1 // Copyright 2018 The PDFium Authors
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/helpers/dump.h"
6
7 #include <limits.h>
8 #include <string.h>
9
10 #include <algorithm>
11 #include <functional>
12 #include <iterator>
13 #include <string>
14 #include <utility>
15
16 #include "public/cpp/fpdf_scopers.h"
17 #include "public/fpdf_doc.h"
18 #include "public/fpdf_transformpage.h"
19 #include "testing/fx_string_testhelpers.h"
20
21 using GetBoxInfoFunc =
22 std::function<bool(FPDF_PAGE, float*, float*, float*, float*)>;
23
24 namespace {
25
ConvertToWString(const unsigned short * buf,unsigned long buf_size)26 std::wstring ConvertToWString(const unsigned short* buf,
27 unsigned long buf_size) {
28 std::wstring result;
29 result.reserve(buf_size);
30 std::copy(buf, buf + buf_size, std::back_inserter(result));
31 return result;
32 }
33
DumpBoxInfo(GetBoxInfoFunc func,const char * box_type,FPDF_PAGE page,int page_idx)34 void DumpBoxInfo(GetBoxInfoFunc func,
35 const char* box_type,
36 FPDF_PAGE page,
37 int page_idx) {
38 FS_RECTF rect;
39 bool ret = func(page, &rect.left, &rect.bottom, &rect.right, &rect.top);
40 if (!ret) {
41 printf("Page %d: No %s.\n", page_idx, box_type);
42 return;
43 }
44 printf("Page %d: %s: %0.2f %0.2f %0.2f %0.2f\n", page_idx, box_type,
45 rect.left, rect.bottom, rect.right, rect.top);
46 }
47
DumpStructureElementAttributeValues(FPDF_STRUCTELEMENT_ATTR_VALUE attr_value,const char * name,int indent)48 void DumpStructureElementAttributeValues(
49 FPDF_STRUCTELEMENT_ATTR_VALUE attr_value,
50 const char* name,
51 int indent) {
52 if (!attr_value) {
53 printf("%*s FPDF_StructElement_Attr_GetValue failed for %s\n", indent, "",
54 name);
55 return;
56 }
57
58 FPDF_OBJECT_TYPE type = FPDF_StructElement_Attr_GetType(attr_value);
59 switch (type) {
60 case FPDF_OBJECT_BOOLEAN: {
61 int value;
62 if (FPDF_StructElement_Attr_GetBooleanValue(attr_value, &value)) {
63 printf("%*s %s: %d\n", indent, "", name, value);
64 } else {
65 printf("%*s %s: Failed FPDF_StructElement_Attr_GetBooleanValue\n",
66 indent, "", name);
67 }
68 break;
69 }
70 case FPDF_OBJECT_NUMBER: {
71 float value;
72 if (FPDF_StructElement_Attr_GetNumberValue(attr_value, &value)) {
73 printf("%*s %s: %f\n", indent, "", name, value);
74 } else {
75 printf("%*s %s: Failed FPDF_StructElement_Attr_GetNumberValue\n",
76 indent, "", name);
77 }
78 break;
79 }
80 case FPDF_OBJECT_STRING:
81 case FPDF_OBJECT_NAME: {
82 static const size_t kBufSize = 1024;
83 unsigned short buffer[kBufSize];
84 unsigned long len;
85 if (FPDF_StructElement_Attr_GetStringValue(attr_value, buffer,
86 sizeof(buffer), &len)) {
87 printf("%*s %s: %ls\n", indent, "", name,
88 ConvertToWString(buffer, len).c_str());
89 } else {
90 printf("%*s %s: Failed FPDF_StructElement_Attr_GetStringValue\n",
91 indent, "", name);
92 }
93 break;
94 }
95 case FPDF_OBJECT_ARRAY: {
96 printf("%*s %s:\n", indent, "", name);
97 int count = FPDF_StructElement_Attr_CountChildren(attr_value);
98 for (int i = 0; i < count; ++i) {
99 DumpStructureElementAttributeValues(
100 FPDF_StructElement_Attr_GetChildAtIndex(attr_value, i), name,
101 indent + 2);
102 }
103 break;
104 }
105 case FPDF_OBJECT_UNKNOWN: {
106 printf("%*s %s: FPDF_OBJECT_UNKNOWN\n", indent, "", name);
107 break;
108 }
109 default: {
110 printf("%*s %s: NOT_YET_IMPLEMENTED: %d\n", indent, "", name, type);
111 break;
112 }
113 }
114 }
115
DumpStructureElementAttributes(FPDF_STRUCTELEMENT_ATTR attr,int indent)116 void DumpStructureElementAttributes(FPDF_STRUCTELEMENT_ATTR attr, int indent) {
117 static const size_t kBufSize = 1024;
118 int count = FPDF_StructElement_Attr_GetCount(attr);
119 for (int i = 0; i < count; ++i) {
120 char name[kBufSize];
121 unsigned long len;
122 if (!FPDF_StructElement_Attr_GetName(attr, i, name, sizeof(name), &len)) {
123 printf("%*s FPDF_StructElement_Attr_GetName failed for %d\n", indent, "",
124 i);
125 continue;
126 }
127
128 DumpStructureElementAttributeValues(
129 FPDF_StructElement_Attr_GetValue(attr, name), name, indent);
130 }
131 }
132
133 } // namespace
134
DumpChildStructure(FPDF_STRUCTELEMENT child,int indent)135 void DumpChildStructure(FPDF_STRUCTELEMENT child, int indent) {
136 static const size_t kBufSize = 1024;
137 unsigned short buf[kBufSize];
138 unsigned long len = FPDF_StructElement_GetType(child, buf, kBufSize);
139 if (len > 0) {
140 printf("%*s S: %ls\n", indent * 2, "", ConvertToWString(buf, len).c_str());
141 }
142
143 int attr_count = FPDF_StructElement_GetAttributeCount(child);
144 for (int i = 0; i < attr_count; i++) {
145 FPDF_STRUCTELEMENT_ATTR child_attr =
146 FPDF_StructElement_GetAttributeAtIndex(child, i);
147 if (!child_attr) {
148 continue;
149 }
150 printf("%*s A[%d]:\n", indent * 2, "", i);
151 DumpStructureElementAttributes(child_attr, indent * 2 + 2);
152 }
153
154 memset(buf, 0, sizeof(buf));
155 len = FPDF_StructElement_GetActualText(child, buf, kBufSize);
156 if (len > 0) {
157 printf("%*s ActualText: %ls\n", indent * 2, "",
158 ConvertToWString(buf, len).c_str());
159 }
160
161 memset(buf, 0, sizeof(buf));
162 len = FPDF_StructElement_GetAltText(child, buf, kBufSize);
163 if (len > 0) {
164 printf("%*s AltText: %ls\n", indent * 2, "",
165 ConvertToWString(buf, len).c_str());
166 }
167
168 memset(buf, 0, sizeof(buf));
169 len = FPDF_StructElement_GetID(child, buf, kBufSize);
170 if (len > 0) {
171 printf("%*s ID: %ls\n", indent * 2, "", ConvertToWString(buf, len).c_str());
172 }
173
174 memset(buf, 0, sizeof(buf));
175 len = FPDF_StructElement_GetLang(child, buf, kBufSize);
176 if (len > 0) {
177 printf("%*s Lang: %ls\n", indent * 2, "",
178 ConvertToWString(buf, len).c_str());
179 }
180
181 const int mcid_count = FPDF_StructElement_GetMarkedContentIdCount(child);
182 for (int i = 0; i < mcid_count; ++i) {
183 int mcid = FPDF_StructElement_GetMarkedContentIdAtIndex(child, i);
184 if (mcid != -1) {
185 printf("%*s MCID%d: %d\n", indent * 2, "", i, mcid);
186 }
187 }
188
189 FPDF_STRUCTELEMENT parent = FPDF_StructElement_GetParent(child);
190 if (parent) {
191 memset(buf, 0, sizeof(buf));
192 len = FPDF_StructElement_GetID(parent, buf, kBufSize);
193 if (len > 0) {
194 printf("%*s Parent ID: %ls\n", indent * 2, "",
195 ConvertToWString(buf, len).c_str());
196 }
197 }
198
199 memset(buf, 0, sizeof(buf));
200 len = FPDF_StructElement_GetTitle(child, buf, kBufSize);
201 if (len > 0) {
202 printf("%*s Title: %ls\n", indent * 2, "",
203 ConvertToWString(buf, len).c_str());
204 }
205
206 memset(buf, 0, sizeof(buf));
207 len = FPDF_StructElement_GetObjType(child, buf, kBufSize);
208 if (len > 0) {
209 printf("%*s Type: %ls\n", indent * 2, "",
210 ConvertToWString(buf, len).c_str());
211 }
212
213 for (int i = 0; i < FPDF_StructElement_CountChildren(child); ++i) {
214 FPDF_STRUCTELEMENT sub_child = FPDF_StructElement_GetChildAtIndex(child, i);
215 // If the child is not an Element then this will return null. This can
216 // happen if the element is things like an object reference or a stream.
217 if (!sub_child) {
218 continue;
219 }
220
221 DumpChildStructure(sub_child, indent + 1);
222 }
223 }
224
DumpPageInfo(FPDF_PAGE page,int page_idx)225 void DumpPageInfo(FPDF_PAGE page, int page_idx) {
226 DumpBoxInfo(&FPDFPage_GetMediaBox, "MediaBox", page, page_idx);
227 DumpBoxInfo(&FPDFPage_GetCropBox, "CropBox", page, page_idx);
228 DumpBoxInfo(&FPDFPage_GetBleedBox, "BleedBox", page, page_idx);
229 DumpBoxInfo(&FPDFPage_GetTrimBox, "TrimBox", page, page_idx);
230 DumpBoxInfo(&FPDFPage_GetArtBox, "ArtBox", page, page_idx);
231 }
232
DumpPageStructure(FPDF_PAGE page,int page_idx)233 void DumpPageStructure(FPDF_PAGE page, int page_idx) {
234 ScopedFPDFStructTree tree(FPDF_StructTree_GetForPage(page));
235 if (!tree) {
236 fprintf(stderr, "Failed to load struct tree for page %d\n", page_idx);
237 return;
238 }
239
240 printf("Structure Tree for Page %d\n", page_idx);
241 for (int i = 0; i < FPDF_StructTree_CountChildren(tree.get()); ++i) {
242 FPDF_STRUCTELEMENT child = FPDF_StructTree_GetChildAtIndex(tree.get(), i);
243 if (!child) {
244 fprintf(stderr, "Failed to load child %d for page %d\n", i, page_idx);
245 continue;
246 }
247 DumpChildStructure(child, 0);
248 }
249 printf("\n\n");
250 }
251
DumpMetaData(FPDF_DOCUMENT doc)252 void DumpMetaData(FPDF_DOCUMENT doc) {
253 static constexpr const char* kMetaTags[] = {
254 "Title", "Author", "Subject", "Keywords",
255 "Creator", "Producer", "CreationDate", "ModDate"};
256 for (const char* meta_tag : kMetaTags) {
257 char meta_buffer[4096];
258 unsigned long len =
259 FPDF_GetMetaText(doc, meta_tag, meta_buffer, sizeof(meta_buffer));
260 if (!len) {
261 continue;
262 }
263
264 auto* meta_string = reinterpret_cast<unsigned short*>(meta_buffer);
265 printf("%-12s = %ls (%lu bytes)\n", meta_tag,
266 GetPlatformWString(meta_string).c_str(), len);
267 }
268 }
269