An overhead camera has taken a photo of the papers on a bureaucrat's desk. Many of the papers overlap, but since the bureaucrat in question is a bit of a neatness freak, all papers are aligned with their edges parallel to the edges of the desk. As a preliminary step in analyzing these papers, other programs have used cues such as paper color, edges, etc., to label each pixel in the image according to which sheet of paper is visible at that particular location. The result will look something like this:
...aaaaaaaa.dd...aaaaaaaa.ee.ccaaaaaaaa....ccaaaabbaa....ccaaaabbaa....ccaaaabbaa....ccaaaaaaaa.................Here, a '.' denotes the desktop, and alphabetic labels 'a', 'b', 'c', etc., denote distinct sheets of paper. These labels are assigned in an arbitrary order but are dense (no letters are skipped during labeling).
Assume that each sheet of paper is rectangular, and that no important information is being lost between the pixels. Also assume that the camera has captured the entire desktop and that there are no sheets that are completely hidden from view.
For each sheet of paper (in character code order), determine whether the visual evidence proves that the entire sheet of paper is visible, or whether it is possible that a part of the paper is hidden under another sheet.
The input consists of several snapshots.
Each snapshot begins with a line containing two integers, and , denoting the number of rows and columns of pixels in the snapshot. For each snapshot, 1 \leq R,C \leq 40.
The and values are followed by lines, each with characters. These characters will be lowercase alphabetic characters or a period ('.'). As noted earlier, alphabetic characters denote visible portions of distinct sheets of paper, and periods denote parts of the desktop underneath.
All snapshots in the input correspond to a valid arrangement of rectangular sheets of paper. There will be no impossible arrangements.
The last snapshot will be followed by a line containing two zeros separated by a space.
For each snapshot, print one line of output. That line begins with the phrase Uncovered: followed by a space. Then, on the same line, list the papers that are definitely completely visible, in order of character code, with no separating spaces.
Example 1