def getclosest(query):
"""
get closest in here...
"""
# this is a set of phonemes
orig = getInventory(query)
sims = {}
for langid in u.invsets.keys():
if langid == query:
continue
# try getting F1 here instead of just intersection.
tgt = getInventory(langid)
score = phoible.getF1(orig, tgt)
#score = getDistinctiveFeatures(orig, tgt, pmap)
#score = getOV(tgt, orig, langs["eng"])
sims[langid] = score
return sims
def getclosest(query):
"""
get closest in here...
"""
# this is a set of phonemes
orig = getInventory(query)
sims = {}
for langid in u.invsets.keys():
if langid == query:
continue
# try getting F1 here instead of just intersection.
tgt = getInventory(langid)
score = phoible.getF1(orig, tgt)
#score = getDistinctiveFeatures(orig, tgt, pmap)
#score = getOV(tgt, orig, langs["eng"])
sims[langid] = score
return sims
# langlist represents a list of languages to use.
if args.langlist:
lines = map(lambda s: s[0], utils.readFile(args.langlist[0], sep=" "))
u.loadinventorysets(lines)
else:
u.loadinventorysets()
if args.getInventory:
getInventory(args.getInventory[0])
elif args.getF1:
l1, l2 = args.getF1
l1set = getInventory(l1)
l2set = getInventory(l2)
print phoible.getF1(l1set, l2set)
elif args.getClosest:
lang = args.getClosest[0]
c = getclosest(lang)
ci = sorted(c.items(), key=lambda p: p[1], reverse=True)
logger.info("Writing to mine." + lang)
with open("mine." + lang, "w") as out:
out.write(lang + "\t1.0\n")
ignored = 0
iset = set()
for l in ci:
if l[1] >= 0:
out.write(l[0] + "\t" + str(l[1]) + "\n")
else:
iset.add(l[0])
ignored += 1
# langlist represents a list of languages to use.
if args.langlist:
lines = map(lambda s: s[0], utils.readFile(args.langlist[0], sep=" "))
u.loadinventorysets(lines)
else:
u.loadinventorysets()
if args.getInventory:
getInventory(args.getInventory[0])
elif args.getF1:
l1, l2 = args.getF1
l1set = getInventory(l1)
l2set = getInventory(l2)
print phoible.getF1(l1set, l2set)
elif args.getClosest:
lang = args.getClosest[0]
c = getclosest(lang)
ci = sorted(c.items(), key=lambda p: p[1], reverse=True)
logger.info("Writing to mine." + lang)
with open("mine." + lang, "w") as out:
out.write(lang + "\t1.0\n")
ignored = 0
iset = set()
for l in ci:
if l[1] >= 0:
out.write(l[0] + "\t" + str(l[1]) + "\n")
else:
iset.add(l[0])
ignored += 1