Пример #1
0
    def get_performers(self, response):
        jsondata = response.json()
        jsondata = jsondata['responseData']
        for performer in jsondata:
            item = PerformerItem()

            item['name'] = performer['stagename']
            item['image'] = performer['placard']
            if not item['image']:
                item['image'] = None
            item['image_blob'] = None
            item['gender'] = performer['gender']
            item['url'] = self.format_link(response, performer['url'])
            item['network'] = 'Vivid'
            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
Пример #2
0
 def get_performers(self, response):
     performers = response.xpath(
         '//div[contains(@class,"model modelfeature")]')
     for performer in performers:
         item = PerformerItem()
         name = performer.xpath('.//h3/a/text()').get()
         if name:
             item['name'] = html.unescape(name.strip().title())
         item['image'] = performer.xpath(
             './div/a/img/@src0_1x').get().strip()
         item['image_blob'] = None
         url = performer.xpath('.//h3/a/@href').get()
         if url:
             item['url'] = url.strip()
         item['network'] = 'MVG Cash'
         item['astrology'] = ''
         item['bio'] = ''
         item['birthday'] = ''
         item['birthplace'] = ''
         item['cupsize'] = ''
         item['ethnicity'] = ''
         item['eyecolor'] = ''
         item['fakeboobs'] = ''
         item['gender'] = 'Female'
         item['haircolor'] = ''
         item['height'] = ''
         item['measurements'] = ''
         item['nationality'] = ''
         item['piercings'] = ''
         item['tattoos'] = ''
         item['weight'] = ''
         yield item
 def get_performers(self, response):
     jsondata = response.json()['data']['actors']['edges']
     for jsonrow in jsondata:
         item = PerformerItem()
         item['name'] = jsonrow['node']['stageName']
         item['url'] = "https://arxbucks.com"
         item['image'] = jsonrow['node']['thumbnailUrl']
         item['gender'] = jsonrow['node']['gender'].title()
         if item['gender'] == "Transsexual":
             item['gender'] = "Trans"
         # ~ item['image'] = jsonrow['node']['thumbnailUrl']
         # ~ if "https" not in item['image']:
         item['image'] = None
         item['image_blob'] = None
         item['birthday'] = jsonrow['node']['dob']
         if not item['birthday']:
             item['birthday'] = ''
         item['network'] = "ARX Bucks"
         item['bio'] = ''
         item['astrology'] = ''
         item['birthplace'] = ''
         item['ethnicity'] = ''
         item['haircolor'] = ''
         item['weight'] = ''
         item['height'] = ''
         item['measurements'] = ''
         item['tattoos'] = ''
         item['piercings'] = ''
         item['cupsize'] = ''
         item['fakeboobs'] = ''
         item['eyecolor'] = ''
         item['nationality'] = ''
         if item['name']:
             yield item
    def get_performers(self, response):
        meta = response.meta
        jsondata = response.json()['data']['talent']['list']['result']['edges']
        for jsonrow in jsondata:
            item = PerformerItem()
            item['name'] = jsonrow['node']['name']
            item['height'] = str(jsonrow['node']['dimensions']['height']) + "cm"
            item['weight'] = str(jsonrow['node']['dimensions']['weight']) + "kg"
            if jsonrow['node']['dimensions']['measurements']['cup']:
                cup = jsonrow['node']['dimensions']['measurements']['cup']
                cupvalue = re.search(r'(\d{2,3})', cup).group(1)
                cupsize = re.search(r'([A-Za-z]+)', cup).group(1)
                item['cupsize'] = str(round(int(cupvalue) / 2.54)) + cupsize

            if jsonrow['node']['dimensions']['measurements']['waist'] and jsonrow['node']['dimensions']['measurements']['hips']:
                waist = str(round(int(jsonrow['node']['dimensions']['measurements']['waist']) / 2.54))
                hips = str(round(int(jsonrow['node']['dimensions']['measurements']['hips']) / 2.54))
                item['measurements'] = item['cupsize'] + "-" + waist + "-" + hips
            item['network'] = "Fit 18"
            item['url'] = "https://fit18.com/models/" + jsonrow['node']['talentId']
            item['gender'] = 'Female'
            item['bio'] = ''
            item['birthday'] = ''
            item['astrology'] = ''
            item['birthplace'] = ''
            item['ethnicity'] = ''
            item['nationality'] = ''
            item['haircolor'] = ''
            item['tattoos'] = ''
            item['piercings'] = ''
            item['fakeboobs'] = ''
            item['eyecolor'] = ''
            meta['item'] = item.copy()

            imagequery = {
                "operationName": "BatchFindAssetQuery",
                "variables": {
                    "paths": [
                        "/members/models/" + jsonrow['node']['talentId'] + "/profile-sm.jpg"
                    ]
                },
                "query": "query BatchFindAssetQuery($paths: [String!]!) {\n  asset {\n    batch(input: {paths: $paths}) {\n      result {\nserve {\n uri\n}\n}\n}\n}\n}\n"}
            url = "https://fit18.team18.app/graphql"
            imagequery = json.dumps(imagequery)
            yield Request(url, headers=self.headers, body=imagequery, method="POST", callback=self.get_images, meta=meta)
    def get_performers(self, response):
        jsondata = response.json()
        data = jsondata['template']
        data = data.replace("\n",
                            "").replace("\t",
                                        "").replace("\r",
                                                    "").replace("  ",
                                                                " ").strip()
        data = data.replace("  ", " ").strip()
        data = data.replace("  ", " ").strip()
        data = data.replace("  ", " ").strip()
        sel = Selector(text=data)
        performers = sel.xpath('//thumb-component')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('./@title')
            if name:
                item['name'] = html.unescape(name.get().strip().title())

            image = performer.xpath('./@url-thumb')
            if image:
                item['image'] = self.format_link(response, image.get())
            else:
                item['image'] = None

            item['image_blob'] = None

            url = performer.xpath('./@link-content')
            if url:
                item['url'] = self.format_link(response,
                                               url.get().strip()).replace(
                                                   " ", "%20")
            else:
                item['url'] = response.url

            item['network'] = 'JavHD'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Female'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
Пример #6
0
    def get_performers(self, response):
        performers = response.xpath('//div[@class="updateItem model"]')
        for performerrow in performers:

            item = PerformerItem()
            image = performerrow.xpath('./div/a/img/@src0_1x').get()
            if image:
                item['image'] = "https://www.philavise.com" + image.strip()
            else:
                image = None
            item['image_blob'] = None
            performer = performerrow.xpath('./p/a/text()').get()
            if performer:
                item['name'] = performer.strip()
            else:
                performer = False
            url = performerrow.xpath('./div/a/@href').get()
            if url:
                item['url'] = url.strip()
            else:
                url = False

            item['network'] = "Philavise"
            item['bio'] = ''
            item['gender'] = ''
            item['astrology'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['ethnicity'] = ''
            item['nationality'] = ''
            item['haircolor'] = ''
            item['eyecolor'] = ''
            item['weight'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['tattoos'] = ''
            item['piercings'] = ''
            item['cupsize'] = ''
            item['fakeboobs'] = ''

            if performer and image and url:
                yield item
            item.clear()
Пример #7
0
    def parse_performer(self, response):
        item = PerformerItem()
        jsondata = response.xpath(
            '//script[@type="application/ld+json"]/text()').get()
        jsondata = jsondata.replace("\r\n", "")
        data = json.loads(jsondata.strip())
        item['name'] = data['name']
        item['network'] = 'POVR'
        item['url'] = data['mainEntityOfPage']
        item['image'] = data['image']
        item['image_blob'] = None
        bio = response.xpath(
            '//div[contains(@class, "player__description")]/p/text()')
        if bio:
            item['bio'] = bio.get().strip()
        else:
            item['bio'] = None

        if 'gender' in data:
            item['gender'] = data['gender'].title()
        else:
            item['gender'] = None

        item['birthday'] = None
        if 'birthDate' in data:
            if data['birthDate'] and data['birthDate'] > '1950-01-01':
                item['birthday'] = dateparser.parse(data['birthDate'],
                                                    date_formats=['%Y-%m-%d'],
                                                    settings={
                                                        'TIMEZONE': 'UTC'
                                                    }).isoformat()

        item['astrology'] = None
        item['ethnicity'] = None
        if 'birthPlace' in data:
            item['birthplace'] = data['birthPlace']
        else:
            item['birthplace'] = None

        if 'height' in data:
            item['height'] = data['height']
        else:
            item['height'] = None

        item['nationality'] = None
        item['haircolor'] = None
        item['eyecolor'] = None
        item['weight'] = None
        item['measurements'] = None
        item['cupsize'] = None
        item['tattoos'] = None
        item['piercings'] = None
        item['fakeboobs'] = None

        yield item
    def parse_performerpage(self, response):
        global json
        itemlist = []

        jsondata = json.loads(response.text)
        data = jsondata['data']
        for jsonentry in data:
            item = PerformerItem()

            item['name'] = jsonentry['name'].title().strip()
            item['network'] = 'Teen Core Club'
            item[
                'url'] = "https://www.teencoreclub.com/browsevideos/actor/" + str(
                    jsonentry['id']) + "/" + jsonentry['name'].replace(
                        " ", "%20").strip()
            item['image'] = "https://www.teencoreclub.com" + jsonentry[
                'image'].replace("\\", "").strip()
            item['image_blob'] = None
            item['bio'] = jsonentry['bio']
            if not item['bio']:
                item['bio'] = ''

            item['gender'] = "Female"
            item['birthday'] = ''
            item['astrology'] = ''
            item['birthplace'] = ''
            item['ethnicity'] = ''
            item['nationality'] = ''
            item['haircolor'] = ''
            item['weight'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['tattoos'] = ''
            item['piercings'] = ''
            item['cupsize'] = ''
            item['fakeboobs'] = ''
            item['eyecolor'] = ''

            itemlist.append(item.copy())
            item.clear()

        return itemlist
    def get_performers(self, response):
        performers = response.xpath('//div[@class="update_details"]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath(
                './a[1][contains(@href,"/models/")]/text()').get()
            if name:
                item['name'] = name.strip()
            else:
                item['name'] = ''

            image = performer.xpath('.//img/@src0_3x').get()
            if not image:
                image = performer.xpath('.//img/@src0_2x').get()
            if not image:
                image = performer.xpath('.//img/@src0_1x').get()

            if image:
                item['image'] = "https://www.claudiamarie.com" + image
            else:
                item['image'] = None
            item['image_blob'] = None

            url = performer.xpath(
                './a[1][contains(@href,"/models/")]/@href').get()
            if url:
                item['url'] = url.strip()
            else:
                item['url'] = ''

            item['network'] = "Claudia Marie"

            item['bio'] = ''
            item['gender'] = ''
            item['birthday'] = ''
            item['astrology'] = ''
            item['birthplace'] = ''
            item['ethnicity'] = ''
            item['nationality'] = ''
            item['haircolor'] = ''
            item['weight'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['tattoos'] = ''
            item['piercings'] = ''
            item['cupsize'] = ''
            item['fakeboobs'] = ''
            item['eyecolor'] = ''

            if item['name'] and item['url']:
                yield item
    def get_performers(self, response):
        global json
        performers = json.loads(response.text)

        for performer in performers:
            item = PerformerItem()

            item['name'] = string.capwords(performer['fullName'])
            item['image'] = performer['previewImage960']
            if not item['image']:
                item['image'] = None
            item['image_blob'] = None
            item['url'] = "https://www.mrbigfatdick.com/models/" + performer['permaLink']
            item['height'] = str(performer['height']) + "cm"
            item['weight'] = str(performer['weight']) + "kg"

            if performer['eyes']:
                item['eyecolor'] = performer['eyes'].title()
            else:
                item['eyecolor'] = ''

            if performer['hair']:
                item['haircolor'] = performer['hair'].title()
            else:
                item['haircolor'] = ''

            if "m" in performer['gender']:
                item['gender'] = "Male"
            else:
                item['gender'] = "Female"

            item['network'] = 'MrBigfatdick'

            if performer['description']:
                item['bio'] = performer['description'].title()
            else:
                item['bio'] = ''

            item['astrology'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['fakeboobs'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''

            yield item
    def get_performers(self, response):
        performers = response.xpath(
            '//div[contains(@class,"overflow-hidden") and ./a[contains(@href,"/performers/")]]'
        )
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('.//a/p/text()').get()
            if name:
                item['name'] = name.strip()
            else:
                item['name'] = ''

            image = performer.xpath('.//source/@data-srcset').get()
            if image:
                image = re.search(r'(.*)\?', image).group(1)
                if image:
                    item['image'] = image.strip()

            if not image:
                item['image'] = None
            item['image_blob'] = None

            url = performer.xpath('./div/a/@href').get()
            if url:
                item['url'] = "https://xconfessions.com/" + url.strip()
            else:
                item['url'] = ''

            item['network'] = 'XConfessions'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = ''
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
    def get_performers(self, response):
        performers = response.xpath(
            '//div[@class="actor-list"]/div[@class="item"]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('./a[@class="title"]/text()')
            if name:
                item['name'] = html.unescape(name.get().strip().title())

            image = performer.xpath('./a/@style')
            if image:
                image = image.get()
                image = re.search(r'.*(https.*?\.jpg).*', image).group(1)
                item['image'] = self.format_link(response, image)
            else:
                item['image'] = None

            item['image_blob'] = None

            url = performer.xpath('./a[1]/@href')
            if url:
                item['url'] = self.format_link(response,
                                               url.get().strip()).replace(
                                                   " ", "%20")
            else:
                item['url'] = response.url

            item['network'] = 'POVR'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Female'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
    def get_performers(self, response):
        performers = response.xpath('//div[@class="modelPic"]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath(
                './div[@class="modelName"]/p/a/text()').get()
            if name:
                item['name'] = html.unescape(name.strip().title())

            image = performer.xpath(
                './a/img/@src0_3x|./a/img/@src0_2x|./a/img/@src0_1x|./a/img/@src0'
            ).get()
            if image:
                item['image'] = image.strip()
            else:
                item['image'] = None
            item['image_blob'] = None

            url = performer.xpath('./a[1]/@href').get()
            if url:
                if "sets.php" in url:
                    uri = urlparse(response.url)
                    base = uri.scheme + "://" + uri.netloc
                    url = base + "/" + url.strip()
                item['url'] = url.strip()

            item['network'] = 'Nebraska Coeds'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Female'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
Пример #14
0
    def get_performers(self, response):
        performers = response.xpath(
            '//div[contains(@class, "performer-list-block")]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('./h3/a/text()')
            if name:
                item['name'] = html.unescape(name.get().strip().title())

            image = performer.xpath('./a/img/@src')
            if image:
                image = image.get()
                item['image'] = self.format_link(response, image)
            else:
                item['image'] = None

            item['image_blob'] = None

            url = performer.xpath('./h3/a/@href')
            if url:
                item['url'] = self.format_link(response,
                                               url.get().strip()).replace(
                                                   " ", "%20")
            else:
                item['url'] = response.url

            item['network'] = 'Two Tgirls'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Trans'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
    def get_performers(self, response):
        performers = response.xpath('//div[@class="modelPic"]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('./div/p/a/text()').get()
            if name:
                item['name'] = html.unescape(name.strip().title())

            image = performer.xpath('.//a/img/@src0_2x').get()
            if image:
                item[
                    'image'] = "https://lasvegasamateurs.com/" + image.replace(
                        " ", "%20")
            else:
                item['image'] = None
            item['image'] = None

            item['image_blob'] = None

            url = performer.xpath(
                './a[contains(@href, "/models/")][1]/@href').get()
            if url:
                item['url'] = self.format_link(response, url.strip()).replace(
                    " ", "%20")

            item['network'] = 'Las Vegas Amateurs'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Female'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
Пример #16
0
    def get_performers(self, response):
        meta = response.meta
        performers = response.xpath('//div[@class="item-portrait"]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('.//h4/a/text()').get()
            if name:
                item['name'] = html.unescape(name.strip().title())

            image = performer.xpath('./a/img/@src0_1x').get()
            if image:
                item['image'] = image.strip()
            else:
                item['image'] = None
            item['image_blob'] = None

            url = performer.xpath('./a[1]/@href').get()
            if url:
                item['url'] = url.strip()

            item['network'] = 'Genuine Sin'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            if 'g=m' in meta['pagination']:
                item['gender'] = "Male"
            else:
                item['gender'] = "Female"

            yield item
    def get_performers(self, response):
        performers = response.xpath('//div[@class="model-item"]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('.//h3/a/text()')
            if name:
                item['name'] = html.unescape(name.get().strip().title())

            image = performer.xpath('./div[@class="thumb-wrap"]/img[1]/@src')
            if image:
                item['image'] = self.format_link(response, image.get())
            else:
                item['image'] = None

            item['image_blob'] = None

            url = performer.xpath('.//h3/a/@href')
            if url:
                item['url'] = self.format_link(response,
                                               url.get().strip()).replace(
                                                   " ", "%20")
            else:
                item['url'] = response.url

            item['network'] = 'POV Perv'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Female'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
    def get_performers(self, response):
        response_text = response.text
        response_text = response_text.replace("\\", "")
        response_xpath = Selector(text=response_text)

        performers = response_xpath.xpath('//a[contains(@href,"amatrice=")]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('.//div[@class="pull-left"]/text()').get()
            if name:
                name = unidecode(name)
                item['name'] = html.unescape(name.strip().title())

            image = performer.xpath('.//img/@src').get()
            if image:
                item['image'] = image.strip()
            else:
                item['image'] = None
            item['image_blob'] = None

            url = performer.xpath('./@href').get()
            if url:
                item['url'] = "https://www.lafranceapoil.com/en/" + url.strip()

            item['network'] = 'La France a Poil'

            item['height'] = ''
            item['weight'] = ''
            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Female'
            item['haircolor'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            yield item
Пример #19
0
    def get_performers(self, response):
        performers = response.xpath(
            '//div[contains(@class, "portfolio-item")]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('./a/div/div/text()').get()
            if name:
                item['name'] = string.capwords(html.unescape(name.strip()))

            image = performer.xpath('./a/div/img/@src').get()
            if image:
                item['image'] = self.format_link(response,
                                                 image).replace(" ", "%20")
            else:
                item['image'] = None
            item['image_blob'] = None

            url = performer.xpath('./a/@href').get()
            if url:
                item['url'] = self.format_link(response, url.strip()).replace(
                    " ", "%20")

            item['network'] = 'Model Media US'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Female'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
    def get_performers(self, response):
        performers = response.xpath('//div[contains(@class,"card performer")]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath(
                './/a[contains(@class,"primary")]/text()').get()
            if name:
                name = name.replace("More ", "")
                item['name'] = html.unescape(name.strip().title())

            image = performer.xpath('.//span/img/@src').get()
            if image:
                item['image'] = image.strip()
            else:
                item['image'] = None
            item['image_blob'] = None

            url = performer.xpath(
                './/a[contains(@class,"primary")]/@href').get()
            if url:
                item['url'] = url.strip()

            item['network'] = 'Manojob'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Female'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
    def get_performers(self, response):
        performers = response.xpath('//li[contains(@class,"featured-video")]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('.//h5/a/text()').get()
            if name:
                item['name'] = html.unescape(name.strip().title())
            else:
                item['name'] = ''

            image = performer.xpath('./a/img/@src0_1x').get()
            if image:
                image = "https://www.boxtrucksex.com" + image.strip()
                item['image'] = image.strip()
            else:
                item['image'] = None
            item['image_blob'] = None

            url = performer.xpath('.//h5/a/@href').get()
            if url:
                item['url'] = url.strip()

            item['network'] = 'Box Truck Sex'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Female'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
Пример #22
0
    def get_performers(self, response):
        performers = response.xpath('//ul[@id="lstPornstars"]/li')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('./p[1]/a/text()').get()
            if name:
                item['name'] = string.capwords(html.unescape(name.strip()))

            image = performer.xpath('./div/a/img/@src').get()
            if image:
                item['image'] = self.format_link(response,
                                                 image).replace(" ", "%20")
            else:
                item['image'] = None
            item['image_blob'] = None

            url = performer.xpath('./p[1]/a/@href').get()
            if url:
                item['url'] = self.format_link(response, url.strip()).replace(
                    " ", "%20")

            item['network'] = 'Club Filly'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Female'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
    def get_performers(self, response):
        performers = response.xpath('//div[@class="update_details"]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('./a[1]/text()').get()
            if name:
                item['name'] = html.unescape(name.strip().title())

            image = performer.xpath(
                './/img/@src0_3x|.//img/@src0_2x|.//img/@src0_1x').get()
            if image:
                item['image'] = "http://american-pornstar.com" + image.strip()
            else:
                item['image'] = None
            item['image_blob'] = None

            url = performer.xpath('./a[1]/@href').get()
            if url:
                item['url'] = url.strip()

            item['network'] = 'American Pornstar'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = ''
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
    def get_performers(self, response):
        performers = response.xpath(
            '//div[contains(@class,"text-center pad_bottom_15")]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('./h3/text()').get()
            if name:
                item['name'] = html.unescape(name.strip().title())

            image = performer.xpath('./a/img/@src').get()
            if image:
                item['image'] = image.strip()
            else:
                item['image'] = None
            item['image_blob'] = None

            url = performer.xpath('./a/@href').get()
            if url:
                item['url'] = "https://www.trans500.com" + url.strip()

            item['network'] = 'Josh Stone Productions'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Trans'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
    def get_performers(self, response):
        performers = response.xpath('//div[@class="modelPic"]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('./div[@class="modelName"]/p/a/text()').get()
            if name:
                item['name'] = html.unescape(name.strip().title())

            image = performer.xpath('.//img/@src0_3x|.//img/@src0_2x|.//img/@src0_1x').get()
            if image:
                item['image'] = "https://www.bigbootytgirls.com/" + image.strip()
            else:
                item['image'] = None
            item['image_blob'] = None

            url = performer.xpath('./a/@href').get()
            if url:
                item['url'] = url.strip()

            item['network'] = 'Big Booty Tgirls'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Trans'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
    def get_performers(self, response):
        performers = response.xpath('//div[@class="model"]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('./div/p/text()').get()
            if name:
                item['name'] = html.unescape(name.strip().title())
            else:
                item['name'] = ''

            image = performer.xpath('.//img/@src').get()
            if image:
                item['image'] = image.strip()
            else:
                item['image'] = None
            item['image_blob'] = None

            item['url'] = response.url

            item['network'] = 'Monger in Asia'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Female'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
    def get_performers(self, response):
        performers = response.xpath('//div[@class="update_details"]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('./a[1]/text()').get()
            if name:
                item['name'] = html.unescape(name.strip().title())

            image = performer.xpath('.//img/@src0_2x').get()
            if image:
                item['image'] = self.format_link(response, image.strip())
            else:
                item['image'] = None
            item['image_blob'] = None

            item['url'] = performer.xpath('./a[1]/@href').get().strip()

            item['network'] = 'Older Woman Fun'

            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Female'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
    def get_performers(self, response):
        performers = response.xpath(
            '//div[@class="row"]/a[contains(@href, "/girls")]')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('./div/p/text()').get()
            if name:
                item['name'] = html.unescape(name.strip().title())

            image = performer.xpath('./div/div[@class="img"]/@style').get()
            image = re.search(r'\'(http.*)\'', image)
            if image:
                item['image'] = image.group(1)
            else:
                item['image'] = None
            item['image_blob'] = None

            item['url'] = performer.xpath('./@href').get()
            item['network'] = 'My Sex Mobile'
            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Female'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
Пример #29
0
    def get_performers(self, response):
        performers = response.xpath('//div[@class="modelsgrid"]/div')
        for performer in performers:
            item = PerformerItem()

            name = performer.xpath('./h3/a/text()').get()
            if name:
                item['name'] = html.unescape(name.strip().title())

            image = performer.xpath('./a/img/@src').get()
            if image:
                item['image'] = "https://www.kinkymistresses.com" + image
            else:
                item['image'] = None
            item['image_blob'] = None

            item['url'] = "https://www.kinkymistresses.com/" + performer.xpath(
                './a/@href').get()
            item['network'] = 'Kinky Mistresses'
            item['astrology'] = ''
            item['bio'] = ''
            item['birthday'] = ''
            item['birthplace'] = ''
            item['cupsize'] = ''
            item['ethnicity'] = ''
            item['eyecolor'] = ''
            item['fakeboobs'] = ''
            item['gender'] = 'Female'
            item['haircolor'] = ''
            item['height'] = ''
            item['measurements'] = ''
            item['nationality'] = ''
            item['piercings'] = ''
            item['tattoos'] = ''
            item['weight'] = ''

            yield item
Пример #30
0
    def get_performers(self, response):
        item_list = []
        jsondata = response.json()
        jsondata = jsondata['snippets']
        jsondata = jsondata['snippet-modelsGrid-modelItemsAppend'].lower()
        jsonsel = Selector(text=jsondata)
        performers = jsonsel.xpath(
            '//div[contains(@class,"color_12-shadow-sm-hover")]')
        count = 0
        for performer in performers:
            count = count + 1
            item = PerformerItem()
            item['bio'] = ''
            item['gender'] = ''
            item['birthday'] = ''
            item['astrology'] = ''
            item['birthplace'] = ''
            item['ethnicity'] = ''
            item['nationality'] = ''
            item['haircolor'] = ''
            item['measurements'] = ''
            item['tattoos'] = ''
            item['piercings'] = ''
            item['fakeboobs'] = ''
            item['eyecolor'] = ''
            item['cupsize'] = ''
            item['height'] = ''
            item['weight'] = ''
            item['network'] = "PornCZ"
            name = performer.xpath('./div/h3/a/text()').get()
            if name:
                item['name'] = name.strip().title()

            url = performer.xpath('./a/@href').get()
            if url:
                item['url'] = "https://www.porncz.com/" + url.strip()

            image = performer.xpath('./a/img/@data-src').get()
            if image:
                item['image'] = "https://www.porncz.com" + image.strip()

            item['image_blob'] = None

            descline = performer.xpath('./a/div/p/text()').get()
            if descline:
                descline = descline.replace("-", "").strip()
                if re.search('size:(.*)weight', descline):
                    cupsize = re.search('size:(.*)weight', descline).group(1)
                    if cupsize:
                        item['cupsize'] = cupsize.strip().title()

                if re.search(r'(\d+\ kg)', descline):
                    weight = re.search(r'(\d+\ kg)', descline).group(1)
                    if weight:
                        item['weight'] = weight.strip().title()

                if re.search(r'(\d+\ cm)', descline):
                    height = re.search(r'(\d+\ cm)', descline).group(1)
                    if height:
                        item['height'] = height.strip().title()

            item_list.append(item.copy())
            item.clear()

        return item_list