def main():
quick = False
nofail = False
try:
opts, args = getopt.getopt(sys.argv[1:], "E:srhdqVi", ["help"])
except getopt.GetoptError as err:
print(str(err), file=sys.stderr)
usage()
sys.exit(2)
for o, a in opts:
if o == "-h" or o == "--help":
usage()
sys.exit(0)
elif o == "-E":
settings.extension = a
elif o == "-r":
settings.recurse = True
elif o == "-d":
settings.deep = True
elif o == "-q":
quick = True
elif o == "-i":
nofail = True
elif o == "-V":
print("FoLiA " + folia.FOLIAVERSION + ", library version " + folia.LIBVERSION, file=sys.stderr)
sys.exit(0)
else:
raise Exception("No such option: " + o)
schema = lxml.etree.RelaxNG(folia.relaxng())
if len(args) >= 1:
success = True
for x in sys.argv[1:]:
if x[0] != "-":
if os.path.isdir(x):
r = processdir(x, schema, quick, settings.deep)
elif os.path.isfile(x):
r = validate(x, schema, quick, settings.deep)
else:
print("ERROR: File or directory not found: " + x, file=sys.stderr)
sys.exit(3)
if not r:
success = False
if not success and not nofail:
sys.exit(1)
else:
print("ERROR: No files specified", file=sys.stderr)
sys.exit(2)
def main():
quick = False
try:
opts, args = getopt.getopt(sys.argv[1:], "E:srhqV", ["help"])
except getopt.GetoptError as err:
print(str(err), file=sys.stderr)
usage()
sys.exit(2)
for o, a in opts:
if o == '-h' or o == '--help':
usage()
sys.exit(0)
elif o == '-E':
settings.extension = a
elif o == '-r':
settings.recurse = True
elif o == '-q':
quick = True
elif o == '-V':
print("FoLiA " + folia.FOLIAVERSION + ", library version " + folia.LIBVERSION,file=sys.stderr)
sys.exit(0)
else:
raise Exception("No such option: " + o)
schema = lxml.etree.RelaxNG(folia.relaxng())
if len(args) >= 1:
for x in sys.argv[1:]:
if x[0] != '-':
if os.path.isdir(x):
processdir(x,schema,quick)
elif os.path.isfile(x):
validate(x, schema,quick)
else:
print("ERROR: File or directory not found: " + x,file=sys.stderr)
sys.exit(3)
else:
print("ERROR: No files specified",file=sys.stderr)
sys.exit(2)
for o, a in opts:
if o == '-h' or o == '--help':
usage()
sys.exit(0)
elif o == '-E':
settings.extension = a
elif o == '-r':
settings.recurse = True
else:
raise Exception("No such option: " + o)
schema = lxml.etree.RelaxNG(folia.relaxng())
if len(sys.argv) >= 2:
for x in sys.argv[1:]:
if os.path.isdir(x):
processdir(x,schema)
elif os.path.isfile(x):
validate(x, schema)
else:
print >>sys.stderr, "ERROR: File or directory not found: " + x
sys.exit(3)
else:
print >>sys.stderr,"ERROR: No files specified"
sys.exit(2)
if __name__ == "__main__":
if __name__ == '__main__':
try:
sonardir = sys.argv[1]
foliadir = sys.argv[2]
xsltfile = sys.argv[3]
threads = int(sys.argv[4])
except:
print >> sys.stderr, "Syntax: sonarconversion.py sonardir foliadir dcoi2folia.xsl threads"
sys.exit(2)
#Let XSLT do the basic conversion to HTML
xslt = lxml.etree.parse(xsltfile)
dcoitofoliatransformer = lxml.etree.XSLT(xslt)
schema = lxml.etree.RelaxNG(folia.relaxng())
if foliadir[-1] != '/': foliadir += '/'
try:
os.mkdir(foliadir[:-1])
except:
pass
maxtasksperchild = 10
preindex = True
processor = folia.CorpusProcessor(sonardir, process, threads, 'pos', "",
lambda x: True, maxtasksperchild,
preindex)
for i, _ in enumerate(processor):
progress = round((i + 1) / float(len(processor.index)) * 100, 1)
print "#" + str(i) + " - " + str(progress) + '%'
#!/usr/bin/env python
#-*- coding:utf-8 -*-
import sys
try:
from pynlpl.formats import folia
except ImportError:
print >>sys.stderr,"PyNLPL is required to run this script, obtain it from https://github.com/proycon/pynlpl and either install it or make sure the pynlpl/ dir is symlinked from the directory this script is in."
sys.exit(3)
import codecs
print "Generating Relax NG schema"
folia.relaxng('folia-min.rng')
#!/usr/bin/python
data = codecs.open("folia-min.rng",'r','utf-8').readlines()
data.insert(1,u"""<!--
RelaxNG schema for FoLiA XML v%s
by Maarten van Gompel
Induction of Linguistic Knowledge Research group
Tilburg University
http://ilk.uvt.nl/folia
http://github.com/proycon/folia
Schema version %s
(auto-generated by pynlpl.formats.folia)
Validation examples:
$ xmllint ‒‒relaxng folia.rng foliadocument.xml
$ jing folia.rng foliadocument.xml
#-*- coding:utf-8 -*-
from __future__ import print_function, unicode_literals, division, absolute_import
import sys
try:
from pynlpl.formats import folia
except ImportError:
print(
"PyNLPL is required to run this script, obtain it from https://github.com/proycon/pynlpl and either install it or make sure the pynlpl/ dir is symlinked from the directory this script is in.",
file=sys.stderr)
sys.exit(3)
import io
print("Generating Relax NG schema")
folia.relaxng('folia-min.rng')
#!/usr/bin/python
data = io.open("folia-min.rng", 'r', encoding='utf-8').readlines()
data.insert(
1, """<!--
RelaxNG schema for FoLiA XML v%s
by Maarten van Gompel
Induction of Linguistic Knowledge Research group
Tilburg University
http://ilk.uvt.nl/folia
http://github.com/proycon/folia
Schema version %s
(auto-generated by pynlpl.formats.folia)
def main():
quick = False
nofail = False
try:
opts, args = getopt.getopt(sys.argv[1:], "E:D:WsrhdqVit", ["help"])
except getopt.GetoptError as err:
print(str(err), file=sys.stderr)
usage()
sys.exit(2)
for o, a in opts:
if o == '-h' or o == '--help':
usage()
sys.exit(0)
elif o == '-D':
settings.debug = int(a)
print("DEBUG level set to " + str(settings.debug), file=sys.stderr)
elif o == '-E':
settings.extension = a
elif o == '-r':
settings.recurse = True
elif o == '-t':
settings.stricttextvalidation = True
elif o == '-d':
settings.deep = True
elif o == '-q':
quick = True
elif o == '-i':
nofail = True
elif o == '-W':
settings.warn = False
elif o == '-V':
print("FoLiA " + folia.FOLIAVERSION + ", library version " +
folia.LIBVERSION,
file=sys.stderr)
sys.exit(0)
else:
raise Exception("No such option: " + o)
schema = lxml.etree.RelaxNG(folia.relaxng())
if len(args) >= 1:
success = True
skipnext = False
for x in sys.argv[1:]:
if x in ('-E', '-D'):
skipnext = True
continue
elif x[0] != '-' and not skipnext:
if os.path.isdir(x):
r = processdir(x, schema, quick, settings.deep,
settings.stricttextvalidation,
settings.warn, settings.debug)
elif os.path.isfile(x):
r = validate(x, schema, quick, settings.deep,
settings.stricttextvalidation, settings.warn,
settings.debug)
else:
print("ERROR: File or directory not found: " + x,
file=sys.stderr)
sys.exit(3)
if not r: success = False
if not success and not nofail:
sys.exit(1)
skipnext = False
else:
print("ERROR: No files specified", file=sys.stderr)
sys.exit(2)
