optimization of “_morph” function

[--deleted--]

At the moment, the result of the command “analyze” is stored as a string for each words in a global array, and we split it each times we need to know the morphologies of a word. Instead of doing it repeatedly, we can do it once for each word.

def _morph(rLoc, word, pattern, onlyaffix):
    global analyses
    if word == None:
        return None
    if word not in analyses:
        x = spellchecker.spell(u"<?xml?><query type='analyze'><word>" + word + "</word></query>", rLoc, ())
        if not x:
            return None
        t = x.getAlternatives()
        if not t:
            t = [""]
        analyses[word] = t[0]
    a = analyses[word].split("</a>")[:-1] <----- here is the job done too many times
    [...]

New function :

def _morph(rLoc, word, pattern, onlyaffix):
    global analyses
    if word == None:
        return None
    if word not in analyses:
        x = spellchecker.spell(u"<?xml?><query type='analyze'><word>" + word + "</word></query>", rLoc, ())
        if not x:
            return None
        t = x.getAlternatives()
        if not t:
            t = [""]
        analyses[word] = t[0].split("</a>")[:-1] <----- here is the job done once for each word
    a = analyses[word]
    [...]