Extractor.h

Go to the documentation of this file.

//      Copyright (c) 2024 Eggplant Ltd
//      All Rights Reserved

#pragma once
#pragma warning (disable : 4251)
#pragma warning (disable : 4635)

#include <string>
#include <vector>
using namespace std;

#include "fc.h"
#include "Action.h"
#include "ExtractionCursor.h"
#include "RegExpMatch.h"
#include "CharacterEncoding.h"
#include "VirtualUserInterface.h"

using namespace Facilita;

namespace Facilita {
        
        class FC_API Extractor: public CharacterEncoding
        {
        public:
                Extractor();
                Extractor(const string& data);
                Extractor(const string& path, int dummy);
                Extractor(const Extractor& other);
                virtual ~Extractor();
                Extractor& operator=( const Extractor& other);
                Extractor& operator=( const string& data);
                bool find(ExtractionCursor &cursor, const string& text, bool caseSensitive = true);
                bool rfind(ExtractionCursor &cursor, const string& text, bool caseSensitive = true);
                string extract(ExtractionCursor &cursor, const string& before, const string& after, const string& defaultValue, bool caseSensitive = true);

                string extractFromEnd(ExtractionCursor &cursor, const string& after, const string& before, const string& defaultValue, bool caseSensitive = true);

                string extract(ExtractionCursor &cursor, const StringVector& before, const string& after, const string& defaultValue, bool caseSensitive = true);

                string extractFromEnd(ExtractionCursor &cursor, const StringVector& after, const string& before, const string& defaultValue, bool caseSensitive = true);


                string extract(ExtractionCursor &cursor, const StringVector& before, const string& after, bool caseSensitive = true);


                string extractFromEnd(ExtractionCursor &cursor, const StringVector& after, const string& before, bool caseSensitive = true);


                string extract(ExtractionCursor &cursor, const string& before, const string& after, bool caseSensitive = true);

                string extractFromEnd(ExtractionCursor &cursor, const string& after, const string& before, bool caseSensitive = true);

                RegExpMatchList extractRegExp(ExtractionCursor &cursor, const string& regExp, bool caseSensitive = true);

                string extractRegExp(ExtractionCursor &cursor, const string& beforeRegExp, const string& afterRegExp, bool caseSensitive = true);
                string extractRegExpFromEnd(ExtractionCursor &cursor, const string& afterRegExp, const string& beforeRegExp, bool caseSensitive = true);

                bool verifyContains(const string &text, bool caseSensitive = true);     // check if content contains the pattern

                virtual StringVector extractList(const string& before, const string& after, bool caseSensitive = true, int maxItems = -1);
                virtual StringVector extractList(ExtractionCursor &cursor, const string& before, const string& after, const string& endAt = "", bool caseSensitive = true, int maxItems = -1);
                
                const string path() const { return _path;}
                const string& data() const { return _data;}
        
#ifndef SWIG
                int getCurrentPosition() const { return _currentPosition;}
                int& getCurrentPositionRef(){ return _currentPosition;}
                void setCurrentPosition(int position) {_currentPosition = position;}
                void incrementCurrentPosition(int inc=1) {_currentPosition += inc;}
                
                void rewind() {_currentPosition = _rewindPosition;}
                void advanceRewind() {_rewindPosition = _currentPosition;}
                void setEndSearch() {_endSearchPosition = _currentPosition;}
                
                string stringAtCurrentPosition(int len) const;
                const char* getContentPtr(void) { return _data.data();}
                int findString(int startPos, const string& str, bool caseSensitive = true);
                int rfindString(int startPos, const string& str, bool caseSensitive = true);
                bool findPtr(int startPos, const StringVector& before, const char* pAfterString, bool casesensitive, char** pMatchedString, int& matchedStringPos, int& matchedStringlength);

                bool rfindPtr(int startPos, const StringVector& after, const char* pBeforeString, bool casesensitive, char** pMatchedString, int& matchedStringPos, int& matchedStringlength);

                int findString(int startPos, const char* pBeforeString, bool casesensitive);
                int rfindString(int startPos, const char* str);
                int rfindStringIc(int startPos, const char* str);
                const char* getKeywordValue(const char* keyword, int& valueLen, bool bRemoveQuotes);
                int findStringCheckIc(int startPos, const char* string, const char checkChar, bool skipQuotes, int endPos=0);
                int findStringCheck(int startPos, const char* str, const char checkChar);
                void skipToStringIc(const char* pTag);
                virtual void setCharEncoding(const string& charEncoding);
                virtual string getCharEncoding();
#endif // SWIG
        private:
        #pragma warning (disable : 4251)
                string _data;
                string _path;
                string _charEncoding;
        #pragma warning(default:4251)
                int _currentPosition;                           // used when parsing
                int _rewindPosition;
                int _endSearchPosition;
        };
        
        class FC_API ExtractorWithActionsInterface: public CharacterEncoding
        {
        public:
                virtual ~ExtractorWithActionsInterface() {}
                virtual bool find(ExtractionCursor &cursor, const string& text, bool caseSensitive)= 0;
                virtual bool rfind(ExtractionCursor &cursor, const string& text, bool caseSensitive)= 0;
                
                virtual string extract(ExtractionCursor &cursor, const string& before, const string& after, const string& defaultValue, bool caseSensitive)= 0;
                virtual string extractFromEnd(ExtractionCursor &cursor, const string& after, const string& before, const string& defaultValue, bool caseSensitive) = 0;

                virtual string extract(ExtractionCursor &cursor, const StringVector& before, const string& after, const string& defaultValue, bool caseSensitive)= 0;
                virtual string extractFromEnd(ExtractionCursor &cursor, const StringVector& after, const string& before, const string& defaultValue, bool caseSensitive) = 0;

                virtual string extract(ExtractionCursor &cursor, const StringVector& before, const string& after, bool caseSensitive)= 0;
                virtual string extractFromEnd(ExtractionCursor &cursor, const StringVector& after, const string& before, bool caseSensitive) = 0;

                virtual string extract(ExtractionCursor &cursor, const string& before, const string& after, bool caseSensitive)= 0;
                virtual string extractFromEnd(ExtractionCursor &cursor, const string& after, const string& before, bool caseSensitive) = 0;

                virtual RegExpMatchList extractRegExp(ExtractionCursor &cursor, const string& regExp, bool caseSensitive)= 0;
                virtual string extractRegExp(ExtractionCursor &cursor, const string& beforeRegExp, const string& afterRegExp, bool caseSensitive = true) = 0;
                virtual string extractRegExpFromEnd(ExtractionCursor &cursor, const string& afterRegExp, const string& beforeRegExp, bool caseSensitive = true) = 0;
                virtual bool verifyContains(const string &text, bool caseSensitive) = 0;        // check if content contains the pattern
                virtual bool find(ExtractionCursor &cursor, const string& text, ActionType failAction, bool caseSensitive) = 0;
                virtual bool rfind(ExtractionCursor &cursor, const string& text, ActionType failAction, bool caseSensitive) = 0;


                virtual string extract(ExtractionCursor &cursor, const string& before, const string& after, ActionType failAction, bool caseSensitive) = 0;
                virtual string extractFromEnd(ExtractionCursor &cursor, const string& after, const string& before, ActionType failAction, bool caseSensitive) = 0;

                virtual string extract(ExtractionCursor &cursor, const StringVector& before, const string& after, ActionType failAction, bool caseSensitive) = 0;
                virtual string extractFromEnd(ExtractionCursor &cursor, const StringVector& after, const string& before, ActionType failAction, bool caseSensitive) = 0;
                
                virtual string extractRegExp(ExtractionCursor &cursor, const string& beforeRegExp, const string& afterRegExp, ActionType failAction, bool caseSensitive = true) = 0;
                virtual string extractRegExpFromEnd(ExtractionCursor &cursor, const string& afterRegExp, const string& beforeRegExp, ActionType failAction, bool caseSensitive = true) = 0;
                virtual bool verifyContains(const string &text, ActionType failAction, bool caseSensitive) = 0;
        };

        class FC_API ExtractorWithActions: public Extractor
        {
        public:
                ExtractorWithActions();
                ExtractorWithActions(VirtualUserInterface& vu, const string& data);
                ExtractorWithActions(VirtualUserInterface& vu, const string& path, int dummy);
                ExtractorWithActions(const ExtractorWithActions& other);
                virtual ~ExtractorWithActions();
                ExtractorWithActions& operator=( const ExtractorWithActions& other);
                bool find(ExtractionCursor& cursor, const string& text, bool caseSensitive = true);
                bool rfind(ExtractionCursor &cursor, const string& text, bool caseSensitive = true);
                string extract(ExtractionCursor &cursor, const string& before, const string& after, const string& defaultValue, bool caseSensitive = true);

                string extractFromEnd(ExtractionCursor &cursor, const string& after, const string& before, const string& defaultValue, bool caseSensitive = true);

                string extract(ExtractionCursor &cursor, const StringVector& before, const string& after, const string& defaultValue, bool caseSensitive = true);

                string extractFromEnd(ExtractionCursor &cursor, const StringVector& after, const string& before, const string& defaultValue, bool caseSensitive = true);

                string extract(ExtractionCursor &cursor, const StringVector& before, const string& after, bool caseSensitive = true);

                string extractFromEnd(ExtractionCursor &cursor, const StringVector& after, const string& before, bool caseSensitive = true);

                string extract(ExtractionCursor &cursor, const string& before, const string& after, bool caseSensitive = true);

                string extractFromEnd(ExtractionCursor &cursor, const string& after, const string& before, bool caseSensitive = true);


                RegExpMatchList extractRegExp(ExtractionCursor &cursor, const string& regExp, bool caseSensitive = true);
                string extractRegExp(ExtractionCursor &cursor, const string& beforeRegExp, const string& afterRegExp, bool caseSensitive = true);
                string extractRegExpFromEnd(ExtractionCursor &cursor, const string& afterRegExp, const string& beforeRegExp, bool caseSensitive = true);
                bool verifyContains(const string &text, bool caseSensitive = true);

                bool find(ExtractionCursor &cursor, const string& text, ActionType failAction, bool caseSensitive = true);
                bool rfind(ExtractionCursor &cursor, const string& text, ActionType failAction, bool caseSensitive = true);
                string extract(ExtractionCursor &cursor, const string& before, const string& after, ActionType failAction, bool caseSensitive = true);

                string extractFromEnd(ExtractionCursor &cursor, const string& after, const string& before, ActionType failAction, bool caseSensitive = true);

                string extract(ExtractionCursor &cursor, const StringVector& before, const string& after, ActionType failAction, bool caseSensitive = true);

                string extractFromEnd(ExtractionCursor &cursor, const StringVector& after, const string& before, ActionType failAction, bool caseSensitive = true);

                string extractRegExp(ExtractionCursor &cursor, const string& beforeRegExp, const string& afterRegExp, ActionType failAction, bool caseSensitive = true);
                string extractRegExpFromEnd(ExtractionCursor &cursor, const string& afterRegExp, const string& beforeRegExp, ActionType failAction, bool caseSensitive = true);
                bool verifyContains(const string &text, ActionType failAction, bool caseSensitive);

                VirtualUserInterface* getVU() { return _vu;}
        
        private:
                VirtualUserInterface* _vu;
        };
}