1e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov// Copyright 2015 PDFium Authors. All rights reserved.
2e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov// Use of this source code is governed by a BSD-style license that can be
3e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov// found in the LICENSE file.
4e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
5e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov#include "../../public/fpdf_text.h"
6e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov#include "../../public/fpdfview.h"
7e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov#include "../../testing/embedder_test.h"
8e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov#include "testing/gtest/include/gtest/gtest.h"
9e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
10e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganovnamespace {
11e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
12e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganovstatic bool check_unsigned_shorts(const char* expected,
13e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov                                  const unsigned short* actual,
14e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov                                  size_t length) {
15e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  if (length > strlen(expected) + 1) {
16e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov    return false;
17e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  }
18e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  for (size_t i = 0; i < length; ++i) {
19e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov    if (actual[i] != static_cast<unsigned short>(expected[i])) {
20e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      return false;
21e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov    }
22e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  }
23e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  return true;
24e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov}
25e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
26e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov}  // namespace
27e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
28e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganovclass FPDFTextEmbeddertest : public EmbedderTest {
29e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov};
30e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
31e6986e1e8d4a57987f47c215490cb080a65ee29aSvet GanovTEST_F(FPDFTextEmbeddertest, Text) {
32e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(OpenDocument("testing/resources/hello_world.pdf"));
33e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDF_PAGE page = LoadPage(0);
34e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NE(nullptr, page);
35e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
36e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDF_TEXTPAGE textpage = FPDFText_LoadPage(page);
37e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NE(nullptr, textpage);
38e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
39e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  static const char expected[] = "Hello, world!\r\nGoodbye, world!";
40e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  unsigned short fixed_buffer[128];
41e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  memset(fixed_buffer, 0xbd, sizeof(fixed_buffer));
42e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
43e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Check includes the terminating NUL that is provided.
44e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(sizeof(expected), FPDFText_GetText(textpage, 0, 128, fixed_buffer));
45e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(check_unsigned_shorts(expected, fixed_buffer, sizeof(expected)));
46e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
47e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Count does not include the terminating NUL in the string literal.
48e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(sizeof(expected) - 1, FPDFText_CountChars(textpage));
49e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  for (size_t i = 0; i < sizeof(expected) - 1; ++i) {
50e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov    EXPECT_EQ(expected[i], FPDFText_GetUnicode(textpage, i)) << " at " << i;
51e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  }
52e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
53e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(12.0, FPDFText_GetFontSize(textpage, 0));
54e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(16.0, FPDFText_GetFontSize(textpage, 15));
55e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
56e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  double left = 0.0;
57e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  double right = 0.0;
58e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  double bottom = 0.0;
59e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  double top = 0.0;
60e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_GetCharBox(textpage, 4, &left, &right, &bottom, &top);
61e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NEAR(41.071, left, 0.001);
62e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NEAR(46.243, right, 0.001);
63e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NEAR(49.844, bottom, 0.001);
64e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NEAR(55.520, top, 0.001);
65e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
66e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(4, FPDFText_GetCharIndexAtPos(
67e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      textpage, 42.0, 50.0, 1.0, 1.0));
68e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(-1, FPDFText_GetCharIndexAtPos(
69e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      textpage, 0.0, 0.0, 1.0, 1.0));
70e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(-1, FPDFText_GetCharIndexAtPos(
71e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      textpage, 199.0, 199.0, 1.0, 1.0));
72e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
73e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Test out of range indicies.
74e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(-1, FPDFText_GetCharIndexAtPos(
75e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      textpage, 42.0, 10000000.0, 1.0, 1.0));
76e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(-1, FPDFText_GetCharIndexAtPos(
77e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      textpage, -1.0, 50.0, 1.0, 1.0));
78e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
79e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Count does not include the terminating NUL in the string literal.
80e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(2, FPDFText_CountRects(textpage, 0, sizeof(expected) - 1));
81e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
82e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  left = 0.0;
83e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  right = 0.0;
84e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  bottom = 0.0;
85e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  top = 0.0;
86e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_GetRect(textpage, 1, &left, &top, &right, &bottom);
87e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NEAR(20.847, left, 0.001);
88e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NEAR(135.167, right, 0.001);
89e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NEAR(96.655, bottom, 0.001);
90e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NEAR(116.000, top, 0.001);
91e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
92e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Test out of range indicies set outputs to (0.0, 0.0, 0.0, 0.0).
93e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  left = -1.0;
94e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  right = -1.0;
95e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  bottom = -1.0;
96e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  top = -1.0;
97e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_GetRect(textpage, -1, &left, &top, &right, &bottom);
98e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0.0, left);
99e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0.0, right);
100e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0.0, bottom);
101e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0.0, top);
102e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
103e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  left = -2.0;
104e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  right = -2.0;
105e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  bottom = -2.0;
106e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  top = -2.0;
107e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_GetRect(textpage, 2, &left, &top, &right, &bottom);
108e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0.0, left);
109e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0.0, right);
110e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0.0, bottom);
111e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0.0, top);
112e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
113e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(9, FPDFText_GetBoundedText(
114e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      textpage, 41.0, 56.0, 82.0, 48.0, 0, 0));
115e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
116e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Extract starting at character 4 as above.
117e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  memset(fixed_buffer, 0xbd, sizeof(fixed_buffer));
118e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(1, FPDFText_GetBoundedText(
119e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      textpage, 41.0, 56.0, 82.0, 48.0, fixed_buffer, 1));
120e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(check_unsigned_shorts(expected + 4, fixed_buffer, 1));
121e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0xbdbd, fixed_buffer[1]);
122e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
123e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  memset(fixed_buffer, 0xbd, sizeof(fixed_buffer));
124e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(9, FPDFText_GetBoundedText(
125e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      textpage, 41.0, 56.0, 82.0, 48.0, fixed_buffer, 9));
126e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(check_unsigned_shorts(expected + 4, fixed_buffer, 9));
127e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0xbdbd, fixed_buffer[9]);
128e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
129e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  memset(fixed_buffer, 0xbd, sizeof(fixed_buffer));
130e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(10, FPDFText_GetBoundedText(
131e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      textpage, 41.0, 56.0, 82.0, 48.0, fixed_buffer, 128));
132e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(check_unsigned_shorts(expected + 4, fixed_buffer, 9));
133e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0u, fixed_buffer[9]);
134e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0xbdbd, fixed_buffer[10]);
135e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
136e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_ClosePage(textpage);
137e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  UnloadPage(page);
138e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov}
139e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
140e6986e1e8d4a57987f47c215490cb080a65ee29aSvet GanovTEST_F(FPDFTextEmbeddertest, TextSearch) {
141e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(OpenDocument("testing/resources/hello_world.pdf"));
142e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDF_PAGE page = LoadPage(0);
143e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NE(nullptr, page);
144e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
145e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDF_TEXTPAGE textpage = FPDFText_LoadPage(page);
146e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NE(nullptr, textpage);
147e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
148e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Avoid issues with system wchar_t width vs. FPDF_WideString.
149e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  const unsigned short nope[] = { 'n', 'o', 'p', 'e', '\0' };
150e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  const unsigned short world[] = { 'w', 'o', 'r', 'l', 'd', '\0' };
151e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  const unsigned short world_caps[] = { 'W', 'O', 'R', 'L', 'D', '\0' };
152e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  const unsigned short world_substr[] = { 'o', 'r', 'l', 'd', '\0' };
153e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
154e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // No occurences of "nope" in test page.
155e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDF_SCHHANDLE search = FPDFText_FindStart(textpage, nope, 0, 0);
156e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NE(nullptr, search);
157e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
158e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0, FPDFText_GetSchCount(search));
159e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
160e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Advancing finds nothing.
161e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_FALSE(FPDFText_FindNext(search));
162e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
163e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0, FPDFText_GetSchCount(search));
164e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
165e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Retreating finds nothing.
166e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_FALSE(FPDFText_FindPrev(search));
167e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
168e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0, FPDFText_GetSchCount(search));
169e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_FindClose(search);
170e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
171e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Two occurences of "world" in test page.
172e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  search = FPDFText_FindStart(textpage, world, 0, 2);
173e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NE(nullptr, search);
174e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
175e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Remains not found until advanced.
176e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
177e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0, FPDFText_GetSchCount(search));
178e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
179e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // First occurence of "world" in this test page.
180e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(FPDFText_FindNext(search));
181e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
182e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(5, FPDFText_GetSchCount(search));
183e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
184e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Last occurence of "world" in this test page.
185e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(FPDFText_FindNext(search));
186e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(24, FPDFText_GetSchResultIndex(search));
187e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(5, FPDFText_GetSchCount(search));
188e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
189e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Found position unchanged when fails to advance.
190e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_FALSE(FPDFText_FindNext(search));
191e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(24, FPDFText_GetSchResultIndex(search));
192e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(5, FPDFText_GetSchCount(search));
193e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
194e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Back to first occurence.
195e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(FPDFText_FindPrev(search));
196e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
197e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(5, FPDFText_GetSchCount(search));
198e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
199e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Found position unchanged when fails to retreat.
200e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_FALSE(FPDFText_FindPrev(search));
201e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
202e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(5, FPDFText_GetSchCount(search));
203e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_FindClose(search);
204e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
205e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Exact search unaffected by case sensitiity and whole word flags.
206e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  search = FPDFText_FindStart(
207e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      textpage, world, FPDF_MATCHCASE | FPDF_MATCHWHOLEWORD, 0);
208e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NE(nullptr, search);
209e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(FPDFText_FindNext(search));
210e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
211e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(5, FPDFText_GetSchCount(search));
212e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_FindClose(search);
213e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
214e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Default is case-insensitive, so matching agaist caps works.
215e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  search = FPDFText_FindStart(textpage, world_caps, 0, 0);
216e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NE(nullptr, search);
217e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(FPDFText_FindNext(search));
218e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(7, FPDFText_GetSchResultIndex(search));
219e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(5, FPDFText_GetSchCount(search));
220e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_FindClose(search);
221e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
222e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // But can be made case sensitive, in which case this fails.
223e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  search = FPDFText_FindStart(textpage, world_caps, FPDF_MATCHCASE, 0);
224e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_FALSE(FPDFText_FindNext(search));
225e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0, FPDFText_GetSchResultIndex(search));
226e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0, FPDFText_GetSchCount(search));
227e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_FindClose(search);
228e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
229e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Default is match anywhere within word, so matching substirng works.
230e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  search = FPDFText_FindStart(textpage, world_substr, 0, 0);
231e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(FPDFText_FindNext(search));
232e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(8, FPDFText_GetSchResultIndex(search));
233e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(4, FPDFText_GetSchCount(search));
234e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_FindClose(search);
235e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
236e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // But can be made to mach word boundaries, in which case this fails.
237e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  search = FPDFText_FindStart(textpage, world_substr, FPDF_MATCHWHOLEWORD, 0);
238e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_FALSE(FPDFText_FindNext(search));
239e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // TODO(tsepez): investigate strange index/count values in this state.
240e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_FindClose(search);
241e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
242e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_ClosePage(textpage);
243e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  UnloadPage(page);
244e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov}
245e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
246e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov// Test that the page has characters despite a bad stream length.
247e6986e1e8d4a57987f47c215490cb080a65ee29aSvet GanovTEST_F(FPDFTextEmbeddertest, StreamLengthPastEndOfFile) {
248e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(OpenDocument("testing/resources/bug_57.pdf"));
249e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDF_PAGE page = LoadPage(0);
250e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NE(nullptr, page);
251e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
252e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDF_TEXTPAGE textpage = FPDFText_LoadPage(page);
253e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NE(nullptr, textpage);
254e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(13, FPDFText_CountChars(textpage));
255e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
256e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_ClosePage(textpage);
257e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  UnloadPage(page);
258e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov}
259e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
260e6986e1e8d4a57987f47c215490cb080a65ee29aSvet GanovTEST_F(FPDFTextEmbeddertest, WebLinks) {
261e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(OpenDocument("testing/resources/weblinks.pdf"));
262e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDF_PAGE page = LoadPage(0);
263e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NE(nullptr, page);
264e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
265e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDF_TEXTPAGE textpage = FPDFText_LoadPage(page);
266e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NE(nullptr, textpage);
267e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
268e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDF_PAGELINK pagelink = FPDFLink_LoadWebLinks(textpage);
269e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NE(nullptr, pagelink);
270e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
271e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Page contains two HTTP-style URLs.
272e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(2, FPDFLink_CountWebLinks(pagelink));
273e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
274e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Only a terminating NUL required for bogus links.
275e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(1, FPDFLink_GetURL(pagelink, 2, nullptr, 0));
276e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(1, FPDFLink_GetURL(pagelink, 1400, nullptr, 0));
277e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(1, FPDFLink_GetURL(pagelink, -1, nullptr, 0));
278e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
279e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Query the number of characters required for each link (incl NUL).
280e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(25, FPDFLink_GetURL(pagelink, 0, nullptr, 0));
281e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(26, FPDFLink_GetURL(pagelink, 1, nullptr, 0));
282e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
283e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  static const char expected_url[] = "http://example.com?q=foo";
284e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  unsigned short fixed_buffer[128];
285e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
286e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Retrieve a link with too small a buffer.  Buffer will not be
287e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // NUL-terminated, but must not be modified past indicated length,
288e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // so pre-fill with a pattern to check write bounds.
289e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  memset(fixed_buffer, 0xbd, sizeof(fixed_buffer));
290e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(1, FPDFLink_GetURL(pagelink, 0, fixed_buffer, 1));
291e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(check_unsigned_shorts(expected_url, fixed_buffer, 1));
292e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0xbdbd, fixed_buffer[1]);
293e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
294e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Check buffer that doesn't have space for a terminating NUL.
295e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  memset(fixed_buffer, 0xbd, sizeof(fixed_buffer));
296e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(sizeof(expected_url) - 1, FPDFLink_GetURL(
297e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      pagelink, 0, fixed_buffer, sizeof(expected_url) - 1));
298e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(check_unsigned_shorts(
299e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      expected_url, fixed_buffer, sizeof(expected_url) - 1));
300e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0xbdbd, fixed_buffer[sizeof(expected_url) - 1]);
301e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
302e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Retreive link with exactly-sized buffer.
303e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  memset(fixed_buffer, 0xbd, sizeof(fixed_buffer));
304e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(sizeof(expected_url), FPDFLink_GetURL(
305e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      pagelink, 0, fixed_buffer, sizeof(expected_url)));
306e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(check_unsigned_shorts(
307e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      expected_url, fixed_buffer, sizeof(expected_url)));
308e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0u, fixed_buffer[sizeof(expected_url) - 1]);
309e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0xbdbd, fixed_buffer[sizeof(expected_url)]);
310e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
311e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Retreive link with ample-sized-buffer.
312e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  memset(fixed_buffer, 0xbd, sizeof(fixed_buffer));
313e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(sizeof(expected_url), FPDFLink_GetURL(
314e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      pagelink, 0, fixed_buffer, 128));
315e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_TRUE(check_unsigned_shorts(
316e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov      expected_url, fixed_buffer, sizeof(expected_url)));
317e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0u, fixed_buffer[sizeof(expected_url) - 1]);
318e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0xbdbd, fixed_buffer[sizeof(expected_url)]);
319e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
320e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Each link rendered in a single rect in this test page.
321e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(1, FPDFLink_CountRects(pagelink, 0));
322e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(1, FPDFLink_CountRects(pagelink, 1));
323e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
324e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Each link rendered in a single rect in this test page.
325e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0, FPDFLink_CountRects(pagelink, -1));
326e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0, FPDFLink_CountRects(pagelink, 2));
327e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(0, FPDFLink_CountRects(pagelink, 10000));
328e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
329e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Check boundary of valid link index with valid rect index.
330e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  double left = 0.0;
331e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  double right = 0.0;
332e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  double top = 0.0;
333e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  double bottom = 0.0;
334e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFLink_GetRect(pagelink, 0, 0, &left, &top, &right, &bottom);
335e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NEAR(50.791, left, 0.001);
336e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NEAR(187.963, right, 0.001);
337e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NEAR(97.624, bottom, 0.001);
338e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_NEAR(108.736, top, 0.001);
339e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
340e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Check that valid link with invalid rect index leaves parameters unchanged.
341e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  left = -1.0;
342e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  right = -1.0;
343e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  top = -1.0;
344e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  bottom = -1.0;
345e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFLink_GetRect(pagelink, 0, 1, &left, &top, &right, &bottom);
346e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(-1.0, left);
347e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(-1.0, right);
348e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(-1.0, bottom);
349e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(-1.0, top);
350e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
351e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  // Check that invalid link index leaves parameters unchanged.
352e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  left = -2.0;
353e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  right = -2.0;
354e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  top = -2.0;
355e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  bottom = -2.0;
356e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFLink_GetRect(pagelink, -1, 0, &left, &top, &right, &bottom);
357e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(-2.0, left);
358e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(-2.0, right);
359e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(-2.0, bottom);
360e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  EXPECT_EQ(-2.0, top);
361e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov
362e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFLink_CloseWebLinks(pagelink);
363e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  FPDFText_ClosePage(textpage);
364e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov  UnloadPage(page);
365e6986e1e8d4a57987f47c215490cb080a65ee29aSvet Ganov}
366