russelnickson/Hyper-Linked Tag Cloud

## Hyper-Linked Tag Cloud

          # PROGRAM TO GENERATE HYPER-LINKED TAG CLOUD #

  # Run at command line with three parameters
    # Source File
    # Noise File
    # No.of Tags to Generate


import string,sys,os,webbrowser


f1 = open(sys.argv[1])
f2 = open(sys.argv[2])
try:
    count = int(sys.argv[3])
except IndexError:
    count = 20
content = string.lower(f1.read())
noisecontent = string.lower(f2.read())
workinglist = string.split(content)
cleanlist = []
for item in workinglist:
    temp = item.strip(string.punctuation)
    cleanlist = cleanlist + [temp,]
freq = {}
for item in cleanlist:
    if item in noisecontent:
        continue
    else:
        try:
           freq[item] += 1
        except KeyError:
           freq[item] = 1


def most_common(h):
    t = []
    for key, value in h.items():
        t.append((value, key))
    t.sort(reverse = True)
    return t


print '\nWORD' +'\t\t    ' + 'FREQUENCY'
freq = most_common(freq)
for freque, word in freq[0:count]:
    print "%-15s" % word , '\t' , freque


cwd = os.getcwd()
workinglist = content.split('.')
for index in range(count):
    link = cwd + '/hyper-linked_tagcloud' + str(index) + '.html'
    sen_wit_word = ('',)
    for items in workinglist:
        if freq[index][1] in items:
            sen_wit_word=sen_wit_word + (items,)
            f=open(link,'w')
    f.write('<html><body>')
    for item in sen_wit_word:
        item = item.replace(freq[index][1],'<font color="#ff0000">' + freq[index][1] + '</font>')
        f.write(item + '\n<br/>')
    f.write('</body></html>')
    f.close()


link = cwd + '/hyper-linked_tagcloud.html'
f = open(link,'w')
f.write('<html><body>')
index = 0
for freque, word in freq[0:count]:
    f.write('<font face = "Times New Roman"size = "' + str((freque*10) / freq[0][0]) + '"color="#23bb07">'+'<a href="' + cwd + '/hyper-linked_tagcloud' + str(index) + '.html">' + word + '</a>' + '\t\t\t</font>')
    index += 1
f.write('</body></html>')
f.close()


iecommand = "c:/program files/internet explorer/iexplore.exe %s &"
ie = webbrowser.get(iecommand)
ie.open(link)


                                                # russel nickson

	# PROGRAM TO GENERATE HYPER-LINKED TAG CLOUD #

	# Run at command line with three parameters
	# Source File
	# Noise File
	# No.of Tags to Generate


	import string,sys,os,webbrowser


	f1 = open(sys.argv[1])
	f2 = open(sys.argv[2])
	try:
	count = int(sys.argv[3])
	except IndexError:
	count = 20
	content = string.lower(f1.read())
	noisecontent = string.lower(f2.read())
	workinglist = string.split(content)
	cleanlist = []
	for item in workinglist:
	temp = item.strip(string.punctuation)
	cleanlist = cleanlist + [temp,]
	freq = {}
	for item in cleanlist:
	if item in noisecontent:
	continue
	else:
	try:
	freq[item] += 1
	except KeyError:
	freq[item] = 1


	def most_common(h):
	t = []
	for key, value in h.items():
	t.append((value, key))
	t.sort(reverse = True)
	return t



	print '\nWORD' +'\t\t ' + 'FREQUENCY'
	freq = most_common(freq)
	for freque, word in freq[0:count]:
	print "%-15s" % word , '\t' , freque




	cwd = os.getcwd()
	workinglist = content.split('.')
	for index in range(count):
	link = cwd + '/hyper-linked_tagcloud' + str(index) + '.html'
	sen_wit_word = ('',)
	for items in workinglist:
	if freq[index][1] in items:
	sen_wit_word=sen_wit_word + (items,)
	f=open(link,'w')
	f.write('<html><body>')
	for item in sen_wit_word:
	item = item.replace(freq[index][1],'<font color="#ff0000">' + freq[index][1] + '</font>')
	f.write(item + '\n<br/>')
	f.write('</body></html>')
	f.close()



	link = cwd + '/hyper-linked_tagcloud.html'
	f = open(link,'w')
	f.write('<html><body>')
	index = 0
	for freque, word in freq[0:count]:
	f.write('<font face = "Times New Roman"size = "' + str((freque*10) / freq[0][0]) + '"color="#23bb07">'+'<a href="' + cwd + '/hyper-linked_tagcloud' + str(index) + '.html">' + word + '</a>' + '\t\t\t</font>')
	index += 1
	f.write('</body></html>')
	f.close()



	iecommand = "c:/program files/internet explorer/iexplore.exe %s &"
	ie = webbrowser.get(iecommand)
	ie.open(link)



	# russel nickson