Merge remote-tracking branch 'origin/master'

This commit is contained in:
nikitin712
2023-03-29 18:40:36 +03:00
8 changed files with 170 additions and 86 deletions
+8 -6
View File
@@ -5,28 +5,30 @@ import psycopg2
import wget
browser = webdriver.Chrome(service=Service('C:\Desktop\exe\chromedriver.exe'))
browser.get('https://flawery.ru/moscow/bouquets/event-yanvary25/?min=0&max=max&sorting=discount&filter_24h=on')
browser.get('https://flawery.ru/moscow/bouquets/color-blue/?min=0&max=max&sorting=discount')
soup = BeautifulSoup(browser.page_source, "lxml")
Name = soup.find_all(attrs={"class": "catalog_title"})
Price = soup.find_all(attrs={"class": "catalog_price_now"})
Delivery_Time = soup.find_all(attrs={"class": "catalog_express"})
Discount = soup.find_all(attrs={"class": "catalog_sale"})
Image = soup.find_all(attrs={"class": "catalog_item catalog_item_popup"})
connection = psycopg2.connect(host='localhost', dbname='FHWDB', user='postgres', password='Q1w2e3r4')
cursor = connection.cursor()
create_q = '''CREATE TABLE Parse
(ID serial primary key, Name varchar(100), Price varchar(9), Delivery_Time varchar(25), Discount varchar(10), src varchar(110))'''
(ID serial primary key, Name varchar(100), Price varchar(9), Discount varchar(10), src_of_picture varchar(50))'''
cursor.execute(create_q)
connection.commit()
for j in range(60):
url = 'https://flawery.ru'+Image[j].find('a').find('picture').find('img').attrs['src']
tempf = f"C:\\Users\\user\\Desktop\\FHWDB{j}.jpg"
tempf = f"C:\\Users\\user\\PFDB{j}.jpg"
wget.download(url, tempf)
insert_query = f'''INSERT into public.Parse(Name, Price, Delivery_Time, Discount, src) values ('{Name[j].text}', '{Price[j].text}', '{Delivery_Time[j].text}', '{Discount[j].text}', '{tempf}') '''
insert_query = f'''INSERT into public.Parse(Name, Price, Discount, src_of_picture) values ('{Name[j].text}', '{Price[j].text}', '{Discount[j].text}', '{tempf}') '''
cursor.execute(insert_query)
connection.commit()
cursor.execute("SELECT * from Parse")
print(cursor.fetchall())
cursor.close()
connection.close()
+31 -8
View File
@@ -1,12 +1,35 @@
from bs4 import BeautifulSoup
from selenium import webdriver
from selenium.webdriver.chrome.service import Service
import psycopg2
import wget
s = Service("E:\data\chromedriver.exe")
browser = webdriver.Chrome(service=s)
browser.get("https://ria.ru/")
html_text = browser.page_source
soup = BeautifulSoup(html_text, 'lxml')
news = soup.find_all('div', class_='cell cell-list')
for new in news:
print(new.text)
browser = webdriver.Chrome(service=Service('E:\data\chromedriver.exe'))
browser.get('https://8956food.ru/')
soup = BeautifulSoup(browser.page_source, "lxml")
Name = soup.find_all(attrs={"class": "subtitle"})
Price = soup.find_all(attrs={"class": "price"})
Image = soup.find_all(attrs={"class": "holder-img"})
connection = psycopg2.connect(host='localhost', dbname='dbdata', user='postgres', password='Q1w2e3r4')
cursor = connection.cursor()
create_q = '''CREATE TABLE Food
(ID serial primary key, Name varchar(500), Price varchar(60), src varchar(100))'''
cursor.execute(create_q)
connection.commit()
#https://smartomato.ams3.cdn.digitaloceanspaces.com/uploads/media/photo/769221/dish_large__1.jpg
for j in range(15):
url = Image[j].find('img').attrs['src']
tempf = f"C:\\Users\\Harutyun\\Desktop\\prog\\food{j}.jpg"
wget.download(url, tempf)
insert_query = f'''INSERT into public.Food(Name, Price, src) values ('{Name[j].text}', '{Price[j].text}', '{tempf}') '''
cursor.execute(insert_query)
connection.commit()
cursor.execute("SELECT * from Food")
print(cursor.fetchall())
cursor.close()
connection.close()
@@ -0,0 +1,29 @@
from bs4 import BeautifulSoup
from selenium import webdriver
from selenium.webdriver.chrome.service import Service
from wget import download
import psycopg2
connection = psycopg2.connect(host='localhost', dbname='dbdata',
user='postgres', password='Q1w2e3r4')
cursor=connection.cursor()
s = Service('C:\\Users\\Dasha\\Desktop\\Progect\\chromedriver.exe')
browser = webdriver.Chrome(service=s)
browser = webdriver.Chrome()
url = 'https://www.tretyakovgallery.ru/'
browser.get(url)
html_text=browser.page_source
soup = BeautifulSoup(html_text, 'lxml')
products = soup.find_all(attrs={"class": "building-item"})
for i, product in enumerate(products):
link = product.find("img").attrs.get("src")
tag = product.find("strong")
name = product.find(attrs={"class": "building-item__title"})
description = product.find(attrs={"class": "building-item__description"})
if tag is not None:
download(url + link, f"img\\{i}.jpg")
cursor.execute(f"""INSERT INTO public.parsing(url, workhours, name, description, filename)
VALUES
( '{url + link}', '{tag.text}', '{name.text.strip()}', '{description.text.strip()}', 'img\\{i}.jpg')""")
connection.commit()
cursor.close()
connection.close()
+19 -29
View File
@@ -7,65 +7,55 @@ import time
i = 1
s = Service("С:\DATA\ChromeDriver\chromedriver.exe")
browser = webdriver.Chrome(service=s)
browser.get("https://www.mvideo.ru/noutbuki-planshety-komputery-8/noutbuki-118?from=under_search")
browser.get("https://www.mvideo.ru/noutbuki-planshety-komputery-8/noutbuki-118/f/tolko-v-nalichii=da?from=under_search")
time.sleep(5)
"""
height = browser.execute_script("return document.body.scrollHeight")
time.sleep(4)
while 900*i < height:
height = browser.execute_script("return document.body.scrollHeight")
browser.execute_script(f"window.scrollTo(0,{900*i})")
for i in range(10):
browser.execute_script(f"window.scrollTo(0, {1000*i})")
time.sleep(1)
i+=1
"""
html_text = browser.page_source
soup = BeautifulSoup(html_text, 'lxml')
productst = soup.find_all('a', class_="product-title__text")
pricest = soup.find_all('span', class_="price__main-value")
features = soup.find_all('ul', class_="product-feature-list product-feature-list--undefined")
pictures = soup.find_all('img', class_="product-picture__img product-picture__img--list")
products, prices, diagonals, resolutions, CPUs, RAMs, Graphics_controllers, Volumes = [],[],[],[],[],[],[],[]
for i in range(len(productst)):
products.append(productst[i].text.strip())
prices.append(pricest[i].text.replace("\xa0", "").strip())
for feature in features:
u = feature.text
diagonals.append(u[20:u.find("\"",20)+1])
resolutions.append(u[u.find("\"",20)+2:u.find(".", 23)+1])
CPUs.append(u[u.find("Процессор",20)+9:u.find("ГГц",20)+3].replace("\xa0",""))
RAMs.append(" ".join(u[u.find("(RAM)",20)+5:u.find("Графический контроллер")].split()))
Graphics_controllers.append(u[u.find("Графический контроллер")+22:u.rfind("Объем")-2])
Volumes.append(u[u.rfind("Объем")+6:].replace("\xa0", "").strip())
for i in range(len(Volumes)):
u = Volumes[i]
if u.rfind("D")!=-1: Volumes[i] = u[:u.rfind("D")+1]+ " " + u[u.rfind("D")+1:]
else: Volumes[i] = u[:u.rfind("С")+1]+ " " + u[u.rfind("С")+1:]
features = soup.find_all('span', class_="product-feature-list__value")
pictures = soup.find_all('div', class_="product-picture-container")
connection=psycopg2.connect(host='localhost', dbname='dbdata', user='postgres', password='Q1w2e3r4')
cursor = connection.cursor()
insert = """CREATE TABLE public.laptops(
""" insert = CREATE TABLE public.laptops(
id serial primary key,
product varchar(100),
price varchar(15),
diagonal varchar(5),
resolution varchar(20),
CPU varchar(50),
RAM varchar(10),
RAM varchar(15),
Graphics_Controller varchar(40),
Volume varchar(25),
src varchar(100)
);
"""
insert = """TRUNCATE TABLE public.laptops; ALTER SEQUENCE laptops_id_seq RESTART WITH 1;"""
cursor.execute(insert)
connection.commit()
for i in range(len(products)):
url = "https://"+pictures[i]['src']
for i in range(len(productst)):
url = "https://"+pictures[i].find('img')['src']
filename = f"C:\DATA\Images\{i+1}.jpg"
u = []
for j in range(10):
u.append(features[10*i+j].text)
wget.download(url, filename)
t = pricest[i].text.replace("\xa0", " ")
insert = f"""INSERT INTO public.laptops(
product, price, diagonal, resolution, CPU, RAM, Graphics_Controller, Volume, src)
VALUES
('{products[i]}', '{prices[i]}', '{diagonals[i]}', '{resolutions[i]}', '{CPUs[i]}', '{RAMs[i]}', '{Graphics_controllers[i]}', '{Volumes[i]}', '{filename}');"""
('{productst[i].text.strip()}', '{t.strip()}', '{u[0][:u[0].find("/")]}', '{u[0][u[0].find("/")+1:]}',
'{u[2]}', '{u[4]+" "+u[5]}', '{u[6]}', '{u[8]}', '{filename}');"""
cursor.execute(insert)
connection.commit()
cursor.execute("select * from laptops")
print("Результат", cursor.fetchall())
cursor.close()
connection.close()
@@ -1,42 +0,0 @@
import psycopg2
import wget
from bs4 import BeautifulSoup
from selenium import webdriver
from selenium.webdriver.chrome.service import Service
s = Service("С:\DATA\ChromeDriver\chromedriver.exe")
browser = webdriver.Chrome(service=s)
browser.get("https://online.metro-cc.ru/virtual/assortiment_rioba-4887?from=under_search&page=2")
html_text = browser.page_source
soup = BeautifulSoup(html_text, 'lxml')
productst = soup.find_all('span', class_="product-card-name__text")
pricest = soup.find_all('span', class_="product-price__sum-rubles")
pictures = soup.find_all('img', class_="product-card-photo__image")
connection=psycopg2.connect(host='localhost', dbname='dbdata', user='postgres', password='Q1w2e3r4')
cursor = connection.cursor()
products, prices = [], []
for i in range(len(productst)):
products.append(productst[i].text.replace("\n", "").strip())
prices.append(pricest[i].text+" Р/шт")
insert = """CREATE TABLE public.Chocolate(
id serial primary key,
product varchar(100),
price varchar(25),
src varchar(50)
);
"""
cursor.execute(insert)
connection.commit()
for i in range(len(products)):
url = pictures[i]['src']
filename = f"C:\DATA\Images2\{i+1}.jpg"
wget.download(url, filename)
insert = f"""INSERT INTO public.Chocolate(
product, price, src)
VALUES
('{products[i]}', '{prices[i]}', '{filename}');"""
cursor.execute(insert)
connection.commit()
cursor.execute("select * from Chocolate")
print("Результат", cursor.fetchall())
cursor.close()
connection.close()
+40
View File
@@ -0,0 +1,40 @@
from bs4 import BeautifulSoup
from selenium import webdriver
from selenium.webdriver.chrome.service import Service
import wget
import psycopg2
connection = psycopg2.connect(host='localhost', dbname='dbdata', user='postgres', password='Q1w2e3r4')
cursor = connection.cursor()
s = Service(r"C:\\Documents\\data\\chromedriver.exe")
browser = webdriver.Chrome(service=s)
browser.get('https://www.citilink.ru/catalog/smartfony/')
html_text = browser.page_source
b_soup = BeautifulSoup(html_text, 'lxml')
insert_qwery = """create table phones\n
(id serial primary key, name varchar(100), price varchar(100),delivery varchar(100), pictures varchar(100))"""
cursor.execute(insert_qwery)
connection.commit()
phone_names = b_soup.find_all('a', class_="app-catalog-9gnskf e1259i3g0")
phone_price = b_soup.find_all('span', class_="e1j9birj0 e106ikdt0 app-catalog-j8h82j e1gjr6xo0")
phone_delivery = b_soup.find_all('div', class_="e16h9msl0 app-catalog-1fnzrtt e15ufesp0")
pictures = b_soup.find_all('div', class_="app-catalog-zugobn e1o1y6qd0")
for i in range(15):
url = 'https://www.citilink.ru/catalog/smartfony/'
filename = f"C:\\Users\\mark6\\OneDrive\\Desktop\\БД\\img\\{i}.jpg"
wget.download(url, filename)
ins_qwery = f"""INSERT INTO public.phones(name, price, delivery, pictures)\n
VALUES ('{phone_names[i].text}', '{phone_price[i].text}', '{phone_delivery[i].text}', '{filename}');"""
cursor.execute(ins_qwery)
connection.commit()
cursor.execute("Select * from phones")
print(cursor.fetchall())
cursor.close()
connection.close()
+39
View File
@@ -0,0 +1,39 @@
from bs4 import BeautifulSoup
from selenium import webdriver
from selenium.webdriver.chrome.service import Service
import psycopg2
import wget
connection = psycopg2.connect(host='localhost', dbname='dbdata', user='postgres', password='Q1w2e3r4')
cursor = connection.cursor()
crtable = """ create table Notebooks
(id serial primary key, name varchar(500), description varchar(500), code varchar(200), price varchar(100),
picture varchar(500)) """
cursor.execute(crtable)
connection.commit()
s = Service('C:\\Users\\Yekaterina\\Downloads\\chromedriver_win32\\chromedriver.exe')
browser = webdriver.Chrome(service=s)
browser.get('https://www.citilink.ru/catalog/noutbuki/')
html_text = browser.page_source
soup = BeautifulSoup(html_text, 'lxml')
name = soup.find_all('div', class_='app-catalog-1tp0ino e1an64qs0')
description = soup.find_all('div', class_='app-catalog-1o4umte eevw8x70')
code = soup.find_all('div', class_= 'app-catalog-0 e1dsj6g20')
price = soup.find_all('div', class_= 'app-catalog-0 e1dsj6g20')
picture = soup.find_all('div', class_='app-catalog-0 e1jarwcz0')
for i in range(10):
url = picture[i].find('img').attrs['src']
file = f"C:\\Users\\Yekaterina\\Desktop\\baza\\pic{i}.JPG"
wget.download(url, file)
ins_qwery = f"""insert into public.Notebooks(name, description, code, price, picture)
values ('{name[i].text}', '{description[i].text}', '{code[i].text}', '{price[i].text}','{file}')"""
cursor.execute(ins_qwery)
connection.commit()
cursor.close()
connection.close()
+4 -1
View File
@@ -21,8 +21,11 @@ with psyc.connect(dbname="db_for_parse", user="perfecto") as conn:
img_tag = product.find(attrs={"class": "imagef"}).find("img")
title = img_tag.attrs.get("alt")
image = url + img_tag.attrs.get("src")
price = product.find(attrs={"class": "product-price"}).text
desc = product.find(attrs={"class": "product-desc"}).text
download(url + img_tag.attrs.get("src"), f"images/{i}.jpg")
cursor.execute(f"insert into images(link, name) values ('{image}', '{title}')")
cursor.execute(f"""insert into images(link, name, price, description, file)
values ('{image}', '{title}', '{price}', '{desc}', 'images/{i}.jpg')""")
conn.commit()