/* */ #include "XML2SAXMetalinkProcessor.h" #include #include "BinaryStream.h" #include "MetalinkParserStateMachine.h" #include "Metalinker.h" #include "MetalinkEntry.h" #include "util.h" #include "message.h" #include "DlAbortEx.h" #include "A2STR.h" namespace aria2 { class SessionData { public: SharedHandle _stm; std::deque _charactersStack; SessionData(const SharedHandle& stm):_stm(stm) {} }; static void mlStartElement (void* userData, const xmlChar* srcLocalname, const xmlChar* srcPrefix, const xmlChar* srcNsUri, int numNamespaces, const xmlChar **namespaces, int numAttrs, int numDefaulted, const xmlChar **attrs) { SessionData* sd = reinterpret_cast(userData); std::vector xmlAttrs; size_t index = 0; for(int attrIndex = 0; attrIndex < numAttrs; ++attrIndex, index += 5) { XmlAttr xmlAttr; assert(attrs[index]); xmlAttr.localname = reinterpret_cast(attrs[index]); if(attrs[index+1]) { xmlAttr.prefix = reinterpret_cast(attrs[index+1]); } if(attrs[index+2]) { xmlAttr.nsUri = reinterpret_cast(attrs[index+2]); } const char* valueBegin = reinterpret_cast(attrs[index+3]); const char* valueEnd = reinterpret_cast(attrs[index+4]); xmlAttr.value = std::string(valueBegin, valueEnd); xmlAttrs.push_back(xmlAttr); } assert(srcLocalname); std::string localname = reinterpret_cast(srcLocalname); std::string prefix; std::string nsUri; if(srcPrefix) { prefix = reinterpret_cast(srcPrefix); } if(srcNsUri) { nsUri = reinterpret_cast(srcNsUri); } sd->_stm->beginElement(localname, prefix, nsUri, xmlAttrs); if(sd->_stm->needsCharactersBuffering()) { sd->_charactersStack.push_front(A2STR::NIL); } } static void mlEndElement (void* userData, const xmlChar* srcLocalname, const xmlChar* srcPrefix, const xmlChar* srcNsUri) { SessionData* sd = reinterpret_cast(userData); std::string characters; if(sd->_stm->needsCharactersBuffering()) { characters = sd->_charactersStack.front(); sd->_charactersStack.pop_front(); } std::string localname = reinterpret_cast(srcLocalname); std::string prefix; std::string nsUri; if(srcPrefix) { prefix = reinterpret_cast(srcPrefix); } if(srcNsUri) { nsUri = reinterpret_cast(srcNsUri); } sd->_stm->endElement(localname, prefix, nsUri, characters); } static void mlCharacters(void* userData, const xmlChar* ch, int len) { SessionData* sd = reinterpret_cast(userData); if(sd->_stm->needsCharactersBuffering()) { sd->_charactersStack.front() += std::string(&ch[0], &ch[len]); } } static xmlSAXHandler mySAXHandler = { 0, // internalSubsetSAXFunc 0, // isStandaloneSAXFunc 0, // hasInternalSubsetSAXFunc 0, // hasExternalSubsetSAXFunc 0, // resolveEntitySAXFunc 0, // getEntitySAXFunc 0, // entityDeclSAXFunc 0, // notationDeclSAXFunc 0, // attributeDeclSAXFunc 0, // elementDeclSAXFunc 0, // unparsedEntityDeclSAXFunc 0, // setDocumentLocatorSAXFunc 0, // startDocumentSAXFunc 0, // endDocumentSAXFunc 0, // startElementSAXFunc 0, // endElementSAXFunc 0, // referenceSAXFunc &mlCharacters, // charactersSAXFunc 0, // ignorableWhitespaceSAXFunc 0, // processingInstructionSAXFunc 0, // commentSAXFunc 0, // warningSAXFunc 0, // errorSAXFunc 0, // fatalErrorSAXFunc 0, // getParameterEntitySAXFunc 0, // cdataBlockSAXFunc 0, // externalSubsetSAXFunc XML_SAX2_MAGIC, // unsigned int initialized 0, // void * _private &mlStartElement, // startElementNsSAX2Func &mlEndElement, // endElementNsSAX2Func 0, // xmlStructuredErrorFunc }; SharedHandle MetalinkProcessor::parseFile(const std::string& filename) { _stm.reset(new MetalinkParserStateMachine()); SharedHandle sessionData(new SessionData(_stm)); int retval = xmlSAXUserParseFile(&mySAXHandler, sessionData.get(), filename.c_str()); if(retval != 0) { throw DL_ABORT_EX(MSG_CANNOT_PARSE_METALINK); } return _stm->getResult(); } SharedHandle MetalinkProcessor::parseFromBinaryStream(const SharedHandle& binaryStream) { _stm.reset(new MetalinkParserStateMachine()); size_t bufSize = 4096; unsigned char buf[bufSize]; ssize_t res = binaryStream->readData(buf, 4, 0); if(res != 4) { throw DL_ABORT_EX("Too small data for parsing XML."); } SharedHandle sessionData(new SessionData(_stm)); xmlParserCtxtPtr ctx = xmlCreatePushParserCtxt (&mySAXHandler, sessionData.get(), reinterpret_cast(buf), res, 0); try { off_t readOffset = res; while(1) { ssize_t res = binaryStream->readData(buf, bufSize, readOffset); if(res == 0) { break; } if(xmlParseChunk(ctx, reinterpret_cast(buf), res, 0) != 0) { throw DL_ABORT_EX(MSG_CANNOT_PARSE_METALINK); } readOffset += res; } xmlParseChunk(ctx, reinterpret_cast(buf), 0, 1); } catch(Exception& e) { xmlFreeParserCtxt(ctx); throw; } xmlFreeParserCtxt(ctx); if(!_stm->finished()) { throw DL_ABORT_EX(MSG_CANNOT_PARSE_METALINK); } return _stm->getResult(); } } // namespace aria2