python整合连续数字的练习，包含itertools\groupby用法 

#汉字数字转阿拉伯数字
 1 class ConvertNum:
 2     def __init__(self,cnNum):
 3     　　self.dict =   {u'零':0,u'一':1,u'二':2,u'三':3,u'四':4,u'五':5,u'六':6,u'七':7,u'八':8,u'九':9,u'十':10,u'百':100,u'千':1000,u'万':10000}
 4     　　self.cnNum = cnNum
 5 
 6     def convert(self):
 7         count = 0
 8         result = 0
 9         tmp = 0
10 
11         while count < len(self.cnNum):
12         tmpChr = self.cnNum[count:count+1]
13         tmpNum = self.dict[tmpChr]
14 
15         if tmpNum == 10000:
16             result = result + tmp
17             result = result * tmpNum
18         elif tmpNum >= 10:
19             if tmp==0:
20             　　tmp = 1
21             result = result + tmpNum * tmp
22             tmp = 0
23         else:
24             tmp = tmpNum
25             count = count + 1
26             result = result + tmp
27         return result                    
 1 import re
 2 from convertNum import ConvertNum
 3 from itertools import *

 4 list_a = [u'中华一村',u'中华二村',u'中华三村',u'中华十二村',u'中华五村',u'中点开',u'大地三街',u'大地二街',u'中华七村',u'中华八村',u'中华九村']
 6     patternCombineHz = re.compile(ur'(.*?)([一二三四五六七八九十]+)([村街巷连])$')
 7     l,dicts = [],[]
 8     for item in list_a:
 9         m = patternCombineHz.search(item)     #获取数字信息
10         if m:
11             dict = {'headFootText':(m.group(1),m.group(3)),'numText':m.group(2)}
12             dicts.append(dict)
13         else:
14             l.append(item)
15     dictHead = {}
16     for dict in dicts:     #生成头尾为key，数字为value的字典
17         if dictHead.has_key(dict.get('headFootText')):
18             dictHead[dict.get('headFootText')].append(dict.get('numText'))
19         else:
20             dictHead[dict.get('headFootText')] = [dict.get('numText')]
21     for key,value in dictHead.items():
22         dictNum = {}
23         for i in value:
24             dictNum[ConvertNum(i).convert()] = i
25         sortalbNum = sorted([ConvertNum(i).convert() for i in value])   #汉字转换阿拉伯数字后排序
26         newNum = []
27         #排序后列表元素减去索引号，判断是否等差，并分组生成迭代器，每一组表示连续数字
28         for k,g in groupby(enumerate(sortalbNum),lambda (i,x):x-i):
29             gb = [t[1] for t in g]
30             mingb,maxgb = min(gb),max(gb)
31             if maxgb-mingb>1:
32                 combineStr = u'~'.join(map(lambda num:dictNum.get(num),[mingb,maxgb]))
33                 newNum.append(combineStr)
34             else:
35                 newNum.extend(map(lambda num:dictNum.get(num),gb))
36         # newNum = [dictNum.get(num) for num in sortalbNum]
37         if len(newNum) == 1 and not u'~' in newNum[0]:
38             combineAddr = u'%s%s%s' % (key[0],u'、'.join(newNum),key[1])
39         else:
40             combineAddr = u'%s（%s）%s' % (key[0],u'、'.join(newNum),key[1])
41         l.append(combineAddr)
42 
43     for i in l:
44         print i

output:>>>>>>>>

中点开
中华（一~三、五、七~九、十二）村
大地（二、三）街

posted @ 2014-08-11 15:21  一个坟头  阅读(3160)  评论(0)  编辑  收藏  举报