|
@@ -4,6 +4,26 @@ import codecs
|
|
|
import dataset
|
|
|
|
|
|
|
|
|
+
|
|
|
+def extract_comp(fname):
|
|
|
+ resultstr=""
|
|
|
+ fr=codecs.open(fname,'r','utf-8')
|
|
|
+ lines=fr.readlines()
|
|
|
+ inloop=False
|
|
|
+ for l in lines:
|
|
|
+ if inloop:
|
|
|
+ if '<script async defer crossorigin="anonymous" src="https://connect.' in l:
|
|
|
+ break
|
|
|
+ resultstr+=l+"\n"
|
|
|
+
|
|
|
+ if 'row justify-content-between mt-5' in l:
|
|
|
+ inloop=True
|
|
|
+ resultstr+=l+"\n"
|
|
|
+ continue
|
|
|
+ return resultstr
|
|
|
+
|
|
|
+
|
|
|
+#<script async defer crossorigin="anonymous" src="https://connect.facebook.net/
|
|
|
def proc_file(fname,fname2):
|
|
|
fr=codecs.open(fname,'r','utf-8')
|
|
|
fw=codecs.open(fname2,'w','utf-8')
|
|
@@ -12,6 +32,7 @@ def proc_file(fname,fname2):
|
|
|
# <div class="row mt-5" id="row">
|
|
|
if 'div' in l and 'class' in l and 'row mt-5' in l and 'id="row"' in l:
|
|
|
fw.write('<div class="row mt-5" id="row_choz8">\n')
|
|
|
+ fw.write('</div>\n')
|
|
|
|
|
|
if 'nav-link' in l and '設計師專欄' in l and 'class' in l:
|
|
|
fw.write('<a class="nav-link " onclick="forum();">設計師專欄</a>')
|
|
@@ -47,6 +68,8 @@ for c in cursor:
|
|
|
os.system('curl https://hhh.com.tw/designers/cases/'+str(did)+'/1-page/new-sort/ -o orig.html')
|
|
|
os.system('curl https://hhh.com.tw/designers/details/'+str(did)+'/ -o comp.html')
|
|
|
|
|
|
+ result=extract_comp(curdir+"/comp.html")
|
|
|
+ print(result)
|
|
|
proc_file(curdir+"/orig.html",curdir+"/index.html")
|
|
|
break
|
|
|
|