Parse matrix from XML using Python

Question

Let's say I have a matrix stored in an XML-file in the following format:

<?xml version="1.0"?>

<Matrix>
    <Value Col="0" Row="0">0.19343</Value>
    <Value Col="1" Row="0">0.95079</Value>
    <Value Col="2" Row="0">0.89542</Value>
    <Value Col="0" Row="1">0.14391</Value>
    <Value Col="1" Row="1">0.094629</Value>
    <Value Col="2" Row="1">0.52303</Value>
</Matrix>

What is the best way to parse these values into a numpy array using xml.etree in Python without knowing the size of the dimensions of the matrix? Otherwise, I guess i could just simply do:

import xml.etree.ElementTree as ET
import numpy as np

rowcnt = 2
colcnt = 3

xmltree = ET.parse('some_xmlfile.xml')
matrix = np.zeros(shape=(rowcnt, colcnt))

for m in xmltree.iter('Matrix'):
    for v in m.iter('Value'):
        col = int(v.attrib['Col'])
        row = int(v.attrib['Row'])
        matrix[row, col] = float(v.text)
    print matrix

Answer 1

I'm not claiming that this is the best way to create a numpy array from your XML file, but this should work for an arbitrary number of columns (although the rows must be the same size), and for arbitrarily ordered <Value> elements.

import numpy as np
import xml.etree.ElementTree as ET
from collections import defaultdict

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

data = defaultdict(list)

# group into rows of (col, val) tuples
for val in root.iter('Value'):
    data[int(val.attrib['Row'])].append((int(val.attrib['Col']), val.text))

# sort columns and format into a space separated string
rows = []
for row in data:
    rows.append(' '.join([cols[1] for cols in sorted(data[row])]))

# build array from matrix string
matrix = np.array(np.mat(';'.join(rows)))


>>> matrix
array([[ 0.19343 ,  0.95079 ,  0.89542 ],
       [ 0.14391 ,  0.094629,  0.52303 ]])