Question

I want to remove all strange characters from a string to make it “url safe”. Therefor, I have a function that goes like this:

def urlize(url, safe=u''):
   intab =  u"àáâãäåòóôõöøèéêëçìíîïùúûüÿñ" + safe
   outtab = u"aaaaaaooooooeeeeciiiiuuuuyn" + safe
   trantab = dict((ord(a), b) for a, b in zip(intab, outtab))
   return url.lower().translate(trantab).strip()

This works just great, but now I want to reuse that funcion to allow special characters. For example, the quotation mark.

urlize(u'This is sóme randóm "text" that í wánt to process',u'"')

…and that throws the following error:

Traceback (most recent call last):
  File "<input>", line 1, in <module>
TypeError: expected a character buffer object

I have tried, but did not work:

urlize(u'text',u'\"')
intab =  u"àáâãäåòóôõöøèéêëçìíîïùúûüÿñ%s" , safe

–EDIT–
The full function looks like this

def urlize(url, safe=u''):

    intab =  u"àáâãäåòóôõöøèéêëçìíîïùúûüÿñ" + safe
    outtab = u"aaaaaaooooooeeeeciiiiuuuuyn" + safe
    trantab = dict((ord(a), b) for a, b in zip(intab, outtab))
    translated_url = url.lower().translate(trantab).strip()

    pos = 0
    stop = len(translated_url)
    new_url= ''
    last_division_char = False

    while pos < stop:
        if not translated_url[pos].isalnum() and translated_url[pos] not in safe:
            if (not last_division_char) and (pos != stop -1):
                new_url+='-'
                last_division_char = True
        else:
            new_url+=translated_url[pos]
            last_division_char = False
        pos+=1

    return new_url

–EDIT– Goal

What I want is to normalize text so that I can put it on the url myself, and use it like an Id. For example, if I want to show the products of a category, I’d rather put “ninos-y-bebes” instead of “niños-y-bebés” (spanish for kids and babies). I really don’t want all the áéíóúñ (which are the special characters in spanish) in my url, but I don’t want to get rid of them either. That’s why I would like to replace all characters that looks the same (not 100% all of them, I dont care) and then delete all non alfanumeric characters left.

Answer 1

The unidecode module is a safer option (it will handle other special simbols like “degree”):

>>> from unidecode import unidecode
>>> s = u'This is sóme randóm "text" that í wánt to process'
>>> unidecode(s)
'This is some random "text" that i want to process'
>>> import urllib
>>> urllib.urlencode(dict(x=unidecode(s)))[2:]
'This+is+some+random+%22text%22+that+i+want+to+process'

[ update ]

i think i’m already doing that -> u”aaaaaaooooooeeeeciiiiuuuuyn” – Marco Bruggmann

Fair enough, if you are willing to keep track of every unicode character out there for your translation table (accented characters are not the only issues, there are a whole lot of symbols to rain on your parade).

Worst, many unicode symbols may be visually identical to their ASCII counterparts, leading to hard to diagnose errors.

[ update ]

What about something like:

>>> safe_chars = 'abcdefghijklmnopqrstuvwxyz01234567890-_'
>>> filter(lambda x: x in safe_chars, "i think i'm already doing that")
'ithinkimalreadydoingthat'

[ update ]

@Daenyth I tried it, but I only get errors: from urllib import urlencode => urlencode(‘google.com/’;) => TypeError: not a valid non-string sequence or mapping object – Marco Bruggmann

The urlencode function is intended to produce QUERYSTRING formated output (a=1&b=2&c=3). It expects key/value pairs:

>>> urllib.urlencode(dict(url='google.com/'))
'url=google.com%2F'

>>> help(urllib.urlencode)
Help on function urlencode in module urllib:

urlencode(query, doseq=0)
    Encode a sequence of two-element tuples or dictionary into a URL query string.

    If any values in the query arg are sequences and doseq is true, each
    sequence element is converted to a separate parameter.

    If the query arg is a sequence of two-element tuples, the order of the
    parameters in the output will match the order of parameters in the
    input.
(END)

[ update ]

That will works without a doubt, but what I want is to normalize text so that I can put it on the url myself, and use it like an Id. For example, if I want to show the products of a category, I’d rather put “ninos-y-bebes” instead of “niños-y-bebés” (spanish for kids and babies). I really don’t want all the áéíóúñ (which are the special characters in spanish) in my url, but I don’t want to get rid of them either. That’s why I would like to replace all characters that looks the same (not 100% all of them, I dont care) and then delete all non alfanumeric characters left.

Ok, Marco, what you want is a routine to create the so called slugs, isn’t it?

You can do it in one line:

>>> s = u'This is sóme randóm "text" that í wánt to process'
>>> allowed_chars = 'abcdefghijklmnopqrstuwvxyz01234567890'
>>> ''.join([ x if x in allowed_chars else '-' for x in unidecode(s.lower()) ])
u'this-is-some-random--text--that-i-want-to-process'
>>> s = u"Niños y Bebés"
>>> ''.join([ x if x in allowed_chars else '-' for x in unidecode(s.lower()) ])
u'ninos-y-bebes'
>>> s = u"1ª Categoria, ½ docena"
>>> ''.join([ x if x in allowed_chars else '-' for x in unidecode(s.lower()) ])
u'1a-categoria--1-2-docena'