| 17,23 → 17,23 |
|---|
| infix_1 = u"" |
| infix_2 = u"" |
| for eachword in strings: |
| regex = re.sub(" ","[^ ]* [^ ]*",eachword) |
| regex = re.sub("^","[^ ]*",regex) |
| regex = re.sub("$","[^ ]*",regex) |
| regex = re.sub("<0><1>","[^ ]*",regex) |
| regex = re.sub("<2>","[^ ]*",regex) |
| regex = re.sub(u" ",u"[^ ]* [^ ]*",eachword) |
| regex = re.sub(u"^",u"[^ ]*",regex) |
| regex = re.sub(u"$",u"[^ ]*",regex) |
| regex = re.sub(u"<0><1>",u"[^ ]*",regex) |
| regex = re.sub(u"<2>",u"[^ ]*",regex) |
| if re.match(regex,original): |
| realword = eachword |
| break |
| if realword == u"": |
| return ["[" + original + "]",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""] |
| return [u"[" + original + u"]",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""] |
| else: |
| if re.search("<",realword): |
| beginning = re.sub("<0><1>.*",u"",realword) |
| middle = re.sub(".*<0><1>(.*)<2>.*","\\1",realword) |
| end = re.sub(".*<2>",u"",realword) |
| infix01 = re.sub(".*?" + re.sub("<0><1>","([^ ]*)",re.sub("<2>","[^ ]*",realword)) + ".*?","\\1",original) |
| infix_2 = re.sub(".*?" + re.sub("<2>","([^ ]*)",re.sub("<0><1>","[^ ]*",realword)) + ".*?","\\1",original) |
| if re.search(u"<",realword): |
| beginning = re.sub(u"<0><1>.*",u"",realword) |
| middle = re.sub(u".*<0><1>(.*)<2>.*",ur"\1",realword) |
| end = re.sub(u".*<2>",u"",realword) |
| infix01 = re.sub(u".*?" + re.sub(u"<0><1>",u"([^ ]*)",re.sub(u"<2>",u"[^ ]*",realword)) + u".*?",ur"\1",original) |
| infix_2 = re.sub(u".*?" + re.sub(u"<2>",u"([^ ]*)",re.sub(u"<0><1>",u"[^ ]*",realword)) + u".*?",ur"\1",original) |
| for eachinfix in infixes0: |
| if infix01.startswith(eachinfix): |
| infix0 = eachinfix |
| 47,15 → 47,15 |
|---|
| if infix_1.startswith(eachinfix): |
| infix1 = eachinfix |
| infix_1 = infix_1[len(eachinfix):] |
| if infix_1 != "": |
| if re.search("<0><1><2>",realword): |
| if infix_1 != u"": |
| if re.search(u"<0><1><2>",realword): |
| infix_2 = infix_1 |
| else: |
| return ["[" + original + "]",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""] |
| return [u"[" + original + u"]",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""] |
| gotinfix1 = True |
| break |
| if gotinfix1 == False: |
| if re.search("<0><1><2>",realword): |
| if re.search(u"<0><1><2>",realword): |
| if infix_1 == u"": |
| infix_2 = infix_1 |
| infix1 = u"" |
| 62,7 → 62,7 |
|---|
| elif infix_1 == u"": |
| infix1 = u"" |
| else: |
| return ["[" + original + "]",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""] |
| return [u"[" + original + u"]",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""] |
| gotinfix2 = False |
| for eachinfix in infixes2: |
| if infix_2.startswith(eachinfix): |
| 73,10 → 73,10 |
|---|
| if infix_2.startswith(end): |
| suffixes = infix2[len(end) - 1:] + end |
| elif infix_2 == u"": |
| infix2 = "" |
| infix2 = u"" |
| else: |
| return ["[" + original + "]",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""] |
| # print "0" + unicode(infix0) + " 1" + unicode(infix1) + " 2" + unicode(infix2) |
| return [u"[" + original + u"]",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""] |
| # print u"0" + unicode(infix0) + u" 1" + unicode(infix1) + u" 2" + unicode(infix2) |
| return [realword,infix0,infix1,infix2,u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""] |
| else: |
| return [realword,u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u"",u""] |