depth_parse.py

#!/usr/bin/python
from xml.etree.ElementTree import iterparse
import sys
import matplotlib.pyplot as plt
from numpy.random import normal

source_file="structure.rdf.u8"

avg_sum=0
res_count=0
depth_list = []
def resolvetag(ns, tag, nsmap):
  return '{' + nsmap[ns] + '}' + tag

print "Start processing"
nsmap = {}
for event, elem in iterparse(source_file, events=["start-ns","start"]):
  if event == "start-ns":
      ns, url = elem
      nsmap[ns] = url
  else:
      break
print nsmap

for (event, node) in iterparse(source_file, events=['start','end']):
    if "Topic" in node.tag and event == "start":
        if node.attrib.get(resolvetag("r","id",nsmap)):
            data = node.attrib[resolvetag("r","id",nsmap)]
            depth = len(data.split('/'))
            depth_list.append(depth)
            avg_sum+=depth
            res_count+=1
            print avg_sum," ", res_count
    if event == 'end':
        if node.attrib.get(resolvetag("r","id",nsmap)):
            node.clear()

print "Avarage value:", avg_sum/res_count
print "Float avarage value:", avg_sum/float(res_count)

print "Plotting..."

plt.hist(depth_list)

plt.xlabel("Depth")
plt.ylabel("Count")
plt.show()