rhubarb-lip-sync/tests/stringToolsTests.cpp

96 lines
3.9 KiB
C++
Raw Blame History

This file contains ambiguous Unicode characters

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

#include <gmock/gmock.h>
#include "stringTools.h"
using namespace testing;
using std::string;
using std::wstring;
// splitIntoLines
TEST(splitIntoLines, splitsOnLineBreaks) {
EXPECT_THAT(splitIntoLines("this\nis\r\na\r\ntest"), ElementsAre("this", "is", "a", "test"));
}
TEST(splitIntoLines, handlesEmptyElements) {
EXPECT_THAT(splitIntoLines("\n1\n\n\n2\n"), ElementsAre("", "1", "", "", "2", ""));
EXPECT_THAT(splitIntoLines("\n"), ElementsAre("", ""));
EXPECT_THAT(splitIntoLines(""), ElementsAre(""));
}
// wrapSingleLineString
TEST(wrapSingleLineString, basic) {
const char* lipsum = "Lorem ipsum dolor sit amet, consectetur adipisici elit, sed eiusmod tempor incidunt ut labore et dolore magna aliqua.";
EXPECT_THAT(wrapSingleLineString(lipsum, 30), ElementsAre("Lorem ipsum dolor sit amet,", "consectetur adipisici elit,", "sed eiusmod tempor incidunt ut", "labore et dolore magna aliqua."));
}
TEST(wrapSingleLineString, preciseWrapPosition) {
const char* test = "a b c";
EXPECT_THAT(wrapSingleLineString(test, 5), ElementsAre("a b c"));
EXPECT_THAT(wrapSingleLineString(test, 4), ElementsAre("a b", "c"));
EXPECT_THAT(wrapSingleLineString(test, 3), ElementsAre("a b", "c"));
EXPECT_THAT(wrapSingleLineString(test, 2), ElementsAre("a", "b", "c"));
EXPECT_THAT(wrapSingleLineString(test, 1), ElementsAre("a", "b", "c"));
}
TEST(wrapSingleLineString, overlongLines) {
EXPECT_THAT(wrapSingleLineString("aaa bbbb ccc", 3), ElementsAre("aaa", "bbb", "b", "ccc"));
EXPECT_THAT(wrapSingleLineString("aaa bbbb c", 3), ElementsAre("aaa", "bbb", "b c"));
EXPECT_THAT(wrapSingleLineString("a bbbb c", 5), ElementsAre("a", "bbbb", "c"));
EXPECT_THAT(wrapSingleLineString("aa b", 1), ElementsAre("a", "a", "b"));
}
TEST(wrapSingleLineString, discardingSpacesAtWrapPositionsAndEnd) {
const char* test = " a b c ";
EXPECT_THAT(wrapSingleLineString(test, 20), ElementsAre(" a b c"));
EXPECT_THAT(wrapSingleLineString(test, 6), ElementsAre(" a b", "c"));
EXPECT_THAT(wrapSingleLineString(test, 5), ElementsAre(" a", "b c"));
EXPECT_THAT(wrapSingleLineString(test, 4), ElementsAre(" a", "b c"));
EXPECT_THAT(wrapSingleLineString(test, 3), ElementsAre(" a", "b", "c"));
EXPECT_THAT(wrapSingleLineString(test, 2), ElementsAre("", "a", "b", "c"));
}
TEST(wrapSingleLineString, errorHandling) {
EXPECT_NO_THROW(wrapSingleLineString("test", 1));
EXPECT_NO_THROW(wrapSingleLineString("", 1));
// Throw if lineLength < 1
EXPECT_ANY_THROW(wrapSingleLineString("test", 0));
EXPECT_ANY_THROW(wrapSingleLineString("", 0));
EXPECT_ANY_THROW(wrapSingleLineString("test", -1));
EXPECT_ANY_THROW(wrapSingleLineString("", -1));
// Throw if string contains tabs
EXPECT_ANY_THROW(wrapSingleLineString("a\tb", 10));
// Throw if string contains line breaks
EXPECT_ANY_THROW(wrapSingleLineString("a\nb", 10));
}
// wrapString
TEST(wrapString, basic) {
EXPECT_THAT(wrapString("\n\nLine no 3\n\nLine no 4\n", 8), ElementsAre("", "", "Line no", "3", "", "Line no", "4", ""));
}
// latin1ToWide
TEST(latin1ToWide, basic) {
string pangramLatin1 = "D\350s No\353l o\371 un z\351phyr ha\357 me v\352t de gla\347ons w\374rmiens, je d\356ne d'exquis r\364tis de boeuf au kir \340 l'a\377 d'\342ge m\373r & c\346tera!";
wstring pangramWide = L"Dès Noël où un zéphyr haï me vêt de glaçons würmiens, je dîne d'exquis rôtis de boeuf au kir à l'aÿ d'âge mûr & cætera!";
EXPECT_EQ(pangramWide, latin1ToWide(pangramLatin1));
}
// utf8ToAscii
TEST(utf8ToAscii, string) {
EXPECT_EQ(
"A naive man called was having pina colada and creme brulee.",
utf8ToAscii("A naïve man called 晨 was having piña colada and crème brûlée."));
EXPECT_EQ(string(""), utf8ToAscii(""));
EXPECT_EQ(string("- - - - - - - - - -"), utf8ToAscii("- — ― "));
EXPECT_EQ(string("' ' ' ' \" \" \" \" \" \""), utf8ToAscii(" “ ” „ ‟ « »"));
EXPECT_EQ(string("1 2 3"), utf8ToAscii("¹ ² ³"));
EXPECT_EQ(string("1/4 1/2 3/4"), utf8ToAscii("¼ ½ ¾"));
}