unittests/Support/LocaleTest.cpp

   1 //===- 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
  13 namespace llvm {
  14 namespace sys {
  15 namespace locale {
  16 namespace {
  17
  18 // FIXME: WIN32 implementation is incorrect. We should consider using the one
  19 // from LocaleGeneric.inc for WIN32.
  20 #ifndef _WIN32
  21 TEST(Locale, columnWidth) {
  22   // FIXME: This test fails with MacOSX implementation of columnWidth.
  23 #ifndef __APPLE__
  24   EXPECT_EQ(0, columnWidth(""));
  25   EXPECT_EQ(1, columnWidth(" "));
  26   EXPECT_EQ(1, columnWidth("a"));
  27   EXPECT_EQ(1, columnWidth("~"));
  28
  29   EXPECT_EQ(6, columnWidth("abcdef"));
  30
  31   EXPECT_EQ(-1, columnWidth("\x01"));
  32   EXPECT_EQ(-1, columnWidth("aaaaaaaaaa\x01"));
  33   EXPECT_EQ(-1, columnWidth("\342\200\213")); // 200B ZERO WIDTH SPACE
  34
  35   // 00AD SOFT HYPHEN is displayed on most terminals as a space or a dash. Some
  36   // text editors display it only when a line is broken at it, some use it as a
  37   // line-break hint, but don't display. We choose terminal-oriented
  38   // interpretation.
  39   EXPECT_EQ(1, columnWidth("\302\255"));
  40
  41   EXPECT_EQ(0, columnWidth("\314\200")); // 0300 COMBINING GRAVE ACCENT
  42   EXPECT_EQ(1, columnWidth("\340\270\201")); // 0E01 THAI CHARACTER KO KAI
  43   EXPECT_EQ(2, columnWidth("\344\270\200")); // CJK UNIFIED IDEOGRAPH-4E00
  44
  45   EXPECT_EQ(4, columnWidth("\344\270\200\344\270\200"));
  46   EXPECT_EQ(3, columnWidth("q\344\270\200"));
  47   EXPECT_EQ(3, columnWidth("\314\200\340\270\201\344\270\200"));
  48
  49   // Invalid UTF-8 strings, columnWidth should error out.
  50   EXPECT_EQ(-2, columnWidth("\344"));
  51   EXPECT_EQ(-2, columnWidth("\344\270"));
  52   EXPECT_EQ(-2, columnWidth("\344\270\033"));
  53   EXPECT_EQ(-2, columnWidth("\344\270\300"));
  54   EXPECT_EQ(-2, columnWidth("\377\366\355"));
  55
  56   EXPECT_EQ(-2, columnWidth("qwer\344"));
  57   EXPECT_EQ(-2, columnWidth("qwer\344\270"));
  58   EXPECT_EQ(-2, columnWidth("qwer\344\270\033"));
  59   EXPECT_EQ(-2, columnWidth("qwer\344\270\300"));
  60   EXPECT_EQ(-2, columnWidth("qwer\377\366\355"));
  61
  62   // UTF-8 sequences longer than 4 bytes correspond to unallocated Unicode
  63   // characters.
  64   EXPECT_EQ(-2, columnWidth("\370\200\200\200\200"));     // U+200000
  65   EXPECT_EQ(-2, columnWidth("\374\200\200\200\200\200")); // U+4000000
  66 #endif // __APPLE__
  67 }
  68
  69 TEST(Locale, isPrint) {
  70   EXPECT_FALSE(isPrint(0)); // <control-0000>-<control-001F>
  71   EXPECT_FALSE(isPrint(0x01));
  72   EXPECT_FALSE(isPrint(0x1F));
  73   EXPECT_TRUE(isPrint(' '));
  74   EXPECT_TRUE(isPrint('A'));
  75   EXPECT_TRUE(isPrint('~'));
  76   EXPECT_FALSE(isPrint(0x7F)); // <control-007F>..<control-009F>
  77   EXPECT_FALSE(isPrint(0x90));
  78   EXPECT_FALSE(isPrint(0x9F));
  79
  80   EXPECT_TRUE(isPrint(0xAC));
  81   EXPECT_TRUE(isPrint(0xAD)); // SOFT HYPHEN is displayed on most terminals
  82                               // as either a space or a dash.
  83   EXPECT_TRUE(isPrint(0xAE));
  84
  85   // MacOS implementation doesn't think it's printable.
  86 #ifndef __APPLE__
  87   EXPECT_TRUE(isPrint(0x0377)); // GREEK SMALL LETTER PAMPHYLIAN DIGAMMA
  88 #endif // __APPLE__
  89   EXPECT_FALSE(isPrint(0x0378)); // <reserved-0378>..<reserved-0379>
  90
  91   EXPECT_FALSE(isPrint(0x0600)); // ARABIC NUMBER SIGN
  92
  93   EXPECT_FALSE(isPrint(0x1FFFF)); // <reserved-1F774>..<noncharacter-1FFFF>
  94   EXPECT_TRUE(isPrint(0x20000)); // CJK UNIFIED IDEOGRAPH-20000
  95
  96   EXPECT_FALSE(isPrint(0x10FFFF)); // noncharacter
  97 }
  98
  99 #endif // _WIN32
 100
 101 } // namespace
 102 } // namespace locale
 103 } // namespace sys
 104 } // namespace llvm