Hacks > Spellchecking Webservice

I run a small and simple webservice for spellchecking. The interface is in the REST style: simply pass a query string (GET) or an encoded set of values (POST). It handles several encoding types and has a few styles for the output. The paramaters are:

• word or text - Word (single) or text to spellcheck
• lang - Language code to use. Defaults to en. Currently only english and its variants are supported
• var - Variation to use. Defaults to nothing or american. Sadly that's the only supported value at the moment
• end - Character encoding for input and output. Defaults to utf-8. Valid values are
  • utf-8
  • iso8859-*
  • koi8-r
  • viscii
  • cp1252
• mode - Output mode. Defaults to text. Valid values are:
  • text
  • xml
  • javascript

Formats

Text

Lines beginning with :# can be ignored. If they begin with an !, they contain an error message. Otherwise, it's part of a list of suggestions. Each list is preceded by the word it is for. If there are no suggestions because a word is spelled correctly, it is followed by # Correct. Multiple words are separated by a line consisting entirely of #.
Sample output:
One word:

# smple
sample
simple
sampler
#

# ardvark
aardvark
aardvarks
aardvark's
#
# shoes
# Correct
#

! Error - an error ocurred while intializing pspell

Javascript

Up to three variables may be set:

• error - An error message. Only set if an error occurs
• word - The word that was spellchecked. Only set if checking a single word and no errors occur
• suggestions - If a single word was checked, then this is a one-dimensional array of suggestions. If a block of text is being checked, then this an array of arrays where the first element of a subarray is the word checked and the second is an array of suggestions. If the word is spelled correctly, then instead of an array, the value is null. Only set if no errors occur

word = "smple"
suggestions = [
  "sample",
  "simple",
  "sampler" ]

suggestions = [
  [ "ardvark", 
    [
      "aardvark",
      "aardvarks",
      "aardvark's" ]
  ],
  [ "shoes", 
     null ] 
  ]
]

error = "an error ocurred while intializing pspell"

XML

The root element is words, unless an error occurs. It contains zero or more word sub-elements. Each of them contains a text element with the word that was spellchecked and either an empty correct element or suggestions, which contains zero or more suggestion elements.
If an error occurs, then the only element returned is error with an error message. Sample outputs:
A single word:

<?xml version="1.0" encoding="UTF-8"?>
<words>
  <word>
    <text>smple</text>
    <suggestions>
      <suggestion>sample</suggestion>
      <suggestion>simple</suggestion>
      <suggestion>sampler</suggestion>
    </suggestions>
  </word>
</words>

<?xml version="1.0" encoding="UTF-8"?>
<words>
  <word>
    <text>ardvark</text>
    <suggestions>
      <suggestion>aardvark</suggestion>
      <suggestion>aardvarks</suggestion>
      <suggestion>aardvark's</suggestion>

    </suggestions>
  </word>
  <word>
    <text>shoes</text>
    <correct/>
  </word>
</words>

<error>an error ocurred while intializing pspell</error>

If you find a bug or would like to make a suggestion, please e-mail me.

View source. I know it's messy. This script is licensed under the GPL.