Merge pull request #35 from saxbophone/josh/25-invalid-input-error-handling

Adds exception that is raised when input contains invalid symbols

Author: saxbophone

basest/core/decode.py

@@ -4,6 +4,7 @@
     absolute_import, division, print_function, unicode_literals
 )
 
+from ..exceptions import InvalidInputLengthError
 from .encode import encode_raw
 from .utils import ints_to_symbols, symbols_to_ints, validate_symbol_tables
 
@@ -17,6 +18,12 @@ def decode_raw(input_base, output_base, input_ratio, output_ratio, input_data):
     symbol (so interpretted padding integer for decoding base64 would be 64, as
     base64 input would be in the range 0-63).
     """
+    # raise an exception early if padding was truncated
+    if len(input_data) % input_ratio != 0:
+        raise InvalidInputLengthError(
+            'Decoding requires input length to be an exact multiple of the '
+            'input ratio, or for padding to be used to ensure this.'
+        )
     # create a 'workon' copy of the input data so we don't end up changing it
     input_workon = list(input_data)
     # count number of padding symbols

basest/core/utils.py

@@ -4,7 +4,7 @@
     absolute_import, division, print_function, unicode_literals
 )
 
-from ..exceptions import InvalidSymbolTableError
+from ..exceptions import InvalidInputError, InvalidSymbolTableError
 
 
 def ints_to_symbols(ints, symbol_table):
@@ -19,8 +19,19 @@ def symbols_to_ints(symbols, symbol_table):
     """
     Given an iterable of symbols a list of symbols to convert them from,
     convert them to an iterable of ints and return this.
+
+    Raises InvalidInputError if a symbol that is not in the symbol table is
+    encountered.
     """
-    return [symbol_table.index(s) for s in symbols]
+    try:
+        return [symbol_table.index(s) for s in symbols]
+    except ValueError:
+        """
+        This error is raised when index() doesn't find the given symbol in the
+        symbol table. We're catching it because we want to raise our own
+        exception class for this instead, InvalidInputError.
+        """
+        raise InvalidInputError('Encountered symbol not found in symbol table')
 
 
 def _symbol_table_is_unique(symbol_table, padding_symbol=None):

basest/exceptions.py

@@ -2,6 +2,22 @@
 # -*- coding: utf-8 -*-
 
 
+class InvalidSymbolTableError(ValueError):
+    """
+    This exception is raised when the symbol table and/or padding symbol
+    supplied to an encoding/decoding operation are invalid.
+    """
+    pass
+
+
+class InvalidInputError(ValueError):
+    """
+    This exception is raised when an encoding or decoding function receives
+    input data containing symbols which are not in the relevant symbol table.
+    """
+    pass
+
+
 class ImproperUsageError(ValueError):
     """
     This exception is raised when an attempt is made to encode data using a
@@ -13,9 +29,14 @@ class ImproperUsageError(ValueError):
     pass
 
 
-class InvalidSymbolTableError(ValueError):
+class InvalidInputLengthError(ValueError):
     """
-    This exception is raised when the symbol table and/or padding symbol
-    supplied to an encoding/decoding operation are invalid.
+    This exception is raised when an attempt is made to decode data which is
+    not the correct length (e.g. where the length is not an exact multiple of
+    the input ratio).
+
+    This exception is only valid when performing a decoding operation, as data
+    to be encoded is permitted to be shorter than the input ratio as long as
+    the output base is smaller than the input base.
     """
     pass

tests/core/test_encode_decode.py

@@ -9,7 +9,7 @@
 from ddt import data, ddt, unpack
 
 from basest.core import decode, encode
-from basest.exceptions import InvalidSymbolTableError
+from basest.exceptions import InvalidInputError, InvalidSymbolTableError
 
 
 base64_alphabet = [
@@ -187,6 +187,29 @@ def test_encode_rejects_none_used_in_symbol_tables_and_padding(
                 []
             )
 
+    @data(
+        ([0, 1, 2, 3], [0, 1, 4])
+    )
+    @unpack
+    def test_encode_rejects_symbols_not_found_in_symbol_table(
+        self,
+        input_symbols,
+        input_symbol_table
+    ):
+        """
+        When the encode() function is called with input data that contains
+        symbols which are not found in the input symbol table,
+        InvalidInputError should be raised.
+        """
+        with self.assertRaises(InvalidInputError):
+            encode(
+                4, input_symbol_table,
+                8, list(range(8)),
+                'P',
+                2, 3,
+                input_symbols
+            )
+
     @data(
         # Base-64, using most common alphabet - no padding
         (
@@ -341,6 +364,29 @@ def test_decode_rejects_none_used_in_symbol_tables_and_padding(
                 []
             )
 
+    @data(
+        ([0, 1, 2, 3], [0, 1, 4])
+    )
+    @unpack
+    def test_decode_rejects_symbols_not_found_in_symbol_table(
+        self,
+        input_symbols,
+        input_symbol_table
+    ):
+        """
+        When the decode() function is called with input data that contains
+        symbols which are not found in the input symbol table,
+        InvalidInputError should be raised.
+        """
+        with self.assertRaises(InvalidInputError):
+            decode(
+                4, input_symbol_table,
+                'P',
+                8, list(range(8)),
+                3, 2,
+                input_symbols
+            )
+
     @data(
         # Base-64, using most common alphabet with no padding needed
         (

tests/core/test_encode_decode_raw.py

@@ -9,7 +9,7 @@
 from ddt import data, ddt, unpack
 
 from basest.core import decode_raw, encode_raw
-from basest.exceptions import ImproperUsageError
+from basest.exceptions import ImproperUsageError, InvalidInputLengthError
 
 
 @ddt
@@ -136,6 +136,32 @@ def test_decode_raw_invalid_inputs(self, data_type):
                 input_data=data_type()
             )
 
+    @data(
+        # Base-85 - padding has been truncated, which means it is too short!
+        (
+            85, 256, 5, 4,
+            [13, 74, 17, 83, 81, 12, 45]
+        )
+    )
+    @unpack
+    def test_decode_raw_rejects_input_of_incorrect_length(
+        self,
+        input_base, output_base,
+        input_ratio, output_ratio,
+        input_data
+    ):
+        """
+        When decode_raw() is called with input data which is not of a length
+        exactly divisible by the input ratio, InvalidInputLengthError should be
+        raised.
+        """
+        with self.assertRaises(InvalidInputLengthError):
+            decode_raw(
+                input_base=input_base, output_base=output_base,
+                input_ratio=input_ratio, output_ratio=output_ratio,
+                input_data=input_data
+            )
+
     @data(
         # Base-85 - no padding required
         (256, 85, 4, 5, [99, 97, 98, 98, 97, 103, 101, 115]),