useragent解析代码

本代码采用关键字匹配的方法，过滤出国内主流的浏览器以及对应的内核、操作系统以及硬件类型（主要针对Android手机）

说明：由于好多浏览器试图去兼容其他类型的浏览器，所以会在UA中写血多其他兼容的浏览器信息，

所以本代码中对这种类型的浏览器做了一下判断，选择了最外层的浏览器：

比如QQBrowser使用IE内核；而Maxthon兼容Chrome，Chrome兼容safari；

内核也是如此。

class UAParse:
        def  __init__(self):
                self.keywordsmap={'TencentTraveler':1,'QQBrowser':2,'Maxthon':3,'BIDUBrowser':4,'360SE':5,'TheWorld':6,'qihu theworld':7,'SE 2.X':8
                        ,'Firefox':9,'Safari':10,'Chrome':11,'MSIE':12,'Opera':13
                        ,'BdMobile':14,'MQQBrowser':15,'UCWEB':16,'NokiaBrowser':17,'UC':18
                        ,'Iceweasel':19,'Mobile':20,'K-MeleonCCFME':21
                        # kernel
                        ,'Trident':101,'AppleWebKit':102,'Presto':103,'Gecko':104,'KHTML':105
                        #OS
                        ,'SymbianOS':201,'Mac OS X':202,'Android':203,'Windows NT':204,'Linux':205,
                }
                self.pattern='(%s)([0-9/. ]*)'%('|'.join(self.keywordsmap.keys()))
                self.cpat=re.compile(self.pattern)

        def uaparse(self,useragent):
                browser=''
                print useragent
                midx=[100,200,300]
                info={'brw':'','brwv':'','knl':'','knlv':'','os':'','osv':'','hard':''}
                for m in self.cpat.finditer(useragent):
                        if not m.group(1):
                                continue
                        cidx=self.keywordsmap[m.group(1)]
                        if cidx<100:
                                if midx[0]>cidx:
                                        midx[0]=cidx
                                        info['brw']=m.group(1)
                                        info['brwv']=m.group(2)
                        elif cidx<200:
                                if midx[1]>cidx:
                                        midx[1]=cidx
                                        info['knl']=m.group(1)
                                        info['knlv']=m.group(2)
                        elif cidx<300:
                                if midx[2]>cidx:
                                        midx[2]=cidx
                                        info['os']=m.group(1)
                                        info['osv']=m.group(2)
                                        if info['os']=='Android':
                                                pos=useragent[m.end(2):].find(' Build')
                                                if pos>0:
                                                        hard = re.search('([0-9a-zA-Z_-]+)( Build/)',useragent[useragent.rfind(';'):])
                                                        if hard:
                                                                info['hard']=hard.group(1)
                                                else:
                                                        hard = re.search('; ([0-9a-zA-Z_ -]+)([0-9a-zA-Z_ /-]+\))',useragent[useragent.rfind(';'):])
                                                        if hard:
                                                                info['hard']=hard.group(1)
                                        elif info['os']=='Mac OS X':
                                                hard = re.search('\( *([a-zA-Z]+) *;',useragent[:m.start(1)])
                                                if hard:
                                                        info['hard']=hard.group(1)
                                                version = re.search('OS ([0-9_]+)',useragent[:m.start(1)])
                                                if version:
                                                        info['osv'] = version.group(1)
                                        elif info['os']=='Linux':
                                                version = re.search('(Linux) ([0-9a-zA-Z_-]+)',useragent[m.start(1):])
                                                if version:
                                                        info['osv'] = version.group(2)

                info['brwv']=info['brwv'].strip(' /')
                info['knlv']=info['knlv'].strip(' /')
                info['osv']=info['osv'].strip(' /')
                #print info
                print '%s=>%s\t%s=>%s\t%s=>%s\t%s'%(info['brw'],info['brwv'],info['knl'],info['knlv'],info['os'],info['osv'],info['hard'])


def test():
        obj = UAParse()
        obj.uaparse('Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/535.12 (KHTML, like Gecko) Maxthon/3.4.1.1000 Chrome/18.0.966.0 Safari/535.12')
        obj.uaparse('MQQBrowser/3.7/Mozilla/5.0 (Linux; U; Android 2.3.5; zh-cn; GT-N7000 Build/GINGERBREAD) AppleWebKit/533.1 (KHTML, like Gecko) Version/4.0 Mobile Safari/533.1')
        obj.uaparse('MQQBrowser/2.7 Mozilla/5.0 (iPad; U; CPU OS 4_3_5 like Mac OS X; zh-cn) AppleWebKit/533.17.9 (KHTML, like Gecko) Mobile/8L1 Safari/7534.48.3')
        obj.uaparse('Mozilla/5.0 (X11; U; Linux mips64; zh-CN; rv:1.9.0.11) Gecko/2009061212 Iceweasel/3.0.6 (Debian-3.0.6-1)')