Rewritten Expat Metalink XML parser.

pull/2/head
Tatsuhiro Tsujikawa 2011-11-10 23:28:12 +09:00
parent 440f29aca3
commit 3b4a368554
5 changed files with 96 additions and 133 deletions

View File

@ -2,7 +2,7 @@
/* /*
* aria2 - The high speed download utility * aria2 - The high speed download utility
* *
* Copyright (C) 2006 Tatsuhiro Tsujikawa * Copyright (C) 2011 Tatsuhiro Tsujikawa
* *
* This program is free software; you can redistribute it and/or modify * This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by * it under the terms of the GNU General Public License as published by
@ -32,50 +32,52 @@
* files in the program, then also delete it here. * files in the program, then also delete it here.
*/ */
/* copyright --> */ /* copyright --> */
#include "ExpatMetalinkProcessor.h" #include "ExpatXmlParser.h"
#include <cstdio> #include <cstdio>
#include <cstring> #include <cstring>
#include <deque>
#include "DefaultDiskWriter.h" #include <expat.h>
#include "MetalinkParserStateMachine.h"
#include "Metalinker.h" #include "a2io.h"
#include "MetalinkEntry.h" #include "BinaryStream.h"
#include "util.h"
#include "message.h"
#include "DlAbortEx.h"
#include "MetalinkParserState.h"
#include "A2STR.h"
#include "error_code.h"
#include "BufferedFile.h" #include "BufferedFile.h"
#include "ParserStateMachine.h"
#include "A2STR.h"
#include "a2functional.h" #include "a2functional.h"
#include "XmlAttr.h"
namespace aria2 { namespace aria2 {
namespace { namespace {
class SessionData { struct SessionData {
public:
SharedHandle<MetalinkParserStateMachine> stm_;
std::deque<std::string> charactersStack_; std::deque<std::string> charactersStack_;
ParserStateMachine* psm_;
SessionData(const SharedHandle<MetalinkParserStateMachine>& stm):stm_(stm) {} SessionData(ParserStateMachine* psm)
: psm_(psm)
{}
}; };
} // namespace } // namespace
namespace { namespace {
template<typename InputIterator> template<typename InputIterator>
void splitNsName void splitNsName
(std::string& localname, std::string& prefix, std::string& nsUri, (const char** localname,
InputIterator first, InputIterator last) const char** nsUri,
InputIterator first,
InputIterator last)
{ {
std::pair<Scip, Scip> nsNamePair; InputIterator sep = std::find(first, last, '\t');
util::divide(nsNamePair, first, last, '\t'); if(sep == last) {
if(nsNamePair.second.first == nsNamePair.second.second) { *localname = first;
localname.assign(nsNamePair.first.first, nsNamePair.first.second);
} else { } else {
nsUri.assign(nsNamePair.first.first, nsNamePair.first.second); *localname = sep+1;
localname.assign(nsNamePair.second.first, nsNamePair.second.second); size_t nsUriLen = sep-first;
char* temp = new char[nsUriLen+1];
memcpy(temp, first, nsUriLen);
temp[nsUriLen] = '\0';
*nsUri = temp;
} }
} }
} // namespace } // namespace
@ -84,29 +86,34 @@ namespace {
void mlStartElement(void* userData, const char* nsName, const char** attrs) void mlStartElement(void* userData, const char* nsName, const char** attrs)
{ {
SessionData* sd = reinterpret_cast<SessionData*>(userData); SessionData* sd = reinterpret_cast<SessionData*>(userData);
std::vector<XmlAttr> xmlAttrs; std::vector<XmlAttr> xmlAttrs;
if(attrs) { if(attrs) {
const char** p = attrs; const char** p = attrs;
while(*p != 0) { while(*p != 0) {
XmlAttr xa;
const char* attrNsName = *p++; const char* attrNsName = *p++;
if(*p == 0) { if(*p == 0) {
break; break;
} }
XmlAttr xa; splitNsName(&xa.localname, &xa.nsUri,
splitNsName(xa.localname, xa.prefix, xa.nsUri,
attrNsName, attrNsName+strlen(attrNsName)); attrNsName, attrNsName+strlen(attrNsName));
const char* value = *p++; const char* value = *p++;
xa.value.assign(value, value+strlen(value)); xa.value = value;
xa.valueLength = strlen(value);
xmlAttrs.push_back(xa); xmlAttrs.push_back(xa);
} }
} }
std::string localname; const char* localname = 0;
std::string prefix; const char* prefix = 0;
std::string nsUri; const char* nsUri = 0;
splitNsName(localname, prefix, nsUri, nsName, nsName+strlen(nsName)); splitNsName(&localname, &nsUri, nsName, nsName+strlen(nsName));
sd->stm_->beginElement(localname, prefix, nsUri, xmlAttrs); sd->psm_->beginElement(localname, prefix, nsUri, xmlAttrs);
if(sd->stm_->needsCharactersBuffering()) { delete [] nsUri;
for(std::vector<XmlAttr>::iterator i = xmlAttrs.begin(),
eoi = xmlAttrs.end(); i != eoi; ++i) {
delete [] (*i).nsUri;
}
if(sd->psm_->needsCharactersBuffering()) {
sd->charactersStack_.push_front(A2STR::NIL); sd->charactersStack_.push_front(A2STR::NIL);
} }
} }
@ -115,18 +122,18 @@ void mlStartElement(void* userData, const char* nsName, const char** attrs)
namespace { namespace {
void mlEndElement(void* userData, const char* nsName) void mlEndElement(void* userData, const char* nsName)
{ {
std::string localname; const char* localname = 0;
std::string prefix; const char* prefix = 0;
std::string nsUri; const char* nsUri = 0;
splitNsName(localname, prefix, nsUri, nsName, nsName+strlen(nsName)); splitNsName(&localname, &nsUri, nsName, nsName+strlen(nsName));
SessionData* sd = reinterpret_cast<SessionData*>(userData); SessionData* sd = reinterpret_cast<SessionData*>(userData);
std::string characters; std::string characters;
if(sd->stm_->needsCharactersBuffering()) { if(sd->psm_->needsCharactersBuffering()) {
characters = sd->charactersStack_.front(); characters = sd->charactersStack_.front();
sd->charactersStack_.pop_front(); sd->charactersStack_.pop_front();
} }
sd->stm_->endElement(localname, prefix, nsUri, characters); sd->psm_->endElement(localname, prefix, nsUri, characters);
delete [] nsUri;
} }
} // namespace } // namespace
@ -134,113 +141,81 @@ namespace {
void mlCharacters(void* userData, const char* ch, int len) void mlCharacters(void* userData, const char* ch, int len)
{ {
SessionData* sd = reinterpret_cast<SessionData*>(userData); SessionData* sd = reinterpret_cast<SessionData*>(userData);
if(sd->stm_->needsCharactersBuffering()) { if(sd->psm_->needsCharactersBuffering()) {
sd->charactersStack_.front().append(&ch[0], &ch[len]); sd->charactersStack_.front().append(&ch[0], &ch[len]);
} }
} }
} // namespace } // namespace
XmlParser::XmlParser(ParserStateMachine* psm)
: psm_(psm)
{}
XmlParser::~XmlParser() {}
namespace { namespace {
XML_Parser createParser(const SharedHandle<SessionData>& sessionData) XML_Parser createParser(SessionData* sd)
{ {
XML_Parser parser = XML_ParserCreateNS(0, static_cast<const XML_Char>('\t')); XML_Parser parser = XML_ParserCreateNS(0, static_cast<const XML_Char>('\t'));
XML_SetUserData(parser, sessionData.get()); XML_SetUserData(parser, sd);
XML_SetElementHandler(parser, &mlStartElement, &mlEndElement); XML_SetElementHandler(parser, &mlStartElement, &mlEndElement);
XML_SetCharacterDataHandler(parser, &mlCharacters); XML_SetCharacterDataHandler(parser, &mlCharacters);
return parser; return parser;
} }
} // namespace } // namespace
namespace { bool XmlParser::parseFile(const char* filename)
void checkError(XML_Parser parser)
{
if(XML_Parse(parser, 0, 0, 1) == XML_STATUS_ERROR) {
throw DL_ABORT_EX2(MSG_CANNOT_PARSE_METALINK,
error_code::METALINK_PARSE_ERROR);
}
SessionData* sessionData =
reinterpret_cast<SessionData*>(XML_GetUserData(parser));
const SharedHandle<MetalinkParserStateMachine>& stm = sessionData->stm_;
if(!stm->finished()) {
throw DL_ABORT_EX2(MSG_CANNOT_PARSE_METALINK,
error_code::METALINK_PARSE_ERROR);
}
if(!stm->getErrors().empty()) {
throw DL_ABORT_EX2(stm->getErrorString(),
error_code::METALINK_PARSE_ERROR);
}
}
} // namespace
SharedHandle<Metalinker>
MetalinkProcessor::parseFile
(const std::string& filename,
const std::string& baseUri)
{ {
BufferedFile* fp = 0; BufferedFile* fp = 0;
auto_delete_d<BufferedFile*> deleter(fp); if(strcmp(filename, DEV_STDIN) == 0) {
if(filename == DEV_STDIN) {
fp = new BufferedFile(stdin); fp = new BufferedFile(stdin);
return parseFile(*fp); auto_delete_d<BufferedFile*> deleter(fp);
return parseFile(fp);
} else { } else {
fp = new BufferedFile(filename, BufferedFile::READ); fp = new BufferedFile(filename, BufferedFile::READ);
return parseFile(*fp, baseUri); auto_delete_d<BufferedFile*> deleter(fp);
return parseFile(fp);
} }
} }
SharedHandle<Metalinker> bool XmlParser::parseFile(BufferedFile* fp)
MetalinkProcessor::parseFile
(BufferedFile& fp, const std::string& baseUri)
{ {
stm_.reset(new MetalinkParserStateMachine());
stm_->setBaseUri(baseUri);
char buf[4096]; char buf[4096];
SessionData sessionData(psm_);
SharedHandle<SessionData> sessionData(new SessionData(stm_)); XML_Parser parser = createParser(&sessionData);
XML_Parser parser = createParser(sessionData);
auto_delete<XML_Parser> deleter(parser, XML_ParserFree); auto_delete<XML_Parser> deleter(parser, XML_ParserFree);
while(1) { while(1) {
size_t res = fp.read(buf, sizeof(buf)); size_t res = fp->read(buf, sizeof(buf));
if(XML_Parse(parser, buf, res, 0) == XML_STATUS_ERROR) { if(XML_Parse(parser, buf, res, 0) == XML_STATUS_ERROR) {
throw DL_ABORT_EX2(MSG_CANNOT_PARSE_METALINK, return false;
error_code::METALINK_PARSE_ERROR);
} }
if(res < sizeof(buf)) { if(res < sizeof(buf)) {
break; break;
} }
} }
checkError(parser); return XML_Parse(parser, 0, 0, 1) != XML_STATUS_ERROR && psm_->finished();
return stm_->getResult();
} }
SharedHandle<Metalinker> bool XmlParser::parseBinaryStream(BinaryStream* bs)
MetalinkProcessor::parseFromBinaryStream
(const SharedHandle<BinaryStream>& binaryStream,
const std::string& baseUri)
{ {
stm_.reset(new MetalinkParserStateMachine()); const ssize_t bufSize = 4096;
stm_->setBaseUri(baseUri);
ssize_t bufSize = 4096;
unsigned char buf[bufSize]; unsigned char buf[bufSize];
SessionData sessionData(psm_);
SharedHandle<SessionData> sessionData(new SessionData(stm_)); XML_Parser parser = createParser(&sessionData);
XML_Parser parser = createParser(sessionData);
auto_delete<XML_Parser> deleter(parser, XML_ParserFree); auto_delete<XML_Parser> deleter(parser, XML_ParserFree);
off_t readOffset = 0; off_t readOffset = 0;
while(1) { while(1) {
ssize_t res = binaryStream->readData(buf, bufSize, readOffset); ssize_t res = bs->readData(buf, bufSize, readOffset);
if(res == 0) { if(res == 0) {
break; break;
} }
if(XML_Parse(parser, reinterpret_cast<const char*>(buf), res, 0) == if(XML_Parse(parser, reinterpret_cast<const char*>(buf), res, 0) ==
XML_STATUS_ERROR) { XML_STATUS_ERROR) {
throw DL_ABORT_EX2(MSG_CANNOT_PARSE_METALINK, return false;
error_code::METALINK_PARSE_ERROR);
} }
readOffset += res; readOffset += res;
} }
checkError(parser); return XML_Parse(parser, 0, 0, 1) != XML_STATUS_ERROR && psm_->finished();
return stm_->getResult();
} }
} // namespace aria2 } // namespace aria2

View File

@ -2,7 +2,7 @@
/* /*
* aria2 - The high speed download utility * aria2 - The high speed download utility
* *
* Copyright (C) 2006 Tatsuhiro Tsujikawa * Copyright (C) 2011 Tatsuhiro Tsujikawa
* *
* This program is free software; you can redistribute it and/or modify * This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by * it under the terms of the GNU General Public License as published by
@ -32,42 +32,33 @@
* files in the program, then also delete it here. * files in the program, then also delete it here.
*/ */
/* copyright --> */ /* copyright --> */
#ifndef D_EXPAT_METALINK_PROCESSOR_H #ifndef D_EXPAT_XML_PARSER_H
#define D_EXPAT_METALINK_PROCESSOR_H #define D_EXPAT_XML_PARSER_H
#include "common.h" #include "common.h"
#include <string> #include <cstdlib>
#include <expat.h>
#include "SharedHandle.h"
#include "A2STR.h"
namespace aria2 { namespace aria2 {
class Metalinker;
class BinaryStream; class BinaryStream;
class MetalinkParserStateMachine; class ParserStateMachine;
class BufferedFile; class BufferedFile;
class MetalinkProcessor { class XmlParser {
private:
SharedHandle<MetalinkParserStateMachine> stm_;
SharedHandle<Metalinker> parseFile
(BufferedFile& fp,
const std::string& baseUri = A2STR::NIL);
public: public:
SharedHandle<Metalinker> parseFile // This object does not delete psm.
(const std::string& filename, XmlParser(ParserStateMachine* psm);
const std::string& baseUri = A2STR::NIL); ~XmlParser();
bool parseFile(const char* filename);
bool parseBinaryStream(BinaryStream* binaryStream);
bool parseMemory(const char* xml, size_t size);
private:
bool parseFile(BufferedFile* fp);
SharedHandle<Metalinker> parseFromBinaryStream ParserStateMachine* psm_;
(const SharedHandle<BinaryStream>& binaryStream,
const std::string& baseUri = A2STR::NIL);
}; };
} // namespace aria2 } // namespace aria2
#endif // D_EXPAT_METALINK_PROCESSOR_H #endif // D_EXPAT_XML_PARSER_H

View File

@ -43,8 +43,6 @@
#include "a2io.h" #include "a2io.h"
#include "BinaryStream.h" #include "BinaryStream.h"
#include "ParserStateMachine.h" #include "ParserStateMachine.h"
#include "message.h"
#include "DlAbortEx.h"
#include "A2STR.h" #include "A2STR.h"
#include "a2functional.h" #include "a2functional.h"
#include "XmlAttr.h" #include "XmlAttr.h"

View File

@ -40,7 +40,7 @@
#ifdef HAVE_LIBXML2 #ifdef HAVE_LIBXML2
# include "Xml2XmlParser.h" # include "Xml2XmlParser.h"
#elif HAVE_LIBEXPAT #elif HAVE_LIBEXPAT
# include "Expat2XmlParser.h" # include "ExpatXmlParser.h"
#endif #endif
#endif // D_XML_PARSER_H #endif // D_XML_PARSER_H

View File

@ -1,4 +1,4 @@
#include "MetalinkProcessor.h" #include "metalink_helper.h"
#include <iostream> #include <iostream>
@ -21,7 +21,6 @@
#include "fmt.h" #include "fmt.h"
#include "RecoverableException.h" #include "RecoverableException.h"
#include "util.h" #include "util.h"
#include "metalink_helper.h"
namespace aria2 { namespace aria2 {