def taobao_spider():
Commodity_base_info = commodity_base_info()
commodity_infos = Commodity_base_info.query.all()
for commodity in commodity_infos:
print(commodity.commodity_name)
search_taobao(commodity.commodity_name, commodity.commodity_type,
commodity.commodity_brand)
def zhihu_review_spider():
Commodity_base_info = commodity_base_info()
Commodity_infos = Commodity_base_info.query.all()
file = open('APP/spider/zhihu_{}.txt'.format(time.strftime('%m%d')), 'x')
for commodity in Commodity_infos:
print(commodity.commodity_name)
text = get_x_zse_86(commodity.commodity_name)
# print(text)
file.write(commodity.commodity_name + ',' + text[2:] + '/n')
zhihu_search_spider(commodity.commodity_type,
commodity.commodity_brand,
commodity.commodity_name, text)
file.close()
def index():
commodity = commodity_base_info()
commodity_list = []
for info in commodity.query.limit(3):
info = vars(info)
# print(info)
commodity_type = info.pop('commodity_type')
name = info.pop('commodity_name')
price = info.pop('commodity_base_price')
img_path = info.pop('img_path')
infos = info.pop('info').replace('>', '').split(';')
# print(img_path)
base_info = {}
n = 0
for i in infos:
if n > 3:
break
n += 1
# print(i)
i = i.split(':')
if len(i) < 2:
continue
# print(i[1])
text = ''
texts = i[1].split(',')
if len(texts) >= 3:
texts = texts[:-2]
# print(texts)
# print(type(text))
# print(text)
for c in texts:
if c != '':
text += c + ';'
base_info[i[0]] = text.replace('，;', '')
info = {
'name': name,
'price': price,
'img_path': img_path,
'commodity_type': commodity_type,
'base_info': base_info
}
# print(info['img_path'])
commodity_list.append(info)
return render_template('index.html', index_list=commodity_list)
def base_infoz():
base_infos = []
base_urls = [
'http://detail.zol.com.cn/cell_phone_index/subcate57_613_list_1.html', # 手机
'http://detail.zol.com.cn/cell_phone_index/subcate57_1795_list_1.html',
'http://detail.zol.com.cn/cell_phone_index/subcate57_1673_list_1.html',
'http://detail.zol.com.cn/cell_phone_index/subcate57_544_list_1.html',
'http://detail.zol.com.cn/cell_phone_index/subcate57_98_list_1.html',
'http://detail.zol.com.cn/cell_phone_index/subcate57_34645_list_1.html',
'http://detail.zol.com.cn/notebook_index/subcate16_160_list_1.html', # 笔记本
'http://detail.zol.com.cn/notebook_index/subcate16_21_list_1.html',
'http://detail.zol.com.cn/notebook_index/subcate16_544_list_1.html',
'http://detail.zol.com.cn/notebook_index/subcate16_223_list_1.html',
'http://detail.zol.com.cn/notebook_index/subcate16_227_list_1.html',
'http://detail.zol.com.cn/notebook_index/subcate16_613_list_1.html',
'http://detail.zol.com.cn/notebook_index/subcate16_1191_list_1.html',
'http://detail.zol.com.cn/notebook_index/subcate16_34645_list_1.html',
'http://detail.zol.com.cn/cpu/amd/', # CPU
'http://detail.zol.com.cn/cpu/intel/',
'http://detail.zol.com.cn/motherboard/gigabyte/', # 主板
'http://detail.zol.com.cn/motherboard/asus/',
'http://detail.zol.com.cn/vga/colorful/', # 显卡
'http://detail.zol.com.cn/vga/galaxy/',
'http://detail.zol.com.cn/memory/corsair/', # 内存
'http://detail.zol.com.cn/memory/gskill/',
'http://detail.zol.com.cn/solid_state_drive/samsung/', # 固态
'http://detail.zol.com.cn/solid_state_drive/toshiba/',
'http://detail.zol.com.cn/power/corsair/', # 电源
'http://detail.zol.com.cn/power/coolermaster/'
]
for base_url in base_urls:
info_list = phone_info_spider(base_url)
for base_info in info_list:
Commodity_info = commodity_base_info()
Commodity_info.commodity_brand = base_info['commodity_brand']
Commodity_info.commodity_name = base_info['commodity_name']
Commodity_info.commodity_type = base_info['commodity_type']
Commodity_info.img_path = base_info['img_path']
Commodity_info.info = base_info['info']
Commodity_info.commodity_base_price = base_info[
'commodity_base_price']
Commodity_info.save()
print(base_info)
base_infos.append(base_info)
def Contrast():
commodity_type = request.args.get('commodity_type')
commodity_name = request.args.get('commodity_name')
commodity_contrast_name = request.args.get('commodity_contrast_name')
contrast_info = request.args.get('contrast_info').split(',')
# print(commodity_type)
# print(commodity_name)
# print(contrast_info)
# return commodity_type+';'+commodity_name
commodity = commodity_base_info()
commodity_review = commodity_review_info()
commodity_price = commodity_price_info()
user = User()
price_list = []
review_list = []
# print(commodity_name)
# print(commodity_type)
res = commodity.query.filter(
commodity_base_info.commodity_type == commodity_type).all()
# print(vars(res[0]))
phone_list = []
for i in res:
phone_list.append(i.commodity_name)
# print(res)
info = commodity.query.filter(
commodity_base_info.commodity_type == commodity_type,
commodity_base_info.commodity_name == commodity_name).all()
info_list = []
# print(info)
info = info[0]
info = vars(info)
info_commodity_type = info.pop('commodity_type')
info_name = info.pop('commodity_name')
info_price = info.pop('commodity_base_price')
info_img_path = info.pop('img_path')
base_info = {}
infos = info.pop('info')
infos = re.sub(r'(击败.*?)，', '', infos)
infos = infos.replace('>', '').split(';')
for i in infos:
# print(i)
i = i.split(':')
if len(i) < 2:
continue
# print(i[1])
info_list.append(i[0])
text = ''
texts = i[1].split(',')
if len(texts) >= 3:
texts = texts[:-2]
# print(texts)
if i[0] in contrast_info:
# print(type(text))
# print(text)
for c in texts:
if c != '':
text += c + ';'
base_info[i[0]] = text.replace('，;', '')
info = {
'name': info_name,
'price': info_price,
'img_path': info_img_path,
'commodity_type': info_commodity_type,
'base_info': base_info,
'phone_list': phone_list,
'info_list': info_list
}
other_info = commodity.query.filter(
commodity_base_info.commodity_type == commodity_type,
commodity_base_info.commodity_name == commodity_contrast_name).all()
other_info = other_info[0]
other_info_commodity_type = other_info.commodity_type
other_info_commodity_name = other_info.commodity_name
other_info_info = other_info.info
other_info_img_path = other_info.img_path
other_info_commodity_base_price = other_info.commodity_base_price
other_infos = re.sub(r'(击败.*?)，', '', other_info_info)
other_infos = other_infos.replace('>', '').split(';')
other_base_info = {}
for i in other_infos:
# print(i)
i = i.split(':')
if len(i) < 2:
continue
# print(i[1])
info_list.append(i[0])
text = ''
texts = i[1].split(',')
if len(texts) >= 3:
texts = texts[:-2]
# print(texts)
if i[0] in contrast_info:
# print(type(text))
# print(text)
for c in texts:
if c != '':
text += c + ';'
other_base_info[i[0]] = text.replace('，;', '')
other_info = {
'name': other_info_commodity_name,
'price': other_info_commodity_base_price,
'img_path': other_info_img_path,
'commodity_type': other_info_commodity_type,
'base_info': other_base_info
}
if 'username' in session:
user_id = user.query.filter(
User.username == session['username']).all()[0].user_id
# print(user_id)
# print(vars(user_collection.query.all()[0]))
# price_infos = commodity_price.query(commodity_price_info,user_collection).outerjoin(user_collection, commodity_price_info.id == user_collection.commodity_info_id and user_collection.user_id == user_id).filter(commodity_price_info.commodity_type == commodity_type,commodity_price_info.commodity_name == commodity_name).order_by(-commodity_price_info.price).all()
price_collection = db.session.query(
commodity_price_info, user_collection).outerjoin(
user_collection, commodity_price_info.id ==
user_collection.commodity_info_id).filter(
commodity_price_info.commodity_type == commodity_type,
commodity_price_info.commodity_name == commodity_name,
user_collection.user_id == user_id).order_by(
-commodity_price_info.price).all()
collection_priceid_list = []
for collection in price_collection:
collection_priceid_list.append(collection[0].id)
price_infos = commodity_price.query.filter(
commodity_price_info.commodity_type == commodity_type,
commodity_price_info.commodity_name == commodity_name).order_by(
-commodity_price_info.price).all()
# print(price_infos)
for price_info in price_infos:
# print(vars(price_info[0]))
# print(vars(price_info[1]))
if price_info.id in collection_priceid_list:
price_list.append({
'price': price_info.price,
'price_title': price_info.price_title,
'price_url': price_info.price_url,
'price_img_path': price_info.price_img_path,
'price_id': price_info.id,
'commodity_name': price_info.commodity_name,
'collection_flag': 1
})
else:
price_list.append({
'price': price_info.price,
'price_title': price_info.price_title,
'price_url': price_info.price_url,
'price_img_path': price_info.price_img_path,
'price_id': price_info.id,
'commodity_name': price_info.commodity_name,
'collection_flag': 0
})
else:
price_infos = commodity_price.query.filter(
commodity_price_info.commodity_type == commodity_type,
commodity_price_info.commodity_name == commodity_name).order_by(
-commodity_price_info.price).all()
price_id = 0
for price_info in price_infos:
price_id += 1
price_list.append({
'price': price_info.price,
'price_title': price_info.price_title,
'price_url': price_info.price_url,
'price_img_path': price_info.price_img_path,
'price_id': price_info.id,
'commodity_name': price_info.commodity_name,
'collection_flag': 0
})
review_infos = commodity_review.query.filter(
commodity_review_info.commodity_type == commodity_type,
commodity_review_info.commodity_name == commodity_name).all()
for review_info in review_infos:
review_list.append({
'review_url':
review_info.review_url,
'review_title':
review_info.review_title.replace('<em>', '').replace('</em>', ''),
'review_img_path':
review_info.review_img_path,
'review_excerpt':
review_info.review_excerpt.replace('<em>',
'').replace('</em>', '')
})
return render_template('commodity_contrast.html',
commodity_type=commodity_type,
commodity_name=commodity_name,
info=info,
price_list=price_list,
review_list=review_list,
other_info=other_info)
def commodity_brand(commodity_type, commodity_brand):
commodity = commodity_base_info()
path_list = []
commodity_list = []
if request.method == 'POST':
key_word = request.form.get('search_keyword')
res = commodity.query.filter(
or_(commodity_base_info.commodity_type.contains(key_word),
commodity_base_info.commodity_brand.contains(key_word),
commodity_base_info.commodity_name.contains(key_word))).all()
path_list.append('搜索')
path_list.append(key_word)
else:
if commodity_brand == '*':
res = commodity.query.filter(
commodity_base_info.commodity_type == commodity_type).all()
path_list.append(commodity_type)
else:
res = commodity.query.filter(
commodity_base_info.commodity_type == commodity_type,
commodity_base_info.commodity_brand == commodity_brand).all()
path_list.append(commodity_type)
path_list.append(commodity_brand)
commodity_sum = len(res)
for info in res:
info = vars(info)
# print(info)
commodity_type = info.pop('commodity_type')
name = info.pop('commodity_name')
price = info.pop('commodity_base_price')
img_path = info.pop('img_path')
infos = info.pop('info').replace('>', '').split(';')
# print(img_path)
base_info = {}
n = 0
for i in infos:
if n > 3:
break
n += 1
# print(i)
i = i.split(':')
if len(i) < 2:
continue
# print(i[1])
text = ''
texts = i[1].split(',')
if len(texts) >= 3:
texts = texts[:-2]
# print(texts)
# print(type(text))
# print(text)
for c in texts:
if c != '':
text += c + ';'
base_info[i[0]] = text.replace('，;', '')
info = {
'name': name,
'price': price,
'img_path': img_path,
'commodity_type': commodity_type,
'base_info': base_info
}
# print(info['img_path'])
commodity_list.append(info)
return render_template('commodity_brand.html',
index_list=commodity_list,
path=path_list,
commodity_sum=commodity_sum)