Esempio n. 1
0
    "-V", "--verbose",
    action="store_true",
    help="verbose output"
    )


args = parser.parse_args()

if args.verbose:
    print >>stderr, "Reading corpus from", args.corpus
    
corpus = HitaextDoc(file=args.corpus)

from_tree = corpus.get_doc_tree("from")
to_tree = corpus.get_doc_tree("to")

from_tree.update()
to_tree.update()

corpus.inject_alignments(from_tree, to_tree)

corpus.alignment.set("method", "id")

corpus.extract_alignments(from_tree, to_tree)

if args.verbose:
    print >>stderr, "Writing corpus to", args.corpus
    
corpus.write(args.corpus)

Esempio n. 2
0
                    help="")

parser.add_argument("-V",
                    "--verbose",
                    action="store_true",
                    help="verbose output")

args = parser.parse_args()

if args.verbose:
    print >> stderr, "Reading corpus from", args.corpus

corpus = HitaextDoc(file=args.corpus)

from_tree = corpus.get_doc_tree("from")
to_tree = corpus.get_doc_tree("to")

from_tree.update()
to_tree.update()

corpus.inject_alignments(from_tree, to_tree)

corpus.alignment.set("method", "id")

corpus.extract_alignments(from_tree, to_tree)

if args.verbose:
    print >> stderr, "Writing corpus to", args.corpus

corpus.write(args.corpus)
Esempio n. 3
0
    if args.verbose or args.test:
        print >>stderr, "Reading Hitaext document", fn
        
    htdoc = HitaextDoc(file=fn)
    
    for side in ("from", "to"):
        path = htdoc.get_filename(side)
        
        if args.verbose or args.test:
            print >>stderr, "Current %s path is %s" %  (side, path)
        
        # a heuristic to deal with windows paths        
        if path[0] in uppercase and path[1] == ":":
            # strip drive letter
            path = path[2:]
        path = path.replace("\\", "/")
        
        path = join(args.dir, basename(path))
        
        if args.verbose or args.test:
            print >>stderr, "Changing %s path to %s" %  (side, path)
        
        if not args.test:
            htdoc.set_filename(side, path)
        
    if args.verbose and not args.test:
        print >>stderr, "Writing Hitaext document\n", fn
    
    if not args.test:
        htdoc.write(fn)
Esempio n. 4
0
    if args.verbose or args.test:
        print >> stderr, "Reading Hitaext document", fn

    htdoc = HitaextDoc(file=fn)

    for side in ("from", "to"):
        path = htdoc.get_filename(side)

        if args.verbose or args.test:
            print >> stderr, "Current %s path is %s" % (side, path)

        # a heuristic to deal with windows paths
        if path[0] in uppercase and path[1] == ":":
            # strip drive letter
            path = path[2:]
        path = path.replace("\\", "/")

        path = join(args.dir, basename(path))

        if args.verbose or args.test:
            print >> stderr, "Changing %s path to %s" % (side, path)

        if not args.test:
            htdoc.set_filename(side, path)

    if args.verbose and not args.test:
        print >> stderr, "Writing Hitaext document\n", fn

    if not args.test:
        htdoc.write(fn)