blackteay/DrissionPage
1
0
Fork 0
mirror of https://gitee.com/g1879/DrissionPage.git synced 2024-12-10 04:00:23 +08:00
Code Issues Projects Releases Wiki Activity

完善注释

Browse Source
This commit is contained in:
g1879 2020-08-13 11:34:16 +08:00
parent 9d28932f45
commit 16847f829d
8 changed files with 177 additions and 60 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

24
DrissionPage/common.py
View File

@ -131,16 +131,27 @@ def get_loc_from_str(loc: str) -> tuple:
def _make_xpath_str(tag: str, arg: str, val: str, mode: str = 'fuzzy') -> str:
"""生成xpath语句"""
"""生成xpath语句 \n
:param tag: 标签名
:param arg: 属性名
:param val: 属性值
:param mode: 'exact' 或 'fuzzy'，对应精确或模糊查找
:return: xpath字符串
"""
tag_name = '' if tag == '*' else f'name()="{tag}" and '
if mode == 'exact':
return f'//*[{tag_name}{arg}={_make_search_str(val)}]'
else:
elif mode == 'fuzzy':
return f"//*[{tag_name}contains({arg},{_make_search_str(val)})]"
else:
raise ValueError("Argument mode can only be 'exact' or 'fuzzy'.")
def _make_search_str(search_str: str) -> str:
"""将"转义，不知何故不能直接用\""""
"""将"转义，不知何故不能直接用\ \n
:param search_str: 查询字符串
:return: 把"转义后的字符串
"""
parts = search_str.split('"')
parts_num = len(parts)
search_str = 'concat('
@ -151,8 +162,11 @@ def _make_search_str(search_str: str) -> str:
return search_str
def translate_loc_to_xpath(loc) -> tuple:
"""把By类型转为xpath或css selector"""
def translate_loc_to_xpath(loc: tuple) -> tuple:
"""把By类型的loc元组转换为css selector或xpath类型的 \n
:param loc: By类型的loc元组
:return: css selector或xpath类型的loc元组
"""
loc_by = 'xpath'
loc_str = None
if loc[0] == 'xpath':

94
DrissionPage/config.py
View File

@ -17,7 +17,7 @@ class OptionsManager(object):
"""管理配置文件内容的类"""
def __init__(self, path: str = None):
"""初始化，读取配置文件，如没有设置临时文件夹，则设置并新建
"""初始化，读取配置文件，如没有设置临时文件夹，则设置并新建 \n
:param path: ini文件的路径，默认读取模块文件夹下的
"""
self.path = path or Path(__file__).parent / 'configs.ini'
@ -30,7 +30,11 @@ class OptionsManager(object):
self.save()
def get_value(self, section: str, item: str) -> Any:
"""获取配置的值"""
"""获取配置的值 \n
:param section: 段名
:param item: 项名
:return: 项值
"""
try:
return eval(self._conf.get(section, item))
except SyntaxError:
@ -39,7 +43,10 @@ class OptionsManager(object):
return None
def get_option(self, section: str) -> dict:
"""把section内容以字典方式返回"""
"""把section内容以字典方式返回 \n
:param section: 段名
:return: 段内容生成的字典
"""
items = self._conf.items(section)
option = dict()
for j in items:
@ -50,14 +57,19 @@ class OptionsManager(object):
return option
def set_item(self, section: str, item: str, value: Any):
"""设置配置值"""
"""设置配置值 \n
:param section: 段名
:param item: 项名
:param value: 项值
:return: 当前对象
"""
self._conf.set(section, item, str(value))
return self
def save(self, path: str = None):
"""保存配置文件
"""保存配置文件 \n
:param path: ini文件的路径，默认保存到模块文件夹下的
:return: None
:return: 当前对象
"""
path = path or Path(__file__).parent / 'configs.ini'
self._conf.write(open(path, 'w', encoding='utf-8'))
@ -66,7 +78,9 @@ class OptionsManager(object):
class DriverOptions(Options):
def __init__(self, read_file=True):
"""初始化，默认从文件读取设置"""
"""初始化，默认从文件读取设置 \n
:param read_file: 是否从默认ini文件中读取配置信息
"""
super().__init__()
self._driver_path = None
if read_file:
@ -89,9 +103,9 @@ class DriverOptions(Options):
return self.binary_location
def save(self, path: str = None):
"""保存设置到文件
"""保存设置到文件 \n
:param path: ini文件的路径，默认保存到模块文件夹下的
:return: None
:return: 当前对象
"""
om = OptionsManager()
options = _chrome_options_to_dict(self)
@ -104,9 +118,9 @@ class DriverOptions(Options):
return self
def remove_argument(self, value: str):
"""移除一个设置
"""移除一个argument项 \n
:param value: 设置项名，有值的设置项传入设置名称即可
:return: None
:return: 当前对象
"""
del_list = []
for argument in self._arguments:
@ -117,22 +131,27 @@ class DriverOptions(Options):
return self
def remove_experimental_option(self, key: str):
"""移除一个实验设置，传入key值删除"""
"""移除一个实验设置，传入key值删除 \n
:param key: 实验设置的名称
:return: 当前对象
"""
if key in self._experimental_options:
self._experimental_options.pop(key)
return self
def remove_all_extensions(self):
"""移除所有插件
因插件是以整个文件储存，难以移除其中一个，故如须设置则全部移除再重设"""
"""移除所有插件 \n
因插件是以整个文件储存，难以移除其中一个，故如须设置则全部移除再重设
:return: 当前对象
"""
self._extensions = []
return self
def set_argument(self, arg: str, value: Union[bool, str]):
"""设置浏览器配置argument属性
"""设置浏览器配置的argument属性 \n
:param arg: 属性名
:param value: 属性值，有值的属性传入值，没有的传入bool
:return: None
:return: 当前对象
"""
self.remove_argument(arg)
if value:
@ -141,27 +160,45 @@ class DriverOptions(Options):
return self
def set_headless(self, on_off: bool = True):
"""设置headless"""
"""设置是否隐藏浏览器界面 \n
:param on_off: 开或关
:return: 当前对象
"""
return self.set_argument('--headless', on_off)
def set_no_imgs(self, on_off: bool = True):
"""设置是否加载图片"""
"""设置是否加载图片 \n
:param on_off: 开或关
:return: 当前对象
"""
return self.set_argument('--blink-settings=imagesEnabled=false', on_off)
def set_no_js(self, on_off: bool = True):
"""设置禁用js"""
"""设置是否禁用js \n
:param on_off: 开或关
:return: 当前对象
"""
return self.set_argument('--disable-javascript', on_off)
def set_mute(self, on_off: bool = True):
"""设置静音"""
"""设置是否静音 \n
:param on_off: 开或关
:return: 当前对象
"""
return self.set_argument('--mute-audio', on_off)
def set_user_agent(self, user_agent: str):
"""设置user agent"""
"""设置user agent \n
:param user_agent: user agent文本
:return: 当前对象
"""
return self.set_argument('user-agent', user_agent)
def set_proxy(self, proxy: str):
"""设置代理"""
"""设置代理 \n
:param proxy: 代理url和端口
:return: 当前对象
"""
return self.set_argument('--proxy-server', proxy)
def set_paths(self,
@ -171,14 +208,14 @@ class DriverOptions(Options):
download_path: str = None,
user_data_path: str = None,
cache_path: str = None):
"""简易设置路径函数
"""快捷的路径设置函数 \n
:param driver_path: chromedriver.exe路径
:param chrome_path: chrome.exe路径
:param debugger_address: 调试浏览器地址，例：127.0.0.1:9222
:param download_path: 下载文件路径
:param user_data_path: 用户数据路径
:param cache_path: 缓存路径
:return: None
:return: 当前对象
"""
def format_path(path: str) -> str:
@ -200,7 +237,10 @@ class DriverOptions(Options):
def _dict_to_chrome_options(options: dict) -> Options:
"""从传入的字典获取浏览器设置，返回ChromeOptions对象"""
"""从传入的字典获取浏览器设置，返回ChromeOptions对象 \n
:param options: 配置信息字典
:return: 保存浏览器配置的ChromeOptions对象
"""
chrome_options = webdriver.ChromeOptions()
if 'debugger_address' in options and options['debugger_address']:
# 控制已打开的浏览器
@ -238,6 +278,10 @@ def _dict_to_chrome_options(options: dict) -> Options:
def _chrome_options_to_dict(options: Union[dict, DriverOptions, None]) -> Union[dict, None]:
"""把chrome配置对象转换为字典 \n
:param options: chrome配置对象，字典或DriverOptions对象
:return: 配置字典
"""
if options is None or isinstance(options, dict):
return options

65
DrissionPage/drission.py
View File

@ -26,7 +26,7 @@ class Drission(object):
session_options: dict = None,
ini_path: str = None,
proxy: dict = None):
"""初始化配置信息，但不生成session和driver实例
"""初始化配置信息，但不生成session和driver实例 \n
:param driver_options: chrome设置，Options类或设置字典
:param session_options: session设置
:param ini_path: ini文件路径'
@ -52,7 +52,7 @@ class Drission(object):
@property
def session(self) -> HTMLSession:
"""获取HTMLSession对象"""
"""返回HTMLSession对象，如为None则按配置信息创建"""
if self._session is None:
self._session = HTMLSession()
attrs = ['headers', 'cookies', 'auth', 'proxies', 'hooks', 'params', 'verify',
@ -67,7 +67,7 @@ class Drission(object):
@property
def driver(self) -> WebDriver:
"""获取WebDriver对象，按传入配置信息初始化"""
"""返回WebDriver对象，如为None则按配置信息创建"""
if self._driver is None:
if isinstance(self._driver_options, dict):
options = _dict_to_chrome_options(self._driver_options)
@ -92,22 +92,33 @@ class Drission(object):
@property
def driver_options(self) -> dict:
"""返回driver配置信息"""
return self._driver_options
@property
def session_options(self) -> dict:
"""返回session配置信息"""
return self._session_options
@session_options.setter
def session_options(self, value: dict):
def session_options(self, value: dict) -> None:
"""设置session配置
:param value: session配置字典
:return: None
"""
self._session_options = value
@property
def proxy(self) -> Union[None, dict]:
"""返回代理信息"""
return self._proxy
@proxy.setter
def proxy(self, proxies: dict = None):
def proxy(self, proxies: dict = None) -> None:
"""设置代理信息 \n
:param proxies: 代理信息字典
:return: None
"""
self._proxy = proxies
if self._session:
self._session.proxies = proxies
@ -121,9 +132,15 @@ class Drission(object):
for cookie in cookies:
self._ensure_add_cookie(cookie)
def cookies_to_session(self, copy_user_agent: bool = False, driver: WebDriver = None, session: Session = None) \
-> None:
"""把driver的cookies复制到session"""
def cookies_to_session(self, copy_user_agent: bool = False,
driver: WebDriver = None,
session: Session = None) -> None:
"""把driver对象的cookies复制到session对象 \n
:param copy_user_agent: 是否复制ua信息
:param driver: 来源driver对象
:param session: 目标session对象
:return: None
"""
driver = driver or self.driver
session = session or self.session
if copy_user_agent:
@ -131,8 +148,15 @@ class Drission(object):
for cookie in driver.get_cookies():
session.cookies.set(cookie['name'], cookie['value'], domain=cookie['domain'])
def cookies_to_driver(self, url: str, driver: WebDriver = None, session: Session = None) -> None:
"""把session的cookies复制到driver"""
def cookies_to_driver(self, url: str,
driver: WebDriver = None,
session: Session = None) -> None:
"""把session对象的cookies复制到driver对象 \n
:param url: 作用域
:param driver: 目标driver对象
:param session: 来源session对象
:return: None
"""
driver = driver or self.driver
session = session or self.session
domain = urlparse(url).netloc
@ -146,7 +170,12 @@ class Drission(object):
self._ensure_add_cookie(cookie_data, driver=driver)
def _ensure_add_cookie(self, cookie, override_domain=None, driver=None) -> None:
"""添加cookie到driver"""
"""添加cookie到driver \n
:param cookie: 要添加的cookie
:param override_domain: 覆盖作用域
:param driver: 操作的driver对象
:return: None
"""
driver = driver or self.driver
if override_domain:
cookie['domain'] = override_domain
@ -171,8 +200,12 @@ class Drission(object):
raise WebDriverException(f"Couldn't add the following cookie to the webdriver\n{cookie}\n")
def _is_cookie_in_driver(self, cookie, driver=None) -> bool:
"""检查cookie是否已经在driver里
只检查name、value、domain，检查domain时比较宽"""
"""检查cookie是否已经在driver里 \n
只检查name、value、domain，检查domain时比较宽 \n
:param cookie: 要检查的cookie
:param driver: 被检查的driver
:return: 返回布尔值
"""
driver = driver or self.driver
for driver_cookie in driver.get_cookies():
if (cookie['name'] == driver_cookie['name'] and
@ -183,7 +216,11 @@ class Drission(object):
return False
def user_agent_to_session(self, driver: WebDriver = None, session: Session = None) -> None:
"""把driver的user-agent复制到session"""
"""把driver的user-agent复制到session \n
:param driver: 来源driver对象
:param session: 目标session对象
:return: None
"""
driver = driver or self.driver
session = session or self.session
selenium_user_agent = driver.execute_script("return navigator.userAgent;")

6
DrissionPage/driver_element.py
View File

@ -181,7 +181,7 @@ class DriverElement(DrissionElement):
"""返回当前元素下级符合条件的子元素，默认返回第一个 \n
示例： \n
- 用loc元组查找： \n
ele.ele((By.CLASS_NAME, 'ele_class')) - 返回所有class为ele_class的子元素 \n
ele.ele((By.CLASS_NAME, 'ele_class')) - 返回第一个class为ele_class的子元素 \n
- 用查询字符串查找： \n
查找方式：属性、tag name和属性、文本、xpath、css selector \n
其中，@表示属性，=表示精确匹配，:表示模糊匹配，无控制字符串时默认搜索该字符串 \n
@ -251,7 +251,7 @@ class DriverElement(DrissionElement):
:param show_errmsg: 出现异常时是否打印信息
:return: DriverElement对象组成的列表
"""
if not isinstance(loc_or_str, tuple) and not isinstance(loc_or_str, str):
if not isinstance(loc_or_str, (tuple, str)):
raise TypeError('Type of loc_or_str can only be tuple or str.')
return self.ele(loc_or_str, mode='all', show_errmsg=show_errmsg, timeout=timeout)
@ -405,7 +405,7 @@ class DriverElement(DrissionElement):
:return: 是否拖拽成功
"""
# x, y：目标坐标点
if isinstance(ele_or_loc, DriverElement) or isinstance(ele_or_loc, WebElement):
if isinstance(ele_or_loc, (DriverElement, WebElement)):
target_x = ele_or_loc.location['x'] + ele_or_loc.size['width'] // 2
target_y = ele_or_loc.location['y'] + ele_or_loc.size['height'] // 2
elif isinstance(ele_or_loc, tuple):

42
DrissionPage/easy_set.py
View File

@ -19,7 +19,7 @@ def set_paths(driver_path: str = None,
user_data_path: str = None,
cache_path: str = None,
check_version: bool = True) -> None:
"""简易设置路径函数
"""快捷的路径设置函数 \n
:param driver_path: chromedriver.exe路径
:param chrome_path: chrome.exe路径
:param debugger_address: 调试浏览器地址，例：127.0.0.1:9222
@ -57,10 +57,10 @@ def set_paths(driver_path: str = None,
def set_argument(arg: str, value: Union[bool, str]) -> None:
"""设置浏览器配置argument属性
"""设置浏览器配置argument属性 \n
:param arg: 属性名
:param value: 属性值，有值的属性传入值，没有的传入bool
:return:
:return: None
"""
do = DriverOptions()
do.remove_argument(arg)
@ -71,37 +71,59 @@ def set_argument(arg: str, value: Union[bool, str]) -> None:
def set_headless(on_off: bool = True) -> None:
"""设置headless"""
"""设置是否隐藏浏览器界面 \n
:param on_off: 开或关
:return: None
"""
set_argument('--headless', on_off)
def set_no_imgs(on_off: bool = True) -> None:
"""设置是否加载图片"""
"""设置是否禁止加载图片 \n
:param on_off: 开或关
:return: None
"""
set_argument('--blink-settings=imagesEnabled=false', on_off)
def set_no_js(on_off: bool = True) -> None:
"""设置禁用js"""
"""设置是否禁用js \n
:param on_off: 开或关
:return: None
"""
set_argument('--disable-javascript', on_off)
def set_mute(on_off: bool = True) -> None:
"""设置静音"""
"""设置是否静音 \n
:param on_off: 开或关
:return: None
"""
set_argument('--mute-audio', on_off)
def set_user_agent(user_agent: str) -> None:
"""设置user agent"""
"""设置user agent \n
:param user_agent: user agent文本
:return: None
"""
set_argument('user-agent', user_agent)
def set_proxy(proxy: str) -> None:
"""设置代理"""
"""设置代理 \n
:param proxy: 代理网址和端口
:return: None
"""
set_argument('--proxy-server', proxy)
def check_driver_version(driver_path: str = None, chrome_path: str = None) -> bool:
"""检查传入的chrome和chromedriver是否匹配"""
"""检查传入的chrome和chromedriver是否匹配 \n
:param driver_path: chromedriver.exe路径
:param chrome_path: chrome.exe路径
:return: 是否匹配
"""
print('正在检测可用性...')
om = OptionsManager()
driver_path = driver_path or om.get_value('paths', 'chromedriver_path') or 'chromedriver'

2
DrissionPage/mix_page.py
View File

@ -236,7 +236,7 @@ class MixPage(Null, SessionPage, DriverPage):
- 接收到元素对象时： \n
返回元素对象对象 \n
- 用loc元组查找： \n
ele.ele((By.CLASS_NAME, 'ele_class')) - 返回所有class为ele_class的子元素 \n
ele.ele((By.CLASS_NAME, 'ele_class')) - 返回第一个class为ele_class的子元素 \n
- 用查询字符串查找： \n
查找方式：属性、tag name和属性、文本、xpath、css selector \n
其中，@表示属性，=表示精确匹配，:表示模糊匹配，无控制字符串时默认搜索该字符串 \n

2
DrissionPage/session_element.py
View File

@ -117,7 +117,7 @@ class SessionElement(DrissionElement):
"""返回当前元素下级符合条件的子元素，默认返回第一个 \n
示例： \n
- 用loc元组查找： \n
ele.ele((By.CLASS_NAME, 'ele_class')) - 返回所有class为ele_class的子元素 \n
ele.ele((By.CLASS_NAME, 'ele_class')) - 返回第一个class为ele_class的子元素 \n
- 用查询字符串查找： \n
查找方式：属性、tag name和属性、文本、xpath、css selector \n
其中，@表示属性，=表示精确匹配，:表示模糊匹配，无控制字符串时默认搜索该字符串 \n

2
DrissionPage/session_page.py
View File

@ -133,7 +133,7 @@ class SessionPage(object):
:param show_errmsg: 出现异常时是否打印信息
:return: SessionElement对象组成的列表
"""
if not isinstance(loc_or_str, tuple) and not isinstance(loc_or_str, str):
if not isinstance(loc_or_str, (tuple, str)):
raise TypeError('Type of loc_or_str can only be tuple or str.')
return self.ele(loc_or_str, mode='all', show_errmsg=True)