def buildMaster(gathered):
master = {}
# Put all the fingerprints into a single dictionary
for assignment in gathered:
for element in assignment.values():
for entry in element:
createOrAppend(master, entry.hash, entry)
return master
def compareAll(assignment, master, matches):
# Calculates inersection between assignments.
for key in assignment.keys():
if key not in master:
continue
data = master[key]
# Check for fingerprints with same hash in the master.
for entry in data:
# If there is a match and its not by the same author, add it.
if entry.auth != assignment[key][0].auth:
createOrAppend(matches, entry.auth, entry)
def vsDB(assignment, matches, db_path, lang):
import sqlite3
conn = sqlite3.connect(db_path)
cur = conn.cursor()
lang, auth = getTableNames(lang)
query_string = "select hash, sline, scol, eline, ecol, auth, path"
query_string += "from " + lang + ", " + auth + " where auth<>? and "
query_string += "hash=? and " + lang + ".fileid=" + auth + ".rowid"
# queryString = 'select hash, sline, scol, eline, ecol, auth, path from ' + lang + ' where auth<>? and hash=?'
for key in assignment.keys():
target = (assignment[key][0].auth, key)
for row in cur.execute(query_string, target):
createOrAppend(matches, row["auth"], Entry(*row))
def processFolder(path, merge, author, options):
"""Returns fingerprints of all files found in folder
Runs recursively on all folders found there of. Adds all fingerprints to
the dictionary its passed."""
#Path juggling.
current = os.getcwd()
os.chdir(path)
targets = os.listdir(".")
for entry in targets:
if os.path.isdir(entry): #If folder, recurse.
processFolder(entry, merge, author, options)
else:
temp = examine(entry, options)
for tup in temp: #Add to dictionary.
entry = Entry(*tup, auth=author, path=os.path.abspath(entry))
createOrAppend(merge, tup.hash, entry)
os.chdir(current)
def processFolder(path, merge, author, options):
"""Returns fingerprints of all files found in folder
Runs recursively on all folders found there of. Adds all fingerprints to
the dictionary its passed."""
#Path juggling.
current = os.getcwd()
os.chdir(path)
targets = os.listdir(".")
for entry in targets:
if os.path.isdir(entry): #If folder, recurse.
processFolder(entry, merge, author, options)
else:
temp = examine(entry, options)
for tup in temp: #Add to dictionary.
entry = Entry(*tup, auth=author, path=os.path.abspath(entry))
createOrAppend(merge, tup.hash, entry)
os.chdir(current)
