numpy.int64' object has no attribute 'replace' while writing xml file using minidom

Question

I've an xml file to be parsed. I did this using minidom parser of python. I had to add an attribute to a particular element which I did after the parsing. Now, I want to write the file back. I am not able to do this.

Below is the error trace which I get for this. I installed numpy 1.8 (win 32 python 2.7) version today. However, I had never tried writing xml file before. Could you please help?

Here's the code snippet:

xmlfile=open(xmlFile,'r')
xmldoc = minidom.parse(xmlFile)
tElements = xmldoc.getElementsByTagName("TEA")
for t in tElements:
    if(t.childNodes):
        print t.nodeType
        dataList = t.childNodes[0].data
        for data, csvData in product(dataList, clusterDataList):
            if(data == csvData[1]):
              t.setAttribute("cluster",csvData[0])
xmlfile.close()
fileWriter=open("sujatha_new.xml",'w')

xmldoc.writexml(fileWriter)
fileWriter.close()

Here's the exception:

Traceback (most recent call last):
  File "C:\Users\w44ylqrl\workspace\Python\Test\T\XMLConverter.py", line 215, in <module>
    addClusterInfo('..\\T\\preprocessed_For_Clustering\\outputs\\sujatha-new.csv', '..\\T\\xml\\sujatha.xml')
  File "C:\Users\w44ylqrl\workspace\Python\Test\T\XMLConverter.py", line 48, in addClusterInfo
    xmldoc.writexml(fileWriter)
  File "C:\Python27\Lib\xml\dom\minidom.py", line 1752, in writexml
    node.writexml(writer, indent, addindent, newl)
  File "C:\Python27\Lib\xml\dom\minidom.py", line 817, in writexml
    node.writexml(writer, indent+addindent, addindent, newl)
  File "C:\Python27\Lib\xml\dom\minidom.py", line 817, in writexml
    node.writexml(writer, indent+addindent, addindent, newl)
  File "C:\Python27\Lib\xml\dom\minidom.py", line 817, in writexml
    node.writexml(writer, indent+addindent, addindent, newl)
  File "C:\Python27\Lib\xml\dom\minidom.py", line 817, in writexml
    node.writexml(writer, indent+addindent, addindent, newl)
  File "C:\Python27\Lib\xml\dom\minidom.py", line 817, in writexml
    node.writexml(writer, indent+addindent, addindent, newl)
  File "C:\Python27\Lib\xml\dom\minidom.py", line 807, in writexml
    _write_data(writer, attrs[a_name].value)
  File "C:\Python27\Lib\xml\dom\minidom.py", line 296, in _write_data
    data = data.replace("&", "&amp;").replace("<", "&lt;"). \
AttributeError: 'numpy.int64' object has no attribute 'replace'

Answer 1

None of the python XML implementations allow you to serialize non-string representations, though some of them will let you assign them, if you want to abuse etree etc. as a treelike data structure. Some (like lxml) won't even let you assign them. Just make it a string:

t.setAttribute("cluster",str(csvData[0]))

and it will work. If you're deserializing these as well, you'll need to convert values back to int after loading.

Answer 2

The error occurs when you are trying to add non-string types into the DOM.

Most likely you are either trying to add a (numpy) integer as an attribute name/value of you are trying to add integer content to an element.

You fix this problem by converting all integers into strings, using something like the str() function.

This can also occur with other types, e.g. numpy.float64 as well, but it is by no means limited to numpy types only, it affects all non-string types.