def get_basic_info(html_info):
"""
获取游戏的基本信息
:param html_info:
:return: dict
"""
result = {}
html = html_info
if html:
screenshot_url = ""
developers = ""
category = ""
pname = ""
download = ""
install_num = ""
version = ""
size = 0
api_level = ""
icon_url = ""
soup = BeautifulSoup(html)
display_name = soup.find("h1", {"class": "app-name"})
introduction = soup.find("div", {"class": "brief-long"})
short_desc = soup.find("span", {"class": "head-content"})
app_tags = soup.find("div", {"class": "nav"})
screenshot_info = soup.find("div", {"class": "section-body"})
star_percent = soup.find("span", {"class": "star-percent"})
params_download_num = soup.find("span", {"class": "download-num"})
params_platform = soup.find("span", {"class": "params-platform"})
data_info = soup.find("a", {"class": "inst-btn-big highspeed"})
if data_info:
download = data_info["data_url"]
data_size = data_info["data_size"]
data_ver_name = data_info["data_versionname"]
# data_ver_code = data_info["data_versioncode"]
data_pkg_name = data_info["data_package"]
icon_url = data_info["data_icon"]
size = data_size or 0
version = data_ver_name
# version_code = data_ver_code or 0
pname = data_pkg_name
# 获取显示名
if display_name:
display_name = display_name.text
# 获取简介
if introduction:
# print introduction
introduction = str(introduction).replace("<br />", "$##$")
introduction = BeautifulSoup(introduction)
introduction = introduction.text.replace("$##$", "\n")
introduction = introduction.replace(u"收起", "")
# 获取标签类型信息
if app_tags:
tags_text = app_tags.text
if tags_text:
category = tags_text.split(">")[1]
# 获取截图url
if screenshot_info:
for img in screenshot_info.findAll('img'):
src = img['src']
if src:
screenshot_url += src + "\n"
#获取下载地址,版本,安装数和大小
# if download_info:
# download = download_info["href"]
if params_download_num:
install_num = params_download_num.text
if params_platform:
api_level = params_platform.text
# 获取评分
if star_percent:
star_percent = star_percent.get("style")
# 获取简短描述
if short_desc:
short_desc = short_desc.text
result["display_name"] = display_name
result["introduction"] = introduction
result["screenshot_url"] = screenshot_url
result["developers"] = developers
result["category"] = category
result["icon_url"] = icon_url
result["pkg_name"] = pname
result["version"] = version
result["url1"] = download
result["language"] = ""
result["version_code"] = 0
result["install_num"] = utils.format_install_num(install_num)
result["size"] = size
result["min_sdk_version"] = utils.format_android_level(api_level)
result["star_num"] = utils.format_star_num(star_percent)
result["short_desc"] = short_desc or ""
return result
def get_basic_info(html_info):
"""
获取游戏的基本信息
:param html_info:
:return: dict
"""
result = {}
html = html_info
if html:
soup = BeautifulSoup(html)
name = soup.find("div", {"class": "det-name-int"})
detail = soup.find("div", {"class": "det-app-data-info"})
app_tags = soup.find(id="J_DetCate")
icon_info = soup.find("div", {"class": "det-icon"})
screenshot_info = soup.findAll("div", {"class": "pic-img-box"})
developers_info = soup.findAll("div", {"class": "det-othinfo-data"})
# download_info = soup.find("a", {"class": "det-down-btn"})
params_size = soup.find("div", {"class": "det-size"})
# params_download_num = soup.find("div", {"class": "det-ins-num"})
version_info = soup.find("div", {"class": "det-othinfo-data"})
star = soup.find("div", {"class": "com-blue-star-num"})
display_name = ""
introduction = ""
screenshot_url = ""
category = ""
pname = ""
download = ""
install_num = ""
language = ""
size = ""
version = ""
version_code = ""
star_num = ""
icon_url = ""
developers = ""
# 获取显示名
if name:
display_name = name.text
if "v" in display_name:
display_name = display_name.split("v")[0]
# 获取简介
if detail:
introduction = detail.text
# 获取标签类型信息
if app_tags:
category = app_tags.text
# 获取icon
if icon_info:
icon_info = icon_info.find('img')
icon_url = icon_info['src']
# 获取截图url
if screenshot_info:
for screen in screenshot_info:
img = screen.find('img')
src = img['data-src']
if src:
screenshot_url += src + "\n"
if developers_info:
developers = developers_info[len(developers_info)-1].text
#获取下载地址,版本,安装数和大小
if params_size:
size = params_size.text
if version_info:
version = version_info.text.replace("V", "")
if star:
star_num = star.text
#获取script游戏包名和下载地址
script_infos = soup.findAll('script')
for script in script_infos:
pkg_infos = script.text
if "var appDetailData = {" in pkg_infos:
# print pkg_infos
search = re.search(r"{([\s\S]*)}", pkg_infos, re.M | re.I)
# print search.group()
if search:
data = search.group()
data = data.replace("orgame", '"orgame"', 1)
data = data.replace("apkName", '"apkName"', 1)
data = data.replace("apkCode", '"apkCode"', 1)
data = data.replace("appId", '"appId"', 1)
data = data.replace("appName", '"appName"', 1)
data = data.replace("iconUrl", '"iconUrl"', 1)
data = data.replace("appScore", '"appScore"', 1)
data = data.replace("downTimes", '"downTimes"', 1)
data = data.replace("downUrl", '"downUrl"', 1)
data = data.replace("tipsUpDown", '"tipsUpDown"', 1)
pkg = json.loads(data)
download = pkg["downUrl"]
pname = pkg["apkName"]
install_num = pkg["downTimes"]
version_code = pkg['apkCode']
result["display_name"] = display_name
result["introduction"] = introduction
result["screenshot_url"] = screenshot_url
result["developers"] = developers
result["category"] = category
result["icon_url"] = icon_url
result["version"] = version
result["pkg_name"] = pname
result["url1"] = download
result["language"] = language
result["version_code"] = version_code
result["install_num"] = utils.format_install_num(install_num)
result["size"] = utils.format_file_size(size)
result["star_num"] = utils.format_star_num(star_num, 2)
result["min_sdk_version"] = 0
return result
def get_basic_info(html_info):
result = {}
html = html_info
pname = ""
download = ""
install_num = ""
language = ""
version = ""
size = ""
display_name = ""
introduction = ""
api_level = ""
if html:
soup = BeautifulSoup(html)
# print soup
short_desc = soup.find("meta", {"name": "description"})
#获取下载地址,版本,安装数和大小
download_info = soup.find("a", {"class": "s_btn s_btn4"})
details = soup.find("ul", {"class": "s_info"})
name = soup.find("h1", {"class": "ff f20 fb fl"})
introduction_info = soup.find("div", {"class": "o-content"})
screenshot_info = soup.find(id="lstImges")
icon_star = soup.find("div", {"class": "s_intro_pic fl"})
screenshot_url = ""
developers = ""
category = ""
icon_url = ""
star_num = ""
if download_info:
download_info = download_info["href"]
if download_info:
download = "http://apk.91.com" + download_info
if details:
list_li = details.findAll("li")
for i, li in enumerate(list_li):
if i == 0:
version = li.text.replace(u"版本：", "").replace(u"历史版本", "")
if i == 1:
install_num = li.text.replace(u"下载次数：", "")
if i == 2:
size = li.text.replace(u"文件大小：", "")
if i == 3:
api_level = li.text
if i == 8:
if li:
developers = li.text.replace(u"开发商：", "")
if i == 9:
if li:
tags = li.findAll("a")
for tag in tags:
category += tag.text + "\n"
if icon_star:
icon = icon_star.find('img')
star = icon_star.find('a')
if icon:
icon_url = icon['src']
if star:
star_num = star['class']
if introduction_info:
introduction = introduction_info.text
if name:
display_name = name.text
if screenshot_info:
for img in screenshot_info.findAll('img'):
src = img['src']
if src:
screenshot_url += src + u"\n"
result["pkg_name"] = pname
result["version"] = version
result["url1"] = download
result["language"] = language
result["display_name"] = display_name
result["introduction"] = introduction
result["screenshot_url"] = screenshot_url
result["developers"] = developers
result["category"] = category
result["icon_url"] = icon_url
result["version_code"] = ""
result["install_num"] = utils.format_install_num(install_num)
result["size"] = utils.format_file_size(size)
result["star_num"] = utils.format_star_num(star_num, 2)
result["min_sdk_version"] = utils.format_android_level(api_level)
result["short_desc"] = short_desc.get("content")
return result
def get_basic_info(html_info):
"""
获取游戏的基本信息
:param html_info:
:return: dict
"""
result = {}
html = html_info
if html:
screenshot_url = ""
developers = ""
category = ""
pname = ""
download = ""
install_num = ""
version = ""
size = 0
api_level = ""
icon_url = ""
soup = BeautifulSoup(html)
display_name = soup.find("h1", {"class": "app-name"})
introduction = soup.find("div", {"class": "brief-long"})
short_desc = soup.find("span", {"class": "head-content"})
app_tags = soup.find("div", {"class": "nav"})
screenshot_info = soup.find("div", {"class": "section-body"})
star_percent = soup.find("span", {"class": "star-percent"})
params_download_num = soup.find("span", {"class": "download-num"})
params_platform = soup.find("span", {"class": "params-platform"})
data_info = soup.find("a", {"class": "inst-btn-big highspeed"})
if data_info:
download = data_info["data_url"]
data_size = data_info["data_size"]
data_ver_name = data_info["data_versionname"]
# data_ver_code = data_info["data_versioncode"]
data_pkg_name = data_info["data_package"]
icon_url = data_info["data_icon"]
size = data_size or 0
version = data_ver_name
# version_code = data_ver_code or 0
pname = data_pkg_name
# 获取显示名
if display_name:
display_name = display_name.text
# 获取简介
if introduction:
# print introduction
introduction = str(introduction).replace("<br />", "$##$")
introduction = BeautifulSoup(introduction)
introduction = introduction.text.replace("$##$", "\n")
introduction = introduction.replace(u"收起", "")
# 获取标签类型信息
if app_tags:
tags_text = app_tags.text
if tags_text:
category = tags_text.split(">")[1]
# 获取截图url
if screenshot_info:
for img in screenshot_info.findAll('img'):
src = img['src']
if src:
screenshot_url += src + "\n"
#获取下载地址,版本,安装数和大小
# if download_info:
# download = download_info["href"]
if params_download_num:
install_num = params_download_num.text
if params_platform:
api_level = params_platform.text
# 获取评分
if star_percent:
star_percent = star_percent.get("style")
# 获取简短描述
if short_desc:
short_desc = short_desc.text
result["display_name"] = display_name
result["introduction"] = introduction
result["screenshot_url"] = screenshot_url
result["developers"] = developers
result["category"] = category
result["icon_url"] = icon_url
result["pkg_name"] = pname
result["version"] = version
result["url1"] = download
result["language"] = ""
result["version_code"] = 0
result["install_num"] = utils.format_install_num(install_num)
result["size"] = size
result["min_sdk_version"] = utils.format_android_level(api_level)
result["star_num"] = utils.format_star_num(star_percent)
result["short_desc"] = short_desc or ""
return result
def get_basic_info(html_info):
result = {}
html = html_info
pname = ""
download = ""
language = ""
version = ""
size = ""
display_name = ""
developer = ""
category = ""
icon_url = ""
star_num = ""
screenshot_url = ""
introduction = ""
if html:
soup = BeautifulSoup(html)
#获取下载地址,版本,安装数和大小
app_info = soup.find("div", {"class": "app-info"})
download_info = soup.find("a", {"class": "download"})
details = soup.find("div", {"class": "details preventDefault"})
introductions = soup.find("p", {"class": "pslide"})
screenshot_info = soup.find(id="J_thumbnail_wrap")
if download_info and "http://app.mi.com" not in download_info:
download = "http://app.mi.com" + download_info['href']
if details:
list_li = details.find("ul").findAll("li")
for index, element in enumerate(list_li):
if index == 1:
size = element.text
if index == 3:
version = element.text
if index == 7:
pname = element.text
if app_info:
name = app_info.find("h3")
icon = app_info.find("img")
categorys = app_info.find("p", {"class": "special-font action"})
developers = app_info.findAll("p")
star = app_info.find("div",
attrs={"class": re.compile(r"star1-hover")})
if name:
display_name = name.text
if icon:
icon_url = icon['src']
if categorys:
category = categorys.text
if "|" in category:
category = category.split("|")
category = category[0].replace(u"分类：", "")
if developers:
if len(developers) == 2:
d = developers[0]
developer = d.text
if star:
star_nums = star['class']
nums = star_nums.split('-', 2)
if len(nums) == 3:
star_num = nums[2]
if screenshot_info:
screenshot_info = screenshot_info.findAll('img')
for img in screenshot_info:
src = img['src']
if src:
screenshot_url += src + "\n"
if introductions:
introductions = str(introductions).replace("<br />", "$##$")
introductions = BeautifulSoup(introductions)
introduction = introductions.text.replace("$##$", "\n")
result["pkg_name"] = pname
result["version"] = version
result["url1"] = download
result["language"] = language
result["display_name"] = display_name
result["introduction"] = introduction
result["screenshot_url"] = screenshot_url
result["developers"] = developer
result["category"] = category
result["icon_url"] = icon_url
result["version_code"] = ""
result["install_num"] = ""
result["size"] = utils.format_file_size(size)
result["star_num"] = utils.format_star_num(star_num)
result["min_sdk_version"] = 0
result["short_desc"] = ""
return result
def get_basic_info(html_info):
result = {}
html = html_info
pname = ""
download = ""
install_num = ""
language = ""
version = ""
size = ""
display_name = ""
introduction = ""
api_level = ""
if html:
soup = BeautifulSoup(html)
# print soup
short_desc = soup.find("meta", {"name": "description"})
#获取下载地址,版本,安装数和大小
download_info = soup.find("a", {"class": "s_btn s_btn4"})
details = soup.find("ul", {"class": "s_info"})
name = soup.find("h1", {"class": "ff f20 fb fl"})
introduction_info = soup.find("div", {"class": "o-content"})
screenshot_info = soup.find(id="lstImges")
icon_star = soup.find("div", {"class": "s_intro_pic fl"})
screenshot_url = ""
developers = ""
category = ""
icon_url = ""
star_num = ""
if download_info:
download_info = download_info["href"]
if download_info:
download = "http://apk.91.com" + download_info
if details:
list_li = details.findAll("li")
for i, li in enumerate(list_li):
if i == 0:
version = li.text.replace(u"版本：", "").replace(u"历史版本", "")
if i == 1:
install_num = li.text.replace(u"下载次数：", "")
if i == 2:
size = li.text.replace(u"文件大小：", "")
if i == 3:
api_level = li.text
if i == 8:
if li:
developers = li.text.replace(u"开发商：", "")
if i == 9:
if li:
tags = li.findAll("a")
for tag in tags:
category += tag.text + "\n"
if icon_star:
icon = icon_star.find('img')
star = icon_star.find('a')
if icon:
icon_url = icon['src']
if star:
star_num = star['class']
if introduction_info:
introduction = introduction_info.text
if name:
display_name = name.text
if screenshot_info:
for img in screenshot_info.findAll('img'):
src = img['src']
if src:
screenshot_url += src + u"\n"
result["pkg_name"] = pname
result["version"] = version
result["url1"] = download
result["language"] = language
result["display_name"] = display_name
result["introduction"] = introduction
result["screenshot_url"] = screenshot_url
result["developers"] = developers
result["category"] = category
result["icon_url"] = icon_url
result["version_code"] = ""
result["install_num"] = utils.format_install_num(install_num)
result["size"] = utils.format_file_size(size)
result["star_num"] = utils.format_star_num(star_num, 2)
result["min_sdk_version"] = utils.format_android_level(api_level)
result["short_desc"] = short_desc.get("content")
return result
def get_basic_info(html_info):
"""
获取游戏的基本信息
:param html_info:
:return: dict
"""
result = {}
html = html_info
if html:
soup = BeautifulSoup(html)
display_name = soup.find(id="app-name")
introduction = soup.find(id="html-brief")
app_tags = soup.find("div", {"class": "app-tags"})
icon_url = soup.find("dt")
short_desc = soup.find("dl", {"class": "clearfix"})
screenshot_url = ""
developers = ""
category = ""
pname = ""
download = ""
install_num = ""
language = ""
version_name = ""
version_code = 0
size = 0
star_num = 0
min_sdk_version = 0
# 获取显示名
if display_name:
display_name = display_name.find("span").text
# 如果获取不到介绍信息，尝试第二种方式
if not introduction:
introduction = soup.find("div", {"class": "infors"})
# 获取简介，截图，开发商和类型
if introduction:
# 获取截图url
scroll = introduction.find(id="scrollbar")
desc = introduction.find("div", {"class": "breif"})
if scroll:
imgs = scroll["data-snaps"]
for img in imgs.split(","):
if img and "icon.png" not in img:
screenshot_url += img + "\n"
else:
for img in introduction.findAll("img"):
src = img['src']
if src and "icon.png" not in src:
screenshot_url += src + "\n"
if desc:
desc_str = str(desc).replace("<br />", "$##$")
desc_str = desc_str.replace("</td>", "$##$</td>")
desc_str = BeautifulSoup(desc_str)
introduction = desc_str.text.replace("$##$", "\n").replace(" ", " ")
introduction = introduction.replace("versioncode", "\nversioncode")
introduction = introduction.replace("updatetime", "\nupdatetime")
else:
introduction = str(introduction).replace("<br />", "$##$")
introduction = introduction.replace("</td>", "$##$</td>")
introduction = introduction.replace("</p>", "$##$</p>")
introduction = BeautifulSoup(introduction)
introduction = introduction.text.replace("$##$", "\n").replace(" ", " ")
introduction = introduction.replace("versioncode", "\nversioncode")
introduction = introduction.replace("updatetime", "\nupdatetime")
#获取游戏包名和下载地址
script_infos = soup.findAll('script')
for script in script_infos:
pkg_infos = script.text
if "var detail = (function () {" in pkg_infos:
search = re.search(r"return {([\s\S]*)};", pkg_infos, re.M | re.I)
# print search
if search:
data = search.group().replace("return", "").replace(";", "").replace("'", "\"")
pkg = json.loads(data)
pname = pkg['pname']
download = pkg['downloadUrl']
version_code = pkg['vcode']
#获取语言，版本,安装数和大小
pf = soup.find("div", {"class": "pf"})
basic_info = soup.find("div", {"class": "base-info"})
if pf:
pf_s3 = pf.findAll("span", {"class": "s-3"}) # 下载量
star = pf.find("span", {"class": "s-1 js-votepanel"}) # 评分
if len(pf_s3) == 2:
install_num = pf_s3[0].text
size = pf_s3[1].text
install_num = install_num.replace(u"下载：", "").replace(u"次", "")
if star:
star_num = star.text
if basic_info:
infos = basic_info.findAll("td")
for info in infos:
text = info.text
# print text
if u"作者：" in text:
developers = text.replace(u"作者：", "")
if u"语言" in text:
language = text.replace(u"语言：", "")
if u"版本" in text:
search = text.split("versioncode")
if len(search) > 0:
version_name = search[0].replace(u"版本：", "")
if u"系统：" in text:
min_sdk_version = utils.format_android_level(text)
# 获取标签类型信息
if app_tags:
tags = app_tags.findAll("a")
for tag in tags:
if "360" in tag.text:
continue
category += tag.text + "\n"
if icon_url:
icon_url = icon_url.find("img")
icon_url = icon_url["src"]
# 获取简短描述
if short_desc:
short_desc = short_desc.find("p")
if short_desc:
short_desc = short_desc.text
short_desc = short_desc.replace(u"【小编点评】", "")
result["display_name"] = display_name
result["introduction"] = introduction
result["screenshot_url"] = screenshot_url
result["developers"] = developers
result["category"] = category
result["icon_url"] = icon_url
result["pkg_name"] = pname
result["version"] = version_name
result["url1"] = download
result["language"] = language
result["version_code"] = version_code
result["install_num"] = utils.format_install_num(install_num)
result["size"] = utils.format_file_size(size)
result["star_num"] = utils.format_star_num(star_num)
result["min_sdk_version"] = min_sdk_version
result["short_desc"] = short_desc or ""
return result
def get_basic_info(html_info):
result = {}
html = html_info
pname = ""
download = ""
install_num = ""
language = ""
version = ""
size = ""
display_name = ""
category = ""
icon_url = ""
star_num = ""
screenshot_url = ""
api_level = ""
if html:
soup = BeautifulSoup(html)
#获取下载地址,版本,安装数和大小
app_version = soup.find(id="appSoftName")
details = soup.find("div", {"class": "code_box_border"})
introduction = soup.find(id="softIntroduce")
screenshot_info = soup.find(id="screenImgUl")
icon_star = soup.find("div", {"class": "detail_content"})
if app_version:
version = app_version.text
if "(" in version:
display_name = version.split("(")[0]
search = re.search(r"\(.*\)", version, re.M | re.I)
version = search.group().replace("(", "").replace(")", "")
if details:
spans = details.findAll("span", {"class": "font14"})
for i, span in enumerate(spans):
if i == 1:
install_num = span.text
if install_num:
install_num = install_num.replace(u"热度", "")
if i == 3:
size = span.text
if i == 4:
category = span.text
if i == 6:
language = span.text
# api_level = details.find("span", {"class": "font14 d_gj_line left"})
api_level = details.find("span", {"class": "font14 detailMiniSdk d_gj_line left"})
if api_level:
api_level = api_level.text
download_info = details.find("a", {"class": "link_btn"})
if download_info:
download = download_info['href']
if "appdown" in download:
pname_info = download.split("/")
download = "http://apk.hiapk.com" + download
if len(pname_info) > 2:
pname = pname_info[2].split("?")[0]
if icon_star:
icon = icon_star.find("img")
star = icon_star.find("div", attrs={"class": re.compile(r"star_bg star_m")})
if icon:
icon_url = icon['src']
if star:
if star:
star_num = star['class']
if screenshot_info:
screenshot_info = screenshot_info.findAll('img')
for img in screenshot_info:
src = img['src']
if src:
screenshot_url += src + "\n"
if introduction:
# print introduction
introduction = introduction.text
result["pkg_name"] = pname
result["version"] = version
result["url1"] = download
result["language"] = language
result["display_name"] = display_name
result["introduction"] = introduction
result["screenshot_url"] = screenshot_url
result["developers"] = ""
result["category"] = category
result["icon_url"] = icon_url
result["version_code"] = ""
result["install_num"] = utils.format_install_num(install_num)
result["size"] = utils.format_file_size(size)
result["star_num"] = utils.format_star_num(star_num, 0.2)
result["min_sdk_version"] = utils.format_android_level(api_level)
result["short_desc"] = ""
return result
def get_basic_info(html_info):
"""
获取游戏的基本信息
:param html_info:
:return: dict
"""
result = {}
html = html_info
if html:
soup = BeautifulSoup(html)
display_name = soup.find(id="app-name")
introduction = soup.find(id="html-brief")
app_tags = soup.find("div", {"class": "app-tags"})
icon_url = soup.find("dt")
short_desc = soup.find("dl", {"class": "clearfix"})
screenshot_url = ""
developers = ""
category = ""
pname = ""
download = ""
install_num = ""
language = ""
version_name = ""
version_code = 0
size = 0
star_num = 0
min_sdk_version = 0
# 获取显示名
if display_name:
display_name = display_name.find("span").text
# 如果获取不到介绍信息，尝试第二种方式
if not introduction:
introduction = soup.find("div", {"class": "infors"})
# 获取简介，截图，开发商和类型
if introduction:
# 获取截图url
scroll = introduction.find(id="scrollbar")
desc = introduction.find("div", {"class": "breif"})
if scroll:
imgs = scroll["data-snaps"]
for img in imgs.split(","):
if img and "icon.png" not in img:
screenshot_url += img + "\n"
else:
for img in introduction.findAll("img"):
src = img['src']
if src and "icon.png" not in src:
screenshot_url += src + "\n"
if desc:
desc_str = str(desc).replace("<br />", "$##$")
desc_str = desc_str.replace("</td>", "$##$</td>")
desc_str = BeautifulSoup(desc_str)
introduction = desc_str.text.replace("$##$", "\n").replace(
" ", " ")
introduction = introduction.replace("versioncode",
"\nversioncode")
introduction = introduction.replace("updatetime",
"\nupdatetime")
else:
introduction = str(introduction).replace("<br />", "$##$")
introduction = introduction.replace("</td>", "$##$</td>")
introduction = introduction.replace("</p>", "$##$</p>")
introduction = BeautifulSoup(introduction)
introduction = introduction.text.replace("$##$", "\n").replace(
" ", " ")
introduction = introduction.replace("versioncode",
"\nversioncode")
introduction = introduction.replace("updatetime",
"\nupdatetime")
#获取游戏包名和下载地址
script_infos = soup.findAll('script')
for script in script_infos:
pkg_infos = script.text
if "var detail = (function () {" in pkg_infos:
search = re.search(r"return {([\s\S]*)};", pkg_infos,
re.M | re.I)
# print search
if search:
data = search.group().replace("return", "").replace(
";", "").replace("'", "\"")
pkg = json.loads(data)
pname = pkg['pname']
download = pkg['downloadUrl']
version_code = pkg['vcode']
#获取语言，版本,安装数和大小
pf = soup.find("div", {"class": "pf"})
basic_info = soup.find("div", {"class": "base-info"})
if pf:
pf_s3 = pf.findAll("span", {"class": "s-3"}) # 下载量
star = pf.find("span", {"class": "s-1 js-votepanel"}) # 评分
if len(pf_s3) == 2:
install_num = pf_s3[0].text
size = pf_s3[1].text
install_num = install_num.replace(u"下载：", "").replace(u"次", "")
if star:
star_num = star.text
if basic_info:
infos = basic_info.findAll("td")
for info in infos:
text = info.text
# print text
if u"作者：" in text:
developers = text.replace(u"作者：", "")
if u"语言" in text:
language = text.replace(u"语言：", "")
if u"版本" in text:
search = text.split("versioncode")
if len(search) > 0:
version_name = search[0].replace(u"版本：", "")
if u"系统：" in text:
min_sdk_version = utils.format_android_level(text)
# 获取标签类型信息
if app_tags:
tags = app_tags.findAll("a")
for tag in tags:
if "360" in tag.text:
continue
category += tag.text + "\n"
if icon_url:
icon_url = icon_url.find("img")
icon_url = icon_url["src"]
# 获取简短描述
if short_desc:
short_desc = short_desc.find("p")
if short_desc:
short_desc = short_desc.text
short_desc = short_desc.replace(u"【小编点评】", "")
result["display_name"] = display_name
result["introduction"] = introduction
result["screenshot_url"] = screenshot_url
result["developers"] = developers
result["category"] = category
result["icon_url"] = icon_url
result["pkg_name"] = pname
result["version"] = version_name
result["url1"] = download
result["language"] = language
result["version_code"] = version_code
result["install_num"] = utils.format_install_num(install_num)
result["size"] = utils.format_file_size(size)
result["star_num"] = utils.format_star_num(star_num)
result["min_sdk_version"] = min_sdk_version
result["short_desc"] = short_desc or ""
return result
def get_basic_info(html_info):
"""
获取游戏的基本信息
:param html_info:
:return: dict
"""
result = {}
html = html_info
if html:
soup = BeautifulSoup(html)
name = soup.find("div", {"class": "det-name-int"})
detail = soup.find("div", {"class": "det-app-data-info"})
app_tags = soup.find(id="J_DetCate")
icon_info = soup.find("div", {"class": "det-icon"})
screenshot_info = soup.findAll("div", {"class": "pic-img-box"})
developers_info = soup.findAll("div", {"class": "det-othinfo-data"})
# download_info = soup.find("a", {"class": "det-down-btn"})
params_size = soup.find("div", {"class": "det-size"})
# params_download_num = soup.find("div", {"class": "det-ins-num"})
version_info = soup.find("div", {"class": "det-othinfo-data"})
star = soup.find("div", {"class": "com-blue-star-num"})
display_name = ""
introduction = ""
screenshot_url = ""
category = ""
pname = ""
download = ""
install_num = ""
language = ""
size = ""
version = ""
version_code = ""
star_num = ""
icon_url = ""
developers = ""
# 获取显示名
if name:
display_name = name.text
if "v" in display_name:
display_name = display_name.split("v")[0]
# 获取简介
if detail:
introduction = detail.text
# 获取标签类型信息
if app_tags:
category = app_tags.text
# 获取icon
if icon_info:
icon_info = icon_info.find('img')
icon_url = icon_info['src']
# 获取截图url
if screenshot_info:
for screen in screenshot_info:
img = screen.find('img')
src = img['data-src']
if src:
screenshot_url += src + "\n"
if developers_info:
developers = developers_info[len(developers_info) - 1].text
#获取下载地址,版本,安装数和大小
if params_size:
size = params_size.text
if version_info:
version = version_info.text.replace("V", "")
if star:
star_num = star.text
#获取script游戏包名和下载地址
script_infos = soup.findAll('script')
for script in script_infos:
pkg_infos = script.text
if "var appDetailData = {" in pkg_infos:
# print pkg_infos
search = re.search(r"{([\s\S]*)}", pkg_infos, re.M | re.I)
# print search.group()
if search:
data = search.group()
data = data.replace("orgame", '"orgame"', 1)
data = data.replace("apkName", '"apkName"', 1)
data = data.replace("apkCode", '"apkCode"', 1)
data = data.replace("appId", '"appId"', 1)
data = data.replace("appName", '"appName"', 1)
data = data.replace("iconUrl", '"iconUrl"', 1)
data = data.replace("appScore", '"appScore"', 1)
data = data.replace("downTimes", '"downTimes"', 1)
data = data.replace("downUrl", '"downUrl"', 1)
data = data.replace("tipsUpDown", '"tipsUpDown"', 1)
pkg = json.loads(data)
download = pkg["downUrl"]
pname = pkg["apkName"]
install_num = pkg["downTimes"]
version_code = pkg['apkCode']
result["display_name"] = display_name
result["introduction"] = introduction
result["screenshot_url"] = screenshot_url
result["developers"] = developers
result["category"] = category
result["icon_url"] = icon_url
result["version"] = version
result["pkg_name"] = pname
result["url1"] = download
result["language"] = language
result["version_code"] = version_code
result["install_num"] = utils.format_install_num(install_num)
result["size"] = utils.format_file_size(size)
result["star_num"] = utils.format_star_num(star_num, 2)
result["min_sdk_version"] = 0
return result
def perpare_data(info, download_url, file_size, ver_code, ver_name, channel,
apk_info):
if not info:
return
doc = info["docV2"]
#if channel == "GG官方":
market_channel = channel
if channel == "samsung":
market_channel = "三星"
game_name = doc["title"]
pkg_name = apk_info["pkg_name"]
game_desc = doc["descriptionHtml"]
if game_desc:
game_desc = game_desc.replace("<br>",
"\n").replace("<p>",
"\n").replace("<p>", "\n")
app_details = doc['details']['appDetails']
game_types = app_details["appCategory"][0]
try:
game_types = google_game_type[game_types]
except:
game_types = '其他'
downloaded_cnts = doc["details"]["appDetails"]["numDownloads"]
developer = doc['details']['appDetails']['developerName']
utils.check_developer(developer)
game_language = "多国语言"
screen_shot_urls = ""
icon_url = ""
images = doc["image"]
if images:
for image in images:
image_type = image["imageType"]
image_url = image["imageUrl"]
if image_type == 4:
icon_url = image_url
if image_type == 1:
screen_shot_urls += image_url + "\n"
is_crack_apk = 1 # 破解版
min_sdk = ""
star_num = doc["aggregateRating"]["starRating"]
now = str(int(time.time()))
label_info = dict()
pkg_info = dict()
game_name = ftoj(game_name)
game_desc = ftoj(game_desc)
#if channel == "GG官方":
g_name = game_name + u"(%s)" % channel
if channel == "samsung":
g_name = game_name + u"(samsung)"
#g_name = game_name + u"(GG官方)"
#adrooy
#g_name = game_name + u"(samsung)"
game_id = utils.gen_label_info_id(g_name)
g_name = g_name.replace(u"(GG官方)", "")
if ver_name:
ver_name = filter(lambda ch: ch in '0123456789.', ver_name)
if 'gameid' in apk_info:
game_id = apk_info['gameid']
label_info["game_id"] = game_id
label_info["game_name"] = g_name
label_info["game_types"] = game_types
label_info["origin_types"] = game_types
label_info["screen_shot_urls"] = screen_shot_urls
label_info["icon_url"] = icon_url
label_info["detail_desc"] = game_desc
label_info["star_num"] = utils.format_star_num(str(star_num), 2)
label_info["download_counts"] = utils.format_install_num(downloaded_cnts)
label_info["game_language"] = game_language
label_info["now"] = now
label_info["file_size"] = file_size
label_info["ver_name"] = ver_name
label_info["developer"] = developer
pkg_info["market_channel"] = market_channel
pkg_info["game_name"] = g_name
pkg_info["pkg_name"] = pkg_name
pkg_info["ver_code"] = ver_code
pkg_info["ver_name"] = ver_name
pkg_info["file_size"] = file_size
pkg_info["download_urls"] = download_url.strip()
pkg_info["game_desc"] = game_desc
pkg_info["game_types"] = game_types
pkg_info["origin_types"] = game_types
pkg_info["downloaded_cnts"] = utils.format_install_num(downloaded_cnts)
pkg_info["game_language"] = game_language
pkg_info["screen_shot_urls"] = screen_shot_urls
pkg_info["icon_url"] = icon_url
pkg_info["now"] = now
pkg_info["is_crack_apk"] = is_crack_apk
if "ggvercode" not in apk_info:
apk_info["ggvercode"] = "null"
apk_id = utils.gen_pkg_info_id(0, pkg_name, ver_name, market_channel,
apk_info["ggvercode"])
pkg_info["apk_id"] = apk_id
pkg_info["game_id"] = game_id
pkg_info[
"url4details"] = "https://play.google.com/store/apps/details?id=%s" % pkg_name
#print apk_id, game_id
#import sys
#sys.exit()
return label_info, pkg_info
def get_basic_info(html_info):
result = {}
html = html_info
pname = ""
download = ""
language = ""
version = ""
size = ""
display_name = ""
developer = ""
category = ""
icon_url = ""
star_num = ""
screenshot_url = ""
introduction = ""
if html:
soup = BeautifulSoup(html)
#获取下载地址,版本,安装数和大小
app_info = soup.find("div", {"class": "app-info"})
download_info = soup.find("a", {"class": "download"})
details = soup.find("div", {"class": "details preventDefault"})
introductions = soup.find("p", {"class": "pslide"})
screenshot_info = soup.find(id="J_thumbnail_wrap")
if download_info and "http://app.mi.com" not in download_info:
download = "http://app.mi.com" + download_info['href']
if details:
list_li = details.find("ul").findAll("li")
for index, element in enumerate(list_li):
if index == 1:
size = element.text
if index == 3:
version = element.text
if index == 7:
pname = element.text
if app_info:
name = app_info.find("h3")
icon = app_info.find("img")
categorys = app_info.find("p", {"class": "special-font action"})
developers = app_info.findAll("p")
star = app_info.find("div", attrs={"class": re.compile(r"star1-hover")})
if name:
display_name = name.text
if icon:
icon_url = icon['src']
if categorys:
category = categorys.text
if "|" in category:
category = category.split("|")
category = category[0].replace(u"分类：", "")
if developers:
if len(developers) == 2:
d = developers[0]
developer = d.text
if star:
star_nums = star['class']
nums = star_nums.split('-', 2)
if len(nums) == 3:
star_num = nums[2]
if screenshot_info:
screenshot_info = screenshot_info.findAll('img')
for img in screenshot_info:
src = img['src']
if src:
screenshot_url += src + "\n"
if introductions:
introductions = str(introductions).replace("<br />", "$##$")
introductions = BeautifulSoup(introductions)
introduction = introductions.text.replace("$##$", "\n")
result["pkg_name"] = pname
result["version"] = version
result["url1"] = download
result["language"] = language
result["display_name"] = display_name
result["introduction"] = introduction
result["screenshot_url"] = screenshot_url
result["developers"] = developer
result["category"] = category
result["icon_url"] = icon_url
result["version_code"] = ""
result["install_num"] = ""
result["size"] = utils.format_file_size(size)
result["star_num"] = utils.format_star_num(star_num)
result["min_sdk_version"] = 0
result["short_desc"] = ""
return result
def perpare_data(info, download_url, file_size, ver_code, ver_name, channel, apk_info):
if not info:
return
doc = info["docV2"]
#if channel == "GG官方":
market_channel = channel
if channel == "samsung":
market_channel = "三星"
game_name = doc["title"]
pkg_name = apk_info["pkg_name"]
game_desc = doc["descriptionHtml"]
if game_desc:
game_desc = game_desc.replace("<br>", "\n").replace("<p>", "\n").replace("<p>", "\n")
app_details = doc['details']['appDetails']
game_types = app_details["appCategory"][0]
try:
game_types = google_game_type[game_types]
except:
game_types = '其他'
downloaded_cnts = doc["details"]["appDetails"]["numDownloads"]
developer = doc['details']['appDetails']['developerName']
utils.check_developer(developer)
game_language = "多国语言"
screen_shot_urls = ""
icon_url = ""
images = doc["image"]
if images:
for image in images:
image_type = image["imageType"]
image_url = image["imageUrl"]
if image_type == 4:
icon_url = image_url
if image_type == 1:
screen_shot_urls += image_url + "\n"
is_crack_apk = 1 # 破解版
min_sdk = ""
star_num = doc["aggregateRating"]["starRating"]
now = str(int(time.time()))
label_info = dict()
pkg_info = dict()
game_name = ftoj(game_name)
game_desc = ftoj(game_desc)
#if channel == "GG官方":
g_name = game_name + u"(%s)" % channel
if channel == "samsung":
g_name = game_name + u"(samsung)"
#g_name = game_name + u"(GG官方)"
#adrooy
#g_name = game_name + u"(samsung)"
game_id = utils.gen_label_info_id(g_name)
g_name = g_name.replace(u"(GG官方)", "")
if ver_name:
ver_name = filter(lambda ch: ch in '0123456789.', ver_name)
if 'gameid' in apk_info:
game_id = apk_info['gameid']
label_info["game_id"] = game_id
label_info["game_name"] = g_name
label_info["game_types"] = game_types
label_info["origin_types"] = game_types
label_info["screen_shot_urls"] = screen_shot_urls
label_info["icon_url"] = icon_url
label_info["detail_desc"] = game_desc
label_info["star_num"] = utils.format_star_num(str(star_num), 2)
label_info["download_counts"] = utils.format_install_num(downloaded_cnts)
label_info["game_language"] = game_language
label_info["now"] = now
label_info["file_size"] = file_size
label_info["ver_name"] = ver_name
label_info["developer"] = developer
pkg_info["market_channel"] = market_channel
pkg_info["game_name"] = g_name
pkg_info["pkg_name"] = pkg_name
pkg_info["ver_code"] = ver_code
pkg_info["ver_name"] = ver_name
pkg_info["file_size"] = file_size
pkg_info["download_urls"] = download_url.strip()
pkg_info["game_desc"] = game_desc
pkg_info["game_types"] = game_types
pkg_info["origin_types"] = game_types
pkg_info["downloaded_cnts"] = utils.format_install_num(downloaded_cnts)
pkg_info["game_language"] = game_language
pkg_info["screen_shot_urls"] = screen_shot_urls
pkg_info["icon_url"] = icon_url
pkg_info["now"] = now
pkg_info["is_crack_apk"] = is_crack_apk
if "ggvercode" not in apk_info:
apk_info["ggvercode"] = "null"
apk_id = utils.gen_pkg_info_id(0, pkg_name, ver_name, market_channel, apk_info["ggvercode"])
pkg_info["apk_id"] = apk_id
pkg_info["game_id"] = game_id
pkg_info["url4details"] = "https://play.google.com/store/apps/details?id=%s" % pkg_name
#print apk_id, game_id
#import sys
#sys.exit()
return label_info, pkg_info