| 1 | def opendictionaryfile(file): |
| 2 | """ Every line from file Returned as dictionary type""" |
| 3 | dictionary= {} |
| 4 | file = open(file) |
| 5 | for line in file: |
| 6 | dictionary[line]=0 |
| 7 | |
| 8 | return dictionary |
| 9 | |
| 10 | def checkifvalidword(word): |
| 11 | """ Checks if word is a valid word. |
| 12 | Any nonalphabet or ' character in a word is considered not a word. |
| 13 | if word return 1 else 0""" |
| 14 | import re |
| 15 | pat = re.compile('[^a-zA-Z\']') |
| 16 | if pat.search(word) == None: |
| 17 | return 1 |
| 18 | else: |
| 19 | return 0 |
| 20 | |
| 21 | def printmistakes(misspelledwords): |
| 22 | "Prints Spelling mistakes in alphabetical order and number of occurances""" |
| 23 | alphalist = [] |
| 24 | for key in misspelledwords: |
| 25 | alphalist.append(key) |
| 26 | alphalist.sort() |
| 27 | for key in alphalist: |
| 28 | print key, |
| 29 | print ' (' + str(misspelledwords.get(key)) + ')' |
| 30 | |
| 31 | def spellcheckfile(txtfile, dictfile): |
| 32 | """ Spell check txtfile with dictionary of words specifed in dictfile |
| 33 | Prints Misspelled words and # occurances """ |
| 34 | dictionary = opendictionaryfile(dictfile) |
| 35 | misspelledwords = {} |
| 36 | txtfile = open(txtfile) |
| 37 | for line in txtfile: |
| 38 | for word in line.split(' '): |
| 39 | if checkifvalidword(word) == 1: |
| 40 | if (word + '\n' in dictionary) == 0: |
| 41 | misspelledwords[word] = misspelledwords.get(word, 0) + 1 |
| 42 | printmistakes(misspelledwords) |