用 Python lxml 解析 XML

Parse XML with Python lxml

我正在尝试使用 python 库 lxml 解析 XML,并希望结果输出在一个数据框。我对 python 和解析比较陌生,所以请耐心等待我概述问题。我正在尝试解析的原始 xml 可用 here

我有兴趣获得在“invstOrSec”中找到的一些相关标签。下面是一个“invstOrSec”实例的快照,其中标签“curCd”附带的文本是美元。

<?xml version="1.0" encoding="UTF-8"?>
    <invstOrSec>
        <name>NIPPON LIFE INSURANCE</name>
        <lei>549300Y0HHMFW3EVWY08</lei>
        <curCd>USD</curCd>
    <invstOrSec>

这相对简单,我目前的方法是先在字典中定义相关标签,然后在循环中将它们粗化到数据帧中。

    import pandas as pd
    from lxml import etree

    # Declare directory
    os.chdir('C:/Users/A1610222/Desktop/Form NPORT/pkg/sec-edgar-filings/0001548717/NPORT-P/0001752724- 
    20-040624')

    # Set root
    xmlfile = "filing-details.xml"
    tree = etree.parse(xmlfile)
    root = tree.getroot()

    # Remove namespace prefixes
    for elem in root.getiterator():
        elem.tag = etree.QName(elem).localname
   
    # Remove unused namespace declarations
    etree.cleanup_namespaces(root)

    # Set path
    invstOrSec = root.xpath('//invstOrSec')

    # Define tags to extract
    vars = {'invstOrSec' : {'name', 'lei', 'curCd'}

    # Extract holdings data
    sec_info =  pd.DataFrame()
    temp = pd.DataFrame()

    for one in invstOrSec:
        for two in one:
            if two.tag in vars['invstOrSec']:
                temp[two.tag] = [two.text]
        sec_info = sec_info.append(temp)  

这是sec_info

的前三行
name lei curCd
NIPPON LIFE INSURANCE 549300Y0HHMFW3EVWY08 USD
Lloyds Banking Group PLC 549300PPXHEU2JF0AM85 USD
Enbridge Inc 98TPTUM4IVMFCZBCUR27 USD

但是,当货币不是美元时,xml 遵循的结构略有不同。请参阅以下示例。

<?xml version="1.0" encoding="UTF-8"?>
    <invstOrSec>
        <name>ACHMEA BV</name>
        <lei>7245007QUMI1FHIQV531</lei>
        <currencyConditional curCd="EUR" exchangeRt="0.89150400"/>
    <invstOrSec>

这次 curCd 被不同的标签 currencyConditional 替换,它包含与文本相反的属性。我很难解释这些情况,同时让我的代码尽可能通用。我希望我已经设法说明了这个问题。再次,如果这太初级了,请原谅。任何帮助将不胜感激。

这是一个你不应该重新发明轮子的案例;使用其他人开发的工具...

import pandas as pd
import pandas_read_xml as pdx

url = 'https://www.sec.gov/Archives/edgar/data/1548717/000175272420040624/primary_doc.xml'

df = pdx.read_xml(url,['edgarSubmission', 'formData', 'invstOrSecs','invstOrSec'])

#because of the non-US currency column, you have to apply one more contortion:
df['currencyConditional'] = df['currencyConditional'].apply(lambda x: x.get('@curCd') if not isinstance(x,float) else "NA" )
df[['name','lei','curCd','currencyConditional']]

输出(部分,显然)- 注意最后一行:

168     BNP PARIBAS     R0MUWSFPU8MPRO8K5P83    USD     NA
169     Societe Generale    O2RNE8IBXP4R0TD8PU41    USD     NA
170     BARCLAYS PLC    213800LBQA1Y9L22JB70    NaN     GBP