如何在python中将xml文件转换为数据帧或csv输出

Question

我有一个xml文件，希望将内容转换为python中csv文件的数据帧：

<?xml version="1.0" encoding="utf-8"?>
<dashboardreport name="jvm_report" version="7.0.21.1017" reportdate="2018-08-08T10:37:01.510-04:00" description="">
  <source name="CORP_GTM">
    <filters summary="from Jul-30 23:40 to Jul-31 02:40">
      <filter>tf:CustomTimeframe?1533008450802:1533019250802</filter>
    </filters>
  </source>
  <reportheader>
    <reportdetails>
      <user>test1</user>
    </reportdetails>
  </reportheader>
  <data>
    <chartdashlet name="jvm_mem_percent" description="" showabsolutevalues="false">
      <measures structuretype="tree">
        <measure measure="Memory Utilization - Memory Utilization (split by Agent)" color="#800080" aggregation="Maximum" unit="%" thresholds="false" drawingorder="1">
          <measure measure="Memory Utilization - test@server1" color="#7aebd0" aggregation="Maximum" unit="%" thresholds="false">
            <measurement timestamp="1533008460000" avg="11.116939544677734" min="11.007165908813477" max="11.143875122070312" sum="66.7016372680664" count="6"></measurement>
            <measurement timestamp="1533008520000" avg="11.204706827799479" min="11.144883155822754" max="11.268420219421387" sum="67.22824096679688" count="6"></measurement>
          </measure>
          <measure measure="Memory Utilization - test@server2" color="#a6f2e0" aggregation="Maximum" unit="%" thresholds="false">
            <measurement timestamp="1533008460000" avg="11.900418599446615" min="10.386141777038574" max="13.744248390197754" sum="71.40251159667969" count="6"></measurement>
            <measurement timestamp="1533008520000" avg="11.139397939046225" min="10.617960929870605" max="11.427289009094238" sum="66.83638763427734" count="6"></measurement>
          </measure>
          <measure measure="Memory Utilization - test@server3" color="#dd2271" aggregation="Maximum" unit="%" thresholds="false">
            <measurement timestamp="1533008460000" avg="8.395787556966146" min="8.340044021606445" max="8.429450035095215" sum="50.374725341796875" count="6"></measurement>
            <measurement timestamp="1533008520000" avg="8.490419387817383" min="8.456218719482422" max="8.5205659866333" sum="50.9425163269043" count="6"></measurement>
           </measure>
            </measure>
      </measures>
    </chartdashlet>
    <chartdashlet name="jvm_trans_errors" description="" showabsolutevalues="false">
      <measures structuretype="tree"></measures>
    </chartdashlet>
    <chartdashlet name="jvm_trans" description="" showabsolutevalues="false">
      <measures structuretype="tree">
        <measure measure="Count Backend - Count Backend (split by Agent)" color="#8080c0" aggregation="Sum" unit="num" thresholds="false" drawingorder="1">
          <measure measure="Count Backend - test@server1" color="#e44e8d" aggregation="Sum" unit="num" thresholds="false">
            <measurement timestamp="1533010380000" avg="1.0" min="1.0" max="1.0" sum="1.0" count="1"></measurement>
            <measurement timestamp="1533011340000" avg="1.0" min="1.0" max="1.0" sum="10.0" count="10"></measurement>
            <measurement timestamp="1533013080000" avg="1.0" min="1.0" max="1.0" sum="1.0" count="1"></measurement>
            <measurement timestamp="1533013200000" avg="1.0" min="1.0" max="1.0" sum="1.0" count="1"></measurement>
            <measurement timestamp="1533014940000" avg="1.0" min="1.0" max="1.0" sum="2.0" count="2"></measurement>
            <measurement timestamp="1533015780000" avg="1.0" min="1.0" max="1.0" sum="1.0" count="1"></measurement>
            <measurement timestamp="1533018480000" avg="1.0" min="1.0" max="1.0" sum="1.0" count="1"></measurement>
            <measurement timestamp="1533018540000" avg="1.0" min="1.0" max="1.0" sum="2.0" count="2"></measurement>
          </measure>
          <measure measure="Count Backend - test@server2" color="#e5cf4d" aggregation="Sum" unit="num" thresholds="false">
            <measurement timestamp="1533009060000" avg="1.0" min="1.0" max="1.0" sum="10.0" count="10"></measurement>
            <measurement timestamp="1533009120000" avg="1.0" min="1.0" max="1.0" sum="1.0" count="1"></measurement>
            <measurement timestamp="1533009420000" avg="1.0" min="1.0" max="1.0" sum="3.0" count="3"></measurement>
            <measurement timestamp="1533009480000" avg="1.0" min="1.0" max="1.0" sum="5.0" count="5"></measurement>
            <measurement timestamp="1533010020000" avg="1.0" min="1.0" max="1.0" sum="4.0" count="4"></measurement>
            <measurement timestamp="1533010320000" avg="1.0" min="1.0" max="1.0" sum="1200.0" count="1200"></measurement>
          </measure>
          <measure measure="Count Backend - test@server3" color="#dec321" aggregation="Sum" unit="num" thresholds="false">
            <measurement timestamp="1533008460000" avg="1.0" min="1.0" max="1.0" sum="4.0" count="4"></measurement>
            <measurement timestamp="1533008520000" avg="1.0" min="1.0" max="1.0" sum="5.0" count="5"></measurement>
            <measurement timestamp="1533008580000" avg="1.0" min="1.0" max="1.0" sum="9.0" count="9"></measurement>
            <measurement timestamp="1533008640000" avg="1.0" min="1.0" max="1.0" sum="5.0" count="5"></measurement>
          </measure>       
          </measure>
        </measures>
    </chartdashlet>
  </data>
</dashboardreport>

输出需要如下所示：

^{pr2}$

我可以这样做：

doc <- read_xml("C:/test1/test.xml")
  dat<-xml_find_all(doc, ".//measure/measure") %>%
    map_df(function(x) {
      xml_find_all(x, ".//measurement") %>%
        map_df(~as.list(xml_attrs(.))) %>%
        select(-min, -avg, -sum) %>%
        mutate(node=xml_attr(x, "measure"))
    })

我需要用python来做这个，有什么想法吗？在

Answer 1

您应该在Python中使用内置库xml。在

现在，您的标记和属性不是标准的，所以我不得不创建一个函数，这个函数可能是针对您的问题进行硬编码的，但是其他人可以使用它作为指导。在

将此类标记视为您拥有的唯一数据源，并从父标记获取其node属性：

<measurement timestamp="1533008520000" avg="8.490419387817383" min="8.456218719482422" max="8.5205659866333" sum="50.9425163269043" count="6"></measurement>

以下函数应该可以工作，使用Pandas创建数据帧并将其导出到.csv文件：

^{pr2}$

只要用.xml文件更改文件名，它就可以工作了。一旦你有了数据帧，你就可以随心所欲地修改数据的精度、近似值和其他特性。在

Answer 2

一种方法是预处理XML文件，然后将其发送给pandas。我在这个例子中使用ElementTree。在

例如：

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

def getMetrics(file_name):
    tree = ET.parse(file_name)
    root = tree.getroot()
    result = []
    for measure in root.iter('measure'):                         #Get all 'measure' tag
        node = measure.attrib["measure"].split("-")[0].strip()    #Get Node
        for measurement in measure:                              #Get Metrics Information
            if "timestamp" in measurement.attrib:
                result.append(dict(node=node, timestamp=measurement.attrib.get("timestamp"), max=measurement.attrib["max"], count=measurement.attrib["count"]))
    return result

df = pd.DataFrame(getMetrics(filename), columns=["timestamp", "max", "count", "node"])          #Form Dataframe
print(df)

df.to_csv("Your_Output.csv")     #Write to CSV. 

输出：

^{pr2}$

---

按注释编辑。如果要从请求传递xml，请使用ET.fromstring并传递r.content或{}。在

例如：

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

def getMetrics(file_name):
    root = ET.fromstring(file_name)
    result = []
    for measure in root.iter('measure'):                         #Get all 'measure' tag
        node = measure.attrib["measure"].split("-")[0].strip()    #Get Node
        for measurement in measure:                              #Get Metrics Information
            if "timestamp" in measurement.attrib:
                result.append(dict(node=node, timestamp=measurement.attrib.get("timestamp"), max=measurement.attrib["max"], count=measurement.attrib["count"]))
    return result

df = pd.DataFrame(getMetrics(r.content), columns=["timestamp", "max", "count", "node"])          #Form Dataframe
print(df)

Answer 3

这里有一个只使用包含的库和Python3.6的解决方案-不需要pandaps

CSV:

import csv
import xml.etree.ElementTree

e = xml.etree.ElementTree.parse('data.xml').getroot()

with open('out.csv', 'w', newline='') as csv_file:
    csv_writer = csv.writer(csv_file)
    for data in e.iter('measures'):
        measures = data.findall('measure/measure')
        for measure in measures:
            for row in measure:
                csv_writer.writerow([row.get('timestamp'), row.get('max'), row.get('count'), measure.get('measure')])

列：

^{pr2}$