def convert(wng_loc, wng_db_loc, createdb):
''' Convert Gloss WordNet into SQLite
'''
merged_folder = os.path.join(wng_loc, 'merged')
print("Path to glosstag folder: %s" % (merged_folder))
print("Path to output database: %s" % (wng_db_loc))
print("Script to execute: %s" % (DB_INIT_SCRIPT))
if os.path.isfile(wng_db_loc):
print("DB file exists (%s | size: %s)" % (wng_db_loc,os.path.getsize(wng_db_loc)))
answer = input("If you want to overwrite this file, please type CONFIRM: ")
if answer != "CONFIRM":
print("Script aborted!")
exit()
db = SQLiteGWordNet(wng_db_loc)
if createdb:
header('Preparing database file ...')
db.setup(DB_INIT_SCRIPT)
#--
xmlfiles = [
#os.path.join(merged_folder, 'test.xml')
os.path.join(merged_folder, 'adv.xml')
,os.path.join(merged_folder, 'adj.xml')
,os.path.join(merged_folder, 'verb.xml')
,os.path.join(merged_folder, 'noun.xml')
]
header('Importing data from XML to SQLite')
xml2db(xmlfiles, db)
pass
def xml2db(xml_files, db):
''' Convert a XML file of Gloss WordNet into SQLite
'''
t = Timer()
header("Extracting Gloss WordNet (XML)")
xmlgwn = XMLGWordNet()
for xml_file in xml_files:
t.start('Reading file: %s' % xml_file)
xmlgwn.read(xml_file)
t.end("Extraction completed %s" % xml_file)
header("Inserting data into đáng SQLite database")
t.start()
db.insert_synsets(xmlgwn.synsets)
t.end('Insertion completed.')
pass
def main():
print("Script to compare WNSQL30 to WN-NTUMC")
wnntu = WordNetNTUMC(WN_NTUMC_FILE)
wn30 = WordNetSQL(WORDNET_30_PATH)
header("WordNet-NTUMC")
ssntu = wnntu.get_all_synsets()
print("Synset count: %s " % (len(ssntu),))
for ss in ssntu[:5]:
print(ss)
sidntu = set([ ss.synset if not ss.synset.endswith('r') else ss.synset[:-1] + 'a' for ss in ssntu ])
header("WordNet SQL 3.0")
sensemap = wn30.all_senses()
sswn30 = []
wn30sensemap = {}
for sses in sensemap.values():
for ss in sses:
sswn30.append(ss)
wn30sensemap[ss.get_canonical_synsetid()] = ss
print("Synset count: %s " % (len(sswn30),))
for ss in sswn30[:5]:
print( "%s: %s" % (ss.get_canonical_synsetid(), wn30.get_senseinfo_by_sid(ss.sid),) )
sidwn30 = set([ ss.get_canonical_synsetid() for ss in sswn30 ])
header("synsets in WNNTUMC but not in WNSQL30")
sids = sidntu.difference(sidwn30)
print(len(sids))
with open(NTUMC_NEW_SYNSETS, 'w') as ntuout:
for sid in sids:
ntuout.write("%s: %s\n" % (sid, ' | '.join([ x._2 for x in wnntu.get_synset_def(sid) ])))
header("synsets in WNSQL30 but not in WNNTUMC")
sids = sidwn30.difference(sidntu)
print(len(sids))
with open(WN30_NEW_SYNSETS, 'w') as wn30out:
for sid in sids:
wn30out.write("%s: %s\n" % (sid, wn30sensemap[sid].gloss))
pass
def convert(args):
''' Convert Gloss WordNet XML into SQLite format
'''
show_info(args)
if os.path.isfile(args.glossdb) and os.path.getsize(args.glossdb) > 0:
print("DB file exists (%s | size: %s)" %
(args.glossdb, os.path.getsize(args.glossdb)))
answer = input(
"If you want to overwrite this file, please type CONFIRM: ")
if answer != "CONFIRM":
print("Script aborted!")
exit()
db = get_gwn(args)
header('Importing data from XML to SQLite')
t = Timer()
header("Extracting Gloss WordNet (XML)")
xmlgwn = get_gwnxml(args)
header("Inserting data into SQLite database")
t.start()
db.insert_synsets(xmlgwn.synsets)
t.end('Insertion completed.')
pass