Line data Source code
1 : /*
2 : * Copyright 2015 Google Inc.
3 : *
4 : * Use of this source code is governed by a BSD-style license that can be
5 : * found in the LICENSE file.
6 : */
7 :
8 : #include "SkMD5.h"
9 : #include "SkMilestone.h"
10 : #include "SkPDFMetadata.h"
11 : #include "SkPDFTypes.h"
12 : #include <utility>
13 :
14 : #define SKPDF_STRING(X) SKPDF_STRING_IMPL(X)
15 : #define SKPDF_STRING_IMPL(X) #X
16 : #define SKPDF_PRODUCER "Skia/PDF m" SKPDF_STRING(SK_MILESTONE)
17 : #define SKPDF_CUSTOM_PRODUCER_KEY "ProductionLibrary"
18 :
19 0 : static SkString pdf_date(const SkTime::DateTime& dt) {
20 0 : int timeZoneMinutes = SkToInt(dt.fTimeZoneMinutes);
21 0 : char timezoneSign = timeZoneMinutes >= 0 ? '+' : '-';
22 0 : int timeZoneHours = SkTAbs(timeZoneMinutes) / 60;
23 0 : timeZoneMinutes = SkTAbs(timeZoneMinutes) % 60;
24 : return SkStringPrintf(
25 : "D:%04u%02u%02u%02u%02u%02u%c%02d'%02d'",
26 0 : static_cast<unsigned>(dt.fYear), static_cast<unsigned>(dt.fMonth),
27 0 : static_cast<unsigned>(dt.fDay), static_cast<unsigned>(dt.fHour),
28 0 : static_cast<unsigned>(dt.fMinute),
29 0 : static_cast<unsigned>(dt.fSecond), timezoneSign, timeZoneHours,
30 0 : timeZoneMinutes);
31 : }
32 :
33 : namespace {
34 : static const struct {
35 : const char* const key;
36 : SkString SkDocument::PDFMetadata::*const valuePtr;
37 : } gMetadataKeys[] = {
38 : {"Title", &SkDocument::PDFMetadata::fTitle},
39 : {"Author", &SkDocument::PDFMetadata::fAuthor},
40 : {"Subject", &SkDocument::PDFMetadata::fSubject},
41 : {"Keywords", &SkDocument::PDFMetadata::fKeywords},
42 : {"Creator", &SkDocument::PDFMetadata::fCreator},
43 : };
44 : } // namespace
45 :
46 0 : sk_sp<SkPDFObject> SkPDFMetadata::MakeDocumentInformationDict(
47 : const SkDocument::PDFMetadata& metadata) {
48 0 : auto dict = sk_make_sp<SkPDFDict>();
49 0 : for (const auto keyValuePtr : gMetadataKeys) {
50 0 : const SkString& value = metadata.*(keyValuePtr.valuePtr);
51 0 : if (value.size() > 0) {
52 0 : dict->insertString(keyValuePtr.key, value);
53 : }
54 : }
55 0 : if (metadata.fProducer.isEmpty()) {
56 0 : dict->insertString("Producer", SKPDF_PRODUCER);
57 : } else {
58 0 : dict->insertString("Producer", metadata.fProducer);
59 0 : dict->insertString(SKPDF_CUSTOM_PRODUCER_KEY, SKPDF_PRODUCER);
60 : }
61 0 : if (metadata.fCreation.fEnabled) {
62 0 : dict->insertString("CreationDate",
63 0 : pdf_date(metadata.fCreation.fDateTime));
64 : }
65 0 : if (metadata.fModified.fEnabled) {
66 0 : dict->insertString("ModDate", pdf_date(metadata.fModified.fDateTime));
67 : }
68 0 : return dict;
69 : }
70 :
71 0 : SkPDFMetadata::UUID SkPDFMetadata::CreateUUID(
72 : const SkDocument::PDFMetadata& metadata) {
73 : // The main requirement is for the UUID to be unique; the exact
74 : // format of the data that will be hashed is not important.
75 0 : SkMD5 md5;
76 0 : const char uuidNamespace[] = "org.skia.pdf\n";
77 0 : md5.write(uuidNamespace, strlen(uuidNamespace));
78 0 : double msec = SkTime::GetMSecs();
79 0 : md5.write(&msec, sizeof(msec));
80 : SkTime::DateTime dateTime;
81 0 : SkTime::GetDateTime(&dateTime);
82 0 : md5.write(&dateTime, sizeof(dateTime));
83 0 : if (metadata.fCreation.fEnabled) {
84 0 : md5.write(&metadata.fCreation.fDateTime,
85 0 : sizeof(metadata.fCreation.fDateTime));
86 : }
87 0 : if (metadata.fModified.fEnabled) {
88 0 : md5.write(&metadata.fModified.fDateTime,
89 0 : sizeof(metadata.fModified.fDateTime));
90 : }
91 :
92 0 : for (const auto keyValuePtr : gMetadataKeys) {
93 0 : md5.write(keyValuePtr.key, strlen(keyValuePtr.key));
94 0 : md5.write("\037", 1);
95 0 : const SkString& value = metadata.*(keyValuePtr.valuePtr);
96 0 : md5.write(value.c_str(), value.size());
97 0 : md5.write("\036", 1);
98 : }
99 : SkMD5::Digest digest;
100 0 : md5.finish(digest);
101 : // See RFC 4122, page 6-7.
102 0 : digest.data[6] = (digest.data[6] & 0x0F) | 0x30;
103 0 : digest.data[8] = (digest.data[6] & 0x3F) | 0x80;
104 : static_assert(sizeof(digest) == sizeof(UUID), "uuid_size");
105 : SkPDFMetadata::UUID uuid;
106 0 : memcpy(&uuid, &digest, sizeof(digest));
107 0 : return uuid;
108 : }
109 :
110 0 : sk_sp<SkPDFObject> SkPDFMetadata::MakePdfId(const UUID& doc,
111 : const UUID& instance) {
112 : // /ID [ <81b14aafa313db63dbd6f981e49f94f4>
113 : // <81b14aafa313db63dbd6f981e49f94f4> ]
114 0 : auto array = sk_make_sp<SkPDFArray>();
115 : static_assert(sizeof(SkPDFMetadata::UUID) == 16, "uuid_size");
116 0 : array->appendString(
117 0 : SkString(reinterpret_cast<const char*>(&doc), sizeof(UUID)));
118 0 : array->appendString(
119 0 : SkString(reinterpret_cast<const char*>(&instance), sizeof(UUID)));
120 0 : return array;
121 : }
122 :
123 : #define HEXIFY(INPUT_PTR, OUTPUT_PTR, HEX_STRING, BYTE_COUNT) \
124 : do { \
125 : for (int i = 0; i < (BYTE_COUNT); ++i) { \
126 : uint8_t value = *(INPUT_PTR)++; \
127 : *(OUTPUT_PTR)++ = (HEX_STRING)[value >> 4]; \
128 : *(OUTPUT_PTR)++ = (HEX_STRING)[value & 0xF]; \
129 : } \
130 : } while (false)
131 0 : static SkString uuid_to_string(const SkPDFMetadata::UUID& uuid) {
132 : // 8-4-4-4-12
133 : char buffer[36]; // [32 + 4]
134 : static const char gHex[] = "0123456789abcdef";
135 : SkASSERT(strlen(gHex) == 16);
136 0 : char* ptr = buffer;
137 0 : const uint8_t* data = uuid.fData;
138 0 : HEXIFY(data, ptr, gHex, 4);
139 0 : *ptr++ = '-';
140 0 : HEXIFY(data, ptr, gHex, 2);
141 0 : *ptr++ = '-';
142 0 : HEXIFY(data, ptr, gHex, 2);
143 0 : *ptr++ = '-';
144 0 : HEXIFY(data, ptr, gHex, 2);
145 0 : *ptr++ = '-';
146 0 : HEXIFY(data, ptr, gHex, 6);
147 0 : SkASSERT(ptr == buffer + 36);
148 0 : SkASSERT(data == uuid.fData + 16);
149 0 : return SkString(buffer, 36);
150 : }
151 : #undef HEXIFY
152 :
153 : namespace {
154 0 : class PDFXMLObject final : public SkPDFObject {
155 : public:
156 0 : PDFXMLObject(SkString xml) : fXML(std::move(xml)) {}
157 0 : void emitObject(SkWStream* stream,
158 : const SkPDFObjNumMap& omap) const override {
159 0 : SkPDFDict dict("Metadata");
160 0 : dict.insertName("Subtype", "XML");
161 0 : dict.insertInt("Length", fXML.size());
162 0 : dict.emitObject(stream, omap);
163 : static const char streamBegin[] = " stream\n";
164 0 : stream->write(streamBegin, strlen(streamBegin));
165 : // Do not compress this. The standard requires that a
166 : // program that does not understand PDF can grep for
167 : // "<?xpacket" and extract the entire XML.
168 0 : stream->write(fXML.c_str(), fXML.size());
169 : static const char streamEnd[] = "\nendstream";
170 0 : stream->write(streamEnd, strlen(streamEnd));
171 0 : }
172 :
173 : private:
174 : const SkString fXML;
175 : };
176 : } // namespace
177 :
178 0 : static int count_xml_escape_size(const SkString& input) {
179 0 : int extra = 0;
180 0 : for (size_t i = 0; i < input.size(); ++i) {
181 0 : if (input[i] == '&') {
182 0 : extra += 4; // strlen("&") - strlen("&")
183 0 : } else if (input[i] == '<') {
184 0 : extra += 3; // strlen("<") - strlen("<")
185 : }
186 : }
187 0 : return extra;
188 : }
189 :
190 0 : const SkString escape_xml(const SkString& input,
191 : const char* before = nullptr,
192 : const char* after = nullptr) {
193 0 : if (input.size() == 0) {
194 0 : return input;
195 : }
196 : // "&" --> "&" and "<" --> "<"
197 : // text is assumed to be in UTF-8
198 : // all strings are xml content, not attribute values.
199 0 : size_t beforeLen = before ? strlen(before) : 0;
200 0 : size_t afterLen = after ? strlen(after) : 0;
201 0 : int extra = count_xml_escape_size(input);
202 0 : SkString output(input.size() + extra + beforeLen + afterLen);
203 0 : char* out = output.writable_str();
204 0 : if (before) {
205 0 : strncpy(out, before, beforeLen);
206 0 : out += beforeLen;
207 : }
208 : static const char kAmp[] = "&";
209 : static const char kLt[] = "<";
210 0 : for (size_t i = 0; i < input.size(); ++i) {
211 0 : if (input[i] == '&') {
212 0 : strncpy(out, kAmp, strlen(kAmp));
213 0 : out += strlen(kAmp);
214 0 : } else if (input[i] == '<') {
215 0 : strncpy(out, kLt, strlen(kLt));
216 0 : out += strlen(kLt);
217 : } else {
218 0 : *out++ = input[i];
219 : }
220 : }
221 0 : if (after) {
222 0 : strncpy(out, after, afterLen);
223 0 : out += afterLen;
224 : }
225 : // Validate that we haven't written outside of our string.
226 0 : SkASSERT(out == &output.writable_str()[output.size()]);
227 0 : *out = '\0';
228 0 : return output;
229 : }
230 :
231 0 : sk_sp<SkPDFObject> SkPDFMetadata::MakeXMPObject(
232 : const SkDocument::PDFMetadata& metadata,
233 : const UUID& doc,
234 : const UUID& instance) {
235 : static const char templateString[] =
236 : "<?xpacket begin=\"\" id=\"W5M0MpCehiHzreSzNTczkc9d\"?>\n"
237 : "<x:xmpmeta xmlns:x=\"adobe:ns:meta/\"\n"
238 : " x:xmptk=\"Adobe XMP Core 5.4-c005 78.147326, "
239 : "2012/08/23-13:03:03\">\n"
240 : "<rdf:RDF "
241 : "xmlns:rdf=\"http://www.w3.org/1999/02/22-rdf-syntax-ns#\">\n"
242 : "<rdf:Description rdf:about=\"\"\n"
243 : " xmlns:xmp=\"http://ns.adobe.com/xap/1.0/\"\n"
244 : " xmlns:dc=\"http://purl.org/dc/elements/1.1/\"\n"
245 : " xmlns:xmpMM=\"http://ns.adobe.com/xap/1.0/mm/\"\n"
246 : " xmlns:pdf=\"http://ns.adobe.com/pdf/1.3/\"\n"
247 : " xmlns:pdfaid=\"http://www.aiim.org/pdfa/ns/id/\">\n"
248 : "<pdfaid:part>2</pdfaid:part>\n"
249 : "<pdfaid:conformance>B</pdfaid:conformance>\n"
250 : "%s" // ModifyDate
251 : "%s" // CreateDate
252 : "%s" // xmp:CreatorTool
253 : "<dc:format>application/pdf</dc:format>\n"
254 : "%s" // dc:title
255 : "%s" // dc:description
256 : "%s" // author
257 : "%s" // keywords
258 : "<xmpMM:DocumentID>uuid:%s</xmpMM:DocumentID>\n"
259 : "<xmpMM:InstanceID>uuid:%s</xmpMM:InstanceID>\n"
260 : "%s" // pdf:Producer
261 : "%s" // pdf:Keywords
262 : "</rdf:Description>\n"
263 : "</rdf:RDF>\n"
264 : "</x:xmpmeta>\n" // Note: the standard suggests 4k of padding.
265 : "<?xpacket end=\"w\"?>\n";
266 :
267 0 : SkString creationDate;
268 0 : SkString modificationDate;
269 0 : if (metadata.fCreation.fEnabled) {
270 0 : SkString tmp;
271 0 : metadata.fCreation.fDateTime.toISO8601(&tmp);
272 0 : SkASSERT(0 == count_xml_escape_size(tmp));
273 : // YYYY-mm-ddTHH:MM:SS[+|-]ZZ:ZZ; no need to escape
274 0 : creationDate = SkStringPrintf("<xmp:CreateDate>%s</xmp:CreateDate>\n",
275 0 : tmp.c_str());
276 : }
277 0 : if (metadata.fModified.fEnabled) {
278 0 : SkString tmp;
279 0 : metadata.fModified.fDateTime.toISO8601(&tmp);
280 0 : SkASSERT(0 == count_xml_escape_size(tmp));
281 0 : modificationDate = SkStringPrintf(
282 0 : "<xmp:ModifyDate>%s</xmp:ModifyDate>\n", tmp.c_str());
283 : }
284 : SkString title =
285 : escape_xml(metadata.fTitle,
286 : "<dc:title><rdf:Alt><rdf:li xml:lang=\"x-default\">",
287 0 : "</rdf:li></rdf:Alt></dc:title>\n");
288 : SkString author =
289 : escape_xml(metadata.fAuthor, "<dc:creator><rdf:Bag><rdf:li>",
290 0 : "</rdf:li></rdf:Bag></dc:creator>\n");
291 : // TODO: in theory, XMP can support multiple authors. Split on a delimiter?
292 : SkString subject = escape_xml(
293 : metadata.fSubject,
294 : "<dc:description><rdf:Alt><rdf:li xml:lang=\"x-default\">",
295 0 : "</rdf:li></rdf:Alt></dc:description>\n");
296 : SkString keywords1 =
297 : escape_xml(metadata.fKeywords, "<dc:subject><rdf:Bag><rdf:li>",
298 0 : "</rdf:li></rdf:Bag></dc:subject>\n");
299 : SkString keywords2 = escape_xml(metadata.fKeywords, "<pdf:Keywords>",
300 0 : "</pdf:Keywords>\n");
301 : // TODO: in theory, keywords can be a list too.
302 :
303 0 : SkString producer("<pdf:Producer>" SKPDF_PRODUCER "</pdf:Producer>\n");
304 0 : if (!metadata.fProducer.isEmpty()) {
305 : // TODO: register a developer prefix to make
306 : // <skia:SKPDF_CUSTOM_PRODUCER_KEY> a real XML tag.
307 0 : producer = escape_xml(
308 : metadata.fProducer, "<pdf:Producer>",
309 : "</pdf:Producer>\n<!-- <skia:" SKPDF_CUSTOM_PRODUCER_KEY ">"
310 0 : SKPDF_PRODUCER "</skia:" SKPDF_CUSTOM_PRODUCER_KEY "> -->\n");
311 : }
312 :
313 : SkString creator = escape_xml(metadata.fCreator, "<xmp:CreatorTool>",
314 0 : "</xmp:CreatorTool>\n");
315 0 : SkString documentID = uuid_to_string(doc); // no need to escape
316 0 : SkASSERT(0 == count_xml_escape_size(documentID));
317 0 : SkString instanceID = uuid_to_string(instance);
318 0 : SkASSERT(0 == count_xml_escape_size(instanceID));
319 0 : return sk_make_sp<PDFXMLObject>(SkStringPrintf(
320 : templateString, modificationDate.c_str(), creationDate.c_str(),
321 : creator.c_str(), title.c_str(), subject.c_str(), author.c_str(),
322 : keywords1.c_str(), documentID.c_str(), instanceID.c_str(),
323 0 : producer.c_str(), keywords2.c_str()));
324 : }
325 :
326 : #undef SKPDF_CUSTOM_PRODUCER_KEY
327 : #undef SKPDF_PRODUCER
328 : #undef SKPDF_STRING
329 : #undef SKPDF_STRING_IMPL
|