def get_app_info(self, uri):
res = self.br.open(uri)
data = res.get_data()
soup = BeautifulSoup(data, "html5lib")
info = AppInfo()
info.name = soup.find(id="desktopContentBlockId").h1.text
info.category = ""
info.version = ""
info.size = ""
info.updated = ""
info.price = ""
info.os = "iOS"
developer = soup.find(id="desktopContentBlockId").h2.text
info.developer = developer.split("：")[1] if "：" in developer else developer
info.language = ""
desc = soup.find('div', attrs={"class" : "product-review"}).p.prettify()
info.description = desc
artwork = soup.find(id="left-stack").div.img["src"]
info.artwork = artwork
div_images = soup.find('div', attrs={"class" : "screenshots"})
images = div_images.find_all('img')
info.images = [img["src"] for img in images]
info.debug()
return info
def get_app_info(self, uri):
res = self.br.open(uri)
data = res.get_data()
soup = BeautifulSoup(data, "html5lib", from_encoding="utf-8")
info = AppInfo()
# info.name = soup.find('h1', attrs={"class" : "doc-banner-title"}).text
info.name = soup.find('div', attrs={"class" : "document-title"}).text
info.category = ""
info.version = ""
info.size = ""
info.updated = ""
info.price = ""
info.os = "Android"
# info.developer = soup.find('a', attrs={"class" : "doc-header-link"}).text
info.developer = soup.find('a', attrs={"class" : "document-subtitle"}).text
info.language = ""
# desc = soup.find(id="doc-original-text").prettify()
desc = soup.find('div', attrs={"class" : "show-more-content"}).prettify()
info.description = desc
# artwork = soup.find('div', attrs={"class" : "doc-banner-icon"}).img["src"]
artwork = soup.find('div', attrs={"class" : "cover-container"}).img["src"]
info.artwork = artwork
# div_images = soup.find('div', attrs={"class" : "screenshot-carousel-content-container"})
div_images = soup.find('div', attrs={"class" : "thumbnails"})
images = div_images.find_all('img')
info.images = [img["src"] for img in images]
info.debug()
return info
def get_app_info(self, uri):
res = self.br.open(uri)
data = res.get_data()
soup = BeautifulSoup(data, "html5lib")
info = AppInfo()
div_viewbox = soup.find('div', attrs={"class" : "viewbox"})
info.name = div_viewbox.h2.renderContents()
info.category = ""
info.version = ""
info.size = ""
info.updated = ""
info.price = ""
info.os = "iOS"
info.developer = ""
info.language = ""
desc = soup.find('div', attrs={"class" : "content"}).prettify()
info.description = desc
artwork = soup.find('div', attrs={"class" : "picview"}).img["src"]
info.artwork = artwork
div_images = soup.find('div', attrs={"class" : "content"})
images = div_images.find_all('img')
info.images = [img["src"] for img in images]
info.debug()
return info
def get_wp_app_info(self, uri):
res = self.br.open(uri)
data = res.get_data()
soup = BeautifulSoup(data, "html5lib")
info = AppInfo()
div_viewbox = soup.find('div', attrs={"class" : "info"})
info.name = div_viewbox.h1.text.strip()
info.category = ""
info.version = ""
info.size = ""
info.updated = ""
info.price = ""
info.os = "iOS"
info.developer = ""
info.language = ""
desc = soup.find(id="discription").prettify() # not description
info.description = desc
artwork = soup.find('div', attrs={"class" : "deinfo"}).img["src"]
info.artwork = self.get_url(uri, artwork)
div_images = soup.find('div', attrs={"class" : "screenshots-container"})
images = div_images.find_all('img')
info.images = [self.get_url(uri, img["src"]) for img in images]
info.debug()
return info
def get_android_app_info(self, uri):
res = self.br.open(uri)
data = res.get_data()
soup = BeautifulSoup(data, "html5lib")
info = AppInfo()
div_viewbox = soup.find('div', attrs={"class" : "info_title"})
info.name = div_viewbox.text
info.category = ""
info.version = ""
info.size = ""
info.updated = ""
info.price = ""
info.os = "Android"
info.developer = ""
info.language = ""
desc = soup.find('div', attrs={"class" : "rom_introductioncon yingyong_intro"}).prettify()
info.description = desc
artwork = soup.find('div', attrs={"class" : "yingyong_img"}).img["src"]
info.artwork = artwork
div_images = soup.find('div', attrs={"class" : "snapshot_list"})
images = div_images.find_all('img')
info.images = [img["src"] for img in images]
info.debug()
return info
def get_ios_app_info(self, uri):
res = self.br.open(uri)
data = res.get_data()
soup = BeautifulSoup(data, "html5lib")
info = AppInfo()
info.name = soup.find(id="appTitle").text.strip()
info.category = ""
info.version = ""
info.size = ""
info.updated = ""
info.price = ""
info.os = "iOS"
info.developer = ""
info.language = ""
desc = soup.find(id="tab2Content").prettify()
info.description = desc
artwork = soup.find('img', attrs={"class" : "img175"})["src"]
info.artwork = self.get_url(uri, artwork)
div_images = soup.find('div', attrs={"class" : "yyxq_yyjt"})
images = div_images.find_all('img')
info.images = [self.get_url(uri, img["src"]) for img in images]
info.debug()
return info
def get_app_info(self, uri):
res = self.br.open(uri)
data = res.get_data()
soup = BeautifulSoup(data, "html5lib")
info = AppInfo()
div_viewbox = soup.find(id="application")
info.name = div_viewbox.h1.renderContents()
info.category = ""
info.version = ""
info.size = ""
info.updated = ""
info.price = ""
info.os = "Windows Phone"
info.developer = ""
info.language = ""
desc = soup.find(id="appDetails").pre.prettify()
info.description = desc
artwork = soup.find(id="appSummary").img["src"]
info.artwork = artwork
div_images = soup.find(id="screenshots")
images = div_images.find_all('a')
info.images = [img["href"] for img in images]
info.debug()
return info
def new_app_info(self): info = AppInfo() info.name = "Motion Math Zoom" info.category = "Education" info.version = "2.1" info.size = "19.1MB" info.updated = "2011-12-30" info.price = "Free" info.developer = "Motion Math" info.language = "English" info.os = "iOS" info.description = "<p>An animal adventure through the world of numbers! Give your child a chance to play with numbers - they'll have a blast zooming through the number line as they master place value.</p>" info.artwork = "https://lh3.ggpht.com/LDtlOWSOvxXBDhx4XqcBBGkANPlusko8WgB80n62bpnyTUgz34gl9OlOpmflckVNBA=h230" return info