近來LingosHook更新很慢，是因為比較忙，當(dāng)然是工作了。。。（我就不說我的夜生活很豐富。。。）
找到一個HTML分析重復(fù)的問題，就是Tidy和PreProcess兩個過程有重復(fù)操作，導(dǎo)致Dict數(shù)據(jù)定位不準確，終于還是讓我‘想起來’了。。。這樣下個版本應(yīng)該快了～
這里貼下HtmlDictParser對象，此對象用于分析HTML中的Dict數(shù)據(jù)，跟DictObject的區(qū)別是其只分析Dict本身數(shù)據(jù)，如ID，和單詞數(shù)據(jù)外，不再像DictObject對象那樣要分析具體詞典結(jié)果數(shù)據(jù)。HtmlDictParser是LingosHook擺脫Dict限制的主要對象。
#ifndef __HTMLDICTPARSER_H__
#define __HTMLDICTPARSER_H__
#include <map>
#include <vector>
#include "wx/wx.h"
#include "DBAccess.h"
#include "TinyHtmlParser.h"
namespace HtmlDictParser
{
struct TDictConfig
{
int m_iLoadParam;
int m_iStoreParam;
};
typedef std::map<int, TDictConfig> TDictConfigMap;//index + config
struct TDictInfo
{
std::wstring m_strDictID;
std::wstring m_strTitle;
TDictConfig m_stConfig;
};
typedef std::map<std::wstring, int> TDictIDMap;//dictid + dictindex
typedef std::map<int, TDictInfo> TDictIndexMap;//dictindex + info
class CDictInfoObject
{
public:
CDictInfoObject() {}
virtual ~CDictInfoObject() {}
int Init(CDBAccess::TDatabase& db);
int Insert(int index, const TDictInfo& info);
int GetDictIndex(const std::wstring& id) const;
protected:
TDictIDMap _mapDictID;
TDictIndexMap _mapDictIndex;
};
struct TDictResult
{
int m_iDictIndex;
int m_iDictStart;
int m_iDictEnd;
};
typedef std::vector<TDictResult> TDictResultVector;
typedef std::map<std::wstring, TDictResultVector> TDictResultMap;
class CParser
{
public:
CParser() {}
virtual ~CParser() {}
virtual int Init(CDBAccess::TDatabase& db);
virtual int ParserHTML(const std::wstring& html, TDictResultVector& result);
virtual int ParserHTML(CDBAccess::TDatabase &db, const std::wstring& dictid, const std::wstring& html, TinyHtmlParser::CDocumentObject& doc, const TinyHtmlParser::CElementObject* dict, TDictResultMap& result);
virtual int SaveResult(CDBAccess::TDatabase& db, int wordid, const TDictResultMap& result);
virtual int GetResult(CDBAccess::TDatabase& db, int wordid, TDictResultMap& result);
virtual int RemoveResult(CDBAccess::TDatabase& db, int wordid);
virtual int GenHtmlResult(const TDictResultVector& vct, const std::wstring& html, wxString& result) const;
protected:
int CheckDictHtml();
int UpdateDictInfo(CDBAccess::TDatabase &db, const std::wstring& dictid, const std::wstring& html, TinyHtmlParser::CDocumentObject& doc, const TinyHtmlParser::CElementObject* dict);
int UpdateDictInfo(CDBAccess::TDatabase &db, const std::wstring& dictid, const std::wstring& title);
protected:
CDictInfoObject _objDictInfo;
};
}
#endif