dost

@@ -64,8 +64,12 @@ class Trie:

             self.use(tokens)

 if __name__ == "__main__":

-    parser = argparse.ArgumentParser(prog="dost", description="Diverse-order sampling by trie")

+    parser = argparse.ArgumentParser(

+        prog="dost", description="Diverse-order sampling by trie",

+        epilog="Tends to run faster on shorter tokens"

+    )

     parser.add_argument("files", default="-", nargs="*")

+    parser.add_argument("-i", "--in-place", action="store_true")

     parser.add_argument(

         "-s", "--separators", default=" \t",

         help="characters that separate tokens in an input item (default space and tab)"

@@ -75,6 +79,8 @@ if __name__ == "__main__":

         help="structural regexp for a token of an input item, overrides -s"

     )

     args = parser.parse_args()

+    if args.in_place and len(args.files) != 1:

+        sys.exit(1)

     re_flags = re.MULTILINE | re.DOTALL

     if args.token:

         token_pattern = args.token

@@ -87,5 +93,9 @@ if __name__ == "__main__":

         with sys.stdin if filename == "-" else open(filename, "r") as handle:

             for word in re.finditer(entry_pattern, handle.read()):

                 t.add(re.findall(token_pattern, word.group(0)))

+    output_handle = sys.stdout

+    if args.in_place and args.files[0] != "-":

+        output_handle = open(args.files[0], "w")

     for word in t.traverse():

-        print("".join(word))

+        output_handle.write("".join(word) + "\n")

+    output_handle.close()