i'm trying to scrape the restaurant pages on tripadvisor (just to learn how it works)
However, i only get the first page.
What am I missing?
here is the code, thanks!
import scrapy
class TripadvSpider(scrapy.Spider):
name = 'tripadv'
allowed_domains = ['tripadvisor.com']
start_urls = ['https://www.tripadvisor.com/Restaurants-g60795-oa0-Philadelphia_Pennsylvania.html#EATERY_LIST_CONTENTS']
def parse(self, response):
for stores in response.css('div.emrzT'):
yield {
'name' : stores.css('a.bHGqj::text').extract(),
'link' : stores.css('a.bHGqj').xpath("#href").extract()
}
next_page = ('http://tripadvisor.com' + response.css('a.nav').attrib['href']).extract()
##next_page = response.xpath('//a[contains(text(), "Next")]/#href).extract())
#next_page = ('http://tripadvisor.com' + response.css('a:contains("Next")').attrib['href'].extract())
if next_page is not None:
yield response.follow(next_page, callback=self.parse)
#djmystica, Now it's working fine
import scrapy
class TripadvSpider(scrapy.Spider):
name = 'tripadv'
allowed_domains = ['tripadvisor.com']
start_urls = [
'https://www.tripadvisor.com/Restaurants-g60795-oa0-Philadelphia_Pennsylvania.html#EATERY_LIST_CONTENTS']
def parse(self, response):
for stores in response.css('div.emrzT'):
yield {
'name': stores.css('a.bHGqj::text').extract_first(),
'link': stores.css('a.bHGqj').xpath("#href").extract_first()}
#next_page = ('http://tripadvisor.com' +response.css('a.nav').attrib['href']).extract()
next_page = response.xpath('//a[contains(text(), "Next")]/#href').extract_first()
abs_next_page = f'https://www.tripadvisor.com{next_page}'
#next_page = ('http://tripadvisor.com' + response.css('a:contains("Next")').attrib['href'].extract())
if abs_next_page is not None:
yield response.follow(abs_next_page, callback=self.parse)
Is there a way to change the response used in an ItemLoader, for example I am parsing two pages and I pass my_item in meta, the response in the item loader is the first one (where I initially defined it).
I am currently using my_item.add_value('price',response.xpath('//p[#id="price"]').extract_first() to get passed this since I can't use my_item.add_xpath('price', '//p[#id="price"]') because the response is from the initial page
my_item = ItemLoader(item=MyItem(),response=response)
#fill my_item
yield Request(My_url,callback=self.parse_item,meta={'my_item':my_item)
You need something like this:
def parse(self, response):
l = ItemLoader(item=YourItem(), response=response)
l.add_xpath('Field1', '...')
l.add_value('Field2', '...')
item = l.load_item()
yield scrapy.Request(
url=another_url,
callback=self.second,
meta={'item': item}
)
def second(self, response):
l = ItemLoader(item=response.meta["item"], response=response)
l.add_xpath("Field3", '...')
yield l.load_item()
For my job, I built a scrapy spider to quickly check in on ~200-500 website landing pages for clues that the pages are not functioning, outside of just 400-style errors. (e.g. check for the presence of "out of stock" on page.) This check happens across approx. 30 different websites under my purview, all of them using the same page structure.
This has worked fine, every day, for 4 months.
Then, suddenly, and without change to the code, I started getting unpredictable errors, about 4 weeks ago:
url_title = response.css("title::text").extract_first()
AttributeError: 'Response' object has no attribute 'css'
If I run this spider, this error will occur with, say... 3 out of 400 pages.
Then, if immediately run the spider again, those same 3 pages are scraped just fine without error, and 4 totally different pages will return the same error.
Furthermore, if I run the EXACT same spider as below, but replace mapping with just these 7 erroneous landing pages, they are scraped perfectly fine.
Is there something in my code that's not quite right??
I'm going to attach the whole code - sorry in advance!! - I just fear that something I might deem as superfluous may in fact be the cause. So this is the whole thing, but with sensitive data replaced with ####.
I've checked all of the affected pages, and of course the css is valid, and the title is always present.
I've done sudo apt-get update & sudo apt-get dist-upgrade on the server running scrapy, in hopes that this would help. No luck.
import scrapy
from scrapy import signals
from sqlalchemy.orm import sessionmaker
from datetime import date, datetime, timedelta
from scrapy.http.request import Request
from w3lib.url import safe_download_url
from sqlalchemy import and_, or_, not_
import smtplib
from email.MIMEMultipart import MIMEMultipart
from email.MIMEText import MIMEText
from sqlalchemy.engine import create_engine
engine = create_engine('mysql://######:#######localhost/LandingPages', pool_recycle=3600, echo=False)
#conn = engine.connect()
from LandingPageVerifier.models import LandingPagesFacebook, LandingPagesGoogle, LandingPagesSimplifi, LandingPagesScrapeLog, LandingPagesScrapeResults
Session = sessionmaker(bind=engine)
session = Session()
# today = datetime.now().strftime("%Y-%m-%d")
# thisyear = datetime.now().strftime("%Y")
# thismonth = datetime.now().strftime("%m")
# thisday = datetime.now().strftime("%d")
# start = date(year=2019,month=04,day=09)
todays_datetime = datetime(datetime.today().year, datetime.today().month, datetime.today().day)
print todays_datetime
landingpages_today_fb = session.query(LandingPagesFacebook).filter(LandingPagesFacebook.created_on >= todays_datetime).all()
landingpages_today_google = session.query(LandingPagesGoogle).filter(LandingPagesGoogle.created_on >= todays_datetime).all()
landingpages_today_simplifi = session.query(LandingPagesSimplifi).filter(LandingPagesSimplifi.created_on >= todays_datetime).all()
session.close()
#Mix 'em together!
landingpages_today = landingpages_today_fb + landingpages_today_google + landingpages_today_simplifi
#landingpages_today = landingpages_today_fb
#Do some iterating and formatting work
landingpages_today = [(u.ad_url_full, u.client_id) for u in landingpages_today]
#print landingpages_today
landingpages_today = list(set(landingpages_today))
#print 'Unique pages: '
#print landingpages_today
# unique_landingpages = [(u[0]) for u in landingpages_today]
# unique_landingpage_client = [(u[1]) for u in landingpages_today]
# print 'Pages----->', len(unique_landingpages)
class LandingPage004Spider(scrapy.Spider):
name='LandingPage004Spider'
#classmethod
def from_crawler(cls, crawler, *args, **kwargs):
spider = super(LandingPage004Spider, cls).from_crawler(crawler, *args, **kwargs)
#crawler.signals.connect(spider.spider_opened, signals.spider_opened)
crawler.signals.connect(spider.spider_closed, signals.spider_closed)
return spider
def spider_closed(self, spider):
#stats = spider.crawler.stats.get_stats()
stats = spider.crawler.stats.get_value('item_scraped_count'),
Session = sessionmaker(bind=engine)
session = Session()
logitem = LandingPagesScrapeLog(scrape_count = spider.crawler.stats.get_value('item_scraped_count'),
is200 = spider.crawler.stats.get_value('downloader/response_status_count/200'),
is400 = spider.crawler.stats.get_value('downloader/response_status_count/400'),
is403 = spider.crawler.stats.get_value('downloader/response_status_count/403'),
is404 = spider.crawler.stats.get_value('downloader/response_status_count/404'),
is500 = spider.crawler.stats.get_value('downloader/response_status_count/500'),
scrapy_errors = spider.crawler.stats.get_value('log_count/ERROR'),
scrapy_criticals = spider.crawler.stats.get_value('log_count/CRITICAL'),
)
session.add(logitem)
session.commit()
session.close()
#mapping = landingpages_today
handle_httpstatus_list = [200, 302, 404, 400, 500]
start_urls = []
def start_requests(self):
for url, client_id in self.mapping:
yield Request(url, callback=self.parse, meta={'client_id': client_id})
def parse(self, response):
##DEBUG - return all scraped data
#wholepage = response.body.lower()
url = response.url
if 'redirect_urls' in response.request.meta:
redirecturl = response.request.meta['redirect_urls'][0]
if 'utm.pag.ca' in redirecturl:
url_shortener = response.request.meta['redirect_urls'][0]
else:
url_shortener = 'None'
else:
url_shortener = 'None'
client_id = response.meta['client_id']
url_title = response.css("title::text").extract_first()
# pagesize = len(response.xpath('//*[not(descendant-or-self::script)]'))
pagesize = len(response.body)
HTTP_code = response.status
####ERROR CHECK: Small page size
if 'instapage' in response.body.lower():
if pagesize <= 20000:
err_small = 1
else:
err_small = 0
else:
if pagesize <= 35000:
err_small = 1
else:
err_small = 0
####ERROR CHECK: Page contains the phrase 'not found'
if 'not found' in response.xpath('//*[not(descendant-or-self::script)]').extract_first().lower():
#their sites are full of HTML errors, making scrapy unable to notice what is and is not inside a script element
if 'dealerinspire' in response.body.lower():
err_has_not_found = 0
else:
err_has_not_found = 1
else:
err_has_not_found = 0
####ERROR CHECK: Page cotains the phrase 'can't be found'
if "can't be found" in response.xpath('//*[not(self::script)]').extract_first().lower():
err_has_cantbefound = 1
else:
err_has_cantbefound = 0
####ERROR CHECK: Page contains the phrase 'unable to locate'
if 'unable to locate' in response.body.lower():
err_has_unabletolocate = 1
else:
err_has_unabletolocate = 0
####ERROR CHECK: Page contains phrase 'no longer available'
if 'no longer available' in response.body.lower():
err_has_nolongeravailable = 1
else:
err_has_nolongeravailable = 0
####ERROR CHECK: Page contains phrase 'no service specials'
if 'no service specials' in response.body.lower():
err_has_noservicespecials = 1
else:
err_has_noservicespecials = 0
####ERROR CHECK: Page contains phrase 'Sorry, no' to match zero inventory for a search, which normally says "Sorry, no items matching your request were found."
if 'sorry, no ' in response.body.lower():
err_has_sorryno = 1
else:
err_has_sorryno = 0
yield {'client_id': client_id, 'url': url, 'url_shortener': url_shortener, 'url_title': url_title, "pagesize": pagesize, "HTTP_code": HTTP_code, "err_small": err_small, 'err_has_not_found': err_has_not_found, 'err_has_cantbefound': err_has_cantbefound, 'err_has_unabletolocate': err_has_unabletolocate, 'err_has_nolongeravailable': err_has_nolongeravailable, 'err_has_noservicespecials': err_has_noservicespecials, 'err_has_sorryno': err_has_sorryno}
#E-mail settings
def sendmail(recipients,subject,body):
fromaddr = "#######"
toaddr = recipients
msg = MIMEMultipart()
msg['From'] = fromaddr
msg['Subject'] = subject
body = body
msg.attach(MIMEText(body, 'html'))
server = smtplib.SMTP('########)
server.starttls()
server.login(fromaddr, "##########")
text = msg.as_string()
server.sendmail(fromaddr, recipients, text)
server.quit()
`
Expected results is a perfect scrape, with no errors.
Actual results are unpredicatable AttributeErrors, claiming that attribute 'css' can't be found on some pages. But if I scrape those pages individually, using the same script, they scrape just fine.
Sometimes Scrapy can't parse HTML because of markup errors, that's why you can't call response.css(). You can catch these events in your code and analyze broken HTML:
def parse(self, response):
try:
....
your code
.....
except:
with open("Error.htm", "w") as f:
f.write(response.body)
UPDATE You can try to check for empty response:
def parse(self, response):
if not response.body:
yield scrapy.Request(url=response.url, callback=self.parse, meta={'client_id': response.meta["client_id"]})
# your original code
I can't find any solution for using start_requests with rules, also I haven't seen any example on the Internet with this two. My purpose is simple, I wanna redefine start_request function to get an ability catch all exceptions dunring requests and also use meta in requests. This is a code of my spider:
class TestSpider(CrawlSpider):
name = 'test'
allowed_domains = ['www.oreilly.com']
start_urls = ['https://www.oreilly.com/library/view/practical-postgresql/9781449309770/ch04s05.html']
# Base on scrapy doc
def start_requests(self):
for u in self.start_urls:
yield Request(u, callback=self.parse_item, errback=self.errback_httpbin, dont_filter=True)
rules = (
Rule(LinkExtractor(), callback='parse_item', follow=True),
)
def parse_item(self, response):
item = {}
item['title'] = response.xpath('//head/title/text()').extract()
item['url'] = response.url
yield item
def errback_httpbin(self, failure):
self.logger.error('ERRRRROR - {}'.format(failure))
This code scrape only one page. I try to modify it and instead of:
def parse_item(self, response):
item = {}
item['title'] = response.xpath('//head/title/text()').extract()
item['url'] = response.url
yield item
I've tried to use this, based on this answer
def parse_item(self, response):
item = {}
item['title'] = response.xpath('//head/title/text()').extract()
item['url'] = response.url
return self.parse(response)
It seems to work, but it doesn't scrape anything, even if I add parse function to my spider. Does anybody know how to use start_request and rules together? I will be glad any information about this topic. Have a nice coding!
I found a solution, but frankly speaking I don't know how it works but it sertantly does it.
class TSpider(CrawlSpider):
name = 't'
allowed_domains = ['books.toscrapes.com']
start_urls = ['https://books.toscrapes.com']
login_page = 'https://books.toscrapes.com'
rules = (
Rule(LinkExtractor(), callback='parse_item', follow=True),
)
def start_requests(self):
yield Request(url=self.login_page, callback=self.login, errback=self.errback_httpbin, dont_filter=True)
def login(self, response):
return FormRequest.from_response(response)
def parse_item(self, response):
item = {}
item['title'] = response.xpath('//head/title/text()').extract()
item['url'] = response.url
yield item
def errback_httpbin(self, failure):
self.logger.error('ERRRRROR - {}'.format(failure))
To catch errors from your rules you need to define errback for your Rule(). But unfortunately this is not possible now.
You need to parse and yield request by yourself (this way you can use errback) or process each response using middleware.
Here is a solution for handle errback in LinkExtractor
Thanks this dude!
Here is my spider
class Spider(scrapy.Spider):
name = "spider"
start_urls = []
with open("clause/clauses.txt") as f:
for line in f:
start_urls(line)
base_url = "<url>"
start_urls = [base_url + "-".join(url.split()) for url in start_url]
def start_requests(self):
self.log("start_urls - {}".format(self.start_urls))
for url in self.start_urls:
yield scrapy.Request(url, dont_filter=True, priority=2, callback=self.parse)
def parse(self, response):
text_items = response.css("some css").extract()
for text in text_items:
if text == "\n":
continue
yield Item({"text" : text})
yield response.follow(response.css("a::attr(href)").extract_first(), callback=self.parse)
There are 20 start urls, yet Im noticing that only the first 4 urls are actually being called and the rest aren't ever executed. The ideal behavior would be for scrappy to first call all 20 start urls, and then from each continue to the next.
Looks like you have a typo:
start_urls = [base_url + "-".join(url.split()) for url in start_url]
should probably be:
start_urls = [base_url + "-".join(url.split()) for url in start_urls]
Notice the missing s in start_urls.
And I suspect this:
with open("clause/clauses.txt") as f:
for line in f:
start_urls(line)
should be:
with open("clause/clauses.txt") as f:
for line in f:
start_urls.append(line)