bool Package::Unpack()
{
PackagePtr sharedMe = shared_from_this();
// very basic sanity check
xmlNodePtr root = xmlDocGetRootElement(_opf);
string rootName(reinterpret_cast<const char*>(root->name));
rootName.tolower();
if ( rootName != "package" )
{
HandleError(EPUBError::OPFInvalidPackageDocument);
return false; // not an OPF file, innit?
}
if ( _getProp(root, "version").empty() )
{
HandleError(EPUBError::OPFPackageHasNoVersion);
}
InstallPrefixesFromAttributeValue(_getProp(root, "prefix", ePub3NamespaceURI));
// go through children to determine the CFI index of the <spine> tag
static const xmlChar* kSpineName = BAD_CAST "spine";
static const xmlChar* kManifestName = BAD_CAST "manifest";
static const xmlChar* kMetadataName = BAD_CAST "metadata";
_spineCFIIndex = 0;
uint32_t idx = 0;
xmlNodePtr child = xmlFirstElementChild(root);
while ( child != nullptr )
{
idx += 2;
if ( xmlStrEqual(child->name, kSpineName) )
{
_spineCFIIndex = idx;
if ( _spineCFIIndex != 6 )
HandleError(EPUBError::OPFSpineOutOfOrder);
}
else if ( xmlStrEqual(child->name, kManifestName) && idx != 4 )
{
HandleError(EPUBError::OPFManifestOutOfOrder);
}
else if ( xmlStrEqual(child->name, kMetadataName) && idx != 2 )
{
HandleError(EPUBError::OPFMetadataOutOfOrder);
}
child = xmlNextElementSibling(child);
}
if ( _spineCFIIndex == 0 )
{
HandleError(EPUBError::OPFNoSpine);
return false; // spineless!
}
#if EPUB_COMPILER_SUPPORTS(CXX_INITIALIZER_LISTS)
XPathWrangler xpath(_opf, {{"opf", OPFNamespace}, {"dc", DCNamespace}});
#else
XPathWrangler::NamespaceList __m;
__m["opf"] = OPFNamespace;
__m["dc"] = DCNamespace;
XPathWrangler xpath(_opf, __m);
#endif
// simple things: manifest and spine items
xmlNodeSetPtr manifestNodes = nullptr;
xmlNodeSetPtr spineNodes = nullptr;
try
{
manifestNodes = xpath.Nodes("/opf:package/opf:manifest/opf:item");
spineNodes = xpath.Nodes("/opf:package/opf:spine/opf:itemref");
if ( manifestNodes == nullptr )
{
HandleError(EPUBError::OPFNoManifestItems);
}
if ( spineNodes == nullptr )
{
HandleError(EPUBError::OPFNoSpineItems);
}
for ( int i = 0; i < manifestNodes->nodeNr; i++ )
{
auto p = std::make_shared<ManifestItem>(sharedMe);
if ( p->ParseXML(p, manifestNodes->nodeTab[i]) )
{
#if EPUB_HAVE(CXX_MAP_EMPLACE)
_manifest.emplace(p->Identifier(), p);
#else
_manifest[p->Identifier()] = p;
#endif
StoreXMLIdentifiable(p);
}
else
{
// TODO: Need an error here
}
}
// check fallback chains
typedef std::map<string, bool> IdentSet;
IdentSet idents;
for ( auto &pair : _manifest )
{
ManifestItemPtr item = pair.second;
if ( item->FallbackID().empty() )
continue;
idents[item->XMLIdentifier()] = true;
while ( !item->FallbackID().empty() )
{
if ( idents[item->FallbackID()] )
{
HandleError(EPUBError::OPFFallbackChainCircularReference);
break;
}
item = item->Fallback();
}
idents.clear();
}
SpineItemPtr cur;
for ( int i = 0; i < spineNodes->nodeNr; i++ )
{
auto next = std::make_shared<SpineItem>(sharedMe);
if ( next->ParseXML(next, spineNodes->nodeTab[i]) == false )
{
// TODO: need an error code here
continue;
}
// validation of idref
auto manifestFound = _manifest.find(next->Idref());
if ( manifestFound == _manifest.end() )
{
HandleError(EPUBError::OPFInvalidSpineIdref, _Str(next->Idref(), " does not correspond to a manifest item"));
continue;
}
// validation of spine resource type w/fallbacks
ManifestItemPtr manifestItem = next->ManifestItem();
bool isContentDoc = false;
do
{
if ( manifestItem->MediaType() == "application/xhtml+xml" ||
manifestItem->MediaType() == "image/svg" )
{
isContentDoc = true;
break;
}
} while ( (manifestItem = manifestItem->Fallback()) );
if ( !isContentDoc )
HandleError(EPUBError::OPFFallbackChainHasNoContentDocument);
StoreXMLIdentifiable(next);
if ( cur != nullptr )
{
cur->SetNextItem(next);
}
else
{
_spine = next;
}
cur = next;
}
}
catch (const std::system_error& exc)
{
if ( manifestNodes != nullptr )
xmlXPathFreeNodeSet(manifestNodes);
if ( spineNodes != nullptr )
xmlXPathFreeNodeSet(spineNodes);
if ( exc.code().category() == epub_spec_category() )
throw;
return false;
}
catch (...)
{
if ( manifestNodes != nullptr )
xmlXPathFreeNodeSet(manifestNodes);
if ( spineNodes != nullptr )
xmlXPathFreeNodeSet(spineNodes);
return false;
}
xmlXPathFreeNodeSet(manifestNodes);
xmlXPathFreeNodeSet(spineNodes);
// now the metadata, which is slightly more involved due to extensions
xmlNodeSetPtr metadataNodes = nullptr;
xmlNodeSetPtr refineNodes = xmlXPathNodeSetCreate(nullptr);
try
{
shared_ptr<PropertyHolder> holderPtr = std::dynamic_pointer_cast<PropertyHolder>(sharedMe);
metadataNodes = xpath.Nodes("/opf:package/opf:metadata/*");
if ( metadataNodes == nullptr )
HandleError(EPUBError::OPFNoMetadata);
bool foundIdentifier = false, foundTitle = false, foundLanguage = false, foundModDate = false;
string uniqueIDRef = _getProp(root, "unique-identifier");
if ( uniqueIDRef.empty() )
HandleError(EPUBError::OPFPackageUniqueIDInvalid);
for ( int i = 0; i < metadataNodes->nodeNr; i++ )
{
xmlNodePtr node = metadataNodes->nodeTab[i];
PropertyPtr p;
if ( node->ns != nullptr && xmlStrcmp(node->ns->href, BAD_CAST DCNamespace) == 0 )
{
// definitely a main node
p = std::make_shared<Property>(holderPtr);
}
else if ( _getProp(node, "name").size() > 0 )
{
// it's an ePub2 item-- ignore it
continue;
}
else if ( _getProp(node, "refines").empty() )
{
// not refining anything, so it's a main node
p = std::make_shared<Property>(holderPtr);
}
else
{
// by elimination it's refining something-- we'll process it later when we know we've got all the main nodes in there
xmlXPathNodeSetAdd(refineNodes, node);
}
if ( p && p->ParseMetaElement(node) )
{
switch ( p->Type() )
{
case DCType::Identifier:
{
foundIdentifier = true;
if ( !uniqueIDRef.empty() && uniqueIDRef != p->XMLIdentifier() )
HandleError(EPUBError::OPFPackageUniqueIDInvalid);
break;
}
case DCType::Title:
{
foundTitle = true;
break;
}
case DCType::Language:
{
foundLanguage = true;
break;
}
case DCType::Custom:
{
if ( p->PropertyIdentifier() == MakePropertyIRI("modified", "dcterms") )
foundModDate = true;
break;
}
default:
break;
}
AddProperty(p);
StoreXMLIdentifiable(p);
}
}
if ( !foundIdentifier )
HandleError(EPUBError::OPFMissingIdentifierMetadata);
if ( !foundTitle )
HandleError(EPUBError::OPFMissingTitleMetadata);
if ( !foundLanguage )
HandleError(EPUBError::OPFMissingLanguageMetadata);
if ( !foundModDate )
HandleError(EPUBError::OPFMissingModificationDateMetadata);
for ( int i = 0; i < refineNodes->nodeNr; i++ )
{
xmlNodePtr node = refineNodes->nodeTab[i];
string ident = _getProp(node, "refines");
if ( ident.empty() )
{
HandleError(EPUBError::OPFInvalidRefinementAttribute, "Empty IRI for 'refines' attribute");
continue;
}
if ( ident[0] == '#' )
{
ident = ident.substr(1);
}
else
{
// validation only right now
IRI iri(ident);
if ( iri.IsEmpty() )
{
HandleError(EPUBError::OPFInvalidRefinementAttribute, _Str("#", ident, " is not a valid IRI"));
}
else if ( iri.IsRelative() == false )
{
HandleError(EPUBError::OPFInvalidRefinementAttribute, _Str(iri.IRIString(), " is not a relative IRI"));
}
continue;
}
auto found = _xmlIDLookup.find(ident);
if ( found == _xmlIDLookup.end() )
{
HandleError(EPUBError::OPFInvalidRefinementTarget, _Str("#", ident, " does not reference an item in this document"));
continue;
}
PropertyPtr prop = std::dynamic_pointer_cast<Property>(found->second);
if ( prop )
{
// it's a property, so this is an extension
PropertyExtensionPtr extPtr = std::make_shared<PropertyExtension>(prop);
if ( extPtr->ParseMetaElement(node) )
prop->AddExtension(extPtr);
}
else
{
// not a property, so treat this as a plain property
shared_ptr<PropertyHolder> ptr = std::dynamic_pointer_cast<PropertyHolder>(found->second);
if ( ptr )
{
prop = std::make_shared<Property>(ptr);
if ( prop->ParseMetaElement(node) )
ptr->AddProperty(prop);
}
}
}
// now look at the <spine> element for properties
xmlNodePtr spineNode = xmlFirstElementChild(root);
for ( uint32_t i = 2; i < _spineCFIIndex; i += 2 )
spineNode = xmlNextElementSibling(spineNode);
string value = _getProp(spineNode, "page-progression-direction");
if ( !value.empty() )
{
PropertyPtr prop = std::make_shared<Property>(holderPtr);
prop->SetPropertyIdentifier(MakePropertyIRI("page-progression-direction"));
prop->SetValue(value);
AddProperty(prop);
}
}
catch (std::system_error& exc)
{
if ( metadataNodes != nullptr )
xmlXPathFreeNodeSet(metadataNodes);
if ( refineNodes != nullptr )
xmlXPathFreeNodeSet(refineNodes);
if ( exc.code().category() == epub_spec_category() )
throw;
return false;
}
catch (...)
{
if ( metadataNodes != nullptr )
xmlXPathFreeNodeSet(metadataNodes);
if ( refineNodes != nullptr )
xmlXPathFreeNodeSet(refineNodes);
return false;
}
xmlXPathFreeNodeSet(metadataNodes);
xmlXPathFreeNodeSet(refineNodes);
// now any content type bindings
xmlNodeSetPtr bindingNodes = nullptr;
try
{
bindingNodes = xpath.Nodes("/opf:package/opf:bindings/*");
if ( bindingNodes != nullptr )
{
for ( int i = 0; i < bindingNodes->nodeNr; i++ )
{
xmlNodePtr node = bindingNodes->nodeTab[i];
if ( xmlStrcasecmp(node->name, MediaTypeElementName) != 0 )
continue;
////////////////////////////////////////////////////////////
// ePub Publications 3.0 §3.4.16: The `mediaType` Element
// The media-type attribute is required.
string mediaType = _getProp(node, "media-type");
if ( mediaType.empty() )
{
HandleError(EPUBError::OPFBindingHandlerNoMediaType);
throw false;
}
// Each child mediaType of a bindings element must define a unique
// content type in its media-type attribute, and the media type
// specified must not be a Core Media Type.
if ( _contentHandlers[mediaType].empty() == false )
{
// user shouldn't have added manual things yet, but for safety we'll look anyway
for ( auto ptr : _contentHandlers[mediaType] )
{
if ( typeid(*ptr) == typeid(MediaHandler) )
{
HandleError(EPUBError::OPFMultipleBindingsForMediaType);
}
}
}
if ( CoreMediaTypes.find(mediaType) != CoreMediaTypes.end() )
{
HandleError(EPUBError::OPFCoreMediaTypeBindingEncountered);
}
// The handler attribute is required
string handlerID = _getProp(node, "handler");
if ( handlerID.empty() )
{
HandleError(EPUBError::OPFBindingHandlerNotFound);
}
// The required handler attribute must reference the ID [XML] of an
// item in the manifest of the default implementation for this media
// type. The referenced item must be an XHTML Content Document.
ManifestItemPtr handlerItem = ManifestItemWithID(handlerID);
if ( !handlerItem )
{
HandleError(EPUBError::OPFBindingHandlerNotFound);
}
if ( handlerItem->MediaType() != "application/xhtml+xml" )
{
HandleError(EPUBError::OPFBindingHandlerInvalidType, _Str("Media handlers must be XHTML content documents, but referenced item has type '", handlerItem->MediaType(), "'."));
}
// All XHTML Content Documents designated as handlers must have the
// `scripted` property set in their manifest item's `properties`
// attribute.
if ( handlerItem->HasProperty(ItemProperties::HasScriptedContent) == false )
{
HandleError(EPUBError::OPFBindingHandlerNotScripted);
}
// all good-- install it now
_contentHandlers[mediaType].push_back(std::make_shared<MediaHandler>(sharedMe, mediaType, handlerItem->AbsolutePath()));
}
}
}
catch (std::exception& exc)
{
std::cerr << "Exception processing OPF file: " << exc.what() << std::endl;
if ( bindingNodes != nullptr )
xmlXPathFreeNodeSet(bindingNodes);
throw;
}
catch (...)
{
if ( bindingNodes != nullptr )
xmlXPathFreeNodeSet(bindingNodes);
return false;
}
xmlXPathFreeNodeSet(bindingNodes);
// now the navigation tables
for ( auto item : _manifest )
{
if ( !item.second->HasProperty(ItemProperties::Navigation) )
continue;
NavigationList tables = NavTablesFromManifestItem(sharedMe, item.second);
for ( auto table : tables )
{
// have to dynamic_cast these guys to get the right pointer type
shared_ptr<class NavigationTable> navTable = std::dynamic_pointer_cast<class NavigationTable>(table);
#if EPUB_HAVE(CXX_MAP_EMPLACE)
_navigation.emplace(navTable->Type(), navTable);
#else
_navigation[navTable->Type()] = navTable;
#endif
}
}
// lastly, let's set the media support information
InitMediaSupport();
return true;
}
bool Package::Unpack()
{
// very basic sanity check
xmlNodePtr root = xmlDocGetRootElement(_opf);
string rootName(reinterpret_cast<const char*>(root->name));
rootName.tolower();
if ( rootName != "package" )
return false; // not an OPF file, innit?
InstallPrefixesFromAttributeValue(_getProp(root, "prefix", ePub3NamespaceURI));
// go through children to determine the CFI index of the <spine> tag
static const xmlChar* kSpineName = BAD_CAST "spine";
_spineCFIIndex = 0;
xmlNodePtr child = root->children;
while ( child != nullptr )
{
if ( child->type == XML_ELEMENT_NODE )
{
_spineCFIIndex += 2;
if ( xmlStrEqual(child->name, kSpineName) )
break;
}
child = child->next;
}
if ( _spineCFIIndex == 0 )
return false; // spineless!
XPathWrangler xpath(_opf, {{"opf", OPFNamespace}, {"dc", DCNamespace}});
// simple things: manifest and spine items
xmlNodeSetPtr manifestNodes = nullptr;
xmlNodeSetPtr spineNodes = nullptr;
try
{
manifestNodes = xpath.Nodes("/opf:package/opf:manifest/opf:item");
spineNodes = xpath.Nodes("/opf:package/opf:spine/opf:itemref");
if ( manifestNodes == nullptr || spineNodes == nullptr )
throw false; // looks invalid, or at least unusable, to me
for ( int i = 0; i < manifestNodes->nodeNr; i++ )
{
ManifestItem *p = new ManifestItem(manifestNodes->nodeTab[i], this);
_manifest.emplace(p->Identifier(), p);
}
SpineItem* cur = nullptr;
for ( int i = 0; i < spineNodes->nodeNr; i++ )
{
SpineItem* next = new SpineItem(spineNodes->nodeTab[i], this);
if ( cur != nullptr )
{
cur->SetNextItem(next);
}
else
{
_spine.reset(next);
}
cur = next;
}
}
catch (...)
{
if ( manifestNodes != nullptr )
xmlXPathFreeNodeSet(manifestNodes);
if ( spineNodes != nullptr )
xmlXPathFreeNodeSet(spineNodes);
return false;
}
xmlXPathFreeNodeSet(manifestNodes);
xmlXPathFreeNodeSet(spineNodes);
// now the metadata, which is slightly more involved due to extensions
xmlNodeSetPtr metadataNodes = nullptr;
xmlNodeSetPtr refineNodes = xmlXPathNodeSetCreate(nullptr);
try
{
metadataNodes = xpath.Nodes("/opf:package/opf:metadata/*");
if ( metadataNodes == nullptr )
throw false;
std::map<string, class Metadata*> metadataByID;
for ( int i = 0; i < metadataNodes->nodeNr; i++ )
{
xmlNodePtr node = metadataNodes->nodeTab[i];
class Metadata* p = nullptr;
if ( node->ns != nullptr && xmlStrcmp(node->ns->href, BAD_CAST DCNamespace) == 0 )
{
// definitely a main node
p = new class Metadata(node, this);
}
else if ( _getProp(node, "name").size() > 0 )
{
// it's an ePub2 item-- ignore it
continue;
}
else if ( _getProp(node, "refines").empty() )
{
// not refining anything, so it's a main node
p = new class Metadata(node, this);
}
else
{
// by elimination it's refining something-- we'll process it later when we know we've got all the main nodes in there
xmlXPathNodeSetAdd(refineNodes, node);
}
if ( p != nullptr )
{
_metadata.push_back(p);
if ( !p->Identifier().empty() )
metadataByID[p->Identifier()] = p;
}
}
for ( int i = 0; i < refineNodes->nodeNr; i++ )
{
xmlNodePtr node = refineNodes->nodeTab[i];
string ident = _getProp(node, "refines");
if ( ident.empty() )
continue;
if ( ident[0] == '#' )
ident = ident.substr(1);
auto found = metadataByID.find(ident);
if ( found == metadataByID.end() )
continue;
found->second->AddExtension(node, this);
}
}
catch (...)
{
if ( metadataNodes != nullptr )
xmlXPathFreeNodeSet(metadataNodes);
if ( refineNodes != nullptr )
xmlXPathFreeNodeSet(refineNodes);
return false;
}
xmlXPathFreeNodeSet(metadataNodes);
xmlXPathFreeNodeSet(refineNodes);
// now any content type bindings
xmlNodeSetPtr bindingNodes = nullptr;
try
{
bindingNodes = xpath.Nodes("/opf:package/opf:bindings/*");
if ( bindingNodes != nullptr )
{
for ( int i = 0; i < bindingNodes->nodeNr; i++ )
{
xmlNodePtr node = bindingNodes->nodeTab[i];
if ( xmlStrcasecmp(node->name, MediaTypeElementName) != 0 )
continue;
////////////////////////////////////////////////////////////
// ePub Publications 3.0 §3.4.16: The `mediaType` Element
// The media-type attribute is required.
string mediaType = _getProp(node, "media-type");
if ( mediaType.empty() )
{
throw std::invalid_argument("mediaType element has missing or empty media-type attribute.");
}
// Each child mediaType of a bindings element must define a unique
// content type in its media-type attribute, and the media type
// specified must not be a Core Media Type.
if ( _contentHandlers[mediaType].empty() == false )
{
// user shouldn't have added manual things yet, but for safety we'll look anyway
for ( auto ptr : _contentHandlers[mediaType] )
{
if ( typeid(*ptr) == typeid(MediaHandler) )
{
throw std::invalid_argument(_Str("Duplicate media handler found for type '", mediaType, "'."));
}
}
}
if ( CoreMediaTypes.find(mediaType) != CoreMediaTypes.end() )
{
throw std::invalid_argument("mediaType element specifies an EPUB Core Media Type.");
}
// The handler attribute is required
string handlerID = _getProp(node, "handler");
if ( handlerID.empty() )
{
throw std::invalid_argument("mediaType element has missing or empty handler attribute.");
}
// The required handler attribute must reference the ID [XML] of an
// item in the manifest of the default implementation for this media
// type. The referenced item must be an XHTML Content Document.
const ManifestItem* handlerItem = ManifestItemWithID(handlerID);
if ( handlerItem == nullptr )
{
throw std::invalid_argument(_Str("mediaType element references non-existent handler with ID '", handlerID, "'."));
}
if ( handlerItem->MediaType() != "application/xhtml+xml" )
{
throw std::invalid_argument(_Str("Media handlers must be XHTML content documents, but referenced item has type '", handlerItem->MediaType(), "'."));
}
// All XHTML Content Documents designated as handlers must have the
// `scripted` property set in their manifest item's `properties`
// attribute.
if ( handlerItem->HasProperty(ItemProperties::HasScriptedContent) == false )
{
throw std::invalid_argument("Media handlers must have the `scripted` property.");
}
// all good-- install it now
_contentHandlers[mediaType].push_back(new MediaHandler(this, mediaType, handlerItem->AbsolutePath()));
}
}
}
catch (std::exception& exc)
{
std::cerr << "Exception processing OPF file: " << exc.what() << std::endl;
if ( bindingNodes != nullptr )
xmlXPathFreeNodeSet(bindingNodes);
return false;
}
catch (...)
{
if ( bindingNodes != nullptr )
xmlXPathFreeNodeSet(bindingNodes);
return false;
}
xmlXPathFreeNodeSet(bindingNodes);
// now the navigation tables
for ( auto item : _manifest )
{
if ( !item.second->HasProperty(ItemProperties::Navigation) )
continue;
NavigationList tables = NavTablesFromManifestItem(item.second);
for ( auto table : tables )
{
// have to dynamic_cast these guys to get the right pointer type
class NavigationTable* navTable = dynamic_cast<class NavigationTable*>(table);
_navigation.emplace(navTable->Type(), navTable);
}
}
return true;
}
