def __init__(self, seed=None):
# if no seed is provided, return an error
if seed == None:
print "Initializing an Analyzer object requires a seed. Please try again."
return
# initialize an empty dictionary
dict = {}
# ---USER FEEDBACK---
print "\nAttempting to read seed file..."
# open the seed file
f = Util.open_file(seed)
# if the file doesn't exist, return
if not f:
print "Initialization failed; please verify that the seed exists then try again."
return
# begin reading
while True :
# read heuristic line
line = f.readline()
# quit if end of file
if not line : break
# store new heuristic
current_heuristic = Util.strip(line)
# ---USER FEEDBACK---
print "Reading files for heuristic \'" + current_heuristic + "\'..."
# read filenames
next_line = f.readline()
# if there isn't another line, quit - incorrect syntax
if not next_line :
print ("Incorrect seed structure. Exiting")
sys.exit()
# try to store number of files for this heuristic
try:
num_files = int(next_line)
# if an exception is thrown...
except ValueError:
# print out an error and return nothing
print "Seed file is of incorrect format. Please try again."
return
# create document array variable
docs = []
# iterate over files
for i in range(num_files) :
# try to open the file
filename = current_heuristic + "/" + str(i) + ".txt"
new_doc = Document(None, filename)
# if the new document's text is successfuly initialized...
if new_doc.text :
# add it to the array
docs.append(new_doc)
# add new heuristic and docs to dict
dict[current_heuristic] = docs
# store dictionary
self.dict = dict
# ---USER FEEDBACK---
print "Done reading files!\n"
# calculate required values
# initialize dictionaries
self.word_counts = {}
self.log_values = {}
# analyze for each heuristic found
for key in self.dict:
self.consolidate(key)
self.transform(key, True)
# ---USER FEEDBACK---
print "Analyzer object initialized!\n"
