TEST_F(FPDFTextEmbeddertest, TextSearch) {
EXPECT_TRUE(OpenDocument("testing/resources/hello_world.pdf"));
FPDF_PAGE page = LoadPage(0);
EXPECT_NE(nullptr, page);
FPDF_TEXTPAGE textpage = FPDFText_LoadPage(page);
EXPECT_NE(nullptr, textpage);
// Avoid issues with system wchar_t width vs. FPDF_WideString.
const unsigned short nope[] = {'n', 'o', 'p', 'e', '\0'};
const unsigned short world[] = {'w', 'o', 'r', 'l', 'd', '\0'};
const unsigned short world_caps[] = {'W', 'O', 'R', 'L', 'D', '\0'};
const unsigned short world_substr[] = {'o', 'r', 'l', 'd', '\0'};
// No occurences of "nope" in test page.
FPDF_SCHHANDLE search = FPDFText_FindStart(textpage, nope, 0, 0);
EXPECT_NE(nullptr, search);
EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(0, FPDFText_GetSchCount(search));
// Advancing finds nothing.
EXPECT_FALSE(FPDFText_FindNext(search));
EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(0, FPDFText_GetSchCount(search));
// Retreating finds nothing.
EXPECT_FALSE(FPDFText_FindPrev(search));
EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(0, FPDFText_GetSchCount(search));
FPDFText_FindClose(search);
// Two occurences of "world" in test page.
search = FPDFText_FindStart(textpage, world, 0, 2);
EXPECT_NE(nullptr, search);
// Remains not found until advanced.
EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(0, FPDFText_GetSchCount(search));
// First occurence of "world" in this test page.
EXPECT_TRUE(FPDFText_FindNext(search));
EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(5, FPDFText_GetSchCount(search));
// Last occurence of "world" in this test page.
EXPECT_TRUE(FPDFText_FindNext(search));
EXPECT_EQ(24, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(5, FPDFText_GetSchCount(search));
// Found position unchanged when fails to advance.
EXPECT_FALSE(FPDFText_FindNext(search));
EXPECT_EQ(24, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(5, FPDFText_GetSchCount(search));
// Back to first occurence.
EXPECT_TRUE(FPDFText_FindPrev(search));
EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(5, FPDFText_GetSchCount(search));
// Found position unchanged when fails to retreat.
EXPECT_FALSE(FPDFText_FindPrev(search));
EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(5, FPDFText_GetSchCount(search));
FPDFText_FindClose(search);
// Exact search unaffected by case sensitiity and whole word flags.
search = FPDFText_FindStart(textpage, world,
FPDF_MATCHCASE | FPDF_MATCHWHOLEWORD, 0);
EXPECT_NE(nullptr, search);
EXPECT_TRUE(FPDFText_FindNext(search));
EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(5, FPDFText_GetSchCount(search));
FPDFText_FindClose(search);
// Default is case-insensitive, so matching agaist caps works.
search = FPDFText_FindStart(textpage, world_caps, 0, 0);
EXPECT_NE(nullptr, search);
EXPECT_TRUE(FPDFText_FindNext(search));
EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(5, FPDFText_GetSchCount(search));
FPDFText_FindClose(search);
// But can be made case sensitive, in which case this fails.
search = FPDFText_FindStart(textpage, world_caps, FPDF_MATCHCASE, 0);
EXPECT_FALSE(FPDFText_FindNext(search));
EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(0, FPDFText_GetSchCount(search));
FPDFText_FindClose(search);
// Default is match anywhere within word, so matching substirng works.
search = FPDFText_FindStart(textpage, world_substr, 0, 0);
EXPECT_TRUE(FPDFText_FindNext(search));
EXPECT_EQ(8, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(4, FPDFText_GetSchCount(search));
FPDFText_FindClose(search);
// But can be made to mach word boundaries, in which case this fails.
search = FPDFText_FindStart(textpage, world_substr, FPDF_MATCHWHOLEWORD, 0);
EXPECT_FALSE(FPDFText_FindNext(search));
// TODO(tsepez): investigate strange index/count values in this state.
FPDFText_FindClose(search);
FPDFText_ClosePage(textpage);
UnloadPage(page);
}
TEST_F(FPDFTextEmbeddertest, TextSearch) {
EXPECT_TRUE(OpenDocument("hello_world.pdf"));
FPDF_PAGE page = LoadPage(0);
EXPECT_TRUE(page);
FPDF_TEXTPAGE textpage = FPDFText_LoadPage(page);
EXPECT_TRUE(textpage);
std::unique_ptr<unsigned short, pdfium::FreeDeleter> nope =
GetFPDFWideString(L"nope");
std::unique_ptr<unsigned short, pdfium::FreeDeleter> world =
GetFPDFWideString(L"world");
std::unique_ptr<unsigned short, pdfium::FreeDeleter> world_caps =
GetFPDFWideString(L"WORLD");
std::unique_ptr<unsigned short, pdfium::FreeDeleter> world_substr =
GetFPDFWideString(L"orld");
// No occurences of "nope" in test page.
FPDF_SCHHANDLE search = FPDFText_FindStart(textpage, nope.get(), 0, 0);
EXPECT_TRUE(search);
EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(0, FPDFText_GetSchCount(search));
// Advancing finds nothing.
EXPECT_FALSE(FPDFText_FindNext(search));
EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(0, FPDFText_GetSchCount(search));
// Retreating finds nothing.
EXPECT_FALSE(FPDFText_FindPrev(search));
EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(0, FPDFText_GetSchCount(search));
FPDFText_FindClose(search);
// Two occurences of "world" in test page.
search = FPDFText_FindStart(textpage, world.get(), 0, 2);
EXPECT_TRUE(search);
// Remains not found until advanced.
EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(0, FPDFText_GetSchCount(search));
// First occurence of "world" in this test page.
EXPECT_TRUE(FPDFText_FindNext(search));
EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(5, FPDFText_GetSchCount(search));
// Last occurence of "world" in this test page.
EXPECT_TRUE(FPDFText_FindNext(search));
EXPECT_EQ(24, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(5, FPDFText_GetSchCount(search));
// Found position unchanged when fails to advance.
EXPECT_FALSE(FPDFText_FindNext(search));
EXPECT_EQ(24, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(5, FPDFText_GetSchCount(search));
// Back to first occurence.
EXPECT_TRUE(FPDFText_FindPrev(search));
EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(5, FPDFText_GetSchCount(search));
// Found position unchanged when fails to retreat.
EXPECT_FALSE(FPDFText_FindPrev(search));
EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(5, FPDFText_GetSchCount(search));
FPDFText_FindClose(search);
// Exact search unaffected by case sensitiity and whole word flags.
search = FPDFText_FindStart(textpage, world.get(),
FPDF_MATCHCASE | FPDF_MATCHWHOLEWORD, 0);
EXPECT_TRUE(search);
EXPECT_TRUE(FPDFText_FindNext(search));
EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(5, FPDFText_GetSchCount(search));
FPDFText_FindClose(search);
// Default is case-insensitive, so matching agaist caps works.
search = FPDFText_FindStart(textpage, world_caps.get(), 0, 0);
EXPECT_TRUE(search);
EXPECT_TRUE(FPDFText_FindNext(search));
EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(5, FPDFText_GetSchCount(search));
FPDFText_FindClose(search);
// But can be made case sensitive, in which case this fails.
search = FPDFText_FindStart(textpage, world_caps.get(), FPDF_MATCHCASE, 0);
EXPECT_FALSE(FPDFText_FindNext(search));
EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(0, FPDFText_GetSchCount(search));
FPDFText_FindClose(search);
// Default is match anywhere within word, so matching substirng works.
search = FPDFText_FindStart(textpage, world_substr.get(), 0, 0);
EXPECT_TRUE(FPDFText_FindNext(search));
EXPECT_EQ(8, FPDFText_GetSchResultIndex(search));
EXPECT_EQ(4, FPDFText_GetSchCount(search));
FPDFText_FindClose(search);
// But can be made to mach word boundaries, in which case this fails.
search =
FPDFText_FindStart(textpage, world_substr.get(), FPDF_MATCHWHOLEWORD, 0);
EXPECT_FALSE(FPDFText_FindNext(search));
// TODO(tsepez): investigate strange index/count values in this state.
FPDFText_FindClose(search);
FPDFText_ClosePage(textpage);
UnloadPage(page);
}
