Tom Sepez | 26b8a5b | 2015-01-27 12:42:36 -0800 | [diff] [blame] | 1 | // Copyright 2015 PDFium Authors. All rights reserved. |
| 2 | // Use of this source code is governed by a BSD-style license that can be |
| 3 | // found in the LICENSE file. |
| 4 | |
| 5 | #include "../../testing/embedder_test.h" |
| 6 | #include "../../fpdfsdk/include/fpdfview.h" |
| 7 | #include "../../fpdfsdk/include/fpdftext.h" |
| 8 | #include "testing/gtest/include/gtest/gtest.h" |
| 9 | |
Tom Sepez | 526f6d5 | 2015-01-28 15:49:13 -0800 | [diff] [blame] | 10 | namespace { |
| 11 | |
| 12 | static bool check_unsigned_shorts(const char* expected, |
| 13 | const unsigned short* actual, |
| 14 | size_t length) { |
| 15 | if (length > strlen(expected) + 1) { |
| 16 | return false; |
| 17 | } |
| 18 | for (size_t i = 0; i < length; ++i) { |
| 19 | if (actual[i] != static_cast<unsigned short>(expected[i])) { |
| 20 | return false; |
| 21 | } |
| 22 | } |
| 23 | return true; |
| 24 | } |
| 25 | |
| 26 | } // namespace |
| 27 | |
Tom Sepez | 26b8a5b | 2015-01-27 12:42:36 -0800 | [diff] [blame] | 28 | class FPDFTextEmbeddertest : public EmbedderTest { |
| 29 | }; |
| 30 | |
Tom Sepez | 526f6d5 | 2015-01-28 15:49:13 -0800 | [diff] [blame] | 31 | TEST_F(FPDFTextEmbeddertest, Text) { |
| 32 | EXPECT_TRUE(OpenDocument("testing/resources/hello_world.pdf")); |
Tom Sepez | da8189e | 2015-01-30 14:41:50 -0800 | [diff] [blame] | 33 | FPDF_PAGE page = LoadPage(0); |
Tom Sepez | 526f6d5 | 2015-01-28 15:49:13 -0800 | [diff] [blame] | 34 | EXPECT_NE(nullptr, page); |
| 35 | |
| 36 | FPDF_TEXTPAGE textpage = FPDFText_LoadPage(page); |
| 37 | EXPECT_NE(nullptr, textpage); |
| 38 | |
| 39 | const char expected[] = "Hello, world!\r\nGoodbye, world!"; |
| 40 | unsigned short fixed_buffer[128]; |
| 41 | memset(fixed_buffer, 0xbd, sizeof(fixed_buffer)); |
| 42 | |
| 43 | // Check includes the terminating NUL that is provided. |
| 44 | EXPECT_EQ(sizeof(expected), FPDFText_GetText(textpage, 0, 128, fixed_buffer)); |
| 45 | EXPECT_TRUE(check_unsigned_shorts(expected, fixed_buffer, sizeof(expected))); |
| 46 | |
| 47 | // Count does not include the terminating NUL in the string literal. |
| 48 | EXPECT_EQ(sizeof(expected) - 1, FPDFText_CountChars(textpage)); |
| 49 | for (size_t i = 0; i < sizeof(expected) - 1; ++i) { |
| 50 | EXPECT_EQ(expected[i], FPDFText_GetUnicode(textpage, i)) << " at " << i; |
| 51 | } |
| 52 | |
| 53 | EXPECT_EQ(12.0, FPDFText_GetFontSize(textpage, 0)); |
| 54 | EXPECT_EQ(16.0, FPDFText_GetFontSize(textpage, 15)); |
| 55 | |
| 56 | double left = 0.0; |
| 57 | double right = 0.0; |
| 58 | double bottom = 0.0; |
| 59 | double top = 0.0; |
| 60 | FPDFText_GetCharBox(textpage, 4, &left, &right, &bottom, &top); |
| 61 | EXPECT_NEAR(41.071, left, 0.001); |
| 62 | EXPECT_NEAR(46.243, right, 0.001); |
| 63 | EXPECT_NEAR(49.844, bottom, 0.001); |
| 64 | EXPECT_NEAR(55.520, top, 0.001); |
| 65 | |
| 66 | EXPECT_EQ(4, FPDFText_GetCharIndexAtPos( |
| 67 | textpage, 42.0, 50.0, 1.0, 1.0)); |
| 68 | EXPECT_EQ(-1, FPDFText_GetCharIndexAtPos( |
| 69 | textpage, 0.0, 0.0, 1.0, 1.0)); |
| 70 | EXPECT_EQ(-1, FPDFText_GetCharIndexAtPos( |
| 71 | textpage, 199.0, 199.0, 1.0, 1.0)); |
| 72 | |
| 73 | // Test out of range indicies. |
| 74 | EXPECT_EQ(-1, FPDFText_GetCharIndexAtPos( |
| 75 | textpage, 42.0, 10000000.0, 1.0, 1.0)); |
| 76 | EXPECT_EQ(-1, FPDFText_GetCharIndexAtPos( |
| 77 | textpage, -1.0, 50.0, 1.0, 1.0)); |
| 78 | |
| 79 | // Count does not include the terminating NUL in the string literal. |
| 80 | EXPECT_EQ(2, FPDFText_CountRects(textpage, 0, sizeof(expected) - 1)); |
| 81 | |
| 82 | left = 0.0; |
| 83 | right = 0.0; |
| 84 | bottom = 0.0; |
| 85 | top = 0.0; |
| 86 | FPDFText_GetRect(textpage, 1, &left, &top, &right, &bottom); |
| 87 | EXPECT_NEAR(20.847, left, 0.001); |
| 88 | EXPECT_NEAR(135.167, right, 0.001); |
| 89 | EXPECT_NEAR(96.655, bottom, 0.001); |
| 90 | EXPECT_NEAR(116.000, top, 0.001); |
| 91 | |
| 92 | // Test out of range indicies set outputs to (0.0, 0.0, 0.0, 0.0). |
| 93 | left = -1.0; |
| 94 | right = -1.0; |
| 95 | bottom = -1.0; |
| 96 | top = -1.0; |
| 97 | FPDFText_GetRect(textpage, -1, &left, &top, &right, &bottom); |
| 98 | EXPECT_EQ(0.0, left); |
| 99 | EXPECT_EQ(0.0, right); |
| 100 | EXPECT_EQ(0.0, bottom); |
| 101 | EXPECT_EQ(0.0, top); |
| 102 | |
| 103 | left = -2.0; |
| 104 | right = -2.0; |
| 105 | bottom = -2.0; |
| 106 | top = -2.0; |
| 107 | FPDFText_GetRect(textpage, 2, &left, &top, &right, &bottom); |
| 108 | EXPECT_EQ(0.0, left); |
| 109 | EXPECT_EQ(0.0, right); |
| 110 | EXPECT_EQ(0.0, bottom); |
| 111 | EXPECT_EQ(0.0, top); |
| 112 | |
| 113 | EXPECT_EQ(9, FPDFText_GetBoundedText( |
| 114 | textpage, 41.0, 56.0, 82.0, 48.0, 0, 0)); |
| 115 | |
| 116 | // Extract starting at character 4 as above. |
| 117 | memset(fixed_buffer, 0xbd, sizeof(fixed_buffer)); |
| 118 | EXPECT_EQ(1, FPDFText_GetBoundedText( |
| 119 | textpage, 41.0, 56.0, 82.0, 48.0, fixed_buffer, 1)); |
| 120 | EXPECT_TRUE(check_unsigned_shorts(expected + 4, fixed_buffer, 1)); |
| 121 | EXPECT_EQ(0xbdbd, fixed_buffer[1]); |
| 122 | |
| 123 | memset(fixed_buffer, 0xbd, sizeof(fixed_buffer)); |
| 124 | EXPECT_EQ(9, FPDFText_GetBoundedText( |
| 125 | textpage, 41.0, 56.0, 82.0, 48.0, fixed_buffer, 9)); |
| 126 | EXPECT_TRUE(check_unsigned_shorts(expected + 4, fixed_buffer, 9)); |
| 127 | EXPECT_EQ(0xbdbd, fixed_buffer[9]); |
| 128 | |
| 129 | memset(fixed_buffer, 0xbd, sizeof(fixed_buffer)); |
| 130 | EXPECT_EQ(10, FPDFText_GetBoundedText( |
| 131 | textpage, 41.0, 56.0, 82.0, 48.0, fixed_buffer, 128)); |
| 132 | EXPECT_TRUE(check_unsigned_shorts(expected + 4, fixed_buffer, 9)); |
| 133 | EXPECT_EQ(0u, fixed_buffer[9]); |
| 134 | EXPECT_EQ(0xbdbd, fixed_buffer[10]); |
| 135 | |
| 136 | FPDFText_ClosePage(textpage); |
Tom Sepez | 526f6d5 | 2015-01-28 15:49:13 -0800 | [diff] [blame] | 137 | } |
| 138 | |
| 139 | TEST_F(FPDFTextEmbeddertest, TextSearch) { |
| 140 | EXPECT_TRUE(OpenDocument("testing/resources/hello_world.pdf")); |
Tom Sepez | da8189e | 2015-01-30 14:41:50 -0800 | [diff] [blame] | 141 | FPDF_PAGE page = LoadPage(0); |
Tom Sepez | 526f6d5 | 2015-01-28 15:49:13 -0800 | [diff] [blame] | 142 | EXPECT_NE(nullptr, page); |
| 143 | |
| 144 | FPDF_TEXTPAGE textpage = FPDFText_LoadPage(page); |
| 145 | EXPECT_NE(nullptr, textpage); |
| 146 | |
| 147 | // Avoid issues with system wchar_t width vs. FPDF_WideString. |
| 148 | const unsigned short nope[] = { 'n', 'o', 'p', 'e', '\0' }; |
| 149 | const unsigned short world[] = { 'w', 'o', 'r', 'l', 'd', '\0' }; |
| 150 | const unsigned short world_caps[] = { 'W', 'O', 'R', 'L', 'D', '\0' }; |
| 151 | const unsigned short world_substr[] = { 'o', 'r', 'l', 'd', '\0' }; |
| 152 | |
| 153 | // No occurences of "nope" in test page. |
| 154 | FPDF_SCHHANDLE search = FPDFText_FindStart(textpage, nope, 0, 0); |
| 155 | EXPECT_NE(nullptr, search); |
| 156 | EXPECT_EQ(0, FPDFText_GetSchResultIndex(search)); |
| 157 | EXPECT_EQ(0, FPDFText_GetSchCount(search)); |
| 158 | |
| 159 | // Advancing finds nothing. |
| 160 | EXPECT_FALSE(FPDFText_FindNext(search)); |
| 161 | EXPECT_EQ(0, FPDFText_GetSchResultIndex(search)); |
| 162 | EXPECT_EQ(0, FPDFText_GetSchCount(search)); |
| 163 | |
| 164 | // Retreating finds nothing. |
| 165 | EXPECT_FALSE(FPDFText_FindPrev(search)); |
| 166 | EXPECT_EQ(0, FPDFText_GetSchResultIndex(search)); |
| 167 | EXPECT_EQ(0, FPDFText_GetSchCount(search)); |
| 168 | FPDFText_FindClose(search); |
| 169 | |
| 170 | // Two occurences of "world" in test page. |
| 171 | search = FPDFText_FindStart(textpage, world, 0, 2); |
| 172 | EXPECT_NE(nullptr, search); |
| 173 | |
| 174 | // Remains not found until advanced. |
| 175 | EXPECT_EQ(0, FPDFText_GetSchResultIndex(search)); |
| 176 | EXPECT_EQ(0, FPDFText_GetSchCount(search)); |
| 177 | |
| 178 | // First occurence of "world" in this test page. |
| 179 | EXPECT_TRUE(FPDFText_FindNext(search)); |
| 180 | EXPECT_EQ(7, FPDFText_GetSchResultIndex(search)); |
| 181 | EXPECT_EQ(5, FPDFText_GetSchCount(search)); |
| 182 | |
| 183 | // Last occurence of "world" in this test page. |
| 184 | EXPECT_TRUE(FPDFText_FindNext(search)); |
| 185 | EXPECT_EQ(24, FPDFText_GetSchResultIndex(search)); |
| 186 | EXPECT_EQ(5, FPDFText_GetSchCount(search)); |
| 187 | |
| 188 | // Found position unchanged when fails to advance. |
| 189 | EXPECT_FALSE(FPDFText_FindNext(search)); |
| 190 | EXPECT_EQ(24, FPDFText_GetSchResultIndex(search)); |
| 191 | EXPECT_EQ(5, FPDFText_GetSchCount(search)); |
| 192 | |
| 193 | // Back to first occurence. |
| 194 | EXPECT_TRUE(FPDFText_FindPrev(search)); |
| 195 | EXPECT_EQ(7, FPDFText_GetSchResultIndex(search)); |
| 196 | EXPECT_EQ(5, FPDFText_GetSchCount(search)); |
| 197 | |
| 198 | // Found position unchanged when fails to retreat. |
| 199 | EXPECT_FALSE(FPDFText_FindPrev(search)); |
| 200 | EXPECT_EQ(7, FPDFText_GetSchResultIndex(search)); |
| 201 | EXPECT_EQ(5, FPDFText_GetSchCount(search)); |
| 202 | FPDFText_FindClose(search); |
| 203 | |
| 204 | // Exact search unaffected by case sensitiity and whole word flags. |
| 205 | search = FPDFText_FindStart( |
| 206 | textpage, world, FPDF_MATCHCASE | FPDF_MATCHWHOLEWORD, 0); |
| 207 | EXPECT_NE(nullptr, search); |
| 208 | EXPECT_TRUE(FPDFText_FindNext(search)); |
| 209 | EXPECT_EQ(7, FPDFText_GetSchResultIndex(search)); |
| 210 | EXPECT_EQ(5, FPDFText_GetSchCount(search)); |
| 211 | FPDFText_FindClose(search); |
| 212 | |
| 213 | // Default is case-insensitive, so matching agaist caps works. |
| 214 | search = FPDFText_FindStart(textpage, world_caps, 0, 0); |
| 215 | EXPECT_NE(nullptr, search); |
| 216 | EXPECT_TRUE(FPDFText_FindNext(search)); |
| 217 | EXPECT_EQ(7, FPDFText_GetSchResultIndex(search)); |
| 218 | EXPECT_EQ(5, FPDFText_GetSchCount(search)); |
| 219 | FPDFText_FindClose(search); |
| 220 | |
| 221 | // But can be made case sensitive, in which case this fails. |
| 222 | search = FPDFText_FindStart(textpage, world_caps, FPDF_MATCHCASE, 0); |
| 223 | EXPECT_FALSE(FPDFText_FindNext(search)); |
| 224 | EXPECT_EQ(0, FPDFText_GetSchResultIndex(search)); |
| 225 | EXPECT_EQ(0, FPDFText_GetSchCount(search)); |
| 226 | FPDFText_FindClose(search); |
| 227 | |
| 228 | // Default is match anywhere within word, so matching substirng works. |
| 229 | search = FPDFText_FindStart(textpage, world_substr, 0, 0); |
| 230 | EXPECT_TRUE(FPDFText_FindNext(search)); |
| 231 | EXPECT_EQ(8, FPDFText_GetSchResultIndex(search)); |
| 232 | EXPECT_EQ(4, FPDFText_GetSchCount(search)); |
| 233 | FPDFText_FindClose(search); |
| 234 | |
| 235 | // But can be made to mach word boundaries, in which case this fails. |
| 236 | search = FPDFText_FindStart(textpage, world_substr, FPDF_MATCHWHOLEWORD, 0); |
| 237 | EXPECT_FALSE(FPDFText_FindNext(search)); |
| 238 | // TODO(tsepez): investigate strange index/count values in this state. |
| 239 | FPDFText_FindClose(search); |
| 240 | |
| 241 | FPDFText_ClosePage(textpage); |
Tom Sepez | 526f6d5 | 2015-01-28 15:49:13 -0800 | [diff] [blame] | 242 | } |
| 243 | |
Tom Sepez | 26b8a5b | 2015-01-27 12:42:36 -0800 | [diff] [blame] | 244 | // Test that the page has characters despite a bad stream length. |
| 245 | TEST_F(FPDFTextEmbeddertest, StreamLengthPastEndOfFile) { |
| 246 | EXPECT_TRUE(OpenDocument("testing/resources/bug_57.pdf")); |
Tom Sepez | da8189e | 2015-01-30 14:41:50 -0800 | [diff] [blame] | 247 | FPDF_PAGE page = LoadPage(0); |
Tom Sepez | 26b8a5b | 2015-01-27 12:42:36 -0800 | [diff] [blame] | 248 | EXPECT_NE(nullptr, page); |
Tom Sepez | 526f6d5 | 2015-01-28 15:49:13 -0800 | [diff] [blame] | 249 | |
Tom Sepez | 26b8a5b | 2015-01-27 12:42:36 -0800 | [diff] [blame] | 250 | FPDF_TEXTPAGE textpage = FPDFText_LoadPage(page); |
| 251 | EXPECT_NE(nullptr, textpage); |
| 252 | EXPECT_EQ(13, FPDFText_CountChars(textpage)); |
Tom Sepez | 526f6d5 | 2015-01-28 15:49:13 -0800 | [diff] [blame] | 253 | |
| 254 | FPDFText_ClosePage(textpage); |
Tom Sepez | 526f6d5 | 2015-01-28 15:49:13 -0800 | [diff] [blame] | 255 | } |
| 256 | |
| 257 | TEST_F(FPDFTextEmbeddertest, WebLinks) { |
| 258 | EXPECT_TRUE(OpenDocument("testing/resources/weblinks.pdf")); |
Tom Sepez | da8189e | 2015-01-30 14:41:50 -0800 | [diff] [blame] | 259 | FPDF_PAGE page = LoadPage(0); |
Tom Sepez | 526f6d5 | 2015-01-28 15:49:13 -0800 | [diff] [blame] | 260 | EXPECT_NE(nullptr, page); |
| 261 | |
| 262 | FPDF_TEXTPAGE textpage = FPDFText_LoadPage(page); |
| 263 | EXPECT_NE(nullptr, textpage); |
| 264 | |
| 265 | FPDF_PAGELINK pagelink = FPDFLink_LoadWebLinks(textpage); |
| 266 | EXPECT_NE(nullptr, pagelink); |
| 267 | |
| 268 | // Page contains two HTTP-style URLs. |
| 269 | EXPECT_EQ(2, FPDFLink_CountWebLinks(pagelink)); |
| 270 | |
| 271 | // Only a terminating NUL required for bogus links. |
| 272 | EXPECT_EQ(1, FPDFLink_GetURL(pagelink, 2, nullptr, 0)); |
| 273 | EXPECT_EQ(1, FPDFLink_GetURL(pagelink, 1400, nullptr, 0)); |
| 274 | EXPECT_EQ(1, FPDFLink_GetURL(pagelink, -1, nullptr, 0)); |
| 275 | |
| 276 | // Query the number of characters required for each link (incl NUL). |
| 277 | EXPECT_EQ(25, FPDFLink_GetURL(pagelink, 0, nullptr, 0)); |
| 278 | EXPECT_EQ(26, FPDFLink_GetURL(pagelink, 1, nullptr, 0)); |
| 279 | |
| 280 | const char expected_url[] = "http://example.com?q=foo"; |
| 281 | unsigned short fixed_buffer[128]; |
| 282 | |
| 283 | // Retrieve a link with too small a buffer. Buffer will not be |
| 284 | // NUL-terminated, but must not be modified past indicated length, |
| 285 | // so pre-fill with a pattern to check write bounds. |
| 286 | memset(fixed_buffer, 0xbd, sizeof(fixed_buffer)); |
| 287 | EXPECT_EQ(1, FPDFLink_GetURL(pagelink, 0, fixed_buffer, 1)); |
| 288 | EXPECT_TRUE(check_unsigned_shorts(expected_url, fixed_buffer, 1)); |
| 289 | EXPECT_EQ(0xbdbd, fixed_buffer[1]); |
| 290 | |
| 291 | // Check buffer that doesn't have space for a terminating NUL. |
| 292 | memset(fixed_buffer, 0xbd, sizeof(fixed_buffer)); |
| 293 | EXPECT_EQ(sizeof(expected_url) - 1, FPDFLink_GetURL( |
| 294 | pagelink, 0, fixed_buffer, sizeof(expected_url) - 1)); |
| 295 | EXPECT_TRUE(check_unsigned_shorts( |
| 296 | expected_url, fixed_buffer, sizeof(expected_url) - 1)); |
| 297 | EXPECT_EQ(0xbdbd, fixed_buffer[sizeof(expected_url) - 1]); |
| 298 | |
| 299 | // Retreive link with exactly-sized buffer. |
| 300 | memset(fixed_buffer, 0xbd, sizeof(fixed_buffer)); |
| 301 | EXPECT_EQ(sizeof(expected_url), FPDFLink_GetURL( |
| 302 | pagelink, 0, fixed_buffer, sizeof(expected_url))); |
| 303 | EXPECT_TRUE(check_unsigned_shorts( |
| 304 | expected_url, fixed_buffer, sizeof(expected_url))); |
| 305 | EXPECT_EQ(0u, fixed_buffer[sizeof(expected_url) - 1]); |
| 306 | EXPECT_EQ(0xbdbd, fixed_buffer[sizeof(expected_url)]); |
| 307 | |
| 308 | // Retreive link with ample-sized-buffer. |
| 309 | memset(fixed_buffer, 0xbd, sizeof(fixed_buffer)); |
| 310 | EXPECT_EQ(sizeof(expected_url), FPDFLink_GetURL( |
| 311 | pagelink, 0, fixed_buffer, 128)); |
| 312 | EXPECT_TRUE(check_unsigned_shorts( |
| 313 | expected_url, fixed_buffer, sizeof(expected_url))); |
| 314 | EXPECT_EQ(0u, fixed_buffer[sizeof(expected_url) - 1]); |
| 315 | EXPECT_EQ(0xbdbd, fixed_buffer[sizeof(expected_url)]); |
| 316 | |
| 317 | // Each link rendered in a single rect in this test page. |
| 318 | EXPECT_EQ(1, FPDFLink_CountRects(pagelink, 0)); |
| 319 | EXPECT_EQ(1, FPDFLink_CountRects(pagelink, 1)); |
| 320 | |
| 321 | // Each link rendered in a single rect in this test page. |
| 322 | EXPECT_EQ(0, FPDFLink_CountRects(pagelink, -1)); |
| 323 | EXPECT_EQ(0, FPDFLink_CountRects(pagelink, 2)); |
| 324 | EXPECT_EQ(0, FPDFLink_CountRects(pagelink, 10000)); |
| 325 | |
| 326 | // Check boundary of valid link index with valid rect index. |
| 327 | double left = 0.0; |
| 328 | double right = 0.0; |
| 329 | double top = 0.0; |
| 330 | double bottom = 0.0; |
| 331 | FPDFLink_GetRect(pagelink, 0, 0, &left, &top, &right, &bottom); |
| 332 | EXPECT_NEAR(50.791, left, 0.001); |
| 333 | EXPECT_NEAR(187.963, right, 0.001); |
| 334 | EXPECT_NEAR(97.624, bottom, 0.001); |
| 335 | EXPECT_NEAR(108.736, top, 0.001); |
| 336 | |
| 337 | // Check that valid link with invalid rect index leaves parameters unchanged. |
| 338 | left = -1.0; |
| 339 | right = -1.0; |
| 340 | top = -1.0; |
| 341 | bottom = -1.0; |
| 342 | FPDFLink_GetRect(pagelink, 0, 1, &left, &top, &right, &bottom); |
| 343 | EXPECT_EQ(-1.0, left); |
| 344 | EXPECT_EQ(-1.0, right); |
| 345 | EXPECT_EQ(-1.0, bottom); |
| 346 | EXPECT_EQ(-1.0, top); |
| 347 | |
| 348 | // Check that invalid link index leaves parameters unchanged. |
| 349 | left = -2.0; |
| 350 | right = -2.0; |
| 351 | top = -2.0; |
| 352 | bottom = -2.0; |
| 353 | FPDFLink_GetRect(pagelink, -1, 0, &left, &top, &right, &bottom); |
| 354 | EXPECT_EQ(-2.0, left); |
| 355 | EXPECT_EQ(-2.0, right); |
| 356 | EXPECT_EQ(-2.0, bottom); |
| 357 | EXPECT_EQ(-2.0, top); |
| 358 | |
| 359 | FPDFLink_CloseWebLinks(pagelink); |
| 360 | FPDFText_ClosePage(textpage); |
Tom Sepez | 26b8a5b | 2015-01-27 12:42:36 -0800 | [diff] [blame] | 361 | } |