def web_getXTxt001k(bs):
x10, tss = bs.find_all('p'), ''
if x10 == []: x10 = bs.find_all('div')
for x in x10:
if x != None:
css = x.text
if tss.find(css) == -1:
css = zstr.str_fltHtm(css)
if len(css) > 10:
tss = ''.join([tss, '\n', css])
#print(css);print('csn',len(css))
#
#tss=tss+'\np'
if len(tss) < 200: tss = ''
return tss
def web_getXTxt001div(bs,claSgn):
x10,tss=bs.find_all('div'),''
for x in x10:
#print('@x',x)
if x!=None:
x2=x.find('div',class_=claSgn)
else:x2=None
#
if x2!=None:
css=x2.text
if tss.find(css)==-1:
css=zstr.str_fltHtm(css)
tss=''.join([tss,'\n',css])
#print("@::",css,'\ncsn,',len(css))
#print("@::",x2)
#
#tss=tss+'\n'+claSgn
if len(tss)<200:tss=''
return tss
def fb_gid_get4htm(htm):
bs = BeautifulSoup(htm, 'html5lib') # 'lxml'
df = pd.DataFrame(columns=tfsys.gidSgn, dtype=str)
ds = pd.Series(tfsys.gidNil, index=tfsys.gidSgn, dtype=str)
#---1#
# zsys.bs_get_ktag_kstr=['align','right']
zsys.bs_get_ktag_kstr = 'matchid'
x10 = bs.find_all(zweb.bs_get_ktag)
for xc, x in enumerate(x10):
ds = pd.Series(tfsys.gidNil, index=tfsys.gidSgn, dtype=str)
#print('\n@x\n',xc,'#',x.attrs)
home_team = x.find(attrs={'align': 'right'})
guest_team = x.find(attrs={'align': 'left'})
ds['gid'] = home_team.find('span')['id'].split('_')[1]
ds['gset'] = x['gamename']
ds['mplay'] = home_team.text
ds['gplay'] = guest_team.text
clst = zt.lst4objs_txt2(zstr.str_fltHtm(x.text), ['\n', '\t', '%'])
score = clst[5].split('-')
if score[0]:
ds['qj'] = score[0]
if score[1]:
ds['qs'] = score[1]
date = x.find(attrs={'title': re.compile("截止时间:*")})
if date.text == '完场':
ds['kend'] = '1'
ds['tsell'] = date['title'].split('：')[1]
ds['tplay'] = ds['tsell'].split(' ')[0]
ds['tweek'] = x['name']
kwin = fb_kwin4qnum(int(ds['qj']), int(ds['qs']))
ds['kwin'] = str(kwin)
#
df = df.append(ds.T, ignore_index=True)
#---5#
df = df[df['gid'] != '-1']
return df