1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /********************************************************************
4 * COPYRIGHT:
5 * Copyright (c) 1997-2016, International Business Machines Corporation and
6 * others. All Rights Reserved.
7 ********************************************************************/
8
9 #include "unicode/utypes.h"
10
11 #if !UCONFIG_NO_COLLATION
12
13 #include "unicode/coll.h"
14 #include "unicode/tblcoll.h"
15 #include "unicode/unistr.h"
16 #include "unicode/sortkey.h"
17 #include "encoll.h"
18 #include "cmemory.h"
19
CollationEnglishTest()20 CollationEnglishTest::CollationEnglishTest()
21 : myCollation(nullptr)
22 {
23 UErrorCode status = U_ZERO_ERROR;
24 myCollation = Collator::createInstance(Locale::getEnglish(), status);
25 }
26
~CollationEnglishTest()27 CollationEnglishTest::~CollationEnglishTest()
28 {
29 delete myCollation;
30 }
31
32 static const char16_t testSourceCases[][CollationEnglishTest::MAX_TOKEN_LEN] = {
33 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, 0},
34 {static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x006B) /* 'k' */, static_cast<char16_t>(0x002D) /* '-' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0069) /* 'i' */, static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
35 {static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x006B) /* 'k' */, static_cast<char16_t>(0x0020) /* ' ' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0069) /* 'i' */, static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
36 {static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x006B) /* 'k' */, static_cast<char16_t>(0x002D) /* '-' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0069) /* 'i' */, static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
37 {static_cast<char16_t>(0x0048) /* 'H' */, static_cast<char16_t>(0x0065) /* 'e' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x006F) /* 'o' */, 0},
38 {static_cast<char16_t>(0x0041) /* 'A' */, static_cast<char16_t>(0x0042) /* 'B' */, static_cast<char16_t>(0x0043) /* 'C' */, 0},
39 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
40 {static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x006B) /* 'k' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0069) /* 'i' */, static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
41 {static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x006B) /* 'k' */, static_cast<char16_t>(0x002D) /* '-' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0069) /* 'i' */, static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
42 {static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x006B) /* 'k' */, static_cast<char16_t>(0x002D) /* '-' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0069) /* 'i' */, static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
43 {static_cast<char16_t>(0x0070) /* 'p' */, 0x00EA, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0068) /* 'h' */, static_cast<char16_t>(0x0065) /* 'e' */, 0},
44 {static_cast<char16_t>(0x0070) /* 'p' */, 0x00E9, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0068) /* 'h' */, 0x00E9, 0},
45 {0x00C4, static_cast<char16_t>(0x0042) /* 'B' */, 0x0308, static_cast<char16_t>(0x0043) /* 'C' */, 0x0308, 0},
46 {static_cast<char16_t>(0x0061) /* 'a' */, 0x0308, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
47 {static_cast<char16_t>(0x0070) /* 'p' */, 0x00E9, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0068) /* 'h' */, static_cast<char16_t>(0x0065) /* 'e' */, static_cast<char16_t>(0x0072) /* 'r' */, 0},
48 {static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x006F) /* 'o' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0065) /* 'e' */, static_cast<char16_t>(0x0073) /* 's' */, 0},
49 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
50 {static_cast<char16_t>(0x0041) /* 'A' */, 0},
51 {static_cast<char16_t>(0x0041) /* 'A' */, 0},
52 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, 0},
53 {static_cast<char16_t>(0x0074) /* 't' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x006F) /* 'o' */, static_cast<char16_t>(0x006D) /* 'm' */, static_cast<char16_t>(0x0070) /* 'p' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x0065) /* 'e' */, static_cast<char16_t>(0x0070) /* 'p' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0069) /* 'i' */, static_cast<char16_t>(0x006E) /* 'n' */, 0},
54 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, 0},
55 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0023) /* '#' */, static_cast<char16_t>(0x0062) /* 'b' */, 0},
56 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0023) /* '#' */, static_cast<char16_t>(0x0062) /* 'b' */, 0},
57 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
58 {static_cast<char16_t>(0x0041) /* 'A' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0064) /* 'd' */, static_cast<char16_t>(0x0061) /* 'a' */, 0},
59 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0064) /* 'd' */, static_cast<char16_t>(0x0061) /* 'a' */, 0},
60 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0064) /* 'd' */, static_cast<char16_t>(0x0061) /* 'a' */, 0},
61 {0x00E6, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0064) /* 'd' */, static_cast<char16_t>(0x0061) /* 'a' */, 0},
62 {0x00E4, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0064) /* 'd' */, static_cast<char16_t>(0x0061) /* 'a' */, 0},
63 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
64 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
65 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
66 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
67 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
68 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0048) /* 'H' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
69 {static_cast<char16_t>(0x0061) /* 'a' */, 0x0308, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
70 {static_cast<char16_t>(0x0074) /* 't' */, static_cast<char16_t>(0x0068) /* 'h' */, static_cast<char16_t>(0x0069) /* 'i' */, 0x0302, static_cast<char16_t>(0x0073) /* 's' */, 0},
71 {static_cast<char16_t>(0x0070) /* 'p' */, 0x00EA, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0068) /* 'h' */, static_cast<char16_t>(0x0065) /* 'e' */},
72 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
73 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
74 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
75 {static_cast<char16_t>(0x0061) /* 'a' */, 0x00E6, static_cast<char16_t>(0x0063) /* 'c' */, 0},
76 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
77 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
78 {static_cast<char16_t>(0x0061) /* 'a' */, 0x00E6, static_cast<char16_t>(0x0063) /* 'c' */, 0},
79 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
80 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
81 {static_cast<char16_t>(0x0070) /* 'p' */, 0x00E9, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0068) /* 'h' */, 0x00E9, 0} // 49
82 };
83
84 static const char16_t testTargetCases[][CollationEnglishTest::MAX_TOKEN_LEN] = {
85 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
86 {static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x006B) /* 'k' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0069) /* 'i' */, static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
87 {static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x006B) /* 'k' */, static_cast<char16_t>(0x002D) /* '-' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0069) /* 'i' */, static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
88 {static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x006B) /* 'k' */, 0},
89 {static_cast<char16_t>(0x0068) /* 'h' */, static_cast<char16_t>(0x0065) /* 'e' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x006F) /* 'o' */, 0},
90 {static_cast<char16_t>(0x0041) /* 'A' */, static_cast<char16_t>(0x0042) /* 'B' */, static_cast<char16_t>(0x0043) /* 'C' */, 0},
91 {static_cast<char16_t>(0x0041) /* 'A' */, static_cast<char16_t>(0x0042) /* 'B' */, static_cast<char16_t>(0x0043) /* 'C' */, 0},
92 {static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x006B) /* 'k' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0069) /* 'i' */, static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x0064) /* 'd' */, static_cast<char16_t>(0x0073) /* 's' */, 0},
93 {static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x006B) /* 'k' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0069) /* 'i' */, static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x0064) /* 'd' */, static_cast<char16_t>(0x0073) /* 's' */, 0},
94 {static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x006B) /* 'k' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0069) /* 'i' */, static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
95 {static_cast<char16_t>(0x0070) /* 'p' */, 0x00E9, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0068) /* 'h' */, 0x00E9, 0},
96 {static_cast<char16_t>(0x0070) /* 'p' */, 0x00E9, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0068) /* 'h' */, static_cast<char16_t>(0x0065) /* 'e' */, static_cast<char16_t>(0x0072) /* 'r' */, 0},
97 {0x00C4, static_cast<char16_t>(0x0042) /* 'B' */, 0x0308, static_cast<char16_t>(0x0043) /* 'C' */, 0x0308, 0},
98 {static_cast<char16_t>(0x0041) /* 'A' */, 0x0308, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
99 {static_cast<char16_t>(0x0070) /* 'p' */, 0x00E9, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0068) /* 'h' */, static_cast<char16_t>(0x0065) /* 'e' */, 0},
100 {static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x006F) /* 'o' */, 0x0302, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0065) /* 'e' */, 0},
101 {static_cast<char16_t>(0x0041) /* 'A' */, 0x00E1, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
102 {static_cast<char16_t>(0x0041) /* 'A' */, 0x00E1, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
103 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
104 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
105 {static_cast<char16_t>(0x0054) /* 'T' */, static_cast<char16_t>(0x0043) /* 'C' */, static_cast<char16_t>(0x006F) /* 'o' */, static_cast<char16_t>(0x006D) /* 'm' */, static_cast<char16_t>(0x0070) /* 'p' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0072) /* 'r' */, static_cast<char16_t>(0x0065) /* 'e' */, static_cast<char16_t>(0x0050) /* 'P' */, static_cast<char16_t>(0x006C) /* 'l' */, static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0069) /* 'i' */, static_cast<char16_t>(0x006E) /* 'n' */, 0},
106 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0042) /* 'B' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
107 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0023) /* '#' */, static_cast<char16_t>(0x0042) /* 'B' */, 0},
108 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0026) /* '&' */, static_cast<char16_t>(0x0062) /* 'b' */, 0},
109 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0023) /* '#' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
110 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0064) /* 'd' */, static_cast<char16_t>(0x0061) /* 'a' */, 0},
111 {0x00C4, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0064) /* 'd' */, static_cast<char16_t>(0x0061) /* 'a' */, 0},
112 {0x00E4, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0064) /* 'd' */, static_cast<char16_t>(0x0061) /* 'a' */, 0},
113 {0x00C4, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0064) /* 'd' */, static_cast<char16_t>(0x0061) /* 'a' */, 0},
114 {0x00C4, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0064) /* 'd' */, static_cast<char16_t>(0x0061) /* 'a' */, 0},
115 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0023) /* '#' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
116 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
117 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x003D) /* '=' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
118 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
119 {0x00E4, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
120 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0043) /* 'C' */, static_cast<char16_t>(0x0048) /* 'H' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
121 {0x00E4, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
122 {static_cast<char16_t>(0x0074) /* 't' */, static_cast<char16_t>(0x0068) /* 'h' */, 0x00EE, static_cast<char16_t>(0x0073) /* 's' */, 0},
123 {static_cast<char16_t>(0x0070) /* 'p' */, 0x00E9, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0068) /* 'h' */, 0x00E9, 0},
124 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0042) /* 'B' */, static_cast<char16_t>(0x0043) /* 'C' */, 0},
125 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
126 {0x00E4, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
127 {static_cast<char16_t>(0x0061) /* 'a' */, 0x00C6, static_cast<char16_t>(0x0063) /* 'c' */, 0},
128 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0042) /* 'B' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
129 {0x00E4, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
130 {static_cast<char16_t>(0x0061) /* 'a' */, 0x00C6, static_cast<char16_t>(0x0063) /* 'c' */, 0},
131 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0042) /* 'B' */, static_cast<char16_t>(0x0064) /* 'd' */, 0},
132 {0x00E4, static_cast<char16_t>(0x0062) /* 'b' */, static_cast<char16_t>(0x0063) /* 'c' */, 0},
133 {static_cast<char16_t>(0x0070) /* 'p' */, 0x00EA, static_cast<char16_t>(0x0063) /* 'c' */, static_cast<char16_t>(0x0068) /* 'h' */, static_cast<char16_t>(0x0065) /* 'e' */, 0} // 49
134 };
135
136 static const Collator::EComparisonResult results[] = {
137 Collator::LESS,
138 Collator::LESS, /*Collator::GREATER,*/
139 Collator::LESS,
140 Collator::GREATER,
141 Collator::GREATER,
142 Collator::EQUAL,
143 Collator::LESS,
144 Collator::LESS,
145 Collator::LESS,
146 Collator::LESS, /*Collator::GREATER,*/ /* 10 */
147 Collator::GREATER,
148 Collator::LESS,
149 Collator::EQUAL,
150 Collator::LESS,
151 Collator::GREATER,
152 Collator::GREATER,
153 Collator::GREATER,
154 Collator::LESS,
155 Collator::LESS,
156 Collator::LESS, /* 20 */
157 Collator::LESS,
158 Collator::LESS,
159 Collator::LESS,
160 Collator::GREATER,
161 Collator::GREATER,
162 Collator::GREATER,
163 /* Test Tertiary > 26 */
164 Collator::LESS,
165 Collator::LESS,
166 Collator::GREATER,
167 Collator::LESS, /* 30 */
168 Collator::GREATER,
169 Collator::EQUAL,
170 Collator::GREATER,
171 Collator::LESS,
172 Collator::LESS,
173 Collator::LESS,
174 /* test identical > 36 */
175 Collator::EQUAL,
176 Collator::EQUAL,
177 /* test primary > 38 */
178 Collator::EQUAL,
179 Collator::EQUAL, /* 40 */
180 Collator::LESS,
181 Collator::EQUAL,
182 Collator::EQUAL,
183 /* test secondary > 43 */
184 Collator::LESS,
185 Collator::LESS,
186 Collator::EQUAL,
187 Collator::LESS,
188 Collator::LESS,
189 Collator::LESS // 49
190 };
191
192 static const char16_t testBugs[][CollationEnglishTest::MAX_TOKEN_LEN] = {
193 {0x61, 0},
194 {0x41, 0},
195 {0x65, 0},
196 {0x45, 0},
197 {0x00e9, 0},
198 {0x00e8, 0},
199 {0x00ea, 0},
200 {0x00eb, 0},
201 {0x65, 0x61, 0},
202 {0x78, 0}
203 };
204
205 // 0x0300 is grave, 0x0301 is acute
206 // the order of elements in this array must be different than the order in CollationFrenchTest
207 static const char16_t testAcute[][CollationEnglishTest::MAX_TOKEN_LEN] = {
208 {0x65, 0x65, 0},
209 {0x65, 0x65, 0x0301, 0},
210 {0x65, 0x65, 0x0301, 0x0300, 0},
211 {0x65, 0x65, 0x0300, 0},
212 {0x65, 0x65, 0x0300, 0x0301, 0},
213 {0x65, 0x0301, 0x65, 0},
214 {0x65, 0x0301, 0x65, 0x0301, 0},
215 {0x65, 0x0301, 0x65, 0x0301, 0x0300, 0},
216 {0x65, 0x0301, 0x65, 0x0300, 0},
217 {0x65, 0x0301, 0x65, 0x0300, 0x0301, 0},
218 {0x65, 0x0301, 0x0300, 0x65, 0},
219 {0x65, 0x0301, 0x0300, 0x65, 0x0301, 0},
220 {0x65, 0x0301, 0x0300, 0x65, 0x0301, 0x0300, 0},
221 {0x65, 0x0301, 0x0300, 0x65, 0x0300, 0},
222 {0x65, 0x0301, 0x0300, 0x65, 0x0300, 0x0301, 0},
223 {0x65, 0x0300, 0x65, 0},
224 {0x65, 0x0300, 0x65, 0x0301, 0},
225 {0x65, 0x0300, 0x65, 0x0301, 0x0300, 0},
226 {0x65, 0x0300, 0x65, 0x0300, 0},
227 {0x65, 0x0300, 0x65, 0x0300, 0x0301, 0},
228 {0x65, 0x0300, 0x0301, 0x65, 0},
229 {0x65, 0x0300, 0x0301, 0x65, 0x0301, 0},
230 {0x65, 0x0300, 0x0301, 0x65, 0x0301, 0x0300, 0},
231 {0x65, 0x0300, 0x0301, 0x65, 0x0300, 0},
232 {0x65, 0x0300, 0x0301, 0x65, 0x0300, 0x0301, 0}
233 };
234
235 static const char16_t testMore[][CollationEnglishTest::MAX_TOKEN_LEN] = {
236 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0065) /* 'e' */, 0},
237 {0x00E6, 0},
238 {0x00C6, 0},
239 {static_cast<char16_t>(0x0061) /* 'a' */, static_cast<char16_t>(0x0066) /* 'f' */, 0},
240 {static_cast<char16_t>(0x006F) /* 'o' */, static_cast<char16_t>(0x0065) /* 'e' */, 0},
241 {0x0153, 0},
242 {0x0152, 0},
243 {static_cast<char16_t>(0x006F) /* 'o' */, static_cast<char16_t>(0x0066) /* 'f' */, 0},
244 };
245
TestTertiary()246 void CollationEnglishTest::TestTertiary(/* char* par */)
247 {
248 int32_t i = 0;
249 myCollation->setStrength(Collator::TERTIARY);
250 for (i = 0; i < 38 ; i++)
251 {
252 doTest(myCollation, testSourceCases[i], testTargetCases[i], results[i]);
253 }
254
255 int32_t j = 0;
256 for (i = 0; i < 10; i++)
257 {
258 for (j = i+1; j < 10; j++)
259 {
260 doTest(myCollation, testBugs[i], testBugs[j], Collator::LESS);
261 }
262 }
263
264 //test more interesting cases
265 Collator::EComparisonResult expected;
266 const int32_t testMoreSize = UPRV_LENGTHOF(testMore);
267 for (i = 0; i < testMoreSize; i++)
268 {
269 for (j = 0; j < testMoreSize; j++)
270 {
271 if (i < j)
272 expected = Collator::LESS;
273 else if (i == j)
274 expected = Collator::EQUAL;
275 else // (i > j)
276 expected = Collator::GREATER;
277 doTest(myCollation, testMore[i], testMore[j], expected );
278 }
279 }
280
281 }
282
TestPrimary()283 void CollationEnglishTest::TestPrimary(/* char* par */)
284 {
285 int32_t i;
286 myCollation->setStrength(Collator::PRIMARY);
287 for (i = 38; i < 43 ; i++)
288 {
289 doTest(myCollation, testSourceCases[i], testTargetCases[i], results[i]);
290 }
291 }
292
TestSecondary()293 void CollationEnglishTest::TestSecondary(/* char* par */)
294 {
295 int32_t i;
296 myCollation->setStrength(Collator::SECONDARY);
297 for (i = 43; i < 49 ; i++)
298 {
299 doTest(myCollation, testSourceCases[i], testTargetCases[i], results[i]);
300 }
301
302 //test acute and grave ordering (compare to french collation)
303 int32_t j;
304 Collator::EComparisonResult expected;
305 const int32_t testAcuteSize = UPRV_LENGTHOF(testAcute);
306 for (i = 0; i < testAcuteSize; i++)
307 {
308 for (j = 0; j < testAcuteSize; j++)
309 {
310 if (i < j)
311 expected = Collator::LESS;
312 else if (i == j)
313 expected = Collator::EQUAL;
314 else // (i > j)
315 expected = Collator::GREATER;
316 doTest(myCollation, testAcute[i], testAcute[j], expected );
317 }
318 }
319 }
320
runIndexedTest(int32_t index,UBool exec,const char * & name,char *)321 void CollationEnglishTest::runIndexedTest( int32_t index, UBool exec, const char* &name, char* /*par*/ )
322 {
323 if (exec) logln("TestSuite CollationEnglishTest: ");
324 if(myCollation) {
325 switch (index) {
326 case 0: name = "TestPrimary"; if (exec) TestPrimary(/* par */); break;
327 case 1: name = "TestSecondary"; if (exec) TestSecondary(/* par */); break;
328 case 2: name = "TestTertiary"; if (exec) TestTertiary(/* par */); break;
329 default: name = ""; break;
330 }
331 } else {
332 dataerrln("Collator couldn't be instantiated!");
333 name = "";
334 }
335 }
336
337 #endif /* #if !UCONFIG_NO_COLLATION */
338