X-Git-Url: https://git.korewanetadesu.com/?p=python-collate.git;a=blobdiff_plain;f=collate%2F_abcollator.py;h=12575eaed844603f039009a62262fe66b8d29ace;hp=fdd7783fc1eceee8382bd46856e8af7ac787e468;hb=92fc0878bc7b75741a3434d17310e390a9304e70;hpb=7dc546d67cd996c0a54b07261ab1b7bcdf4019ce diff --git a/collate/_abcollator.py b/collate/_abcollator.py index fdd7783..12575ea 100644 --- a/collate/_abcollator.py +++ b/collate/_abcollator.py @@ -1,18 +1,33 @@ -import collate._strings +import collate.strings class Collator(object): + encoding = "ascii" + def cmp(self, string1, string2): """Return negative if a < b, zero if a == b, positive if a > b.""" return cmp(self.key(string1), self.key(string2)) - def sortemekey(self, string, invalid=float('inf')): - keys = [] - for sorteme in collate._strings.sortemes(string): - num, alpha = collate._strings.numeric(sorteme, invalid) - if num == invalid: - keys.append(self.key(alpha)) - else: - keys.append(num) + def key(self, string): + return string + + def words(self, string): + """Split the string along word boundries.""" + if isinstance(string, str): + string = string.decode(self.encoding, 'replace') + return string.split() + + def sortemekey(self, string): + """Return a key based on sortemes of a string. + + If the string is a str instance, it is decoded to a unicode + instance according to the 'encoding' attribute of the + Collator. + """ + if isinstance(string, str): + string = string.decode(self.encoding, 'replace') + # Shove the sortkeyed original string on the end to resolve # ties intelligently. - return (keys, self.key(string)) + return (collate.strings.sortemes(string, self.key), + self.key(string)) +