = field_name

    - # Display records
    - data_table.records.each do |record|
      tr class=cycle('row0', 'row1')
        td
          - # __idx is added by the controller to uniquely identify a record.
          = check_box_tag 'selected_records[]', record["__idx"]
          - visible_fields.each_with_index do |field_name, field_index|
            - # Extends the last column out to the right edge of the table.
            - if field_index == visible_fields.length - 1
              td.last = fix_encoding(record[field_name])
            - else
              td = fix_encoding(record[field_name])

The interesting part of this markup is figuring out when the last column is being rendered and to change the styling slightly so that the row color extends to the right edge of the table box:

td
  border-left: 1px solid #d0d0d0
  white-space: nowrap
  padding-left: 5px
  padding-right: 5px
td.last                         // the last column fills any remaining space
  width: 100%

Also notice the snazzy cycle function of Ruby:

tr class=cycle('row0', 'row1')

which cycles through the class styling strings for the rows, achieving the alternating colors:

tr.row0 // alternating row colors
  background-color: #ffffff
tr.row1
  background-color: #ddffdd

Navigation

The Navigation area lets the user select which parent or child table to navigate to.  If rows have been selected in the user table display, the parent/child records are qualified by the selected records.  In that case, to show unqualified records, the user clicks on Show All Records.  To return to the qualified records, click on the last breadcrumb button, as the navigation "breadcrumb" trail preserves the qualifiers.

Parent Navigation

The selected user table is treated as the child table, with foreign keys to the parent table.  If no rows in the user table have been selected, then all rows of the parent table are displayed.  If one or more rows has been selected (using checkboxes) then all columns involved in the foreign key relationship to the parent table are considered to construct the qualifying "where" clause.  For example, if there are two columns, one for billing address and one for shipping address, that refer to an address, then both address records of the parent (assuming the child has different values for the foreign key ID) will be displayed when navigating to the parent. 

Child Navigation

Here the selected user table is considered to be the parent, and the table selected in the combobox "Navigate to child:" is inspected for foreign key fields that map to the primary key fields of the selected user table.  If no rows in the user table have been selected (using checkboxes) then all child records are displayed.  If rows have been selected, then the qualifier is constructed programmatically to limit the child records to only those records that reference the selected rows.

The Slim Markup

- # table navigation
.navigation
  fieldset
    legend View and Navigation:
    br
    .nav_button
      = f.submit("Show All Records", name: 'navigate_show_all')
    - # Separate div because 'Go' buttons are left padded.
    .nav_options
      br Navigate to parent:
      = select_tag "cbParents", options_from_collection_for_select(@parent_tables, 'id', 'name')
      = f.submit("Go", name: 'navigate_to_parent')
      br Navigate to child:
      = select_tag "cbChildren", options_from_collection_for_select(@child_tables, 'id', 'name')
      = f.submit("Go", name: 'navigate_to_child')

Parent and Child Tables

These two regions of the page are almost identical except for how the pagination is handled and of course the tables and their records that they display.  The checkboxes in these tables don't do anything, yet.

The fun part about this markup is that the tabs are styled columns of a table row, and we utilize a scrollbar for when there are more tabs that fit on the screen.

The Slim Markup (for parent tables only)

- # if we have a selected table and it has parent tables, show the parent tables.
- if !@table_name.nil? && @parent_tables.length > 0
  .tab_region
    .tab_list
      table
        tr
          - @parent_tables.each_with_index do |table_info, index|
            td id="parent_tab#{index}"
              a title="View #{table_info.name}" 
		onclick="select_fk_tab('#parent_tab', '#parent_tab_content', #{index}, #{@parent_tables.count})"
                span = "#{table_info.name}"


    - # table_info isn't used because this data is formatted for the comboboxes.
    - # What we're actually simply interested in here is the index.
    - @parent_tables.each_with_index do |table_info, index|
      .tab_content id="parent_tab_content#{index}"
        .tab_table_data_region
          = render "table_data", data_table: @parent_dataset[index]
          .digg_pagination
            = will_paginate @parent_dataset[index].records, 
              param_name: @parent_dataset[index].table_name+"_page", 
              params: {selected_parent_table_index: index }

Javascript

Finally, we need a tiny amount of Javascript to put it all together, which handles selecting a tab and rendering the page after a post or refresh:

/* Deselect all the parent tabs and hide all the content, then select the specified tab and show the desired content */
function select_fk_tab(tab_selector, content_selector, index, num_tabs)
{
  for (var i=0; i<num_tabs; i++)
  {
    $(tab_selector + i.toString()).removeClass('current');
    $(content_selector + i.toString()).hide();
  }

  $(tab_selector + index.toString()).addClass('current');
  $(content_selector + index.toString()).show();
}

Conclusion

When all is said and done, we have:

  • successfully connected Ruby on Rails with SQL Server
  • dealt with using a separate database for session state
  • successfully used ActiveRecord in a dynamic context
  • implemented a good start for a general purpose, web-based, database navigator
  • learned a lot about Slim and Sass in the process (which was one of my primary goals)

License

This article, along with any associated source code and files, is licensed under The Code Project Open License (CPOL)

Share

About the Author

Marc Clifton

United States United States
Marc is the creator of two open source projects, MyXaml, a declarative (XML) instantiation engine and the Advanced Unit Testing framework, and Interacx, a commercial n-tier RAD application suite.  Visit his website, www.marcclifton.com, where you will find many of his articles and his blog.
 
Marc lives in Philmont, NY.

Comments and Discussions

 
QuestionArticle Title? Pinmvpthatraja9-Dec-13 0:01 
Like this?[^]
thatraja
Code converters | Education Needed
No thanks, I am all stocked up. - Luc Pattyn
When you're wrestling a gorilla, you don't stop when you're tired, you stop when the gorilla is - Henry Minute


General General    News News    Suggestion Suggestion    Question Question    Bug Bug    Answer Answer    Joke Joke    Rant Rant    Admin Admin   

Use Ctrl+Left/Right to switch messages, Ctrl+Up/Down to switch threads, Ctrl+Shift+Left/Right to switch pages.

| Advertise | Privacy | Terms of Use | Mobile
Web01 | 2.8.1411022.1 | Last Updated 3 Nov 2013
Article Copyright 2013 by Marc Clifton
Everything else Copyright © CodeProject, 1999-2014
Layout: fixed | fluid