Scrapy CrawlSpider isn't following the links on a particular page

Question

I have made a spider to crawl a forum that requires a login. I start it off on the login page. The problem occurs with the page that I direct the spider to after the login was successful.

If I open up my rules to accept all links, the spider successfully follows the links on the login page. However, it doesn't follow any of the links on the page that I feed it using Request(). This suggests that it isn't because of screwing up the xpath.

The login appears to work - the page_parse function writes the page source to a text file, and the source is from the page I'm looking for, which can only be reached after logging in. However, the pipeline I have in place to take a screenshot of each page captures the login page, but not this page that I then send it on to.

Here is the spider:

class PLMSpider(CrawlSpider):
    name = 'plm'
    allowed_domains = ["patientslikeme.com"]
    start_urls = [
        "https://www.patientslikeme.com/login"
    ]

    rules = (
        Rule(SgmlLinkExtractor(allow=(r"patientslikeme.com/login")), callback='login_parse', follow=True),
        Rule(SgmlLinkExtractor(restrict_xpaths=("//div[@class='content-section']")), callback='post_parse', follow=False),
        Rule(SgmlLinkExtractor(restrict_xpaths=("//div[@class='pagination']")), callback='page_parse', follow=True),
    )

    def __init__(self, **kwargs):
        ScrapyFileLogObserver(open("debug.log", 'w'), level=logging.DEBUG).start()
        CrawlSpider.__init__(self, **kwargs)

    def post_parse(self, response):
        url = response.url
        log.msg("Post parse attempted for {0}".format(url))
        item = PLMItem()
        item['url'] = url
        return item

    def page_parse(self, response):
        url = response.url
        log.msg("Page parse attempted for {0}".format(url))
        item = PLMItem()
        item['url'] = url
        f = open("body.txt", "w")
        f.write(response.body)
        f.close()
        return item

    def login_parse(self, response):
        log.msg("Login attempted")
        return [FormRequest.from_response(response,
                    formdata={'userlogin[login]': username, 'userlogin[password]': password},
                    callback=self.after_login)]

    def after_login(self, response):
        log.msg("Post login")
        if "Login unsuccessful" in response.body:
            self.log("Login failed", level=log.ERROR)
            return
        else:
            return Request(url="https://www.patientslikeme.com/forum/diabetes2/topics",
               callback=self.page_parse)

And here is my debug log:

2014-03-21 18:22:05+0000 [scrapy] INFO: Scrapy 0.18.2 started (bot: plm)
2014-03-21 18:22:05+0000 [scrapy] DEBUG: Optional features available: ssl, http11
2014-03-21 18:22:05+0000 [scrapy] DEBUG: Overridden settings: {'NEWSPIDER_MODULE': 'plm.spiders', 'ITEM_PIPELINES': {'plm.pipelines.ScreenshotPipeline': 1}, 'DEPTH_LIMIT': 5, 'SPIDER_MODULES': ['plm.spiders'], 'BOT_NAME': 'plm', 'DEPTH_PRIORITY': 1, 'SCHEDULER_MEMORY_QUEUE': 'scrapy.squeue.FifoMemoryQueue', 'SCHEDULER_DISK_QUEUE': 'scrapy.squeue.PickleFifoDiskQueue'}
2014-03-21 18:22:05+0000 [scrapy] DEBUG: Enabled extensions: LogStats, TelnetConsole, CloseSpider, WebService, CoreStats, SpiderState
2014-03-21 18:22:06+0000 [scrapy] DEBUG: Enabled downloader middlewares: HttpAuthMiddleware, DownloadTimeoutMiddleware, UserAgentMiddleware, RetryMiddleware, DefaultHeadersMiddleware, MetaRefreshMiddleware, HttpCompressionMiddleware, RedirectMiddleware, CookiesMiddleware, ChunkedTransferMiddleware, DownloaderStats
2014-03-21 18:22:06+0000 [scrapy] DEBUG: Enabled spider middlewares: HttpErrorMiddleware, OffsiteMiddleware, RefererMiddleware, UrlLengthMiddleware, DepthMiddleware
2014-03-21 18:22:06+0000 [scrapy] DEBUG: Enabled item pipelines: ScreenshotPipeline
2014-03-21 18:22:06+0000 [plm] INFO: Spider opened
2014-03-21 18:22:06+0000 [plm] INFO: Crawled 0 pages (at 0 pages/min), scraped 0 items (at 0 items/min)
2014-03-21 18:22:07+0000 [scrapy] INFO: Screenshooter initiated
2014-03-21 18:22:07+0000 [scrapy] DEBUG: Telnet console listening on 0.0.0.0:6023
2014-03-21 18:22:07+0000 [scrapy] DEBUG: Web service listening on 0.0.0.0:6080
2014-03-21 18:22:08+0000 [plm] DEBUG: Crawled (200) <GET https://www.patientslikeme.com/login> (referer: None)
2014-03-21 18:22:08+0000 [plm] DEBUG: Crawled (200) <GET https://www.patientslikeme.com/login> (referer: https://www.patientslikeme.com/login)
2014-03-21 18:22:08+0000 [scrapy] INFO: Login attempted
2014-03-21 18:22:08+0000 [plm] DEBUG: Filtered duplicate request: <GET https://www.patientslikeme.com/login> - no more duplicates will be shown (see DUPEFILTER_CLASS)
2014-03-21 18:22:09+0000 [plm] DEBUG: Redirecting (302) to <GET https://www.patientslikeme.com/profile/activity/all> from <POST https://www.patientslikeme.com/login>
2014-03-21 18:22:10+0000 [plm] DEBUG: Crawled (200) <GET https://www.patientslikeme.com/profile/activity/all> (referer: https://www.patientslikeme.com/login)
2014-03-21 18:22:10+0000 [scrapy] INFO: Post login
2014-03-21 18:22:10+0000 [plm] DEBUG: Crawled (200) <GET https://www.patientslikeme.com/forum/diabetes2/topics> (referer: https://www.patientslikeme.com/profile/activity/all)
2014-03-21 18:22:10+0000 [scrapy] INFO: Page parse attempted for https://www.patientslikeme.com/forum/diabetes2/topics
2014-03-21 18:22:10+0000 [scrapy] INFO: Screenshot attempted for https://www.patientslikeme.com/forum/diabetes2/topics
2014-03-21 18:22:15+0000 [plm] DEBUG: Scraped from <200 https://www.patientslikeme.com/forum/diabetes2/topics>

    {'url': 'https://www.patientslikeme.com/forum/diabetes2/topics'}
2014-03-21 18:22:15+0000 [plm] INFO: Closing spider (finished)
2014-03-21 18:22:15+0000 [plm] INFO: Dumping Scrapy stats:
    {'downloader/request_bytes': 2068,
     'downloader/request_count': 5,
     'downloader/request_method_count/GET': 4,
     'downloader/request_method_count/POST': 1,
     'downloader/response_bytes': 53246,
     'downloader/response_count': 5,
     'downloader/response_status_count/200': 4,
     'downloader/response_status_count/302': 1,
     'finish_reason': 'finished',
     'finish_time': datetime.datetime(2014, 3, 21, 18, 22, 15, 177000),
     'item_scraped_count': 1,
     'log_count/DEBUG': 13,
     'log_count/INFO': 8,
     'request_depth_max': 3,
     'response_received_count': 4,
     'scheduler/dequeued': 5,
     'scheduler/dequeued/memory': 5,
     'scheduler/enqueued': 5,
     'scheduler/enqueued/memory': 5,
     'start_time': datetime.datetime(2014, 3, 21, 18, 22, 6, 377000)}
2014-03-21 18:22:15+0000 [plm] INFO: Spider closed (finished)

Thanks for any help you can give.

---- EDIT ----

I have tried to implement Paul t.'s suggestion. Unfortunately, I'm getting the following error:

    Traceback (most recent call last):
      File "C:\Python27\lib\site-packages\scrapy\crawler.py", line 93, in start
        if self.start_crawling():
      File "C:\Python27\lib\site-packages\scrapy\crawler.py", line 168, in start_crawling
        return self.start_crawler() is not None
      File "C:\Python27\lib\site-packages\scrapy\crawler.py", line 158, in start_crawler
        crawler.start()
      File "C:\Python27\lib\site-packages\twisted\internet\defer.py", line 1213, in unwindGenerator
        return _inlineCallbacks(None, gen, Deferred())
    --- <exception caught here> ---
      File "C:\Python27\lib\site-packages\twisted\internet\defer.py", line 1070, in _inlineCallbacks
        result = g.send(result)
      File "C:\Python27\lib\site-packages\scrapy\crawler.py", line 74, in start
        yield self.schedule(spider, batches)
      File "C:\Python27\lib\site-packages\scrapy\crawler.py", line 61, in schedule
        requests.extend(batch)
    exceptions.TypeError: 'Request' object is not iterable

Since it isn't identifying a particular part of the spider that's to blame, I'm struggling to work out where the problem is.

---- EDIT 2 ----

The problem was being caused by the start_requests function provided by Paul t., which used return rather than yield. If I change it to yield, it works perfectly.

Answer 1

My suggestion is to trick CrawlSpider with:

• a manually crafted request to the login page,
• performing the login,
• and only then do as if CrawlSpider was starting with start_urls , using CrawlSpider's "magic"

Here's an illustration of that:

class PLMSpider(CrawlSpider):
    name = 'plm'
    allowed_domains = ["patientslikeme.com"]

    # pseudo-start_url
    login_url = "https://www.patientslikeme.com/login"

    # start URLs used after login
    start_urls = [
        "https://www.patientslikeme.com/forum/diabetes2/topics",
    ]

    rules = (
        # you want to do the login only once, so it's probably cleaner
        # not to ask the CrawlSpider to follow links to the login page
        #Rule(SgmlLinkExtractor(allow=(r"patientslikeme.com/login")), callback='login_parse', follow=True),

        # you can also deny "/login" to be safe
        Rule(SgmlLinkExtractor(restrict_xpaths=("//div[@class='content-section']"),
                               deny=('/login',)),
             callback='post_parse', follow=False),

        Rule(SgmlLinkExtractor(restrict_xpaths=("//div[@class='pagination']"),
                               deny=('/login',)),
             callback='page_parse', follow=True),
    )

    def __init__(self, **kwargs):
        ScrapyFileLogObserver(open("debug.log", 'w'), level=logging.DEBUG).start()
        CrawlSpider.__init__(self, **kwargs)

    # by default start_urls pages will be sent to the parse method,
    # but parse() is rather special in CrawlSpider
    # so I suggest you create your own initial login request "manually"
    # and ask for it to be parsed by your specific callback
    def start_requests(self):
        yield Request(self.login_url, callback=self.login_parse)

    # you've got the login page, send credentials
    # (so far so good...)
    def login_parse(self, response):
        log.msg("Login attempted")
        return [FormRequest.from_response(response,
                    formdata={'userlogin[login]': username, 'userlogin[password]': password},
                    callback=self.after_login)]

    # so we got a response to the login thing
    # if we're good,
    # just do as if we were starting the crawl now,
    # basically doing what happens when you use start_urls
    def after_login(self, response):
        log.msg("Post login")
        if "Login unsuccessful" in response.body:
            self.log("Login failed", level=log.ERROR)
            return
        else:
            return [Request(url=u) for u in self.start_urls]
            # alternatively, you could even call CrawlSpider's start_requests() method directly
            # that's probably cleaner
            #return super(PLMSpider, self).start_requests()

    def post_parse(self, response):
        url = response.url
        log.msg("Post parse attempted for {0}".format(url))
        item = PLMItem()
        item['url'] = url
        return item

    def page_parse(self, response):
        url = response.url
        log.msg("Page parse attempted for {0}".format(url))
        item = PLMItem()
        item['url'] = url
        f = open("body.txt", "w")
        f.write(response.body)
        f.close()
        return item

    # if you want the start_urls pages to be parsed,
    # you need to tell CrawlSpider to do so by defining parse_start_url attribute
    # https://github.com/scrapy/scrapy/blob/master/scrapy/contrib/spiders/crawl.py#L38
    parse_start_url = page_parse

Answer 2

Your login page is parsed by method parse_start_url . You should redefine the method to parse the login page. Have a look at documentation .