legacywebarchive.cpp

linux下开源浏览器WebKit的源码,市面上的很多商用浏览器都是移植自WebKit

        return false;
    }
    
    return extract(plist.get());
}

bool LegacyWebArchive::extract(CFDictionaryRef dictionary)
{
    ASSERT(dictionary);
    if (!dictionary) {
        LOG(Archives, "LegacyWebArchive - Null root CFDictionary, aborting invalid WebArchive");
        return false;
    }
    
    CFDictionaryRef mainResourceDict = static_cast<CFDictionaryRef>(CFDictionaryGetValue(dictionary, LegacyWebArchiveMainResourceKey));
    if (!mainResourceDict) {
        LOG(Archives, "LegacyWebArchive - No main resource in archive, aborting invalid WebArchive");
        return false;
    }
    if (CFGetTypeID(mainResourceDict) != CFDictionaryGetTypeID()) {
        LOG(Archives, "LegacyWebArchive - Main resource is not the expected CFDictionary, aborting invalid WebArchive");
        return false;
    }
    
    setMainResource(createResource(mainResourceDict));
    if (!mainResource()) {
        LOG(Archives, "LegacyWebArchive - Failed to parse main resource from CFDictionary or main resource does not exist, aborting invalid WebArchive");
        return false;
    }
    
    CFArrayRef subresourceArray = static_cast<CFArrayRef>(CFDictionaryGetValue(dictionary, LegacyWebArchiveSubresourcesKey));
    if (subresourceArray && CFGetTypeID(subresourceArray) != CFArrayGetTypeID()) {
        LOG(Archives, "LegacyWebArchive - Subresources is not the expected Array, aborting invalid WebArchive");
        return false;
    }
    
    if (subresourceArray) {
        CFIndex count = CFArrayGetCount(subresourceArray);
        for (CFIndex i = 0; i < count; ++i) {
            CFDictionaryRef subresourceDict = static_cast<CFDictionaryRef>(CFArrayGetValueAtIndex(subresourceArray, i));
            if (CFGetTypeID(subresourceDict) != CFDictionaryGetTypeID()) {
                LOG(Archives, "LegacyWebArchive - Subresource is not expected CFDictionary, aborting invalid WebArchive");
                return false;
            }
            addSubresource(createResource(subresourceDict));
        }
    }
    
    CFArrayRef subframeArray = static_cast<CFArrayRef>(CFDictionaryGetValue(dictionary, LegacyWebArchiveSubframeArchivesKey));
    if (subframeArray && CFGetTypeID(subframeArray) != CFArrayGetTypeID()) {
        LOG(Archives, "LegacyWebArchive - Subframe archives is not the expected Array, aborting invalid WebArchive");
        return false;
    }
    
    if (subframeArray) {
        CFIndex count = CFArrayGetCount(subframeArray);
        for (CFIndex i = 0; i < count; ++i) {
            CFDictionaryRef subframeDict = static_cast<CFDictionaryRef>(CFArrayGetValueAtIndex(subframeArray, i));
            if (CFGetTypeID(subframeDict) != CFDictionaryGetTypeID()) {
                LOG(Archives, "LegacyWebArchive - Subframe array is not expected CFDictionary, aborting invalid WebArchive");
                return false;
            }
            
            RefPtr<LegacyWebArchive> subframeArchive = create();
            if (subframeArchive->extract(subframeDict))
                addSubframeArchive(subframeArchive.release());
            else
                LOG(Archives, "LegacyWebArchive - Invalid subframe archive skipped");
        }
    }
    
    return true;
}

RetainPtr<CFDataRef> LegacyWebArchive::rawDataRepresentation()
{
    RetainPtr<CFDictionaryRef> propertyList = createPropertyListRep(this);
    if (!propertyList) {
        LOG(Archives, "LegacyWebArchive - Failed to create property list for archive, returning no data");
        return 0;
    }

    RetainPtr<CFWriteStreamRef> stream(AdoptCF, CFWriteStreamCreateWithAllocatedBuffers(0, 0));

    CFWriteStreamOpen(stream.get());
    CFPropertyListWriteToStream(propertyList.get(), stream.get(), kCFPropertyListBinaryFormat_v1_0, 0);

    RetainPtr<CFDataRef> plistData(AdoptCF, static_cast<CFDataRef>(CFWriteStreamCopyProperty(stream.get(), kCFStreamPropertyDataWritten)));

    CFWriteStreamClose(stream.get());

    if (!plistData) {
        LOG(Archives, "LegacyWebArchive - Failed to convert property list into raw data, returning no data");
        return 0;
    }

    return plistData;
}

#if !PLATFORM(MAC)
// FIXME: Is it possible to parse in a Cocoa-style resource response manually, 
// without NSKeyed(Un)Archiver, manipulating plists directly?
// If so, the code that does it will go here.  
// In the meantime, Mac will continue to NSKeyed(Un)Archive the response as it always has
ResourceResponse createResourceResponseFromMacArchivedData(CFDataRef responseData)
{
    return ResourceResponse();
}

RetainPtr<CFDataRef> propertyListDataFromResourceResponse(const ResourceResponse& response)
{
    // FIXME: Write out the "new" format described in ::createResourceResponseFromPropertyListData() up above
    return 0;
}
#endif

PassRefPtr<LegacyWebArchive> LegacyWebArchive::create(Node* node)
{
    ASSERT(node);
    if (!node)
        return create();
        
    Document* document = node->document();
    Frame* frame = document ? document->frame() : 0;
    if (!frame)
        return create();
        
    Vector<Node*> nodeList;
    String markupString = createMarkup(node, IncludeNode, &nodeList);
    Node::NodeType nodeType = node->nodeType();
    if (nodeType != Node::DOCUMENT_NODE && nodeType != Node::DOCUMENT_TYPE_NODE)
        markupString = frame->documentTypeString() + markupString;

    return create(markupString, frame, nodeList);
}

PassRefPtr<LegacyWebArchive> LegacyWebArchive::create(Frame* frame)
{
    ASSERT(frame);
    
    DocumentLoader* documentLoader = frame->loader()->documentLoader();

    if (!documentLoader)
        return 0;
        
    Vector<PassRefPtr<LegacyWebArchive> > subframeArchives;
    
    unsigned children = frame->tree()->childCount();
    for (unsigned i = 0; i < children; ++i) {
        RefPtr<LegacyWebArchive> childFrameArchive = create(frame->tree()->child(i));
        if (childFrameArchive)
            subframeArchives.append(childFrameArchive.release());
    }

    Vector<PassRefPtr<ArchiveResource> > subresources;
    documentLoader->getSubresources(subresources);

    return LegacyWebArchive::create(documentLoader->mainResource(), subresources, subframeArchives);
}

PassRefPtr<LegacyWebArchive> LegacyWebArchive::create(Range* range)
{
    if (!range)
        return 0;
    
    Node* startContainer = range->startContainer();
    if (!startContainer)
        return 0;
        
    Document* document = startContainer->document();
    if (!document)
        return 0;
        
    Frame* frame = document->frame();
    if (!frame)
        return 0;
    
    Vector<Node*> nodeList;
    
    // FIXME: This is always "for interchange". Is that right? See the previous method.
    String markupString = frame->documentTypeString() + createMarkup(range, &nodeList, AnnotateForInterchange);

    return create(markupString, frame, nodeList);
}

PassRefPtr<LegacyWebArchive> LegacyWebArchive::create(const String& markupString, Frame* frame, Vector<Node*>& nodes)
{
    ASSERT(frame);
    
    const ResourceResponse& response = frame->loader()->documentLoader()->response();
    KURL responseURL = response.url();
    
    // it's possible to have a response without a URL here
    // <rdar://problem/5454935>
    if (responseURL.isNull())
        responseURL = KURL("");
        
    PassRefPtr<ArchiveResource> mainResource = ArchiveResource::create(utf8Buffer(markupString), responseURL, response.mimeType(), "UTF-8", frame->tree()->name());

    Vector<PassRefPtr<LegacyWebArchive> > subframeArchives;
    Vector<PassRefPtr<ArchiveResource> > subresources;
    HashSet<KURL> uniqueSubresources;
    
    Vector<Node*>::iterator it = nodes.begin();
    Vector<Node*>::iterator end = nodes.end();
    
    for (; it != end; ++it) {
        Frame* childFrame;
        if (((*it)->hasTagName(HTMLNames::frameTag) || (*it)->hasTagName(HTMLNames::iframeTag) || (*it)->hasTagName(HTMLNames::objectTag)) &&
             (childFrame = static_cast<HTMLFrameOwnerElement*>(*it)->contentFrame())) {
            RefPtr<LegacyWebArchive> subframeArchive;
            subframeArchive = LegacyWebArchive::create(childFrame->document());
            
            if (subframeArchive)
                subframeArchives.append(subframeArchive);
            else
                LOG_ERROR("Unabled to archive subframe %s", childFrame->tree()->name().string().utf8().data());
        } else {
            ListHashSet<KURL> subresourceURLs;
            (*it)->getSubresourceURLs(subresourceURLs);
            
            DocumentLoader* documentLoader = frame->loader()->documentLoader();
            ListHashSet<KURL>::iterator iterEnd = subresourceURLs.end();
            for (ListHashSet<KURL>::iterator iter = subresourceURLs.begin(); iter != iterEnd; ++iter) {
                const KURL& subresourceURL = *iter;
                if (uniqueSubresources.contains(subresourceURL))
                    continue;

                uniqueSubresources.add(subresourceURL);

                RefPtr<ArchiveResource> resource = documentLoader->subresource(subresourceURL);
                if (resource) {
                    subresources.append(resource.release());
                    continue;
                }

                CachedResource *cachedResource = cache()->resourceForURL(subresourceURL);
                if (cachedResource) {
                    resource = ArchiveResource::create(cachedResource->data(), subresourceURL, cachedResource->response());
                    if (resource) {
                        subresources.append(resource.release());
                        continue;
                    }
                }

                // FIXME: should do something better than spew to console here
                LOG_ERROR("Failed to archive subresource for %s", subresourceURL.string().utf8().data());
            }
        }
    }

    // Add favicon if one exists for this page
    if (iconDatabase() && iconDatabase()->isEnabled()) {
        const String& iconURL = iconDatabase()->iconURLForPageURL(responseURL);
        if (!iconURL.isEmpty() && iconDatabase()->iconDataKnownForIconURL(iconURL)) {
            if (Image* iconImage = iconDatabase()->iconForPageURL(responseURL, IntSize(16, 16))) {
                RefPtr<ArchiveResource> resource = ArchiveResource::create(iconImage->data(), KURL(iconURL), "image/x-icon", "", "");
                subresources.append(resource.release());
            }
        }
    }

    return create(mainResource, subresources, subframeArchives);
}

PassRefPtr<LegacyWebArchive> LegacyWebArchive::createFromSelection(Frame* frame)
{
    if (!frame)
        return 0;
    
    RefPtr<Range> selectionRange = frame->selection()->toNormalizedRange();
    Vector<Node*> nodeList;
    String markupString = frame->documentTypeString() + createMarkup(selectionRange.get(), &nodeList, AnnotateForInterchange);
    
    RefPtr<LegacyWebArchive> archive = create(markupString, frame, nodeList);
    
    if (!frame->document() || !frame->document()->isFrameSet())
        return archive.release();
        
    // Wrap the frameset document in an iframe so it can be pasted into
    // another document (which will have a body or frameset of its own). 
    String iframeMarkup = String::format("<iframe frameborder=\"no\" marginwidth=\"0\" marginheight=\"0\" width=\"98%%\" height=\"98%%\" src=\"%s\"></iframe>", 
                                         frame->loader()->documentLoader()->response().url().string().utf8().data());
    RefPtr<ArchiveResource> iframeResource = ArchiveResource::create(utf8Buffer(iframeMarkup), blankURL(), "text/html", "UTF-8", String());

    Vector<PassRefPtr<ArchiveResource> > subresources;

    Vector<PassRefPtr<LegacyWebArchive> > subframeArchives;
    subframeArchives.append(archive);
    
    archive = LegacyWebArchive::create(iframeResource.release(), subresources, subframeArchives);
    
    return archive.release();
}

}