def parse(self, response):
numbers = []
images_base64 = re.findall(r'background-image:url\("data:image/png;base64,(.*?)"\)', str(response.text))[0]
css_content_label = re.findall(r"\.(.*?) { background-position-x:(.*?)px }", str(response.text))
num_map = dict()
for cont in css_content_label:
num_map[cont[0] + " sprite"] = cont[1]
img_fp = BytesIO(base64.b64decode(str(images_base64).encode("utf-8")))
img = Image.open(img_fp)
right_border_list = self.split_img_number(img)
col_list = response.xpath("//div[@class='col-md-1']")
for col in col_list:
class_names = col.xpath("./div/@class").extract()
tags_nums = []
for tag in class_names:
tag_position_x = num_map[tag]
for index, border_x in enumerate(right_border_list):
cur_pos_x = abs(int(tag_position_x))
if index + 1 >= len(right_border_list):
break
if border_x <= cur_pos_x < right_border_list[index + 1]:
tags_nums.append(str(index))
break
numbers.append(int("".join(tags_nums)))
item = GlidedskyItem(numbers=numbers)
yield item
def parse(self, response):
base_font = TTFont(r"glidedsky/data/font.ttf")
base_uni_list = base_font.getGlyphOrder()[1:]
origin_dict = {
"five": "0",
"four": "1",
"two": "2",
"three": "3",
"zero": "4",
"one": "5",
"nine": "6",
"six": "7",
"eight": "8",
"seven": "9",
}
str2num_dict = {
"zero": "0",
"one": "1",
"two": "2",
"three": "3",
"four": "4",
"five": "5",
"six": "6",
"seven": "7",
"eight": "8",
"nine": "9",
}
num_map = dict()
ttf_base64 = re.findall(r"data:font;charset=utf-8;base64,(.*?)\)",
response.text)[0]
base64_data = base64.decodebytes(ttf_base64.encode())
online_font = TTFont(BytesIO(base64_data))
online_uni_list = online_font.getGlyphOrder()[1:]
for uni2 in online_uni_list:
obj2 = online_font["glyf"][uni2]
for uni1 in base_uni_list:
obj1 = base_font["glyf"][uni1]
if obj1 == obj2:
num_map[str2num_dict[uni2]] = origin_dict[uni1]
num = response.xpath("//div[@class='col-md-1']/text()").extract()
numbers = []
for one in num:
new_num = ""
for i in one.strip():
new_num += num_map[i]
numbers.append(int(new_num))
item = GlidedskyItem(numbers=numbers)
yield item
def parse(self, response):
numbers = []
css_label = "".join(response.xpath("//style/text()").extract())
before_dict = dict()
css_content_label = re.findall(r'\.(.*?):before { content:"(.*?)" }',
str(css_label))
for cont in css_content_label:
before_dict[cont[0]] = cont[1]
css_del_names = re.findall(r"\.(.*?) { margin-right:-1em }",
str(css_label))
css_left_label = re.findall(r"\.(.*?) { left:(.*?)em }",
str(css_label))
count_dict = dict()
for count in css_left_label:
count_dict[count[0]] = count[1]
col_list = response.xpath("//div[@class='col-md-1']")
for col in col_list:
class_names = col.xpath("./div/@class").extract()
values = col.xpath("./div/text()").extract()
if len(class_names) >= 3:
move_num = ["0", "0", "0"]
index = 0
for class_name, value in zip(class_names, values):
if class_name not in css_del_names:
if class_name in count_dict and class_name not in css_del_names:
move_num[index +
int(count_dict[class_name])] = value
else:
move_num[index] = value
index += 1
new_number = "".join(move_num)
numbers.append(int(new_number))
elif len(class_names) == 1:
new_number = before_dict[class_names[0]]
numbers.append(int(new_number))
else:
for class_name in class_names:
if class_name in before_dict.keys():
new_number = before_dict[class_name]
numbers.append(int(new_number))
break
item = GlidedskyItem(numbers=numbers)
yield item
def parse(self, response):
num_str_list = response.xpath(
"//div[@class='col-md-1']/text()").extract()
ttf_base64 = re.findall(r"data:font;charset=utf-8;base64,(.*?)\)",
response.text)[0]
base64_data = base64.decodebytes(ttf_base64.encode())
font = TTFont(BytesIO(base64_data))
uni_list = font.getGlyphOrder()[1:11]
camp_list = font.getBestCmap()
def str2num(string):
number = ""
for s in string.strip():
unicode_str = s.encode("unicode-escape").decode()
sixteen_ary = unicode_str.replace("\\u", "0x")
num = uni_list.index(camp_list[int(sixteen_ary, 16)])
number += str(num)
return int(number)
numbers = [str2num(one.strip()) for one in num_str_list]
item = GlidedskyItem(numbers=numbers)
yield item
def parse(self, response):
num = response.xpath("//div[@class='col-md-1']/text()").extract()
item = GlidedskyItem(numbers=[int(str(one.strip())) for one in num])
yield item
def parse(self, response):
numbers = json.loads(response.text).get("items", [])
item = GlidedskyItem(numbers=numbers)
yield item