import sys, re

if len(sys.argv) != 3:
	print 'Usage: \nplacestats.py <fin_from_ragebuilder> <fout_raw_not_in_place_list>'
	exit(2)
	
infilename = sys.argv[1]
outfilename = sys.argv[2]

fin=open(infilename)
fout=open(outfilename,'r+')

#process the ragebuilder output to grab only those that dont load in place 
for aline in fin:
	if re.match(r'.*can not use in-place loading', aline):
		fout.write(aline)
fin.close

fout.seek(0)
		
#lists
names = []
schemas = []

#set
frequency = {}
fileTypeFrequency = {}

#now parsing something like this:
#Line 308: File memory:$03F63BF0,6004,0:packfile1:\des_cropduster.ctyp, schema CMapTypes (0xd98bb561) can not use in-place loading
linenum = 0
for line in fout:
	linelist = line.split(',')
	
	linenum += 1
	if len(linelist) < 3:
		print 'Error here ' , linenum, line, linelist
		filename = 'error'
	else:
		filename = re.sub('0:packfile1:', '', linelist[2])
		filename = re.sub(r'^\\', '', filename)
		filename = re.sub('0:', '', filename)
	if len(linelist) < 4:
		print 'Error here ' , linenum, line, linelist
		schema = 'error'
	else:
		schema = re.sub('\(0x.*', '', linelist[3])
		schema = re.sub('\r\n|\n', '', schema)
		schema = re.sub('schema ', '', schema)
		schema = re.sub(' ', '', schema)
	
	names.append(filename);
	schemas.append(schema);
	

	
	if schema in frequency:
		numUses = frequency[schema]
		frequency[schema] = numUses + 1
	else:
		frequency[schema] = 1
		
fout.close()
	
namesset = set(names)
nameslist = list(namesset)
nameslist.sort()
schemaset = set(schemas)

for fname in namesset:
	fnsplit = fname.split('.')
	filetype = fnsplit[len(fnsplit)-1]
	if filetype in fileTypeFrequency:
		numUses = fileTypeFrequency[filetype]
		fileTypeFrequency[filetype] = numUses + 1
	else:
		fileTypeFrequency[filetype] = 1

print len(nameslist) , 'files can\'t load in place:'
for x in nameslist:
	print x
print '\ndue to' ,len(schemaset), 'schemas:', schemaset
print '\nfrequency of mismatched schemas:'
for w in sorted(frequency, key=frequency.get, reverse=True):
	print w, frequency[w]
print '\n', len(nameslist) , 'files can\'t load in place freq:\n', 
for h in sorted(fileTypeFrequency, key=fileTypeFrequency.get, reverse=True):
	print h, fileTypeFrequency[h]