學習urllib.request和beautifulsoup,並從dribbble和behance上爬取了一些圖片,記錄一下
一、urllib.request
1. url的構造
構造請求的url遇到的主要問題是如何翻頁的問題,dribbble網站是下拉到底自動加載下一頁,地址欄的url沒有變化,如下:
但是通過檢查,我們可以發現request url裏關於page的字段,如下:
因此,我們構造如下的url:
for i in range(25): # 最多25頁
url = 'https://dribbble.com/shots?page=' + str(i + 1) + '&per_page=24'
2. header的構造
不同網頁需要的header的內容不一樣,參照檢查裏request header來構造。例如dribbble需要Referer,即從哪一個頁面跳轉到這個當前頁面的,一般填寫網站相關頁面網址就可以。
headers = {"Accept": "text/html,application/xhtml+xml,application/xml;",
"Referer": "https://dribbble.com/",
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/69.0.3493.3 Safari/537.36"}
3. urllib.request獲取頁面內容
用url和header實例化一個urllib.request.Request(url, headers),然後url.request.urlopen()訪問網頁獲取數據,使用read()函數即可讀取頁面內容。
def open_url(url):
# 將Request類實例化並傳入url爲初始值,然後賦值給req
headers = {"Accept": "text/html,application/xhtml+xml,application/xml;",
"Referer": "https://dribbble.com/",
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/69.0.3493.3 Safari/537.36"}
req = urllib.request.Request(url, headers=headers)
# 訪問url,並將頁面的二進制數據賦值給page
res = urllib.request.urlopen(req)
# 將page中的內容轉換爲utf-8編碼
html = res.read().decode('utf-8')
return html
這裏需要注意的是,有的頁面返回的數據是“text/html; charset=utf-8”格式,直接decode('utf-8')編碼即可,而有的頁面返回的是“application/json; charset=utf-8”格式數據,例如behance:
此時就需要json.loads()來獲取數據,得到的是列表,用操作列表的方式拿到html數據:
html = json.loads(res.read())
return html['html']
二、BeautifulSoup
BeautifulSoup將複雜的html文檔轉換爲樹形結構,每一個節點都是一個對象。
1.創建對象
soup = BeautifulSoup(open_url(url), 'html.parser')
‘html.parser’是解析器,BeautifulSoup支持Python標準庫中的HTML解析器,還支持一些第三方的解析器,如果我們不安裝它,則 Python 會使用 Python默認的解析器,lxml 解析器更加強大,速度更快,推薦安裝,常見解析器:
2. 標籤選擇器
標籤選擇篩選功能弱但是速度快,通過這種“soup.標籤名” 我們就可以獲得這個標籤的內容,但通過這種方式獲取標籤,如果文檔中有多個這樣的標籤,返回的結果是第一個標籤的內容
# 獲取p標籤
soup.p
# 獲取p標籤的屬性的兩種方法
soup.p.attrs['name']
soup.p['name']
# 獲取第一個p標籤的內容
soup.p.string
# 獲取p標籤下所有子標籤,返回一個列表
soup.p.contents
# 獲取p標籤下所有子標籤,返回一個迭代器
for i,child in enumerate(soup.p.children):
print(i,child)
# 獲取父節點的信息
soup.a.parent
# 獲取祖先節點
list(enumerate(soup.a.parents))
# 獲取後面的兄弟節點
soup.a.next_siblings
# 獲取前面的兄弟節點
soup.a.previous_siblings
# 獲取下一個兄弟標籤
soup.a.next_sibling
# 獲取上一個兄弟標籤
souo.a.previous_sinbling
3. 標準選擇器
find_all(name,attrs,recursive,text,**kwargs)可以根據標籤名,屬性,內容查找文檔,返回一個迭代器,例如:
# 獲取所有class爲js-project-module--picture的所有img標籤,並選擇每個標籤的src構成一個列表
image.src = [item['src'] for item in soup.find_all('img', {"class": "js-project-module--picture"})]
# .string獲取div的內容,strip()去除前後空格
desc = soup.find_all('div', {"class": "js-basic-info-description"})
if desc:
image.desc = [item.string.strip() for item in desc]
find(name,attrs,recursive,text,**kwargs),返回匹配的第一個元素
其他一些類似的用法:
find_parents()返回所有祖先節點,find_parent()返回直接父節點
find_next_siblings()返回後面所有兄弟節點,find_next_sibling()返回後面第一個兄弟節點
find_previous_siblings()返回前面所有兄弟節點,find_previous_sibling()返回前面第一個兄弟節點
find_all_next()返回節點後所有符合條件的節點, find_next()返回第一個符合條件的節點
find_all_previous()返回節點後所有符合條件的節點, find_previous()返回第一個符合條件的節點
三、從dribbble爬取圖片完整代碼
1.批量獲取圖片頁面鏈接
# -*- coding: utf-8 -*-
import random
import urllib.request
from bs4 import BeautifulSoup
import os
import time
def open_url(url):
headers = {"Accept": "text/html,application/xhtml+xml,application/xml;",
"Referer": "https://dribbble.com/",
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/69.0.3493.3 Safari/537.36"}
req = urllib.request.Request(url, headers=headers)
res = urllib.request.urlopen(req)
html = res.read().decode('utf-8')
return html
# 打開/創建“dribbble_list.txt”文件,O_CREAT:不存在即創建、O_WRONLY:只寫、O_APPEND:追加
fd = os.open('dribbble_list.txt', os.O_CREAT | os.O_WRONLY | os.O_APPEND)
for i in range(25):
url = 'https://dribbble.com/shots?page=' + str(i + 1) + '&per_page=24'
soup = BeautifulSoup(open_url(url), 'html.parser')
srcs = soup.find_all('a', {"class": "dribbble-link"})
src_list = [src['href'] for src in srcs]
for src in src_list:
os.write(fd, bytes(src, 'UTF-8'))
os.write(fd, bytes('\n', 'UTF-8'))
time.sleep(random.random()*5)
2. 獲取圖片和信息
import os
import random
import urllib.request
import re
import time
from bs4 import BeautifulSoup
class Image:
title = ''
src = ''
desc = []
tags = []
colors = []
view = []
like = []
save = []
def open_url(url):
headers = {"Accept": "text/html,application/xhtml+xml,application/xml;",
"Referer": "https://dribbble.com/shots",
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/69.0.3493.3 Safari/537.36"}
try:
req = urllib.request.Request(url, headers=headers)
res = urllib.request.urlopen(req)
html = res.read().decode('utf-8')
except:
return None
return html
def get_number(x):
return int(re.sub('\D', "", x))
def get_img_info(html):
# 實例化一張圖
image = Image()
soup = BeautifulSoup(html, 'html.parser')
# 標題
image.title = soup.find('div', {"class": "slat-header"}).find('h1').string.strip()
# 地址
image.src = soup.find('div', {"class": "detail-shot"}).find('img')['src']
# 描述
desc = soup.find('div', {"class": "shot-desc"})
if desc:
image.desc = [item.string.strip() for item in desc.find_all(text=True)]
# 標籤
image.tags = [item.string for item in soup.find_all('a', {"rel": "tag"})]
# 顏色
image.colors = [item.string for item in soup.find_all('a', {"style": re.compile('background-color.*')})]
# 瀏覽量
view = soup.find('div', {"class": "shot-views"})
if view:
image.view = [str(get_number(item)) for item in view.stripped_strings]
# 喜歡
like = soup.find('div', {"class": "shot-likes"})
if like:
image.like = [str(get_number(item)) for item in like.stripped_strings]
# 收藏
save = soup.find('div', {"class": "shot-saves"})
if save:
image.save = [str(get_number(item)) for item in save.stripped_strings]
return image
def save_text(root_path, img, num):
text = {
'src': img.src,
'desc': ';'.join(img.desc),
'tags': ';'.join(img.tags),
'colors': ';'.join(img.colors),
'score': ';'.join([img.title, ''.join(img.view), ''.join(img.like), ''.join(img.save)])
}
text_list = ['src', 'desc', 'tags', 'colors', 'score']
for item in text_list:
save_path = root_path + item + '.txt'
fd = os.open(save_path, os.O_CREAT | os.O_WRONLY | os.O_APPEND)
write_str = str(num).zfill(3) + ' ' + text[item] + '\n'
os.write(fd, bytes(write_str, 'UTF-8'))
os.close(fd)
def read_dribbble_data(data_folder):
import pandas as pd
import os
columns = ['url']
df = pd.read_csv(os.path.join(data_folder, 'dribbble_list.txt'), names=columns)
return df
def to_url(img_url):
return 'https://dribbble.com{img_url}'.format(img_url=img_url)
if __name__ == '__main__':
data_folder = './'
df = read_dribbble_data(data_folder)
urls = map(to_url, df['url'].values)
for i, url in enumerate(urls):
print(url)
# 獲取並解析網頁
html = open_url(url)
if html:
image = get_img_info(open_url(url))
# 獲取並保存圖片
# save_path_img = 'img/' + image.title + '.jpg'
save_path_img = 'img/' + str(i+556).zfill(3) + '.jpg'
urllib.request.urlretrieve(image.src, save_path_img)
# 保存“標題 地址 描述 標籤 顏色 瀏覽量 喜歡 收藏”
save_path_text_root = 'dribbble_text/'
save_text(save_path_text_root, img=image, num=i+556)
time.sleep(random.random()*5)
四、從behance爬取圖片完整代碼
1. 批量獲取圖片頁面鏈接
# -*- coding: utf-8 -*-
import random
import urllib.request
from bs4 import BeautifulSoup
import os
import time
import json
def open_url(url):
headers = {"Accept": "*/*",
"Referer": "https://www.behance.net/search?field=48&content=projects&sort=appreciations&time=week",
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/69.0.3493.3 Safari/537.36",
"Host": "www.behance.net",
"Connection": "keep-alive",
"X-BCP": "523bc8eb-c6a4-4eeb-a73d-0bf9ec1c06d9",
"X-NewRelic-ID": "VgUFVldbGwACXFJSBAUF",
"X-Requested-With": "XMLHttpRequest"}
req = urllib.request.Request(url, headers=headers)
res = urllib.request.urlopen(req)
html = json.loads(res.read())
return html['html']
fd = os.open('behance_list.txt', os.O_CREAT | os.O_WRONLY | os.O_APPEND)
for i in range(200):
url = 'https://www.behance.net/search?ordinal=' + str((i+100) * 48) + '&per_page=48&field=48&content=projects&sort=appreciations&time=week&location_id=×tamp=0&mature=0'
print(url)
soup = BeautifulSoup(open_url(url), 'html.parser')
srcs = soup.find_all('a', {"class": "js-project-cover-image-link"})
src_list = [src['href'] for src in srcs]
for src in src_list:
os.write(fd, bytes(src, 'UTF-8'))
os.write(fd, bytes('\n', 'UTF-8'))
time.sleep(random.random()*5)
os.close(fd)
2. 獲取圖片和信息
# -*- coding: utf-8 -*-
import os
import random
import urllib.request
import re
import time
from bs4 import BeautifulSoup
class Image:
title = ''
src = []
desc = []
tags = []
data = []
def open_url(url):
headers = {"Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8",
"Referer": "https://www.behance.net/gallery/70675447/YELLOWSTONE",
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/69.0.3493.3 Safari/537.36",
"Host": "www.behance.net",
"Connection": "keep-alive",
"Upgrade-Insecure-Requests": 1,
"Cookie": "巴啦啦小魔仙全身變"
}
try:
req = urllib.request.Request(url, headers=headers)
res = urllib.request.urlopen(req)
html = res.read().decode('utf-8')
except:
return None
return html
def get_number(x):
return int(re.sub('\D', "", x))
def get_img_info(html):
# 實例化一張圖
image = Image()
soup = BeautifulSoup(html, 'html.parser')
# 地址
image.src = [item['src'] for item in soup.find_all('img', {"class": "js-project-module--picture"})]
# 標題
image.title = soup.find('div', {"class": "js-project-title"}).string.strip()
# 描述
desc = soup.find_all('div', {"class": "js-basic-info-description"})
if desc:
image.desc = [item.string.strip() for item in desc]
# 標籤
tags = soup.find_all('a', {"class": "object-tag"})
if tags:
image.tags = [item.string.strip() for item in tags]
# 瀏覽 點贊 評論
data = soup.find_all('div', {"class": "project-stat"})
if data:
image.data = [item.string.strip() for item in data][:2]
return image
def save_text(root_path, img, num):
text = {
'title': image.title.replace(' ', '_'),
'score': ' '.join(img.data),
'desc': ';' + (';'.join(img.desc)).replace('\n', ';'),
'tags': ';' + ';'.join(img.tags),
'src': ';' + ';'.join(img.src)
}
text_list = ['title', 'score', 'desc', 'tags', 'src']
for item in text_list:
save_path = root_path + item + '.txt'
fd = os.open(save_path, os.O_CREAT | os.O_WRONLY | os.O_APPEND)
write_str = str(num).zfill(5) + ' ' + text[item] + '\n'
os.write(fd, bytes(write_str, 'UTF-8'))
os.close(fd)
def read_dribbble_data(data_folder):
import pandas as pd
import os
columns = ['url']
df = pd.read_csv(os.path.join(data_folder, 'behance_list.txt'), names=columns)
return df
if __name__ == '__main__':
data_folder = './'
urls = read_dribbble_data(data_folder)['url'].values
for i, url in enumerate(urls):
print(url)
# 獲取並解析網頁
html = open_url(url)
if html:
image = get_img_info(open_url(url))
# 獲取並保存圖片
for j, src in enumerate(image.src):
save_path_img = './behance_img/' + str(i).zfill(5) + '_' + str(j).zfill(3) + src[-4:]
urllib.request.urlretrieve(src, save_path_img)
time.sleep(random.random()*3)
# 保存“標題 瀏覽量 喜歡 收藏 描述 標籤 ”
save_path_text_root = './behance_text/'
save_text(save_path_text_root, img=image, num=i)
time.sleep(random.random()*5)