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 TEST(Locale, columnWidth) {
  19   EXPECT_EQ(0, columnWidth(""));
  20   EXPECT_EQ(1, columnWidth(" "));
  21   EXPECT_EQ(1, columnWidth("a"));
  22   EXPECT_EQ(1, columnWidth("~"));
  23
  24   EXPECT_EQ(6, columnWidth("abcdef"));
  25
  26   EXPECT_EQ(-1, columnWidth("\x01"));
  27   EXPECT_EQ(-1, columnWidth("aaaaaaaaaa\x01"));
  28   EXPECT_EQ(-1, columnWidth("\342\200\213")); // 200B ZERO WIDTH SPACE
  29
  30   EXPECT_EQ(0, columnWidth("\314\200")); // 0300 COMBINING GRAVE ACCENT
  31   EXPECT_EQ(1, columnWidth("\340\270\201")); // 0E01 THAI CHARACTER KO KAI
  32   EXPECT_EQ(2, columnWidth("\344\270\200")); // CJK UNIFIED IDEOGRAPH-4E00
  33
  34   EXPECT_EQ(4, columnWidth("\344\270\200\344\270\200"));
  35   EXPECT_EQ(3, columnWidth("q\344\270\200"));
  36   EXPECT_EQ(3, columnWidth("\314\200\340\270\201\344\270\200"));
  37
  38   // Invalid UTF-8 strings, columnWidth should error out.
  39   EXPECT_EQ(-2, columnWidth("\344"));
  40   EXPECT_EQ(-2, columnWidth("\344\270"));
  41   EXPECT_EQ(-2, columnWidth("\344\270\033"));
  42   EXPECT_EQ(-2, columnWidth("\344\270\300"));
  43   EXPECT_EQ(-2, columnWidth("\377\366\355"));
  44
  45   EXPECT_EQ(-2, columnWidth("qwer\344"));
  46   EXPECT_EQ(-2, columnWidth("qwer\344\270"));
  47   EXPECT_EQ(-2, columnWidth("qwer\344\270\033"));
  48   EXPECT_EQ(-2, columnWidth("qwer\344\270\300"));
  49   EXPECT_EQ(-2, columnWidth("qwer\377\366\355"));
  50
  51   // UTF-8 sequences longer than 4 bytes correspond to unallocated Unicode
  52   // characters.
  53   EXPECT_EQ(-2, columnWidth("\370\200\200\200\200"));     // U+200000
  54   EXPECT_EQ(-2, columnWidth("\374\200\200\200\200\200")); // U+4000000
  55 }
  56
  57 TEST(Locale, isPrint) {
  58   EXPECT_EQ(false, isPrint(0)); // <control-0000>-<control-001F>
  59   EXPECT_EQ(false, isPrint(0x01));
  60   EXPECT_EQ(false, isPrint(0x1F));
  61   EXPECT_EQ(true, isPrint(' '));
  62   EXPECT_EQ(true, isPrint('A'));
  63   EXPECT_EQ(true, isPrint('~'));
  64   EXPECT_EQ(false, isPrint(0x7F)); // <control-007F>..<control-009F>
  65   EXPECT_EQ(false, isPrint(0x90));
  66   EXPECT_EQ(false, isPrint(0x9F));
  67
  68   EXPECT_EQ(true, isPrint(0xAC));
  69   EXPECT_EQ(false, isPrint(0xAD)); // SOFT HYPHEN
  70   EXPECT_EQ(true, isPrint(0xAE));
  71
  72   EXPECT_EQ(true, isPrint(0x0377)); // GREEK SMALL LETTER PAMPHYLIAN DIGAMMA
  73   EXPECT_EQ(false, isPrint(0x0378)); // <reserved-0378>..<reserved-0379>
  74
  75   EXPECT_EQ(false, isPrint(0x0600)); // ARABIC NUMBER SIGN
  76
  77   EXPECT_EQ(false, isPrint(0x1FFFF)); // <reserved-1F774>..<noncharacter-1FFFF>
  78   EXPECT_EQ(true, isPrint(0x20000)); // CJK UNIFIED IDEOGRAPH-20000
  79
  80   EXPECT_EQ(false, isPrint(0x10FFFF)); // noncharacter
  81 }
  82
  83 } // namespace
  84 } // namespace locale
  85 } // namespace sys
  86 } // namespace llvm