Created 7 October 2009

This was a presentation I gave at the DevDays Boston conference in 2009. I don’t have much text to go with it, but these are the slides. If you need them larger, there is a zip file of .pngs.

~ Spell corrector ~

Joel’s suggestion was to explain Peter Norvig’s Spell Corrector line-by-line. I’ve made a few small edits for expository purposes.

Here’s the code:

import re, collections

def words(text):
    return re.findall('[a-z]+', text.lower()) 

def train(words):
    model = collections.defaultdict(int)
    for w in words:
        model[w] += 1
    return model

NWORDS = train(words(file('big.txt').read()))

alphabet = 'abcdefghijklmnopqrstuvwxyz'

def edits1(word):
    s = [(word[:i], word[i:]) for i in range(len(word) + 1)]
    deletes    = [a + b[1:] for a, b in s if b]
    transposes = [a + b[1] + b[0] + b[2:] for a, b in s if len(b)>1]
    replaces   = [a + c + b[1:] for a, b in s for c in alphabet if b]
    inserts    = [a + c + b     for a, b in s for c in alphabet]
    return set(deletes + transposes + replaces + inserts)

def known_edits2(word):
    return set(e2 for e1 in edits1(word) for e2 in edits1(e1) if e2 in NWORDS)

def known(words):
    return set(w for w in words if w in NWORDS)

def correct(word):
    candidates = known([word]) or known(edits1(word)) or known_edits2(word) or [word]
    return max(candidates, key=NWORDS.get)

if __name__ == '__main__':
    import sys
    for w in sys.argv[1:]:
        print "%s -> %s" % (w, correct(w))

~ Templating ~

Here’s the completed Nango code:

import re

class Nango1(object):

    def __init__(self, text):
        # Convert {{foo}} into %(foo)s
        text = re.sub(r"{{(.+?)}}", r"%(\1)s", text)
        self.text = text

    def render(self, data):
        return self.text % NangoContext(data)


class NangoContext(object):

    def __init__(self, data):
        self.data = data

    def __getitem__(self, key):
        if "." in key:
            dots = key.split('.')
            value = self[dots[0]]
            for dot in dots[1:]:
                try:
                    value = getattr(value, dot)
                except AttributeError:
                    value = value[dot]
                if callable(value):
                    value = value()
        else:
            value = self.data[key]
        return value