Rework is/to ASCII and implement printable ASCII is/to functions.

Add Utils tests.
2024-11-22 10:31:41 +01:00 · 2016-11-02 18:23:55 +02:00 · 2016-11-02 18:23:55 +02:00 · 5bb0ae0ceb
commit 5bb0ae0ceb
parent 11b6155c64
2 changed files with 269 additions and 13 deletions
--- a/ssh-audit.py
+++ b/ssh-audit.py
@ -1914,28 +1914,58 @@ class Utils(object):
 		if isinstance(v, str):
 			return v
 		elif isinstance(v, text_type):
-			return v.encode(enc)
+			return v.encode(enc)  # PY2 only
 		elif isinstance(v, binary_type):
-			return v.decode(enc)
+			return v.decode(enc)  # PY3 only
 		raise cls._type_err(v, 'native text')
 	
+	@classmethod
+	def _is_ascii(cls, v, char_filter=lambda x: x <= 127):
+		# type: (Union[text_type, str], Callable[[int], bool]) -> bool
+		r = False
+		if isinstance(v, (text_type, str)):
+			for c in v:
+				i = cls.ctoi(c)
+				if not char_filter(i):
+					return r
+			r = True
+		return r
+	
+	@classmethod
+	def _to_ascii(cls, v, char_filter=lambda x: x <= 127, errors='replace'):
+		# type: (Union[text_type, str], Callable[[int], bool], str) -> str
+		if isinstance(v, (text_type, str)):
+			r = bytearray()
+			for c in v:
+				i = cls.ctoi(c)
+				if char_filter(i):
+					r.append(i)
+				else:
+					if errors == 'ignore':
+						continue
+					r.append(63)
+			return cls.to_ntext(r.decode('ascii'))
+		raise cls._type_err(v, 'ascii')
+	
 	@classmethod
 	def is_ascii(cls, v):
 		# type: (Union[text_type, str]) -> bool
-		try:
-			if isinstance(v, (text_type, str)):
-				v.encode('ascii')
-				return True
-		except UnicodeEncodeError:
-			pass
-		return False
+		return cls._is_ascii(v)
 	
 	@classmethod
 	def to_ascii(cls, v, errors='replace'):
 		# type: (Union[text_type, str], str) -> str
-		if isinstance(v, (text_type, str)):
-			return cls.to_ntext(v.encode('ascii', errors))
-		raise cls._type_err(v, 'ascii')
+		return cls._to_ascii(v, errors=errors)
+	
+	@classmethod
+	def is_print_ascii(cls, v):
+		# type: (Union[text_type, str]) -> bool
+		return cls._is_ascii(v, lambda x: x >= 32 and x <= 126)
+	
+	@classmethod
+	def to_print_ascii(cls, v, errors='replace'):
+		# type: (Union[text_type, str], str) -> str
+		return cls._to_ascii(v, lambda x: x >= 32 and x <= 126, errors)
 	
 	@classmethod
 	def unique_seq(cls, seq):
@ -1952,6 +1982,14 @@ class Utils(object):
 		else:
 			return [x for x in seq if x not in seen and not _seen_add(x)]
 	
+	@classmethod
+	def ctoi(cls, c):
+		# type: (Union[text_type, str, int]) -> int
+		if isinstance(c, (text_type, str)):
+			return ord(c[0])
+		else:
+			return c
+	
 	@staticmethod
 	def parse_int(v):
 		# type: (Any) -> int
--- a/test/test_utils.py
+++ b/test/test_utils.py
@ -0,0 +1,218 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+import sys
+import pytest
+
+
+# pylint: disable=attribute-defined-outside-init
+class TestUtils(object):
+	@pytest.fixture(autouse=True)
+	def init(self, ssh_audit):
+		self.utils = ssh_audit.Utils
+		self.PY3 = sys.version_info >= (3,)
+	
+	def test_to_bytes_py2(self):
+		if self.PY3:
+			return
+		# binary_type (native str, bytes as str)
+		assert self.utils.to_bytes('fran\xc3\xa7ais') == 'fran\xc3\xa7ais'
+		assert self.utils.to_bytes(b'fran\xc3\xa7ais') == 'fran\xc3\xa7ais'
+		# text_type (unicode)
+		assert self.utils.to_bytes(u'fran\xe7ais') == 'fran\xc3\xa7ais'
+		# other
+		with pytest.raises(TypeError):
+			self.utils.to_bytes(123)
+	
+	def test_to_bytes_py3(self):
+		if not self.PY3:
+			return
+		# binary_type (bytes)
+		assert self.utils.to_bytes(b'fran\xc3\xa7ais') == b'fran\xc3\xa7ais'
+		# text_type (native str as unicode, unicode)
+		assert self.utils.to_bytes('fran\xe7ais') == b'fran\xc3\xa7ais'
+		assert self.utils.to_bytes(u'fran\xe7ais') == b'fran\xc3\xa7ais'
+		# other
+		with pytest.raises(TypeError):
+			self.utils.to_bytes(123)
+	
+	def test_to_utext_py2(self):
+		if self.PY3:
+			return
+		# binary_type (native str, bytes as str)
+		assert self.utils.to_utext('fran\xc3\xa7ais') == u'fran\xe7ais'
+		assert self.utils.to_utext(b'fran\xc3\xa7ais') == u'fran\xe7ais'
+		# text_type (unicode)
+		assert self.utils.to_utext(u'fran\xe7ais') == u'fran\xe7ais'
+		# other
+		with pytest.raises(TypeError):
+			self.utils.to_utext(123)
+	
+	def test_to_utext_py3(self):
+		if not self.PY3:
+			return
+		# binary_type (bytes)
+		assert self.utils.to_utext(b'fran\xc3\xa7ais') == u'fran\xe7ais'
+		# text_type (native str as unicode, unicode)
+		assert self.utils.to_utext('fran\xe7ais') == 'fran\xe7ais'
+		assert self.utils.to_utext(u'fran\xe7ais') == u'fran\xe7ais'
+		# other
+		with pytest.raises(TypeError):
+			self.utils.to_utext(123)
+	
+	def test_to_ntext_py2(self):
+		if self.PY3:
+			return
+		# str (native str, bytes as str)
+		assert self.utils.to_ntext('fran\xc3\xa7ais') == 'fran\xc3\xa7ais'
+		assert self.utils.to_ntext(b'fran\xc3\xa7ais') == 'fran\xc3\xa7ais'
+		# text_type (unicode)
+		assert self.utils.to_ntext(u'fran\xe7ais') == 'fran\xc3\xa7ais'
+		# other
+		with pytest.raises(TypeError):
+			self.utils.to_ntext(123)
+	
+	def test_to_ntext_py3(self):
+		if not self.PY3:
+			return
+		# str (native str)
+		assert self.utils.to_ntext('fran\xc3\xa7ais') == 'fran\xc3\xa7ais'
+		assert self.utils.to_ntext(u'fran\xe7ais') == 'fran\xe7ais'
+		# binary_type (bytes)
+		assert self.utils.to_utext(b'fran\xc3\xa7ais') == 'fran\xe7ais'
+		# other
+		with pytest.raises(TypeError):
+			self.utils.to_ntext(123)
+	
+	def test_is_ascii_py2(self):
+		if self.PY3:
+			return
+		# text_type (unicode)
+		assert self.utils.is_ascii(u'francais') is True
+		assert self.utils.is_ascii(u'fran\xe7ais') is False
+		# str
+		assert self.utils.is_ascii('francais') is True
+		assert self.utils.is_ascii('fran\xc3\xa7ais') is False
+		# other
+		assert self.utils.is_ascii(123) is False
+	
+	def test_is_ascii_py3(self):
+		if not self.PY3:
+			return
+		# text_type (str)
+		assert self.utils.is_ascii('francais') is True
+		assert self.utils.is_ascii(u'francais') is True
+		assert self.utils.is_ascii('fran\xe7ais') is False
+		assert self.utils.is_ascii(u'fran\xe7ais') is False
+		# other
+		assert self.utils.is_ascii(123) is False
+	
+	def test_to_ascii_py2(self):
+		if self.PY3:
+			return
+		# text_type (unicode)
+		assert self.utils.to_ascii(u'francais') == 'francais'
+		assert self.utils.to_ascii(u'fran\xe7ais') == 'fran?ais'
+		assert self.utils.to_ascii(u'fran\xe7ais', 'ignore') == 'franais'
+		# str
+		assert self.utils.to_ascii('francais') == 'francais'
+		assert self.utils.to_ascii('fran\xc3\xa7ais') == 'fran??ais'
+		assert self.utils.to_ascii('fran\xc3\xa7ais', 'ignore') == 'franais'
+		with pytest.raises(TypeError):
+			self.utils.to_ascii(123)
+	
+	def test_to_ascii_py3(self):
+		if not self.PY3:
+			return
+		# text_type (str)
+		assert self.utils.to_ascii('francais') == 'francais'
+		assert self.utils.to_ascii(u'francais') == 'francais'
+		assert self.utils.to_ascii('fran\xe7ais') == 'fran?ais'
+		assert self.utils.to_ascii('fran\xe7ais', 'ignore') == 'franais'
+		assert self.utils.to_ascii(u'fran\xe7ais') == 'fran?ais'
+		assert self.utils.to_ascii(u'fran\xe7ais', 'ignore') == 'franais'
+		with pytest.raises(TypeError):
+			self.utils.to_ascii(123)
+	
+	def test_is_print_ascii_py2(self):
+		if self.PY3:
+			return
+		# text_type (unicode)
+		assert self.utils.is_print_ascii(u'francais') is True
+		assert self.utils.is_print_ascii(u'francais\n') is False
+		assert self.utils.is_print_ascii(u'fran\xe7ais') is False
+		assert self.utils.is_print_ascii(u'fran\xe7ais\n') is False
+		# str
+		assert self.utils.is_print_ascii('francais') is True
+		assert self.utils.is_print_ascii('francais\n') is False
+		assert self.utils.is_print_ascii('fran\xc3\xa7ais') is False
+		# other
+		assert self.utils.is_print_ascii(123) is False
+	
+	def test_is_print_ascii_py3(self):
+		if not self.PY3:
+			return
+		# text_type (str)
+		assert self.utils.is_print_ascii('francais') is True
+		assert self.utils.is_print_ascii('francais\n') is False
+		assert self.utils.is_print_ascii(u'francais') is True
+		assert self.utils.is_print_ascii(u'francais\n') is False
+		assert self.utils.is_print_ascii('fran\xe7ais') is False
+		assert self.utils.is_print_ascii(u'fran\xe7ais') is False
+		# other
+		assert self.utils.is_print_ascii(123) is False
+	
+	def test_to_print_ascii_py2(self):
+		if self.PY3:
+			return
+		# text_type (unicode)
+		assert self.utils.to_print_ascii(u'francais') == 'francais'
+		assert self.utils.to_print_ascii(u'francais\n') == 'francais?'
+		assert self.utils.to_print_ascii(u'fran\xe7ais') == 'fran?ais'
+		assert self.utils.to_print_ascii(u'fran\xe7ais\n') == 'fran?ais?'
+		assert self.utils.to_print_ascii(u'fran\xe7ais', 'ignore') == 'franais'
+		assert self.utils.to_print_ascii(u'fran\xe7ais\n', 'ignore') == 'franais'
+		# str
+		assert self.utils.to_print_ascii('francais') == 'francais'
+		assert self.utils.to_print_ascii('francais\n') == 'francais?'
+		assert self.utils.to_print_ascii('fran\xc3\xa7ais') == 'fran??ais'
+		assert self.utils.to_print_ascii('fran\xc3\xa7ais\n') == 'fran??ais?'
+		assert self.utils.to_print_ascii('fran\xc3\xa7ais', 'ignore') == 'franais'
+		assert self.utils.to_print_ascii('fran\xc3\xa7ais\n', 'ignore') == 'franais'
+		with pytest.raises(TypeError):
+			self.utils.to_print_ascii(123)
+	
+	def test_to_print_ascii_py3(self):
+		if not self.PY3:
+			return
+		# text_type (str)
+		assert self.utils.to_print_ascii('francais') == 'francais'
+		assert self.utils.to_print_ascii('francais\n') == 'francais?'
+		assert self.utils.to_print_ascii(u'francais') == 'francais'
+		assert self.utils.to_print_ascii(u'francais\n') == 'francais?'
+		assert self.utils.to_print_ascii('fran\xe7ais') == 'fran?ais'
+		assert self.utils.to_print_ascii('fran\xe7ais\n') == 'fran?ais?'
+		assert self.utils.to_print_ascii('fran\xe7ais', 'ignore') == 'franais'
+		assert self.utils.to_print_ascii('fran\xe7ais\n', 'ignore') == 'franais'
+		assert self.utils.to_print_ascii(u'fran\xe7ais') == 'fran?ais'
+		assert self.utils.to_print_ascii(u'fran\xe7ais\n') == 'fran?ais?'
+		assert self.utils.to_print_ascii(u'fran\xe7ais', 'ignore') == 'franais'
+		assert self.utils.to_print_ascii(u'fran\xe7ais\n', 'ignore') == 'franais'
+		with pytest.raises(TypeError):
+			self.utils.to_print_ascii(123)
+	
+	def test_ctoi(self):
+		assert self.utils.ctoi(123) == 123
+		assert self.utils.ctoi('ABC') == 65
+	
+	def test_parse_int(self):
+		assert self.utils.parse_int(123) == 123
+		assert self.utils.parse_int('123') == 123
+		assert self.utils.parse_int(-123) == -123
+		assert self.utils.parse_int('-123') == -123
+		assert self.utils.parse_int('abc') == 0
+	
+	def test_unique_seq(self):
+		assert self.utils.unique_seq((1, 2, 2, 3, 3, 3)) == (1, 2, 3)
+		assert self.utils.unique_seq((3, 3, 3, 2, 2, 1)) == (3, 2, 1)
+		assert self.utils.unique_seq([1, 2, 2, 3, 3, 3]) == [1, 2, 3]
+		assert self.utils.unique_seq([3, 3, 3, 2, 2, 1]) == [3, 2, 1]