compress.coffee

   1 orig_file = 'wordlist.txt'
   2 compressed_file = 'wordlist_compressed.js'
   3 caps = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
   4 fs = require 'fs'
   5
   6 fs.readFile orig_file, 'utf8', (err, data) ->
   7         if err?
   8                 console.log "Error reading #{orig_file}: #{err}"
   9                 process.exit(1)
  10         else
  11                 fs.writeFile compressed_file, (compress data.split '\n'), (err) ->
  12                         if err?
  13                                 console.log "Error writing to #{compressed_file}: #{err}"
  14                                 process.exit(1)
  15
  16 compress = (words) ->
  17         out = 'parse_word_list(['
  18         chunk = ''
  19         prev = ''
  20         for next in words
  21                 match = prev.length
  22                 while match and next.substr(0, match) isnt prev.substr(0, match)
  23                         --match
  24                 # now match is the number of prefix characters this word has in common with the previous
  25                 chunk += caps[prev.length - match]
  26                 if chunk.length > 2000
  27                         out += "'" + chunk.substr(1) + "',"
  28                         chunk = '.'
  29                 chunk += next.substr match
  30                 prev = next
  31         return out + "'" + chunk.substr(1) + "']);"