Python uses lxml with fileinput

Question

Having a simple xml

<?xml version="1.0" encoding="UTF-8" ?>
<root>
    <child>abc</child>
</root>

I wanted to parse it from a file, and this works well:

with open('tst.xml') as test_xml:
    for _, element in lxml.etree.iterparse(test_xml, tag='child'):
        print element.text # prints abc as expected

However, I tried to modify the script to allow it to parse the xml from either the file or from stdinand failed:

fi = fileinput.input('tst.xml')
for _, element in lxml.etree.iterparse(fi, tag='child'):
    print element.text

# File "iterparse.pxi", line 371, in lxml.etree.iterparse.__init__ (src/lxml/lxml.etree.c:97283)
# File "apihelpers.pxi", line 1411, in lxml.etree._encodeFilename (src/lxml/lxml.etree.c:22515)
# TypeError: Argument must be string or unicode.

I'm not sure what I'm doing wrong. Is a FileInput object not a file-like object in python?

Answer 1

Without deep research, it seems that the reason for the exception is that the class FileInputdoes not provide a method read. To achieve my goal, I eventually wrote my own shell:

class FileInput(object):
    def __init__(self, filename=None, *args, **kwargs):
        self.file = open(filename, *args, **kwargs) if filename and filename != "-" else sys.stdin

    def __enter__(self):
        return self.file

    def __exit__(self, type, value, traceback):
        if self.file is not sys.stdin:
            self.file.close()

    def __getattr__(self, name):
        return getattr(self.file, name)

I will wait for a better answer though.

Answer 2

fileinput, :

if filename == '-':   # or, if we don't have a filename argument
    f = sys.stdin
else:
    f = open(filename, 'r')