def purify_html(request):
"""
过滤掉多余css,修正图片一些属性,替换超链接
:param request:
:return:
"""
html = request.GET.get('html', None)
base_url = request.GET.get('baseUrl', None)
css_path = request.GET.get('cssPath', None)
# 若果domain已经在数据库了就找到，没有就新建
domain_name = get_top_domain(base_url)
site_set = Site.objects.filter(domain_name=domain_name)
if site_set.exists():
s = site_set[0]
else:
s = Site()
s.domain_name = domain_name
# 保存二级页面的主要显示区域的css
if css_path.strip():
if not s.second:
second = SecondModifications()
second.css_path = css_path.strip()
second.save()
s.second = second
else:
second = s.second
second.css_path = css_path.strip()
second.save()
s.save()
res = purify(html, base_url)
if not res:
return HttpResponse(status=404)
return HttpResponse(res)
def save(request):
"""
存储导航项,所选区域的css路径
:param request:
:return:
"""
if request.method == "POST":
kind = request.POST.get('type', 'contents')
url = request.POST.get('url', None)
# 如果抓取二级页面内容
if kind == 'contents':
navs_str = request.POST.get('navs', None)
if navs_str:
navs = json.loads(navs_str, 'utf-8')
domain_name = get_top_domain(url)
site_set = Site.objects.filter(domain_name=domain_name)
if site_set.exists():
s = site_set[0]
s.nav_first.clear()
else:
s = Site()
s.domain_name = domain_name.strip()
s.save()
for nav in navs:
if nav['name'].strip() and nav['url'].strip():
n = NavBlock()
n.father_id = 0
n.name = nav['name'].strip()
n.url = nav['url'].strip()
n.save()
for second in nav['secondNavs']:
sn = NavBlock()
sn.father_id = n.id
sn.name = second['name'].strip()
sn.url = second['url'].strip()
sn.save()
# 添加二级nav
s.nav_first.add(sn)
# 添加一级nav
s.nav_first.add(n)
return HttpResponse(status=200)
# 如果抓取图片
elif kind == 'imgs':
url_str = request.POST.get('imgUrl', None)
img_urls = json.loads(url_str, 'utf-8')
domain_name = get_top_domain(url)
site_set = Site.objects.filter(domain_name=domain_name)
if site_set.exists():
s = site_set[0]
if s.sliders:
s.sliders.clear()
else:
s = Site()
s.domain_name = domain_name.strip()
s.save()
for url in img_urls:
if url.strip():
slider = Slider()
slider.url = url.strip()
slider.save()
s.sliders.add(slider)
return HttpResponse(status=200)
elif kind == 'css':
domain_name = get_top_domain(url)
css_path = request.POST.get('cssPath', None)
site_set = Site.objects.filter(domain_name=domain_name)
if site_set.exists():
s = site_set[0]
else:
s = Site()
s.domain_name = domain_name
# 保存二级页面的主要显示区域的css
if css_path.strip():
if not s.second:
second = SecondModifications()
second.css_path = css_path.strip()
second.save()
s.second = second
else:
second = s.second
second.css_path = css_path.strip()
second.save()
s.save()
return HttpResponse(status=200)
elif kind == 'status':
uid = request.POST.get('id', None)
if uid:
site_set = Site.objects.filter(id=uid)
if site_set.exists():
site = site_set[0]
site.is_active = not site.is_active
site.save()
return HttpResponse(status=200)
elif request.method == "DELETE":
request_delete = QueryDict(request.body)
kind = request_delete.get('type', 'imgs')
url = request_delete.get('url', None)
if kind == 'imgs':
domain_name = get_top_domain(url)
site_set = Site.objects.filter(domain_name=domain_name)
if site_set.exists():
s = site_set[0]
s.sliders.clear()
return HttpResponse(status=204)
else:
return HttpResponse(status=404)
return HttpResponse(status=403)