#include "config.h"
#if ENABLE(MHTML)
#include "MHTMLArchive.h"
#include "Document.h"
#include "MHTMLParser.h"
#include "MIMETypeRegistry.h"
#include "MainFrame.h"
#include "Page.h"
#include "PageSerializer.h"
#include "QuotedPrintable.h"
#include "SchemeRegistry.h"
#include "SharedBuffer.h"
#include <time.h>
#include <wtf/CryptographicallyRandomNumber.h>
#include <wtf/DateMath.h>
#include <wtf/GregorianDateTime.h>
#include <wtf/StdLibExtras.h>
#include <wtf/text/Base64.h>
#include <wtf/text/StringBuilder.h>
#if HAVE(SYS_TIME_H)
#include <sys/time.h>
#endif
namespace WebCore {
const char* const quotedPrintable = "quoted-printable";
const char* const base64 = "base64";
static String generateRandomBoundary()
{
const size_t randomValuesLength = 10;
char randomValues[randomValuesLength];
cryptographicallyRandomValues(&randomValues, randomValuesLength);
StringBuilder stringBuilder;
stringBuilder.append("----=_NextPart_000_");
for (size_t i = 0; i < randomValuesLength; ++i) {
if (i == 2)
stringBuilder.append('_');
else if (i == 6)
stringBuilder.append('.');
stringBuilder.append(lowerNibbleToASCIIHexDigit(randomValues[i]));
stringBuilder.append(upperNibbleToASCIIHexDigit(randomValues[i]));
}
return stringBuilder.toString();
}
static String replaceNonPrintableCharacters(const String& text)
{
StringBuilder stringBuilder;
for (size_t i = 0; i < text.length(); ++i) {
if (isASCIIPrintable(text[i]))
stringBuilder.append(text[i]);
else
stringBuilder.append('?');
}
return stringBuilder.toString();
}
MHTMLArchive::MHTMLArchive()
{
}
MHTMLArchive::~MHTMLArchive()
{
clearAllSubframeArchives();
}
Ref<MHTMLArchive> MHTMLArchive::create()
{
return adoptRef(*new MHTMLArchive);
}
RefPtr<MHTMLArchive> MHTMLArchive::create(const URL& url, SharedBuffer& data)
{
if (!SchemeRegistry::shouldTreatURLSchemeAsLocal(url.protocol().toString()))
return nullptr;
MHTMLParser parser(&data);
RefPtr<MHTMLArchive> mainArchive = parser.parseArchive();
if (!mainArchive)
return nullptr;
for (size_t i = 0; i < parser.frameCount(); ++i) {
RefPtr<MHTMLArchive> archive = parser.frameAt(i);
for (size_t j = 1; j < parser.frameCount(); ++j) {
if (i != j)
archive->addSubframeArchive(*parser.frameAt(j));
}
for (size_t j = 0; j < parser.subResourceCount(); ++j)
archive->addSubresource(*parser.subResourceAt(j));
}
return mainArchive;
}
Ref<SharedBuffer> MHTMLArchive::generateMHTMLData(Page* page)
{
Vector<PageSerializer::Resource> resources;
PageSerializer pageSerializer(resources);
pageSerializer.serialize(*page);
String boundary = generateRandomBoundary();
String endOfResourceBoundary = makeString("--", boundary, "\r\n");
GregorianDateTime now;
now.setToCurrentLocalTime();
String dateString = makeRFC2822DateString(now.weekDay(), now.monthDay(), now.month(), now.year(), now.hour(), now.minute(), now.second(), now.utcOffset() / 60);
StringBuilder stringBuilder;
stringBuilder.append("From: <Saved by WebKit>\r\n");
stringBuilder.append("Subject: ");
stringBuilder.append(replaceNonPrintableCharacters(page->mainFrame().document()->title()));
stringBuilder.append("\r\nDate: ");
stringBuilder.append(dateString);
stringBuilder.append("\r\nMIME-Version: 1.0\r\n");
stringBuilder.append("Content-Type: multipart/related;\r\n");
stringBuilder.append("\ttype=\"");
stringBuilder.append(page->mainFrame().document()->suggestedMIMEType());
stringBuilder.append("\";\r\n");
stringBuilder.append("\tboundary=\"");
stringBuilder.append(boundary);
stringBuilder.append("\"\r\n\r\n");
ASSERT(stringBuilder.toString().isAllASCII());
CString asciiString = stringBuilder.toString().utf8();
auto mhtmlData = SharedBuffer::create();
mhtmlData->append(asciiString.data(), asciiString.length());
for (auto& resource : resources) {
stringBuilder.clear();
stringBuilder.append(endOfResourceBoundary);
stringBuilder.append("Content-Type: ");
stringBuilder.append(resource.mimeType);
const char* contentEncoding = nullptr;
if (MIMETypeRegistry::isSupportedJavaScriptMIMEType(resource.mimeType) || MIMETypeRegistry::isSupportedNonImageMIMEType(resource.mimeType))
contentEncoding = quotedPrintable;
else
contentEncoding = base64;
stringBuilder.append("\r\nContent-Transfer-Encoding: ");
stringBuilder.append(contentEncoding);
stringBuilder.append("\r\nContent-Location: ");
stringBuilder.append(resource.url);
stringBuilder.append("\r\n\r\n");
asciiString = stringBuilder.toString().utf8();
mhtmlData->append(asciiString.data(), asciiString.length());
const char* data = resource.data->data();
size_t dataLength = resource.data->size();
Vector<char> encodedData;
if (!strcmp(contentEncoding, quotedPrintable)) {
quotedPrintableEncode(data, dataLength, encodedData);
mhtmlData->append(encodedData.data(), encodedData.size());
mhtmlData->append("\r\n", 2);
} else {
ASSERT(!strcmp(contentEncoding, base64));
base64Encode(data, dataLength, encodedData);
const size_t maximumLineLength = 76;
size_t index = 0;
size_t encodedDataLength = encodedData.size();
do {
size_t lineLength = std::min(encodedDataLength - index, maximumLineLength);
mhtmlData->append(encodedData.data() + index, lineLength);
mhtmlData->append("\r\n", 2);
index += maximumLineLength;
} while (index < encodedDataLength);
}
}
asciiString = makeString("--", boundary, "--\r\n").utf8();
mhtmlData->append(asciiString.data(), asciiString.length());
return mhtmlData;
}
}
#endif