冗余matplotlib.text.Text对象

Question

这个Python 3.8.10和matplotlib 3.5.1代码片段：

import matplotlib.pyplot as plt
import matplotlib.text as text

plt.axis('off')
print(plt.findobj(match=text.Text))
plt.show()

生成一个空的查看器窗口：

​

​

这张名单上有很多副本：

[Text(0.5, 0, ''), Text(1, 0, ''), Text(0, 0, ''), Text(0, 1, ''), Text(0, 0, ''), Text(0, 1, ''), Text(0, 0, ''), Text(0, 1, ''), Text(0, 0, ''), Text(0, 1, ''), Text(0, 0, ''), Text(0, 1, ''), Text(0, 0, ''), Text(0, 1, ''), Text(0, 0.5, ''), Text(0, 0.5, ''), Text(0, 0, ''), Text(1, 0, ''), Text(0, 0, ''), Text(1, 0, ''), Text(0, 0, ''), Text(1, 0, ''), Text(0, 0, ''), Text(1, 0, ''), Text(0, 0, ''), Text(1, 0, ''), Text(0, 0, ''), Text(1, 0, ''), Text(0.5, 1.0, ''), Text(0.0, 1.0, ''), Text(1.0, 1.0, '')]

他们是什么?有必要吗？如何防止它们被创造出来？怎样才能摆脱他们？

我的目标是交互式地添加和删除文本标签。这些无法删除的初始文本对象的存在使其变得更加复杂。理想情况下，我希望有一个函数来删除所有的文本标签。我创建了一组基于鼠标单击坐标的上下文相关标签，并在下一次单击时删除并重新生成整个集合。

Answer 1

创建这些文本对象是因为使用plt.axis("off")，您创建了一个具有标准x-/y-滴答和相应的滴答标签(即文本对象)的axis对象。对于matplotlib，我假设要设置一个不可见的轴，它必须存在，但是我还没有查看源代码来验证这一点。

import matplotlib.pyplot as plt
import matplotlib.text as text

text_before = plt.findobj(match=text.Text)
plt.axis('off')
text_after = plt.findobj(match=text.Text)

print("before ", text_before)
print("after ", text_after)

#plt.show()

输出：

before  []    
after  [Text(0.5, 0, ''), Text(1, 0, ''), Text(0, 0, ''), Text(0, 1, ''), Text(0, 0, ''), Text(0, 1, ''), Text(0, 0, ''), Text(0, 1, ''), Text(0, 0, ''), Text(0, 1, ''), Text(0, 0, ''), Text(0, 1, ''), Text(0, 0, ''), Text(0, 1, ''), Text(0, 0.5, ''), Text(0, 0.5, ''), Text(0, 0, ''), Text(1, 0, ''), Text(0, 0, ''), Text(1, 0, ''), Text(0, 0, ''), Text(1, 0, ''), Text(0, 0, ''), Text(1, 0, ''), Text(0, 0, ''), Text(1, 0, ''), Text(0, 0, ''), Text(1, 0, ''), Text(0.5, 1.0, ''), Text(0.0, 1.0, ''), Text(1.0, 1.0, '')]

即使只是试图定位当前轴，也会生成以前不存在的图形和轴：

print("before", plt.get_fignums())
plt.gca()
print("after", plt.get_fignums())

输出：

before []
after [1]

我认为没有必要删除文本对象，因为它们不会影响最终图像的呈现，而且对于图像生成来说是相当惰性的。而且似乎也没有通过matplotlib实现删除轴上固有的文本艺术家，比如勾标和标题(以及表示脊柱和蜱的Line2D对象)：

import matplotlib.pyplot as plt
import matplotlib.text as text

#the axis object itself can be removed
ax = plt.gca()
#the axis-inherent objects title and tick labels (and not tested here spine and tick Line2D objects)
#cannot be removed
ax_title = plt.title("This is the title") 
_, (ax_label, _, _) = plt.xticks(range(3), list("ABC"))
#text added by user can be removed
ax_text_visible = ax.annotate("This text is visible", (0.30, 0.20), xycoords="axes fraction")
ax_text_invisible = ax.annotate("This text is not visible", (0.20, 0.50), xycoords="axes fraction", visible=False)
ax_text_outside = ax.annotate("This text is outside the canvas", (20, 50), xycoords="axes fraction")
ax_text_not_annotation = plt.text(0.3, 0.7, "Not an annotation")
#test generated objects for existence of the remove method 
test_list = [ax, ax_title, ax_label, ax_text_visible, ax_text_invisible, ax_text_outside, ax_text_not_annotation]
#generate a list of user-created Text objects that ignores axis-inherent text objects
removable_objects = [item for item in test_list if item._remove_method]
for item in removable_objects:
    print(item, type(item))

输出：

AxesSubplot(0.125,0.11;0.775x0.77) <class 'matplotlib.axes._subplots.AxesSubplot'>
Annotation(0.3, 0.2, 'This text is visible') <class 'matplotlib.text.Annotation'>
Annotation(0.2, 0.5, 'This text is not visible') <class 'matplotlib.text.Annotation'>
Annotation(20, 50, 'This text is outside the canvas') <class 'matplotlib.text.Annotation'>
Text(0.3, 0.7, 'Not an annotation') <class 'matplotlib.text.Text'>

因此，我们可以使用remove()方法识别用户放置在轴上的所有文本(不包括标题和勾标)：

#test that all removable user text is also found when screening with ax.findobj(match=text.Text)
matched_text = [item for item in ax.findobj(match=text.Text) if item._remove_method] 
print(set(matched_text) == set(removable_objects[1:])) 
>>>True