之前在转换数据集格式的时候需要将json转换到xml文件，用lxml包进行操作非常方便。

1. 写xml文件

a) 用etree和objectify

from lxml import etree, objectify

E = objectify.ElementMaker(annotate=False)
anno_tree = E.annotation(
 E.folder('VOC2014_instance'),
 E.filename("test.jpg"),
 E.source(
 E.database('COCO'),
 E.annotation('COCO'),
 E.image('COCO'),
 E.url("http://test.jpg")
 ),
 E.size(
 E.width(800),
 E.height(600),
 E.depth(3)
 ),
 E.segmented(0),
)

etree.ElementTree(anno_tree).write("text.xml", pretty_print=True)

输出的test.xml文件内容如下：

"

如果需要在anno_tree的基础上加其他标签的话用append即可:

E2 = objectify.ElementMaker(annotate=False)
anno_tree2 = E2.object(
 E.name("person"),
 E.bndbox(
 E.xmin(100),
 E.ymin(200),
 E.xmax(300),
 E.ymax(400)
 ),
 E.difficult(0)
)
anno_tree.append(anno_tree2)

上面的输出就变成了:

<annotation>
 <folder>VOC2014_instance/person</folder>
 <filename>test.jpg</filename>
 <source>
 <database>COCO</database>
 <annotation>COCO</annotation>
 <image>COCO</image>
 <url>http://test.jpg</url>
 </source>
 <size>
 <width>800</width>
 <height>600</height>
 <depth>3</depth>
 </size>
 <segmented>0</segmented>
 <object>
 <name>person</name>
 <bndbox>
 <xmin>100</xmin>
 <ymin>200</ymin>
 <xmax>300</xmax>
 <ymax>400</ymax>
 </bndbox>
 <difficult>0</difficult>
 </object>
</annotation>

b) 用etree和SubElement

annotation = etree.Element("annotation")
etree.SubElement(annotation, "folder").text = "VOC2014_instance"
etree.SubElement(annotation, "filename").text = "test.jpg"
source = etree.SubElement(annotation, "source")
etree.SubElement(source, "database").text = "COCO"
etree.SubElement(source, "annotation").text = "COCO"
etree.SubElement(source, "image").text = "COCO"
etree.SubElement(source, "url").text = "http://test.jpg"
size = etree.SubElement(annotation, "size")
etree.SubElement(size, "width").text ='800' # 必须用string
etree.SubElement(size, "height").text = '600'
etree.SubElement(size, "depth").text = '3'
etree.SubElement(annotation, "segmented").text = '0'
key_object = etree.SubElement(annotation, "object")
etree.SubElement(key_object, "name").text = “person”
bndbox = etree.SubElement(key_object, "bndbox")
etree.SubElement(bndbox, "xmin").text = str(100)
etree.SubElement(bndbox, "ymin").text = str(200)
etree.SubElement(bndbox, "xmax").text = str(300)
etree.SubElement(bndbox, "ymax").text = str(400)
etree.SubElement(key_object, "difficult").text = '0'
doc = etree.ElementTree(annotation)
doc.write(open("test.xml", "w"), pretty_print=True)

2. 读xml

这里可以用xpath直接提取所需的元素的值。比如想要获取上面test.xml文件的x, y坐标：

tree = etree.parse("test.xml")
# get bbox
for bbox in tree.xpath('//bndbox'): # 获取bndbox元素的内容
 for corner in bbox.getchildren(): # 便利bndbox元素下的子元素
 print corner.text # string类型