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

修改prev，next，parent用法，和旧版不兼容

Browse Source
This commit is contained in:
g1879 2021-12-19 00:58:43 +08:00
parent 32194aa886
commit 8784f7dd48
4 changed files with 77 additions and 84 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

103
DrissionPage/base.py
View File

@ -55,22 +55,23 @@ class BaseElement(BaseParser):
def inner_ele(self) -> Union[WebElement, HtmlElement]:
return self._inner_ele
@property
def next(self):
"""返回后一个兄弟元素"""
return self.nexts()
def next(self, index: int = 1):
"""返回后面的一个兄弟元素，可指定第几个 \n
:param index: 后面第几个兄弟元素
:return: 兄弟元素
"""
nexts = self.nexts(total=1, begin=index)
return nexts[0] if nexts else None
# ----------------以下属性或方法由后代实现----------------
@property
def tag(self):
return
@property
def parent(self):
return
def parent(self, level: int = 1):
pass
@property
def prev(self):
def prev(self, index: int = 1):
return
@property
@ -78,22 +79,25 @@ class BaseElement(BaseParser):
return True
@abstractmethod
def nexts(self, num: int = 1):
def nexts(self, total: int = None, begin: int = 1):
pass
class DrissionElement(BaseElement):
"""DriverElement 和 SessionElement的基类，但不是ShadowRootElement的基类"""
@property
def parent(self):
@abstractmethod
def parent(self, level: int = 1):
"""返回父级元素"""
return self.parents()
pass
@property
def prev(self):
"""返回前一个兄弟元素"""
return self.prevs()
def prev(self, index: int = 1):
"""返回前面的一个兄弟元素，可指定第几个 \n
:param index: 前面第几个
:return: 兄弟元素
"""
prevs = self.prevs(total=1, begin=index)
return prevs[0] if prevs else None
@property
def link(self) -> str:
@ -125,60 +129,57 @@ class DrissionElement(BaseElement):
else:
texts = [x if isinstance(x, str) else x.text for x in self.eles('xpath:./text() | *')]
return [format_html(x.strip(' ').rstrip('\n')) for x in texts if x and sub('[\n\t ]', '', x) != '']
return [format_html(x.strip(' ').rstrip('\n')) for x in texts if x and sub('[\r\n\t ]', '', x) != '']
def nexts(self, num: int = 1, mode: str = 'ele'):
"""返回后面第num个兄弟元素或节点 \n
:param num: 后面第几个兄弟元素或节点
def nexts(self, total: int = None, begin: int = 1, mode: str = 'ele'):
"""返回后面若干个兄弟元素或节点组成的列表，total为None返回所有 \n
:param total: 获取多少个元素或节点
:param begin: 从第几个开始获取，从1起
:param mode: 'ele', 'node' 或 'text'，匹配元素、节点、或文本节点
:return: SessionElement对象
"""
return self._get_brother(num, mode, 'next')
return self._get_brothers(begin=begin, total=total, mode=mode, direction='next')
def prevs(self, num: int = 1, mode: str = 'ele'):
"""返回前面第num个兄弟元素或节点 \n
:param num: 前面第几个兄弟元素或节点
def prevs(self, total: int = None, begin: int = 1, mode: str = 'ele'):
"""返回前面若干个兄弟元素或节点组成的列表，total为None返回所有 \n
:param total: 获取多少个元素或节点
:param begin: 从第几个开始获取，从1起
:param mode: 'ele', 'node' 或 'text'，匹配元素、节点、或文本节点
:return: SessionElement对象
"""
return self._get_brother(num, mode, 'prev')
return self._get_brothers(begin=begin, total=total, mode=mode, direction='prev')
def _get_brother(self, num: int = 1, mode: str = 'ele', direction: str = 'next'):
"""返回前面第num个兄弟节点或元素 \n
:param num: 前面第几个兄弟节点或元素
def _get_brothers(self, begin: int = 1, total: int = None, mode: str = 'ele', direction: str = 'next'):
"""按要求返回兄弟元素或节点组成的列表 \n
:param begin: 从第几个兄弟节点或元素开始
:param total: 获取多少个
:param mode: 'ele', 'node' 或 'text'，匹配元素、节点、或文本节点
:param direction: 'next' 或 'prev'，查找的方向
:return: DriverElement对象或字符串
"""
# 查找节点的类型
if mode == 'ele':
node_txt = '*'
elif mode == 'node':
node_txt = 'node()'
elif mode == 'text':
node_txt = 'text()'
else:
node_txt = {'ele': '*', 'node': 'node()', 'text': 'text()'}.get(mode)
if not node_txt:
raise ValueError(f"mode参数只能是'node'、'ele'或'text'，现在是：'{mode}'。")
# 查找节点的方向
if direction == 'next':
direction_txt = 'following'
elif direction == 'prev':
direction_txt = 'preceding'
else:
direction_txt = {'next': 'following', 'prev': 'preceding'}.get(direction)
if not direction_txt:
raise ValueError(f"direction参数只能是'next'或'prev'，现在是：'{direction}'。")
timeout = 0 if direction == 'prev' else .5
# 获取节点
ele_or_node = self._ele(f'xpath:./{direction_txt}-sibling::{node_txt}[{num}]', timeout=timeout)
# 获取所有节点
nodes = self._ele(f'xpath:./{direction_txt}-sibling::{node_txt}', timeout=timeout, single=False)
# 跳过元素间的换行符
while isinstance(ele_or_node, str) and sub('[\n\t ]', '', ele_or_node) == '':
num += 1
ele_or_node = self._ele(f'xpath:./{direction_txt}-sibling::{node_txt}[{num}]', timeout=timeout)
if direction == 'next':
end = None if not total or total >= len(nodes) else begin + total - 1
begin -= 1
else:
begin = None if not total or total >= len(nodes) else begin - total - 1
end = None
return ele_or_node
return [e for e in nodes[begin:end] if not (isinstance(e, str) and sub('[ \n\t\r]', '', e) == '')]
# ----------------以下属性或方法由后代实现----------------
@property
@ -193,9 +194,9 @@ class DrissionElement(BaseElement):
def raw_text(self):
return
@abstractmethod
def parents(self, num: int = 1):
pass
# @abstractmethod
# def parents(self, num: int = 1):
# pass
@abstractmethod
def attr(self, attr: str):

13
DrissionPage/driver_element.py
View File

@ -91,17 +91,12 @@ class DriverElement(DrissionElement):
"""返回未格式化处理的元素内文本"""
return self.inner_ele.get_attribute('innerText')
@property
def parent(self):
"""返回父级元素"""
return self.parents()
def parents(self, num: int = 1):
"""返回上面第num级父元素 \n
:param num: 第几级父元素
def parent(self, level: int = 1) -> 'DriverElement':
"""返回上面第level级父元素 \n
:param level: 第几级父元素
:return: DriverElement对象
"""
loc = 'xpath', f'.{"/.." * num}'
loc = 'xpath', f'.{"/.." * level}'
return self.ele(loc, timeout=0)
def attr(self, attr: str) -> str:

13
DrissionPage/session_element.py
View File

@ -70,17 +70,12 @@ class SessionElement(DrissionElement):
"""返回未格式化处理的元素内文本"""
return str(self._inner_ele.text_content())
@property
def parent(self):
"""返回父级元素"""
return self.parents()
def parents(self, num: int = 1):
"""返回上面第num级父元素 \n
:param num: 第几级父元素
def parent(self, level: int = 1) -> 'SessionElement':
"""返回上面第level级父元素 \n
:param level: 第几级父元素
:return: SessionElement对象
"""
return self.ele(f'xpath:..{"/.." * (num - 1)}')
return self.ele(f'xpath:..{"/.." * (level - 1)}')
def attr(self, attr: str) -> Union[str, None]:
"""返回attribute属性值 \n

32
DrissionPage/shadow_root_element.py
View File

@ -45,26 +45,28 @@ class ShadowRootElement(BaseElement):
"""返回内部的html文本"""
return self.inner_ele.get_attribute('innerHTML')
@property
def parent(self) -> DriverElement:
"""shadow-root所依赖的父元素"""
return self.parent_ele
def parents(self, num: int = 1) -> DriverElement:
"""返回上面第num级父元素 \n
:param num: 第几级父元素
def parent(self, level: int = 1) -> DriverElement:
"""返回上面第level级父元素 \n
:param level: 第几级父元素
:return: DriverElement对象
"""
loc = 'xpath', f'.{"/.." * (num - 1)}'
return self.parent_ele.ele(loc, timeout=0.1)
if level == 1:
return self.parent_ele
else:
loc = 'xpath', f'.{"/.." * (level - 1)}'
return self.parent_ele.ele(loc, timeout=0.1)
def nexts(self, num: int = 1) -> DriverElement:
"""返回后面第num个兄弟元素 \n
:param num: 后面第几个兄弟元素
def nexts(self, total: int = None, begin: int = 1) -> DriverElement:
"""返回后面若干个兄弟元素或节点组成的列表，total为None返回所有 \n
:param total: 获取多少个元素或节点
:param begin: 从第几个开始获取，从1起
:return: DriverElement对象
"""
loc = 'css selector', f':nth-child({num})'
return self.parent_ele.ele(loc, timeout=0.1)
loc = 'css selector', f':nth-child(n)'
eles = self.parent_ele.eles(loc, timeout=0.1)
end = None if not total or total >= len(eles) else begin + total - 1
return eles[begin - 1:end]
def ele(self,
loc_or_str: Union[Tuple[str, str], str],