def parseFile(infile,
lemodel,
tagmodel,
parsemodel,
folderpref="mate/parses/"):
timestamp = datetime.datetime.now().strftime('%Y-%m-%d_%H:%M')
if folderpref: prelimfolder = folderpref + "_prelim/"
else: prelimfolder = folderpref + timestamp + "_prelim/"
parsefile = parsing(infile=infile,
lemodel=lemodel,
tagmodel=tagmodel,
parsemodel=parsemodel,
outfolder=prelimfolder,
memory=memory) # , depparse=False
#parsefile="mate/parses/2016-09-22_01:18/Louise_Liotard_F_85_et_Jeanne_Mallet_F_75_SO-2-one-word-per-line.conll14_parse"
print "retokenizing..."
newname = retokeniser(parsefile, addtoout="_retok")
print "retokenization done"
if folderpref: outfolder = folderpref + "/"
else: outfolder = folderpref + timestamp + "/"
createNonExistingFolders(outfolder)
emptyname = makeEmpty(newname, outfolder=outfolder)
parsefile = parsing(infile=emptyname,
lemodel=modeldir + args.get("lemmodel", None),
tagmodel=modeldir + args.get("tagmodel", None),
parsemodel=modeldir + args.get("parsemodel", None),
outfolder=outfolder,
memory="40G")
def parseSentenceFile(sentencefile,
modelfolder="mate/platinum.2016-10-20_01:34/models/",
lemodel="LemModel",
tagmodel="TagModel",
parsemodel="ParseModel",
outfolder=".",
memory="4G",
removePunct=False,
specialCharWords=[],
degrade=False):
"""
removed parsertype from options!
"""
print "\n\nparseSentenceFile", sentencefile
emptyConll = emptyFromSentence(sentencefile,
specialCharWords=specialCharWords,
outfolder=outfolder)
print "made empty", emptyConll
parsedfile = parsing(emptyConll,
lemodel=modelfolder + lemodel,
tagmodel=modelfolder + tagmodel,
parsemodel=modelfolder + parsemodel,
outfolder=outfolder,
memory=memory)
print parsedfile
if removePunct: removePuncsFromConllfile(parsedfile)
if degrade: degradeConllfile(parsedfile)
return parsedfile
def parseSentenceFile(sentencefile,
modelfolder="mate/platinum.2016-10-20_01:34/models/",
lemodel="LemModel",
tagmodel="TagModel",
parsemodel="ParseModel",
outfolder=".",
memory="40G",
removePunct=False,
useTokDic=False,
degrade=False):
"""
removed parsertype from options!
"""
emptyConll = emptyFromSentence(sentencefile, useTokDic=useTokDic)
parsedfile = parsing(emptyConll,
lemodel=modelfolder + lemodel,
tagmodel=modelfolder + tagmodel,
parsemodel=modelfolder + parsemodel,
outfolder=outfolder,
memory=memory)
print parsedfile
if removePunct: removePuncsFromConllfile(parsedfile)
if degrade: degradeConllfile(parsedfile)