Song search app with dynamic tables

Question

I decided to make a Ruby app to search for songs, which uses the tinysong API. Things got really cryptic and weird in the view. I managed to get dynamic tables that adjust to data length, but I feel like there must be a much better way to accomplish this output.

require "rubygems"
require 'json'
require "httparty"

def search(search_terms)
  search_string = search_terms.split(' ').join('+')
  response = HTTParty.get("http://tinysong.com/s/#{search_string}?format=json&key=API_KEY_OMITTED_FOR_REVIEW")
  song_hash = JSON.parse(response.body)
  #puts "#{song_hash}"
end



def view_song_search(song_hash)
  term_width       = %x'tput cols'.to_i # Executes terminal
  total_char_limit = term_width - 9
  field_limit      = total_char_limit / 3

  # Sets width of each column to the width of it's highest data.
  artist_width = album_width = song_width = 0
  song_hash.each do |song|
    if song["ArtistName"].length > artist_width then artist_width = song["ArtistName"].length end
    if song["AlbumName"].length > album_width then album_width = song["AlbumName"].length end
    if song["SongName"].length > song_width then song_width = song["SongName"].length end
  end


  # Sets the minimum column width equal to header width.
  if artist_width < "Artist".length then artist_width = "Artist".length end
  if album_width < "Album".length then album_width = "Album".length end
  if song_width < "Song".length then song_width = "Song".length end

  # If total width is higher than terminal width, reduce total width to fit.
  if artist_width + album_width + song_width > total_char_limit
    artist_width = album_width = song_width = field_limit
  end

  # Where things get cryptic.
  printf("+%s+%s+%s+\n", "-" * (artist_width + 1), "-" * (album_width + 2), "-" * (song_width + 2))
  printf("\e[7m|%-#{artist_width}s | %-#{album_width}s | %-#{song_width}s |\e[0m\n", "Artist", "Album", "Song")
  printf("+%s+%s+%s+\n", "-" * (artist_width + 1), "-" * (album_width + 2), "-" * (song_width + 2))

  song_hash.each do |song|
    # Sets data of each column to not exceed column width.
    if song["ArtistName"].length > artist_width || 
       song["AlbumName"].length > album_width ||
       song["SongName"].length > song_width
      artist_name = (song["ArtistName"])[0...artist_width]
      album_name = (song["AlbumName"])[0...album_width]
      song_name = (song["SongName"])[0...song_width]
    else
      artist_name = song["ArtistName"]
      album_name  = song["AlbumName"]
      song_name   = song["SongName"]
    end


    printf("|%s%s | %s%s | %s%s |\n",artist_name, ' ' * (artist_width - artist_name.length),
                                     album_name, ' ' * (album_width  - album_name.length),
                                     song_name,   ' ' * (song_width   - song_name.length))
    printf("+%s-%s-%s+\n", "-" * (artist_width + 1), "-" * (album_width + 2), "-" * (song_width + 2))
  end
end

puts "Enter your search terms: "
view_song_search(search(gets.chomp))

And here is an example of the output:

Answer 1

I'm going to assume that search returns a data structure like this:

data = [
  { 'ArtistName' => 'Crash Test Dummies',
    'AlbumName' => 'God Shuffled His Feet',
    'SongName' => 'Mmmm_Mmmmm...',
  },
  { 'ArtistName' => 'John Powell',
    'AlbumName' => 'How to Train Your Dragon',
    'SongName' => 'Test Drive',
  },
  …
]

In that case, using song_hash as the name of the parameter to view_song_search is misleading, since it's actually an array.

There's no reason why view_song_search needs to know anything about songs, or why it should be tied to artists, albums, and songs at all. You should write a generic table-formatting function, like this:

puts table(data, [['ArtistName', 'Artist'],
                  ['AlbumName', 'Album'],
                  ['SongName', 'Song']])

Here's an implementation of such a function. Remarks:

• A large chunk of the code is the block in the middle to shrink overwide tables. I've tried to scale each column proportionally instead of giving one-third to each column.
• The heading row resembles a data row in many ways, and should be handled using a common codepath.
• The horizontal rules are printed many times; build it just once.
• Use sprintf '%-*s', width, text to space-pad the text to the desired width.

def table(hashes, headings, width_limit=%x'tput cols'.to_i - 5)
  attrs = headings.map { |h| h.first }
  headings = Hash[headings]

  # Minimum widths to accommodate the heading and data of each column,
  # excluding leading space, trailing space, and vertical dividers
  col_widths = attrs.map do |attr|
    [headings[attr].size, hashes.map { |h| h[attr].size }.max].max
  end

  # If the desired width, including delimiters, exceeds the limit,
  # reduce column widths proportionally
  desired_width = col_widths.inject(:+) + 3 * col_widths.size + 1
  if desired_width > width_limit
    col_widths.each_with_index do |width, c|
      col_widths[c] = (width.to_f * width_limit / desired_width).to_i - 2
    end

    # Readjust last column for round-off errors
    desired_width = col_widths.inject(:+) + 3 * col_widths.size + 1
    col_widths[-1] += width_limit - desired_width
  end

  output = hrule = '+' + col_widths.map { |width| '-' * (width + 2) + '+' }.join
  ([headings] + hashes).each_with_index do |row, r|
    output += "\n"
    output += "\e[7m" if r == 0         # Highlight heading row
    output += '|'
    attrs.each_with_index do |attr, c|
      output += sprintf(' %-*s |', col_widths[c], row[attr][0...col_widths[c]])
    end
    output += "\e[0m" if r == 0         # Turn off highlight
    output += "\n" + hrule
  end
  output
end