PassRefPtrWillBeRawPtr<MHTMLArchive> MHTMLParser::parseArchiveWithHeader(MIMEHeader* header)
{
if (!header) {
WTF_LOG_ERROR("Failed to parse MHTML part: no header.");
return nullptr;
}
RefPtrWillBeRawPtr<MHTMLArchive> archive = MHTMLArchive::create();
if (!header->isMultipart()) {
// With IE a page with no resource is not multi-part.
bool endOfArchiveReached = false;
RefPtrWillBeRawPtr<ArchiveResource> resource = parseNextPart(*header, String(), String(), endOfArchiveReached);
if (!resource)
return nullptr;
archive->setMainResource(resource);
return archive;
}
// Skip the message content (it's a generic browser specific message).
skipLinesUntilBoundaryFound(m_lineReader, header->endOfPartBoundary());
bool endOfArchive = false;
while (!endOfArchive) {
RefPtrWillBeRawPtr<MIMEHeader> resourceHeader = MIMEHeader::parseHeader(&m_lineReader);
if (!resourceHeader) {
WTF_LOG_ERROR("Failed to parse MHTML, invalid MIME header.");
return nullptr;
}
if (resourceHeader->contentType() == "multipart/alternative") {
// Ignore IE nesting which makes little sense (IE seems to nest only some of the frames).
RefPtrWillBeRawPtr<MHTMLArchive> subframeArchive = parseArchiveWithHeader(resourceHeader.get());
if (!subframeArchive) {
WTF_LOG_ERROR("Failed to parse MHTML subframe.");
return nullptr;
}
bool endOfPartReached = skipLinesUntilBoundaryFound(m_lineReader, header->endOfPartBoundary());
ASSERT_UNUSED(endOfPartReached, endOfPartReached);
// The top-frame is the first frame found, regardless of the nesting level.
if (subframeArchive->mainResource())
addResourceToArchive(subframeArchive->mainResource(), archive.get());
archive->addSubframeArchive(subframeArchive);
continue;
}
RefPtrWillBeRawPtr<ArchiveResource> resource = parseNextPart(*resourceHeader, header->endOfPartBoundary(), header->endOfDocumentBoundary(), endOfArchive);
if (!resource) {
WTF_LOG_ERROR("Failed to parse MHTML part.");
return nullptr;
}
addResourceToArchive(resource.get(), archive.get());
}
return archive.release();
}
PassRefPtrWillBeRawPtr<MHTMLArchive> MHTMLArchive::create(const KURL& url, SharedBuffer* data)
{
// For security reasons we only load MHTML pages from local URLs.
if (!SchemeRegistry::shouldTreatURLSchemeAsLocal(url.protocol()))
return nullptr;
MHTMLParser parser(data);
RefPtrWillBeRawPtr<MHTMLArchive> mainArchive = parser.parseArchive();
if (!mainArchive)
return nullptr; // Invalid MHTML file.
// Since MHTML is a flat format, we need to make all frames aware of all resources.
for (size_t i = 0; i < parser.frameCount(); ++i) {
RefPtrWillBeRawPtr<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.release();
}
