blob: cb973f02b983d324b8d439fada2074e18b6c6f07 [file] [log] [blame]
Alexander Kornienkoece0bec2013-08-07 00:07:07 +00001//===- unittests/Support/LocaleTest.cpp - Locale.h tests ------------------===//
2//
3// The LLVM Compiler Infrastructure
4//
5// This file is distributed under the University of Illinois Open Source
6// License. See LICENSE.TXT for details.
7//
8//===----------------------------------------------------------------------===//
9
10#include "llvm/Support/Locale.h"
11#include "gtest/gtest.h"
12
13namespace llvm {
14namespace sys {
15namespace locale {
16namespace {
17
Alexander Kornienko31a4f1f2013-08-07 00:41:18 +000018// FIXME: WIN32 implementation is incorrect. We should consider using the one
19// from LocaleGeneric.inc for WIN32.
20#ifndef _WIN32
Alexander Kornienkoece0bec2013-08-07 00:07:07 +000021TEST(Locale, columnWidth) {
22 EXPECT_EQ(0, columnWidth(""));
23 EXPECT_EQ(1, columnWidth(" "));
24 EXPECT_EQ(1, columnWidth("a"));
25 EXPECT_EQ(1, columnWidth("~"));
26
27 EXPECT_EQ(6, columnWidth("abcdef"));
28
29 EXPECT_EQ(-1, columnWidth("\x01"));
30 EXPECT_EQ(-1, columnWidth("aaaaaaaaaa\x01"));
31 EXPECT_EQ(-1, columnWidth("\342\200\213")); // 200B ZERO WIDTH SPACE
32
33 EXPECT_EQ(0, columnWidth("\314\200")); // 0300 COMBINING GRAVE ACCENT
34 EXPECT_EQ(1, columnWidth("\340\270\201")); // 0E01 THAI CHARACTER KO KAI
35 EXPECT_EQ(2, columnWidth("\344\270\200")); // CJK UNIFIED IDEOGRAPH-4E00
36
37 EXPECT_EQ(4, columnWidth("\344\270\200\344\270\200"));
38 EXPECT_EQ(3, columnWidth("q\344\270\200"));
39 EXPECT_EQ(3, columnWidth("\314\200\340\270\201\344\270\200"));
40
Alexander Kornienko31a4f1f2013-08-07 00:41:18 +000041 // FIXME: MacOS implementation of columnWidth seems to not handle incorrect
42 // UTF-8 sequences.
43#ifndef __MACOSX__
Alexander Kornienkoece0bec2013-08-07 00:07:07 +000044 // Invalid UTF-8 strings, columnWidth should error out.
45 EXPECT_EQ(-2, columnWidth("\344"));
46 EXPECT_EQ(-2, columnWidth("\344\270"));
47 EXPECT_EQ(-2, columnWidth("\344\270\033"));
48 EXPECT_EQ(-2, columnWidth("\344\270\300"));
49 EXPECT_EQ(-2, columnWidth("\377\366\355"));
50
51 EXPECT_EQ(-2, columnWidth("qwer\344"));
52 EXPECT_EQ(-2, columnWidth("qwer\344\270"));
53 EXPECT_EQ(-2, columnWidth("qwer\344\270\033"));
54 EXPECT_EQ(-2, columnWidth("qwer\344\270\300"));
55 EXPECT_EQ(-2, columnWidth("qwer\377\366\355"));
56
57 // UTF-8 sequences longer than 4 bytes correspond to unallocated Unicode
58 // characters.
59 EXPECT_EQ(-2, columnWidth("\370\200\200\200\200")); // U+200000
60 EXPECT_EQ(-2, columnWidth("\374\200\200\200\200\200")); // U+4000000
Alexander Kornienko31a4f1f2013-08-07 00:41:18 +000061#endif // __MACOSX__
Alexander Kornienkoece0bec2013-08-07 00:07:07 +000062}
63
64TEST(Locale, isPrint) {
65 EXPECT_EQ(false, isPrint(0)); // <control-0000>-<control-001F>
66 EXPECT_EQ(false, isPrint(0x01));
67 EXPECT_EQ(false, isPrint(0x1F));
68 EXPECT_EQ(true, isPrint(' '));
69 EXPECT_EQ(true, isPrint('A'));
70 EXPECT_EQ(true, isPrint('~'));
71 EXPECT_EQ(false, isPrint(0x7F)); // <control-007F>..<control-009F>
72 EXPECT_EQ(false, isPrint(0x90));
73 EXPECT_EQ(false, isPrint(0x9F));
74
75 EXPECT_EQ(true, isPrint(0xAC));
Alexander Kornienko31a4f1f2013-08-07 00:41:18 +000076 // FIXME: Figure out if we want to treat SOFT HYPHEN as printable character.
77#ifndef __MACOSX__
Alexander Kornienkoece0bec2013-08-07 00:07:07 +000078 EXPECT_EQ(false, isPrint(0xAD)); // SOFT HYPHEN
Alexander Kornienko31a4f1f2013-08-07 00:41:18 +000079#endif // __MACOSX__
Alexander Kornienkoece0bec2013-08-07 00:07:07 +000080 EXPECT_EQ(true, isPrint(0xAE));
81
Alexander Kornienko31a4f1f2013-08-07 00:41:18 +000082 // MacOS implementation doesn't think it's printable.
83#ifndef __MACOSX__
Alexander Kornienkoece0bec2013-08-07 00:07:07 +000084 EXPECT_EQ(true, isPrint(0x0377)); // GREEK SMALL LETTER PAMPHYLIAN DIGAMMA
Alexander Kornienko31a4f1f2013-08-07 00:41:18 +000085#endif // __MACOSX__
Alexander Kornienkoece0bec2013-08-07 00:07:07 +000086 EXPECT_EQ(false, isPrint(0x0378)); // <reserved-0378>..<reserved-0379>
87
88 EXPECT_EQ(false, isPrint(0x0600)); // ARABIC NUMBER SIGN
89
90 EXPECT_EQ(false, isPrint(0x1FFFF)); // <reserved-1F774>..<noncharacter-1FFFF>
91 EXPECT_EQ(true, isPrint(0x20000)); // CJK UNIFIED IDEOGRAPH-20000
92
93 EXPECT_EQ(false, isPrint(0x10FFFF)); // noncharacter
94}
95
Alexander Kornienko31a4f1f2013-08-07 00:41:18 +000096#endif // _WIN32
97
Alexander Kornienkoece0bec2013-08-07 00:07:07 +000098} // namespace
99} // namespace locale
100} // namespace sys
101} // namespace llvm