add lzw

algs4/binarystdout.py

@@ -38,6 +38,11 @@ def write_bit(cls, bit):
         if cls.n == 8:
             cls.clear_buffer()
 
+    @classmethod
+    def write_str(cls, s):
+        for i in range(len(s)):
+            cls.write_byte(ord(s[i]))
+
     @classmethod
     def clear_buffer(cls):
         if cls.n == 0:

algs4/lzw.py

@@ -0,0 +1,76 @@
+"""
+ *  Execution:    python lzw.py - < input.txt   (compress)
+ *  Execution:    python lzw.py + < input.txt   (expand)
+ *  Data files:   https://algs4.cs.princeton.edu/55compression/abraLZW.txt
+ *                https://algs4.cs.princeton.edu/55compression/ababLZW.txt
+ *
+ *  Compress or expand binary input from standard input using LZW.
+ *
+ *  WARNING: STARTING WITH ORACLE JAVA 6, UPDATE 7 the SUBSTRING
+ *  METHOD TAKES TIME AND SPACE LINEAR IN THE SIZE OF THE EXTRACTED
+ *  SUBSTRING (INSTEAD OF CONSTANT SPACE AND TIME AS IN EARLIER
+ *  IMPLEMENTATIONS).
+ *
+ *  See <a href = "http://java-performance.info/changes-to-string-java-1-7-0_06/">this article</a>
+ *  for more details.
+ *
+"""
+
+from algs4.binarystdin import BinaryStdin
+from algs4.binarystdout import BinaryStdout
+from algs4.tst import TST
+
+
+class LZW:
+    R = 256
+    L = 4096
+    W = 12
+
+    @classmethod
+    def compress(cls):
+        input = BinaryStdin.read_str()
+        st = TST()
+        for i in range(cls.R):
+            st.put(chr(i), i)
+        code = cls.R+1
+        while len(input) > 0:
+            s = st.long_prefix_of(input)
+            BinaryStdout.write_bits(st.get(s), cls.W)
+            t = len(s)
+            if t < len(input) and code < cls.L:
+                st.put(input[:t+1], code)
+                code += 1
+            input = input[t:]
+        BinaryStdout.write_bits(cls.R, cls.W)
+        BinaryStdout.close()
+
+    @classmethod
+    def expand(cls):
+        st = ["" for _ in range(cls.L)]
+        for i in range(cls.R):
+            st[i] = chr(i)
+        st[i] = " "
+        i += 1
+        codeword = BinaryStdin.read_int_r(cls.W)
+        val = st[codeword]
+        while True:
+            BinaryStdout.write_str(val)
+            codeword = BinaryStdin.read_int_r(cls.W)
+            if codeword == cls.R:
+                break
+            s = st[codeword]
+            if i == codeword:
+                s = val + val[0]
+            if i < cls.L:
+                st[i] = val + s[0]
+                i += 1
+            val = s
+        BinaryStdout.close()
+
+
+if __name__ == '__main__':
+    import sys
+    if sys.argv[1] == "-":
+        LZW.compress()
+    else:
+        LZW.expand()