# basename = args[0]
# doc = annodisAnnot(basename+".xml")
# feats = FeatureMap(basename+".features")
# prep = Preprocess( basename+".txt.prep.xml")
if options.merge:
feats = FeatureMap("", empty = True)
feats.init_from_dir(".", suffix = options.merge)
basename = "no base file to consider"
else:
if True:
basename = args[0]
feats = FeatureMap(basename + ".features", weird = options.weird)
feats.index(["m#%s" % INDEX1, "m#%s" % INDEX2])
if allfuncs != {}:
doc = annodisAnnot(basename + ".xml")
prep = Preprocess(basename + ".txt.prep.xml")
doc.add_preprocess(prep)
if options.voisins:
doc._voisins = get_voisins_dict(table, doc._vocab)
for entry in doc._voisins:
doc._voisins[entry] = dict(doc._voisins[entry])
else:
print >> sys.stderr, "Usage: script file-basename ?", args
sys.exit(0)
if options.merge:
feats.index(["m#%s" % INDEX1, "m#%s" % INDEX2, "m#FILE"])
else:
feats.index(["m#%s" % INDEX1, "m#%s" % INDEX2])
# basename = args[0]
# doc = annodisAnnot(basename+".xml")
# feats = FeatureMap(basename+".features")
# prep = Preprocess( basename+".txt.prep.xml")
if options.merge:
feats = FeatureMap("", empty=True)
feats.init_from_dir(".", suffix=options.merge)
basename = "no base file to consider"
else:
if True:
basename = args[0]
feats = FeatureMap(basename + ".features", weird=options.weird)
feats.index(["m#%s" % INDEX1, "m#%s" % INDEX2])
if allfuncs != {}:
doc = annodisAnnot(basename + ".xml")
prep = Preprocess(basename + ".txt.prep.xml")
doc.add_preprocess(prep)
if options.voisins:
doc._voisins = get_voisins_dict(table, doc._vocab)
for entry in doc._voisins:
doc._voisins[entry] = dict(doc._voisins[entry])
else:
print >> sys.stderr, "Usage: script file-basename ?", args
sys.exit(0)
if options.merge:
feats.index(["m#%s" % INDEX1, "m#%s" % INDEX2, "m#FILE"])
else:
feats.index(["m#%s" % INDEX1, "m#%s" % INDEX2])
import codecs
import re
import xml.etree.ElementTree as ET
from AnnodisReader import annodisAnnot, Preprocess
from Lookup import Gazetteer
def includes(pos1,pos2):
a1,b1=pos1
a2,b2=pos2
return a1<=a2 and b2<=b1
if __name__=="__main__":
try:
doc=annodisAnnot(sys.argv[1])
except:
print "ERROR reading file:", sys.argv[1]
sys.exit(0)
try:
prep=Preprocess(sys.argv[1].split(".xml")[0]+".txt.prep.xml")
doc.add_preprocess(prep)
except:
print "ERROR reading prepocessed file for", sys.argv[1]
sys.exit(0)
lexicon=Gazetteer(sys.argv[2])
txt=doc.text()
lookup=lexicon.tag(txt)