def buildCollection(dbh, p, collectionName):
''' Build the collection specified'''
# Create the collection
try:
if p.dropCollection==True:
if p.verbose==True:
print "---- Dropping Collection."
print "---- Creating Collection."
dbh.drop_collection(collectionName)
dbh.create_collection(collectionName)
except Exception, e:
handleErrors(p, e)
def mongoUpdate(p, collection, word, slang):
""" Inserts the word and its phonetic representation into a new mongo document. """
# The query condition for the update
q = {"word": word, "wordlen": len(word)}
# Just in case the word doesn't already exist, create a new dictionary item with it, so that it maps to the emoticon/slang
try:
exists = collection.find(q).count()
if exists == 0:
collection.insert(q)
except Exception, e:
handleErrors(p, e)
def main(configFile):
''' Holds it all together '''
# Get the config information into a single object. p also gets passed into the listener
p = getConfigParameters(configFile)
# Handle the mongo connection and authentication
c, dbh, collection, emoCollection = getMongoHandles(p)
# Open the english dictionary file and loop it
try:
f = open(os.path.join(p.sourcePath, p.enPlainFile), 'r')
except Exception, e:
handleErrors(p, e)
def buildIndexes(p, collection, collHandle):
''' Build the indexes specified'''
# Create indexes
try:
if p.verbose==True:
print "---- Create Plain Indexes."
for index in collection['plain']:
collHandle.create_index([(index, ASCENDING)])
if p.verbose==True:
print "---- Index Created On: %s." %index
except Exception, e:
handleErrors(p, e)
def mongoInserter(p, collection, word, pho):
''' Inserts the word and its phonetic representation into a new mongo document. '''
record = {'word' : word,
'pho' : pho,
'wordlen' : len(word),
'pholen' : len(pho)}
try:
if collection.find(record).count() > 0:
res = 0
else:
res = collection.insert(record)
except Exception, e:
handleErrors(p, e)
res = None
def main(configFile, file="norm"):
""" Holds it all together """
# Get the config information into a single object. p also gets passed into the listener
p = getConfigParameters(configFile)
# Handle the mongo connection and authentication
c, dbh, collection, emoCollection = getMongoHandles(p)
if file == "norm":
fileToProcess = p.enNormalisedFile
elif file == "slang":
fileToProcess = p.slangFile
# Open the english dictionary file and loop it
try:
f = open(os.path.join(p.sourcePath, fileToProcess), "r")
except Exception, e:
print e
handleErrors(p, e)
print "---- Index Created On: %s." %index
except Exception, e:
handleErrors(p, e)
# Create compound indexes
try:
if p.verbose==True:
print "---- Create Compound Indexes."
for index in collection['compound']:
if p.verbose==True:
print "---- Index Created On: %s." %index
except Exception, e:
handleErrors(p, e)
#------------------------------------------------------------------------
def buildCollection(dbh, p, collectionName):
''' Build the collection specified'''
# Create the collection
try:
if p.dropCollection==True:
if p.verbose==True:
print "---- Dropping Collection."
print "---- Creating Collection."
dbh.drop_collection(collectionName)
dbh.create_collection(collectionName)
except Exception, e:
exists = collection.find(q).count()
if exists == 0:
collection.insert(q)
except Exception, e:
handleErrors(p, e)
# Adds to the list if slang isn't already present
up = {'$addToSet':{'emo':token}}
try:
response = collection.update(q, up)
except Exception, e:
response = None
handleErrors(p, e)
return response
#----------------------------------------------------------------------------------------
def getDespaced(slang):
''' Removes spaces and returns slang(s) as a list'''
# Add the existing slang to a list
slangs = [slang]
# Remove all spaces from the slang term and add that to the list too
while slang.find(' ') >= 0:
slang = slang.replace(' ', '')
