Integrates changes from Wycliffe-USA#62 into new dev dir structure. C…

…onnects Wycliffe-USA#25.
bbriggs · Jan 23, 2016 · 88ce273 · 88ce273
1 parent 463ad15
commit 88ce273
Show file tree

Hide file tree

Showing 10 changed files with 109,952 additions and 78,851 deletions.
diff --git a/webserver/data/asv_plain.txt b/webserver/data/asv_plain.txt
diff --git a/webserver/data/asv_tokenized.txt b/webserver/data/asv_tokenized.txt
diff --git a/webserver/data/spanish_plain.txt b/webserver/data/spanish_plain.txt
diff --git a/webserver/data/spanish_tokenized.txt b/webserver/data/spanish_tokenized.txt
diff --git a/webserver/tests/data/asv_plain.txt b/webserver/tests/data/asv_plain.txt
diff --git a/webserver/tests/data/asv_tokenized.txt b/webserver/tests/data/asv_tokenized.txt
diff --git a/webserver/tests/data/spanish_plain.txt b/webserver/tests/data/spanish_plain.txt
diff --git a/webserver/tests/data/spanish_plain.txt.out b/webserver/tests/data/spanish_plain.txt.out
diff --git a/webserver/tests/data/spanish_tokenized.txt b/webserver/tests/data/spanish_tokenized.txt
diff --git a/webserver/tools/unbound2plain.py b/webserver/tools/unbound2plain.py
@@ -15,7 +15,7 @@ def unbound2plain(infile,outfile):
         with open(infile) as source:
             for line in source:
                 if not line.startswith('#'):
-                    line = re.sub(r'^[\d\sOn]+\t','', line)
+                    line = re.sub(r'^[\d\sON]+\t','', line)
                     line = re.sub(r'\r\n$','\n',line)
                     tmp_file.write(line)
     source.close()
@@ -32,4 +32,4 @@ def main(infile,outfile):
     parser.add_argument('infile', metavar='source', type=str, nargs=1, help='the Unbound formatted text. Usually ends in _8tf8.txt')
     parser.add_argument('outfile', metavar='dest', type=str, nargs=1, help='file to write for output. Be sure to include your correct suffix (ie, .txt)')
     args = parser.parse_args()
-    main(args.infile[0],args.outfile[0])
+    main(args.infile[0],args.outfile[0])