SkPDF: Optionally output PDF/A-2b archive format.
Note: this format does not yet pass validation tests.
Add skia_pdf_generate_pdfa GYP flag. Default to off for now.
PDF/A files are not reproducable, so they make correctness
testing harder.
Turn the Metadata struct into te SkPDFMetadata struct. This
splits out a lot of functionality around both kinds of metadata.
When PDF/A is used, add an ID entry to the trailer.
Add SkPDFObjNumMap::addObjectRecursively.
Test with
GYP_DEFINES=skia_pdf_generate_pdfa=1 bin/sync-and-gyp
ninja -C out/Release dm
out/Release/dm --config pdf --src skp gm -w /tmp/dm
With skia_pdf_generate_pdfa=0, all PDFs generated from GMs and
SKPs are identical. With skia_pdf_generate_pdfa=1, all PDFs
generated from GMs and SKPs render identically in Pdfium.
BUG=skia:3110
Review URL: https://codereview.chromium.org/1394263003
diff --git a/gyp/common_variables.gypi b/gyp/common_variables.gypi
index 57fad1d..c84a8e5 100644
--- a/gyp/common_variables.gypi
+++ b/gyp/common_variables.gypi
@@ -225,6 +225,7 @@
'skia_win_ltcg%': 1,
'skia_osx_deployment_target%': '<(skia_osx_deployment_target)',
'skia_pdf%': '<(skia_pdf)',
+ 'skia_pdf_generate_pdfa%': 0, # emit larger PDF/A-2b file
'skia_profile_enabled%': '<(skia_profile_enabled)',
'skia_shared_lib%': '<(skia_shared_lib)',
'skia_opencl%': '<(skia_opencl)',
diff --git a/gyp/pdf.gyp b/gyp/pdf.gyp
index ee30488..77d1922 100644
--- a/gyp/pdf.gyp
+++ b/gyp/pdf.gyp
@@ -40,6 +40,7 @@
skia_os in ["win", "android", "linux", "chromeos", "mac"]',
{ 'dependencies': [ 'sfntly.gyp:sfntly' ] }
],
+ [ 'skia_pdf_generate_pdfa', { 'defines': ['SK_PDF_GENERATE_PDFA'] } ],
[ 'skia_android_framework', {
# Add SFTNLY support for PDF (which in turns depends on ICU)
'include_dirs': [
diff --git a/gyp/pdf.gypi b/gyp/pdf.gypi
index f114fa7..888613e 100644
--- a/gyp/pdf.gypi
+++ b/gyp/pdf.gypi
@@ -29,6 +29,8 @@
'<(skia_src_path)/pdf/SkPDFFormXObject.h',
'<(skia_src_path)/pdf/SkPDFGraphicState.cpp',
'<(skia_src_path)/pdf/SkPDFGraphicState.h',
+ '<(skia_src_path)/pdf/SkPDFMetadata.cpp',
+ '<(skia_src_path)/pdf/SkPDFMetadata.h',
'<(skia_src_path)/pdf/SkPDFResourceDict.cpp',
'<(skia_src_path)/pdf/SkPDFResourceDict.h',
'<(skia_src_path)/pdf/SkPDFShader.cpp',
diff --git a/src/doc/SkDocument_PDF.cpp b/src/doc/SkDocument_PDF.cpp
index fb22f18..ff7a038 100644
--- a/src/doc/SkDocument_PDF.cpp
+++ b/src/doc/SkDocument_PDF.cpp
@@ -13,6 +13,7 @@
#include "SkPDFTypes.h"
#include "SkPDFUtils.h"
#include "SkStream.h"
+#include "SkPDFMetadata.h"
class SkPDFDict;
@@ -30,15 +31,18 @@
SkPDFObject* docCatalog,
int64_t objCount,
int32_t xRefFileOffset,
- SkPDFDict* info) {
+ SkPDFObject* info /* take ownership */,
+ SkPDFObject* id /* take ownership */) {
SkPDFDict trailerDict;
- // TODO(vandebo): Linearized format will take a Prev entry too.
- // TODO(vandebo): PDF/A requires an ID entry.
+ // TODO(http://crbug.com/80908): Linearized format will take a
+ // Prev entry too.
trailerDict.insertInt("Size", int(objCount));
trailerDict.insertObjRef("Root", SkRef(docCatalog));
SkASSERT(info);
- trailerDict.insertObjRef("Info", SkRef(info));
-
+ trailerDict.insertObjRef("Info", info);
+ if (id) {
+ trailerDict.insertObject("ID", id);
+ }
stream->writeText("trailer\n");
trailerDict.emitObject(stream, objNumMap, substitutes);
stream->writeText("\nstartxref\n");
@@ -162,49 +166,8 @@
}
}
-struct Metadata {
- SkTArray<SkDocument::Attribute> fInfo;
- SkAutoTDelete<const SkTime::DateTime> fCreation;
- SkAutoTDelete<const SkTime::DateTime> fModified;
-};
-
-static SkString pdf_date(const SkTime::DateTime& dt) {
- int timeZoneMinutes = SkToInt(dt.fTimeZoneMinutes);
- char timezoneSign = timeZoneMinutes >= 0 ? '+' : '-';
- int timeZoneHours = SkTAbs(timeZoneMinutes) / 60;
- timeZoneMinutes = SkTAbs(timeZoneMinutes) % 60;
- return SkStringPrintf(
- "D:%04u%02u%02u%02u%02u%02u%c%02d'%02d'",
- static_cast<unsigned>(dt.fYear), static_cast<unsigned>(dt.fMonth),
- static_cast<unsigned>(dt.fDay), static_cast<unsigned>(dt.fHour),
- static_cast<unsigned>(dt.fMinute),
- static_cast<unsigned>(dt.fSecond), timezoneSign, timeZoneHours,
- timeZoneMinutes);
-}
-
-SkPDFDict* create_document_information_dict(const Metadata& metadata) {
- SkAutoTUnref<SkPDFDict> dict(new SkPDFDict);
- static const char* keys[] = {
- "Title", "Author", "Subject", "Keywords", "Creator" };
- for (const char* key : keys) {
- for (const SkDocument::Attribute& keyValue : metadata.fInfo) {
- if (keyValue.fKey.equals(key)) {
- dict->insertString(key, keyValue.fValue);
- }
- }
- }
- dict->insertString("Producer", "Skia/PDF");
- if (metadata.fCreation) {
- dict->insertString("CreationDate", pdf_date(*metadata.fCreation.get()));
- }
- if (metadata.fModified) {
- dict->insertString("ModDate", pdf_date(*metadata.fModified.get()));
- }
- return dict.detach();
-}
-
static bool emit_pdf_document(const SkTDArray<const SkPDFDevice*>& pageDevices,
- const Metadata& metadata,
+ const SkPDFMetadata& metadata,
SkWStream* stream) {
if (pageDevices.isEmpty()) {
return false;
@@ -222,9 +185,37 @@
pages.push(page.detach());
}
- SkTDArray<SkPDFDict*> pageTree;
SkAutoTUnref<SkPDFDict> docCatalog(new SkPDFDict("Catalog"));
+ SkAutoTUnref<SkPDFObject> infoDict(
+ metadata.createDocumentInformationDict());
+
+ SkAutoTUnref<SkPDFObject> id, xmp;
+#ifdef SK_PDF_GENERATE_PDFA
+ SkPDFMetadata::UUID uuid = metadata.uuid();
+ // We use the same UUID for Document ID and Instance ID since this
+ // is the first revision of this document (and Skia does not
+ // support revising existing PDF documents).
+ // If we are not in PDF/A mode, don't use a UUID since testing
+ // works best with reproducible outputs.
+ id.reset(SkPDFMetadata::CreatePdfId(uuid, uuid));
+ xmp.reset(metadata.createXMPObject(uuid, uuid));
+ docCatalog->insertObjRef("Metadata", xmp.detach());
+
+ // sRGB is specified by HTML, CSS, and SVG.
+ SkAutoTUnref<SkPDFDict> outputIntent(new SkPDFDict("OutputIntent"));
+ outputIntent->insertName("S", "GTS_PDFA1");
+ outputIntent->insertString("RegistryName", "http://www.color.org");
+ outputIntent->insertString("OutputConditionIdentifier",
+ "sRGB IEC61966-2.1");
+ SkAutoTUnref<SkPDFArray> intentArray(new SkPDFArray);
+ intentArray->appendObject(outputIntent.detach());
+ // Don't specify OutputIntents if we are not in PDF/A mode since
+ // no one has ever asked for this feature.
+ docCatalog->insertObject("OutputIntents", intentArray.detach());
+#endif
+
+ SkTDArray<SkPDFDict*> pageTree;
SkPDFDict* pageTreeRoot;
generate_page_tree(pages, &pageTree, &pageTreeRoot);
docCatalog->insertObjRef("Pages", SkRef(pageTreeRoot));
@@ -233,28 +224,13 @@
docCatalog->insertObjRef("Dests", dests.detach());
}
- /* TODO(vandebo): output intent
- SkAutoTUnref<SkPDFDict> outputIntent = new SkPDFDict("OutputIntent");
- outputIntent->insertName("S", "GTS_PDFA1");
- outputIntent->insertString("OutputConditionIdentifier", "sRGB");
- SkAutoTUnref<SkPDFArray> intentArray(new SkPDFArray);
- intentArray->appendObject(SkRef(outputIntent.get()));
- docCatalog->insertObject("OutputIntent", intentArray.detach());
- */
-
// Build font subsetting info before proceeding.
SkPDFSubstituteMap substitutes;
perform_font_subsetting(pageDevices, &substitutes);
- SkAutoTUnref<SkPDFDict> infoDict(
- create_document_information_dict(metadata));
SkPDFObjNumMap objNumMap;
- if (objNumMap.addObject(infoDict)) {
- infoDict->addResources(&objNumMap, substitutes);
- }
- if (objNumMap.addObject(docCatalog.get())) {
- docCatalog->addResources(&objNumMap, substitutes);
- }
+ objNumMap.addObjectRecursively(infoDict, substitutes);
+ objNumMap.addObjectRecursively(docCatalog.get(), substitutes);
size_t baseOffset = stream->bytesWritten();
emit_pdf_header(stream);
SkTDArray<int32_t> offsets;
@@ -286,7 +262,7 @@
stream->writeText(" 00000 n \n");
}
emit_pdf_footer(stream, objNumMap, substitutes, docCatalog.get(), objCount,
- xRefFileOffset, infoDict);
+ xRefFileOffset, infoDict.detach(), id.detach());
// The page tree has both child and parent pointers, so it creates a
// reference cycle. We must clear that cycle to properly reclaim memory.
@@ -404,7 +380,7 @@
SkTDArray<const SkPDFDevice*> fPageDevices;
SkAutoTUnref<SkCanvas> fCanvas;
SkScalar fRasterDpi;
- Metadata fMetadata;
+ SkPDFMetadata fMetadata;
};
} // namespace
///////////////////////////////////////////////////////////////////////////////
diff --git a/src/pdf/SkPDFBitmap.cpp b/src/pdf/SkPDFBitmap.cpp
index 1978040..52b65c0 100644
--- a/src/pdf/SkPDFBitmap.cpp
+++ b/src/pdf/SkPDFBitmap.cpp
@@ -364,19 +364,17 @@
namespace {
class PDFDefaultBitmap : public SkPDFObject {
public:
- void emitObject(SkWStream* stream,
+ void emitObject(SkWStream* stream,
const SkPDFObjNumMap& objNumMap,
- const SkPDFSubstituteMap& substitutes) const override {
- emit_image_xobject(stream, fImage, false, fSMask, objNumMap, substitutes);
+ const SkPDFSubstituteMap& subs) const override {
+ emit_image_xobject(stream, fImage, false, fSMask, objNumMap, subs);
}
void addResources(SkPDFObjNumMap* catalog,
- const SkPDFSubstituteMap& substitutes) const override {
+ const SkPDFSubstituteMap& subs) const override {
if (fSMask.get()) {
- SkPDFObject* obj = substitutes.getSubstitute(fSMask.get());
+ SkPDFObject* obj = subs.getSubstitute(fSMask.get());
SkASSERT(obj);
- if (catalog->addObject(obj)) {
- obj->addResources(catalog, substitutes);
- }
+ catalog->addObjectRecursively(obj, subs);
}
}
PDFDefaultBitmap(const SkImage* image, SkPDFObject* smask)
diff --git a/src/pdf/SkPDFMetadata.cpp b/src/pdf/SkPDFMetadata.cpp
new file mode 100644
index 0000000..6a68a48
--- /dev/null
+++ b/src/pdf/SkPDFMetadata.cpp
@@ -0,0 +1,347 @@
+/*
+ * Copyright 2015 Google Inc.
+ *
+ * Use of this source code is governed by a BSD-style license that can be
+ * found in the LICENSE file.
+ */
+
+#include "SkPDFMetadata.h"
+#include "SkMD5.h"
+#include "SkPDFTypes.h"
+
+SkPDFMetadata::UUID SkPDFMetadata::uuid() const {
+ // The main requirement is for the UUID to be unique; the exact
+ // format of the data that will be hashed is not important.
+ SkMD5 md5;
+ const char uuidNamespace[] = "org.skia.pdf\n";
+ md5.write(uuidNamespace, strlen(uuidNamespace));
+ SkMSec msec = SkTime::GetMSecs();
+ md5.write(&msec, sizeof(msec));
+ SkTime::DateTime dateTime;
+ SkTime::GetDateTime(&dateTime);
+ md5.write(&dateTime, sizeof(dateTime));
+ if (fCreation) {
+ md5.write(fCreation.get(), sizeof(fCreation));
+ }
+ if (fModified) {
+ md5.write(fModified.get(), sizeof(fModified));
+ }
+ for (const auto& kv : fInfo) {
+ md5.write(kv.fKey.c_str(), kv.fKey.size());
+ md5.write("\037", 1);
+ md5.write(kv.fValue.c_str(), kv.fValue.size());
+ md5.write("\036", 1);
+ }
+ SkMD5::Digest digest;
+ md5.finish(digest);
+ // See RFC 4122, page 6-7.
+ digest.data[6] = (digest.data[6] & 0x0F) | 0x30;
+ digest.data[8] = (digest.data[6] & 0x3F) | 0x80;
+ static_assert(sizeof(digest) == sizeof(UUID), "uuid_size");
+ SkPDFMetadata::UUID uuid;
+ memcpy(&uuid, &digest, sizeof(digest));
+ return uuid;
+}
+
+SkPDFObject* SkPDFMetadata::CreatePdfId(const UUID& doc, const UUID& instance) {
+ // /ID [ <81b14aafa313db63dbd6f981e49f94f4>
+ // <81b14aafa313db63dbd6f981e49f94f4> ]
+ SkAutoTUnref<SkPDFArray> array(new SkPDFArray);
+ static_assert(sizeof(UUID) == 16, "uuid_size");
+ array->appendString(
+ SkString(reinterpret_cast<const char*>(&doc), sizeof(UUID)));
+ array->appendString(
+ SkString(reinterpret_cast<const char*>(&instance), sizeof(UUID)));
+ return array.detach();
+}
+
+static SkString pdf_date(const SkTime::DateTime& dt) {
+ int timeZoneMinutes = SkToInt(dt.fTimeZoneMinutes);
+ char timezoneSign = timeZoneMinutes >= 0 ? '+' : '-';
+ int timeZoneHours = SkTAbs(timeZoneMinutes) / 60;
+ timeZoneMinutes = SkTAbs(timeZoneMinutes) % 60;
+ return SkStringPrintf(
+ "D:%04u%02u%02u%02u%02u%02u%c%02d'%02d'",
+ static_cast<unsigned>(dt.fYear), static_cast<unsigned>(dt.fMonth),
+ static_cast<unsigned>(dt.fDay), static_cast<unsigned>(dt.fHour),
+ static_cast<unsigned>(dt.fMinute),
+ static_cast<unsigned>(dt.fSecond), timezoneSign, timeZoneHours,
+ timeZoneMinutes);
+}
+
+SkPDFObject* SkPDFMetadata::createDocumentInformationDict() const {
+ SkAutoTUnref<SkPDFDict> dict(new SkPDFDict);
+ static const char* keys[] = {
+ "Title", "Author", "Subject", "Keywords", "Creator"};
+ for (const char* key : keys) {
+ for (const SkDocument::Attribute& keyValue : fInfo) {
+ if (keyValue.fKey.equals(key)) {
+ dict->insertString(key, keyValue.fValue);
+ }
+ }
+ }
+ dict->insertString("Producer", "Skia/PDF");
+ if (fCreation) {
+ dict->insertString("CreationDate", pdf_date(*fCreation.get()));
+ }
+ if (fModified) {
+ dict->insertString("ModDate", pdf_date(*fModified.get()));
+ }
+ return dict.detach();
+}
+
+#ifdef SK_PDF_GENERATE_PDFA
+// Improvement on SkStringPrintf to allow for arbitrarily long output.
+// TODO: replace SkStringPrintf.
+static SkString sk_string_printf(const char* format, ...) {
+#ifdef SK_BUILD_FOR_WIN
+ va_list args;
+ va_start(args, format);
+ char buffer[1024];
+ int length = _vsnprintf_s(buffer, sizeof(buffer), _TRUNCATE, format, args);
+ va_end(args);
+ if (length >= 0 && length < (int)sizeof(buffer)) {
+ return SkString(buffer, length);
+ }
+ va_start(args, format);
+ length = _vscprintf(format, args);
+ va_end(args);
+
+ SkString string((size_t)length);
+ va_start(args, format);
+ SkDEBUGCODE(int check = ) _vsnprintf_s(string.writable_str(), length + 1,
+ _TRUNCATE, format, args);
+ va_end(args);
+ SkASSERT(check == length);
+ SkASSERT(string[length] == '\0');
+ return skstd::move(string);
+#else // C99/C++11 standard vsnprintf
+ // TODO: When all compilers support this, remove windows-specific code.
+ va_list args;
+ va_start(args, format);
+ char buffer[1024];
+ int length = vsnprintf(buffer, sizeof(buffer), format, args);
+ va_end(args);
+ if (length < 0) {
+ return SkString();
+ }
+ if (length < (int)sizeof(buffer)) {
+ return SkString(buffer, length);
+ }
+ SkString string((size_t)length);
+ va_start(args, format);
+ SkDEBUGCODE(int check = )
+ vsnprintf(string.writable_str(), length + 1, format, args);
+ va_end(args);
+ SkASSERT(check == length);
+ SkASSERT(string[length] == '\0');
+ return skstd::move(string);
+#endif
+}
+
+static const SkString get(const SkTArray<SkDocument::Attribute>& info,
+ const char* key) {
+ for (const auto& keyValue : info) {
+ if (keyValue.fKey.equals(key)) {
+ return keyValue.fValue;
+ }
+ }
+ return SkString();
+}
+
+#define HEXIFY(INPUT_PTR, OUTPUT_PTR, HEX_STRING, BYTE_COUNT) \
+ do { \
+ for (int i = 0; i < (BYTE_COUNT); ++i) { \
+ uint8_t value = *(INPUT_PTR)++; \
+ *(OUTPUT_PTR)++ = (HEX_STRING)[value >> 4]; \
+ *(OUTPUT_PTR)++ = (HEX_STRING)[value & 0xF]; \
+ } \
+ } while (false)
+static SkString uuid_to_string(const SkPDFMetadata::UUID& uuid) {
+ // 8-4-4-4-12
+ char buffer[36]; // [32 + 4]
+ static const char gHex[] = "0123456789abcdef";
+ SkASSERT(strlen(gHex) == 16);
+ char* ptr = buffer;
+ const uint8_t* data = uuid.fData;
+ HEXIFY(data, ptr, gHex, 4);
+ *ptr++ = '-';
+ HEXIFY(data, ptr, gHex, 2);
+ *ptr++ = '-';
+ HEXIFY(data, ptr, gHex, 2);
+ *ptr++ = '-';
+ HEXIFY(data, ptr, gHex, 2);
+ *ptr++ = '-';
+ HEXIFY(data, ptr, gHex, 6);
+ SkASSERT(ptr == buffer + 36);
+ SkASSERT(data == uuid.fData + 16);
+ return SkString(buffer, 36);
+}
+#undef HEXIFY
+
+namespace {
+class PDFXMLObject : public SkPDFObject {
+public:
+ PDFXMLObject(SkString xml) : fXML(skstd::move(xml)) {}
+ void emitObject(SkWStream* stream,
+ const SkPDFObjNumMap& omap,
+ const SkPDFSubstituteMap& smap) const override {
+ SkPDFDict dict("Metadata");
+ dict.insertName("Subtype", "XML");
+ dict.insertInt("Length", fXML.size());
+ dict.emitObject(stream, omap, smap);
+ static const char streamBegin[] = " stream\n";
+ stream->write(streamBegin, strlen(streamBegin));
+ // Do not compress this. The standard requires that a
+ // program that does not understand PDF can grep for
+ // "<?xpacket" and extracť the entire XML.
+ stream->write(fXML.c_str(), fXML.size());
+ static const char streamEnd[] = "\nendstream";
+ stream->write(streamEnd, strlen(streamEnd));
+ }
+
+private:
+ const SkString fXML;
+};
+} // namespace
+
+static int count_xml_escape_size(const SkString& input) {
+ int extra = 0;
+ for (size_t i = 0; i < input.size(); ++i) {
+ if (input[i] == '&') {
+ extra += 4; // strlen("&") - strlen("&")
+ } else if (input[i] == '<') {
+ extra += 3; // strlen("<") - strlen("<")
+ }
+ }
+ return extra;
+}
+
+const SkString escape_xml(const SkString& input,
+ const char* before = nullptr,
+ const char* after = nullptr) {
+ if (input.size() == 0) {
+ return input;
+ }
+ // "&" --> "&" and "<" --> "<"
+ // text is assumed to be in UTF-8
+ // all strings are xml content, not attribute values.
+ size_t beforeLen = before ? strlen(before) : 0;
+ size_t afterLen = after ? strlen(after) : 0;
+ int extra = count_xml_escape_size(input);
+ SkString output(input.size() + extra + beforeLen + afterLen);
+ char* out = output.writable_str();
+ if (before) {
+ strncpy(out, before, beforeLen);
+ out += beforeLen;
+ }
+ static const char kAmp[] = "&";
+ static const char kLt[] = "<";
+ for (size_t i = 0; i < input.size(); ++i) {
+ if (input[i] == '&') {
+ strncpy(out, kAmp, strlen(kAmp));
+ out += strlen(kAmp);
+ } else if (input[i] == '<') {
+ strncpy(out, kLt, strlen(kLt));
+ out += strlen(kLt);
+ } else {
+ *out++ = input[i];
+ }
+ }
+ if (after) {
+ strncpy(out, after, afterLen);
+ out += afterLen;
+ }
+ // Validate that we haven't written outside of our string.
+ SkASSERT(out == &output.writable_str()[output.size()]);
+ *out = '\0';
+ return skstd::move(output);
+}
+
+SkPDFObject* SkPDFMetadata::createXMPObject(const UUID& doc,
+ const UUID& instance) const {
+ static const char templateString[] =
+ "<?xpacket begin=\"\" id=\"W5M0MpCehiHzreSzNTczkc9d\"?>\n"
+ "<x:xmpmeta xmlns:x=\"adobe:ns:meta/\"\n"
+ " x:xmptk=\"Adobe XMP Core 5.4-c005 78.147326, "
+ "2012/08/23-13:03:03\">\n"
+ "<rdf:RDF "
+ "xmlns:rdf=\"http://www.w3.org/1999/02/22-rdf-syntax-ns#\">\n"
+ "<rdf:Description rdf:about=\"\"\n"
+ " xmlns:xmp=\"http://ns.adobe.com/xap/1.0/\"\n"
+ " xmlns:dc=\"http://purl.org/dc/elements/1.1/\"\n"
+ " xmlns:xmpMM=\"http://ns.adobe.com/xap/1.0/mm/\"\n"
+ " xmlns:pdf=\"http://ns.adobe.com/pdf/1.3/\"\n"
+ " xmlns:pdfaid=\"http://www.aiim.org/pdfa/ns/id/\">\n"
+ "<pdfaid:part>2</pdfaid:part>\n"
+ "<pdfaid:conformance>B</pdfaid:conformance>\n"
+ "%s" // ModifyDate
+ "%s" // CreateDate
+ "%s" // MetadataDate
+ "%s" // xmp:CreatorTool
+ "<dc:format>application/pdf</dc:format>\n"
+ "%s" // dc:title
+ "%s" // dc:description
+ "%s" // author
+ "%s" // keywords
+ "<xmpMM:DocumentID>uuid:%s</xmpMM:DocumentID>\n"
+ "<xmpMM:InstanceID>uuid:%s</xmpMM:InstanceID>\n"
+ "<pdf:Producer>Skia/PDF</pdf:Producer>\n"
+ "%s" // pdf:Keywords
+ "</rdf:Description>\n"
+ "</rdf:RDF>\n"
+ "</x:xmpmeta>\n" // Note: the standard suggests 4k of padding.
+ "<?xpacket end=\"w\"?>\n";
+
+ SkString creationDate;
+ SkString modificationDate;
+ SkString metadataDate;
+ if (fCreation) {
+ SkString tmp;
+ fCreation->toISO8601(&tmp);
+ SkASSERT(0 == count_xml_escape_size(tmp));
+ // YYYY-mm-ddTHH:MM:SS[+|-]ZZ:ZZ; no need to escape
+ creationDate = sk_string_printf("<xmp:CreateDate>%s</xmp:CreateDate>\n",
+ tmp.c_str());
+ }
+ if (fModified) {
+ SkString tmp;
+ fModified->toISO8601(&tmp);
+ SkASSERT(0 == count_xml_escape_size(tmp));
+ modificationDate = sk_string_printf(
+ "<xmp:ModifyDate>%s</xmp:ModifyDate>\n", tmp.c_str());
+ metadataDate = sk_string_printf(
+ "<xmp:MetadataDate>%s</xmp:MetadataDate>\n", tmp.c_str());
+ }
+
+ SkString title =
+ escape_xml(get(fInfo, "Title"), "<dc:title><rdf:Alt><rdf:li>",
+ "</rdf:li></rdf:Alt></dc:title>\n");
+ SkString author =
+ escape_xml(get(fInfo, "Author"), "<dc:creator><rdf:Bag><rdf:li>",
+ "</rdf:li></rdf:Bag></dc:creator>\n");
+ // TODO: in theory, XMP can support multiple authors. Split on a delimiter?
+ SkString subject = escape_xml(get(fInfo, "Subject"),
+ "<dc:description><rdf:Alt><rdf:li>",
+ "</rdf:li></rdf:Alt></dc:description>\n");
+ SkString keywords1 =
+ escape_xml(get(fInfo, "Keywords"), "<dc:subject><rdf:Bag><rdf:li>",
+ "</rdf:li></rdf:Bag></dc:subject>\n");
+ SkString keywords2 = escape_xml(get(fInfo, "Keywords"), "<pdf:Keywords>",
+ "</pdf:Keywords>\n");
+
+ // TODO: in theory, keywords can be a list too.
+ SkString creator = escape_xml(get(fInfo, "Creator"), "<xmp:CreatorTool>",
+ "</xmp:CreatorTool>\n");
+ SkString documentID = uuid_to_string(doc); // no need to escape
+ SkASSERT(0 == count_xml_escape_size(documentID));
+ SkString instanceID = uuid_to_string(instance);
+ SkASSERT(0 == count_xml_escape_size(instanceID));
+ return new PDFXMLObject(sk_string_printf(
+ templateString, modificationDate.c_str(), creationDate.c_str(),
+ metadataDate.c_str(), creator.c_str(), title.c_str(),
+ subject.c_str(), author.c_str(), keywords1.c_str(),
+ documentID.c_str(), instanceID.c_str(), keywords2.c_str()));
+}
+
+#endif // SK_PDF_GENERATE_PDFA
diff --git a/src/pdf/SkPDFMetadata.h b/src/pdf/SkPDFMetadata.h
new file mode 100644
index 0000000..3475ac2
--- /dev/null
+++ b/src/pdf/SkPDFMetadata.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright 2015 Google Inc.
+ *
+ * Use of this source code is governed by a BSD-style license that can be
+ * found in the LICENSE file.
+ */
+
+#ifndef SkPDFMetadata_DEFINED
+#define SkPDFMetadata_DEFINED
+
+#include "SkDocument.h"
+#include "SkTime.h"
+
+class SkPDFObject;
+
+struct SkPDFMetadata {
+ SkTArray<SkDocument::Attribute> fInfo;
+ SkAutoTDelete<const SkTime::DateTime> fCreation;
+ SkAutoTDelete<const SkTime::DateTime> fModified;
+
+ SkPDFObject* createDocumentInformationDict() const;
+
+ struct UUID {
+ uint8_t fData[16];
+ };
+ UUID uuid() const;
+ static SkPDFObject* CreatePdfId(const UUID& doc, const UUID& instance);
+
+#ifdef SK_PDF_GENERATE_PDFA
+ SkPDFObject* createXMPObject(const UUID& doc, const UUID& instance) const;
+#endif // SK_PDF_GENERATE_PDFA
+};
+
+#endif // SkPDFMetadata_DEFINED
diff --git a/src/pdf/SkPDFTypes.cpp b/src/pdf/SkPDFTypes.cpp
index 4cd48f4..faa0837 100644
--- a/src/pdf/SkPDFTypes.cpp
+++ b/src/pdf/SkPDFTypes.cpp
@@ -177,9 +177,7 @@
return; // These have no resources.
case Type::kObjRef: {
SkPDFObject* obj = substituteMap.getSubstitute(fObject);
- if (objNumMap->addObject(obj)) {
- obj->addResources(objNumMap, substituteMap);
- }
+ objNumMap->addObjectRecursively(obj, substituteMap);
return;
}
case Type::kObject:
@@ -500,6 +498,13 @@
return true;
}
+void SkPDFObjNumMap::addObjectRecursively(SkPDFObject* obj,
+ const SkPDFSubstituteMap& subs) {
+ if (obj && this->addObject(obj)) {
+ obj->addResources(this, subs);
+ }
+}
+
int32_t SkPDFObjNumMap::getObjectNumber(SkPDFObject* obj) const {
int32_t* objectNumberFound = fObjectNumbers.find(obj);
SkASSERT(objectNumberFound);
diff --git a/src/pdf/SkPDFTypes.h b/src/pdf/SkPDFTypes.h
index ec527fc..0f02998 100644
--- a/src/pdf/SkPDFTypes.h
+++ b/src/pdf/SkPDFTypes.h
@@ -346,6 +346,12 @@
*/
bool addObject(SkPDFObject* obj);
+ /** Add the passed object to the catalog, as well as all its dependencies.
+ * @param obj The object to add. If nullptr, this is a noop.
+ * @param subs Will be passed to obj->addResources().
+ */
+ void addObjectRecursively(SkPDFObject* obj, const SkPDFSubstituteMap& subs);
+
/** Get the object number for the passed object.
* @param obj The object of interest.
*/