Fix/Improve handling of codeword issues

dahuffman/huffmancodec.py

@@ -1,5 +1,6 @@
 import collections
 import itertools
+import math
 from io import IOBase
 import sys
 from heapq import heappush, heappop, heapify
@@ -67,20 +68,30 @@ def ensure_dir(path: Union[str, Path]) -> Path:
 
 class CodeTable:
     """
-    Code table: mapping a symbol to codes (and vice versa).
+    Code table: mapping a symbol to codewords (and vice versa).
 
     The symbols are the things you want to encode, usually characters in a string
     or byte sequence, but it can be anything hashable.
-    The codes are the corresponding bit sequences, represented as a tuple (bits, value)
+    The codewords are the corresponding bit sequences, represented as a tuple (bits, value)
     where `bits` is the number of bits and `value` the integer interpretation of these bits.
     """
 
+    # TODO: use something like namedtuple or class with slots for codewords instead of tuples?
+
     def __init__(self, symbol_code_map: dict):
         self._symbol_map = {}
         self._code_map = {}
         for symbol, (bits, value) in symbol_code_map.items():
-            assert isinstance(bits, int) and bits >= 1, f"Invalid bit count {bits}"
-            assert isinstance(value, int) and value >= 0, f"Invalid code value {value}"
+            if not (
+                isinstance(bits, int)
+                and bits >= 1
+                and isinstance(value, int)
+                and value >= 0
+                and math.log2(max(value, 1)) < bits
+            ):
+                raise ValueError(
+                    "Invalid code: {b} bits, value {v}".format(b=bits, v=value)
+                )
             self._symbol_map[symbol] = (bits, value)
             self._code_map[(bits, value)] = symbol
         # TODO check if code table is actually a prefix code

tests/test_dahuffman.py

@@ -1,18 +1,50 @@
 # coding=utf-8
 import io
 import re
+import textwrap
 from io import StringIO
 from pathlib import Path
 
 import pytest
 
 from dahuffman import HuffmanCodec
-from dahuffman.huffmancodec import PrefixCodec, _EOF
+from dahuffman.huffmancodec import PrefixCodec, _EOF, CodeTable
 
 
 # TODO test streaming
 
 
+class TestCodeTable:
+    def test_basic(self):
+        table = CodeTable({"a": (1, 0), "b": (1, 1)})
+        assert table.get_code("a") == (1, 0)
+        assert table.get_code("b") == (1, 1)
+        assert table.get_symbol(1, 0) == "a"
+        assert table.get_symbol(1, 1) == "b"
+        assert len(table) == 2
+
+    @pytest.mark.parametrize("codes", [
+        {"a": (0, 0), "b": (1, 1)},
+        {"a": (1, 0), "b": (1, -1)},
+        {"a": (1, 2), "b": (1, 1)},
+    ])
+    def test_invalid(self, codes):
+        with pytest.raises(ValueError):
+            CodeTable(codes)
+
+    def test_print(self):
+        table = CodeTable({"a": (2, 0), "b": (3, 7)})
+        out = StringIO()
+        table.print(out)
+        assert out.getvalue() == textwrap.dedent(
+            """\
+            Bits Code Value Symbol
+               2 00       0 'a'
+               3 111      7 'b'
+            """
+        )
+
+
 def test_prefix_codec():
     code_table = {"A": (2, 0), "B": (2, 1), _EOF: (2, 3)}
     codec = PrefixCodec(code_table)