如何将 XML 文件转换为漂亮的 pandas 数据帧？

Question

假设我有这样的 XML：

<author type="XXX" language="EN" gender="xx" feature="xx" web="foobar.com">
    <documents count="N">
        <document KEY="e95a9a6c790ecb95e46cf15bee517651" web="www.foo_bar_exmaple.com"><![CDATA[A large text with lots of strings and punctuations symbols [...]
]]>
        </document>
        <document KEY="bc360cfbafc39970587547215162f0db" web="www.foo_bar_exmaple.com"><![CDATA[A large text with lots of strings and punctuations symbols [...]
]]>
        </document>
        <document KEY="19e71144c50a8b9160b3f0955e906fce" web="www.foo_bar_exmaple.com"><![CDATA[A large text with lots of strings and punctuations symbols [...]
]]>
        </document>
        <document KEY="21d4af9021a174f61b884606c74d9e42" web="www.foo_bar_exmaple.com"><![CDATA[A large text with lots of strings and punctuations symbols [...]
]]>
        </document>
    </documents>
</author>

我想读取这个 XML 文件并将其转换为 pandas DataFrame:

key                                         type     language    feature            web                         data
e95324a9a6c790ecb95e46cf15bE232ee517651      XXX        EN          xx      www.foo_bar_exmaple.com     A large text with lots of strings and punctuations symbols [...]
bc360cfbafc39970587547215162f0db             XXX        EN          xx      www.foo_bar_exmaple.com     A large text with lots of strings and punctuations symbols [...]
19e71144c50a8b9160b3cvdf2324f0955e906fce     XXX        EN          xx      www.foo_bar_exmaple.com     A large text with lots of strings and punctuations symbols [...]
21d4af9021a174f61b8erf284606c74d9e42         XXX        EN          xx      www.foo_bar_exmaple.com     A large text with lots of strings and punctuations symbols [...]

这是我已经尝试过的方法，但我遇到了一些错误，可能有更有效的方法来完成此任务：

from lxml import objectify
import pandas as pd

path = 'file_path'
xml = objectify.parse(open(path))
root = xml.getroot()
root.getchildren()[0].getchildren()
df = pd.DataFrame(columns=('key','type', 'language', 'feature', 'web', 'data'))

for i in range(0,len(xml)):
    obj = root.getchildren()[i].getchildren()
    row = dict(zip(['key','type', 'language', 'feature', 'web', 'data'], [obj[0].text, obj[1].text]))
    row_s = pd.Series(row)
    row_s.name = i
    df = df.append(row_s)

谁能为我提供更好的方法来解决这个问题？

Answer 1

您可以轻松地使用 xml（来自 Python 标准库）转换为 pandas.DataFrame。这是我会做的（当从文件中读取时，将 xml_data 替换为您的文件或文件对象的名称）：

import pandas as pd
import xml.etree.ElementTree as ET
import io

def iter_docs(author):
    author_attr = author.attrib
    for doc in author.iter('document'):
        doc_dict = author_attr.copy()
        doc_dict.update(doc.attrib)
        doc_dict['data'] = doc.text
        yield doc_dict

xml_data = io.StringIO(u'''YOUR XML STRING HERE''')

etree = ET.parse(xml_data) #create an ElementTree object 
doc_df = pd.DataFrame(list(iter_docs(etree.getroot())))

如果您的原始文档中有多个作者或者您的 XML 的根目录不是 author，那么我将添加以下生成器：

def iter_author(etree):
    for author in etree.iter('author'):
        for row in iter_docs(author):
            yield row

并将doc_df = pd.DataFrame(list(iter_docs(etree.getroot())))更改为doc_df = pd.DataFrame(list(iter_author(etree)))

看看 ElementTree tutorial provided in the xml library documentation.

Answer 2

这是将 xml 数据框转换为 pandas 数据框的另一种方法。例如，我从一个字符串中解析 xml，但这个逻辑也适用于读取文件。

import pandas as pd
import xml.etree.ElementTree as ET

xml_str = '<?xml version="1.0" encoding="utf-8"?>\n<response>\n <head>\n  <code>\n   200\n  </code>\n </head>\n <body>\n  <data id="0" name="All Categories" t="2018052600" tg="1" type="category"/>\n  <data id="13" name="RealEstate.com.au [H]" t="2018052600" tg="1" type="publication"/>\n </body>\n</response>'

etree = ET.fromstring(xml_str)
dfcols = ['id', 'name']
df = pd.DataFrame(columns=dfcols)

for i in etree.iter(tag='data'):
    df = df.append(
        pd.Series([i.get('id'), i.get('name')], index=dfcols),
        ignore_index=True)

df.head()

Answer 3

您也可以通过创建元素字典然后直接转换为数据框来进行转换：

import xml.etree.ElementTree as ET
import pandas as pd

# Contents of test.xml
# <?xml version="1.0" encoding="utf-8"?> <tags>   <row Id="1" TagName="bayesian" Count="4699" ExcerptPostId="20258" WikiPostId="20257" />   <row Id="2" TagName="prior" Count="598" ExcerptPostId="62158" WikiPostId="62157" />   <row Id="3" TagName="elicitation" Count="10" />   <row Id="5" TagName="open-source" Count="16" /> </tags>

root = ET.parse('test.xml').getroot()

tags = {"tags":[]}
for elem in root:
    tag = {}
    tag["Id"] = elem.attrib['Id']
    tag["TagName"] = elem.attrib['TagName']
    tag["Count"] = elem.attrib['Count']
    tags["tags"]. append(tag)

df_users = pd.DataFrame(tags["tags"])
df_users.head()

Answer 4

建议使用 xmltodict 库。它很好地处理了您的 xml 文本，我已经用它来摄取一个包含将近一百万条记录的 xml 文件。

Answer 5

从 v1.3 开始，您可以简单地使用：

pandas.read_xml(path_or_file)

如何将 XML 文件转换为漂亮的 pandas 数据帧？

How to convert an XML file to nice pandas dataframe?

python

xml

parsing

dataframe

pandas