def misspelledCheckIntegrity(misspelledList):
wordRe = re.compile(wordMatchRe)
for i, word in enumerate(misspelledList):
try:
if not wordRe.match(word):
raiseError(misspelledFormattingError + str(i))
except TypeError:
raiseError(misspelledFormattingError + str(i))
def makeJsonReq(self, url, postdata):
"""
postdata: dict
"""
cj, content = self.makeHttpReq(url, postdata)
ret = json.loads(content)
if ret["error"]:
errors.raiseError(ret["error"])
return ret
def getMisspelledWords(misspelledFilename):
#read in the input file
try:
f = open(misspelledFilename, 'r')
except IOError:
raiseError(misspelledFileNotFoundError)
misspelled = f.read()
misspelledWordsList = misspelled.split('\n')
verify.misspelledCheckIntegrity(misspelledWordsList)
return misspelledWordsList
def getWordFrequencies(frequenciesFilename):
try:
f = open(frequenciesFilename, 'r')
except IOError:
raiseError(frequenciesFileNotFoundError)
frequencies = f.read()
frequencyList = frequencies.split('\n')
for i, item in enumerate(frequencyList):
frequencyList[i] = item.split(',')
verify.frequenciesCheckIntegrity(frequencyList)
sorted(frequencyList, key=lambda item: int(item[1]))
return frequencyList
def frequenciesCheckIntegrity(frequencyList):
wordRe = re.compile(wordMatchRe)
for i, item in enumerate(frequencyList):
#check that list is formatted correctly
if not isinstance(item, list) or len(item) != 2:
raiseError(frequenciesFormattingError + str(i))
word = item[0]
frequency = item[1]
#check that first entry is a word
if not wordRe.match(word):
raiseError(frequenciesFormattingError + str(i))
#check that second entry is an int
try:
int(frequency)
except ValueError:
raiseError(frequenciesFormattingError + str(i))
def onAnyEvent(self, *args, **kw):
sid = syncer_tls.sid
if not self.validate(sid):
errors.raiseError(errors.sessionnotfound)
return True
def formatSuggestion(word, suggestedWords):
return "* " + word + ": " + str(suggestedWords) + "\n"
def suggestWords(misspelledWords, wordFrequencies, suggestionsFilename):
#open suggestions file
suggestions = open(suggestionsFilename, 'w')
for word in misspelledWords:
#find an array of the suggested words sorted by frequency
suggestedWords = getSuggestedWords(word, wordFrequencies)
suggestions.write(formatSuggestion(word, suggestedWords))
if __name__ == "__main__":
try:
misspelledFilename = str(sys.argv[1])
except IndexError:
raiseError(misspelledFileNotEnteredError)
#try to read in the misspelled words file
misspelledWords = parse.getMisspelledWords(misspelledFilename)
wordFrequenciesFilename = "word_frequency.csv"
#try to read in the word frequencies file
wordFrequencies = parse.getWordFrequencies(wordFrequenciesFilename)
suggestionsFilename = "suggestions.txt"
suggestWords(misspelledWords, wordFrequencies, suggestionsFilename)
