def pgc_zip(zip_filename, pgc_filenames):
zip_arch = zipfile.ZipFile(zip_filename, "w")
arch_dir = os.path.splitext( os.path.basename(zip_filename))[0]
for corpus_filename in multiglob(pgc_filenames):
# add corpus to archive
arch_filename = os.path.join( arch_dir,
os.path.basename(corpus_filename) )
zip_arch.write(corpus_filename, arch_filename)
corpus = ParallelGraphCorpus(inf=corpus_filename,
graph_loading=LOAD_NONE)
for gb in corpus._graphbanks():
gb_filename = gb.get_file_path()
# add graphbank files to archive
arch_filename = os.path.join( arch_dir,
os.path.basename(gb_filename) )
zip_arch.write(gb_filename, arch_filename)
zip_arch.close()
def pgc_zip(zip_filename, pgc_filenames):
zip_arch = zipfile.ZipFile(zip_filename, "w")
arch_dir = os.path.splitext(os.path.basename(zip_filename))[0]
for corpus_filename in multiglob(pgc_filenames):
# add corpus to archive
arch_filename = os.path.join(arch_dir,
os.path.basename(corpus_filename))
zip_arch.write(corpus_filename, arch_filename)
corpus = ParallelGraphCorpus(inf=corpus_filename,
graph_loading=LOAD_NONE)
for gb in corpus._graphbanks():
gb_filename = gb.get_file_path()
# add graphbank files to archive
arch_filename = os.path.join(arch_dir,
os.path.basename(gb_filename))
zip_arch.write(gb_filename, arch_filename)
zip_arch.close()
parser.add_argument(
"-f", "--format",
action="store_true",
help="output indented XML"
)
parser.add_argument(
"-V", "--verbose",
action="store_true",
help="verbose ouput to stderr"
)
args = parser.parse_args()
pgc_fns = multiglob(args.file)
def log(s):
if args.verbose:
print >>sys.stderr, "***", s
log("Reading corpus from " + pgc_fns[0])
corpus = ParallelGraphCorpus(inf=pgc_fns[0])
for fn in pgc_fns[1:]:
log("Joining corpus from " + fn)
# __iadd__ also checks if another corpus is compatible w.r.t. relations
# and meta-data
corpus += ParallelGraphCorpus(inf=fn)
"--config",
metavar="FILE",
help="configuration file to set up a corpus aligner")
parser.add_argument("-x",
"--clear",
action="store_true",
help="remove all existing alignments")
parser.add_argument("-i",
"--in-place",
action="store_true",
help="modify input file(s)")
args = parser.parse_args()
if args.config:
config = imp.load_source("config", args.config)
corpus_aligner = set_up_corpus_aligner(config)
else:
from daeso_nl.ga.corpus import CorpusAligner
corpus_aligner = CorpusAligner()
for inf in multiglob(args.pgc_files):
corpus = ParallelGraphCorpus(inf=inf)
corpus_aligner.align(corpus, clear=args.clear)
if args.in_place:
corpus.write(outf=inf, pprint=True)
else:
corpus.write(pprint=True)
parser.add_argument(
"-x", "--clear",
action="store_true",
help="remove all existing alignments"
)
parser.add_argument(
"-i", "--in-place",
action="store_true",
help="modify input file(s)"
)
args = parser.parse_args()
if args.config:
config = imp.load_source("config", args.config)
corpus_aligner = set_up_corpus_aligner(config)
else:
from daeso_nl.ga.corpus import CorpusAligner
corpus_aligner = CorpusAligner()
for inf in multiglob(args.pgc_files):
corpus = ParallelGraphCorpus(inf=inf)
corpus_aligner.align(corpus, clear=args.clear)
if args.in_place:
corpus.write(outf=inf, pprint=True)
else:
corpus.write(pprint=True)
help="parallel graph corpus filename, "
"or quoted file name pattern for parallel graph corpora")
parser.add_argument("-f",
"--format",
action="store_true",
help="output indented XML")
parser.add_argument("-V",
"--verbose",
action="store_true",
help="verbose ouput to stderr")
args = parser.parse_args()
pgc_fns = multiglob(args.file)
def log(s):
if args.verbose:
print >> sys.stderr, "***", s
log("Reading corpus from " + pgc_fns[0])
corpus = ParallelGraphCorpus(inf=pgc_fns[0])
for fn in pgc_fns[1:]:
log("Joining corpus from " + fn)
# __iadd__ also checks if another corpus is compatible w.r.t. relations
# and meta-data
