Huffman Coding Trees

------- ------ -----

Huffman Coding is a method of compressing data. It is
based on specialized binary trees, which we will call
Huffman Coding Trees.
A Huffman Coding Tree for (compressing) the string
`abaacb' is made as follows. First, make single node
trees for each character appearing in the string, and
assign each tree a `weight' equal to the number of times
the character appears in the string. We get the three
single node trees:
a/3 b/2 c/1
where `a', `b', and `'c' are the names of the single
node trees and 3, 2, and 1 are their weights.
Next we select the two trees with the lowest weights and
replace them both by a single larger tree made by adding
a root node whose two children are the two trees being
replaced. In this case we get
(b)(c)/3
where the name of a tree with two children is written:
(left-child-name)(right-child-name)
and the weight of the new tree is the sum of the weights
of its children. In this case `b' is the left child,
`c' is the right child, and 3=2+1 is the new weight.
After this first tree replacement we are left with two
trees:
a/3 (b)(c)/3
We repeat the replacement step until there is only one
tree:
(a)((b)(c))/6
which is a Huffman Coding Tree of the string `abaacb'.
In this problem you are to compute a Huffman Coding Tree
for arbitrary strings of lower case letters. There is
some ambiguity in the above process we want to remove to
make the judging easier, so we require you to have the
first letter in any node's left child's name be less, in
dictionary order, than the first letter in the node's
right child's name. Thus `(b)(c)' is acceptable but
`(c)(b)' is not, and `(a)((b)(c))' is acceptable but
`((b)(c))(a)' is not.
Superfluous Note
----------- ----
The following is some background information not needed
to solve this problem.
The edges in a Huffman Coding Tree can be labelled with
0 and 1, where 0 is placed on edges to left children,
and 1 on edges to right children, as follows
.
0/ \1
a .
0/ \1
b c
The Huffman code of a letter is just the digits on the
path from the root to the letter. In this case:
a 0 b 10 c 11
The set of Huffman codes for all the letters in a
Huffman Coding Tree is a `prefix code', meaning that no
code in the set is the prefix of any other code in the
set. This `0' is not a prefix of `10' or `11'. As a
consequence, the letters in the original string can be
replaced by their Huffman codes:
abaacb 010001110
and the resulting bit string can be unambiguously de-
coded into the original letter string.

Input
-----
One line for each string. The characters of the line
are just the characters of the string. The string will
consist only of lower case letters, and will not be
longer than 80 letters.
The input will end with an end of file.

Output
------
One line for each string, beginning with `String #:' and
ending with the name of a Huffman Coding Tree for the
string. This tree must obey the rule we gave above to
reduce ambiguity. If there is more than one tree obey-
ing this rule, output only one tree.
Sample Input
------ -----
abaacb
abbcccddddeeeeeffffffggggggghhhhhhhh

Sample Output
------ ------
String 1: (a)((b)(c))
String 2: (((((a)(b))(c))(f))((d)(e)))((g)(h))

File: huffman.txt
Author: Bob Walton <walton@deas.harvard.edu>
Date: Thu Oct 11 13:14:29 EDT 2001
The authors have placed this file in the public domain;
they make no warranty and accept no liability for this
file.
RCS Info (may not be true date or author):
$Author: walton $
$Date: 2001/10/11 17:18:49 $
$RCSfile: huffman.txt,v $
$Revision: 1.3 $