Mercurial > p > roundup > code

annotate test/test_token.py @ 8299:43899d99fc4d

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
refactor(ruff): multiple changes to clear ruff issues Fix a couple of missing returns of booleans for security checks. Turns an implicit return None into an explicit return False. Fix loop index variable being reassigned inside loop by renaming index variable. 2 instances. Consolidate 2 isinstance calls to 1 with tuple class argument. Replace dict(list comprehension) with dict conprehension. Variable renames. Removal of unused variable. Whitespace fixes. sort imports
author John Rouillard <rouilj@ieee.org>
date Wed, 22 Jan 2025 10:10:39 -0500
parents 9a74dfeb8620
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
470
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
1 #
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
2 # Copyright (c) 2001 Richard Jones
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
3 # This module is free software, and you may redistribute it and/or modify
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
4 # under the same terms as Python, so long as this copyright message and
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
5 # disclaimer are retained in their original form.
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
6 #
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
7 # This module is distributed in the hope that it will be useful,
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
8 # but WITHOUT ANY WARRANTY; without even the implied warranty of
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
9 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
10
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
11 import unittest, time
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
12
7181
6971c9249c6d Fix missed roundup.token import.
John Rouillard <rouilj@ieee.org>
parents: 5037
diff changeset
13 from roundup.token_r import token_split
470
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
14
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
15 class TokenTestCase(unittest.TestCase):
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
16 def testValid(self):
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
17 l = token_split('hello world')
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
18 self.assertEqual(l, ['hello', 'world'])
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
19
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
20 def testIgnoreExtraSpace(self):
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
21 l = token_split('hello world ')
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
22 self.assertEqual(l, ['hello', 'world'])
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
23
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
24 def testQuoting(self):
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
25 l = token_split('"hello world"')
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
26 self.assertEqual(l, ['hello world'])
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
27 l = token_split("'hello world'")
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
28 self.assertEqual(l, ['hello world'])
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
29
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
30 def testEmbedQuote(self):
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
31 l = token_split(r'Roch\'e Compaan')
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
32 self.assertEqual(l, ["Roch'e", "Compaan"])
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
33 l = token_split('address="1 2 3"')
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
34 self.assertEqual(l, ['address=1 2 3'])
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
35
7859
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
36 def testEmbedEscapeQuote(self):
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
37 l = token_split(r'"Roch\'e Compaan"')
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
38 self.assertEqual(l, ["Roch'e Compaan"])
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
39
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
40 l = token_split(r'"Roch\"e Compaan"')
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
41 self.assertEqual(l, ['Roch"e Compaan'])
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
42
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
43 l = token_split(r'sql "COLLATE = \"utf8mb4_unicode_ci\";"')
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
44 self.assertEqual(l, ["sql", 'COLLATE = "utf8mb4_unicode_ci";'])
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
45
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
46 l = token_split(r'''sql 'COLLATE = "utf8mb4_unicode_ci";' ''')
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
47 self.assertEqual(l, ["sql", 'COLLATE = "utf8mb4_unicode_ci";'])
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
48
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
49 l = token_split(r'''sql 'COLLATE = \"utf8mb4_unicode_ci\";' ''')
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
50 self.assertEqual(l, ["sql", 'COLLATE = "utf8mb4_unicode_ci";'])
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
51
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
52 l = token_split(r'''sql 'COLLATE = \'utf8mb4_unicode_ci\';' ''')
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
53 self.assertEqual(l, ["sql", "COLLATE = 'utf8mb4_unicode_ci';"])
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
54
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
55 l = token_split(r'''sql 'new\nline\rneed \ttab' ''')
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
56 self.assertEqual(l, ["sql", "new\nline\rneed \ttab"])
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
57
470
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
58 def testEscaping(self):
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
59 l = token_split('"Roch\'e" Compaan')
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
60 self.assertEqual(l, ["Roch'e", "Compaan"])
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
61 l = token_split(r'hello\ world')
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
62 self.assertEqual(l, ['hello world'])
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
63 l = token_split(r'\\')
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
64 self.assertEqual(l, ['\\'])
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
65 l = token_split(r'\n')
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
66 self.assertEqual(l, ['\n'])
7859
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
67 l = token_split(r'\r')
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
68 self.assertEqual(l, ['\r'])
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
69 l = token_split(r'\t')
9a74dfeb8620 feat: can use escaped tokens inside quotes including quotes.
John Rouillard <rouilj@ieee.org>
parents: 7181
diff changeset
70 self.assertEqual(l, ['\t'])
470
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
71
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
72 def testBadQuote(self):
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
73 self.assertRaises(ValueError, token_split, '"hello world')
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
74 self.assertRaises(ValueError, token_split, "Roch'e Compaan")
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
75
9f7320624bc2 Added better tokenising to roundup-admin - handles spaces and stuff.
Richard Jones <richard@users.sourceforge.net>
parents:
diff changeset
76 # vim: set filetype=python ts=4 sw=4 et si
Roundup Issue Tracker: http://roundup-tracker.org/