//
// Copyright (c) 2016, Sequence Logic
//
#ifndef OCR_BASE
#define OCR_BASE

#include <string>
#include <vector>
#include <mutex>
#include <iostream>
#include <string>
#include <fstream>
#include <ctime>
#include <vector>
#include <sstream>

#include "Mutex.h"


#define DEFAULT_THREAD_COUNT 1
#define DEFAULT_TMP_DIR "/tmp/slocr/"

namespace sequencelogic
{
    struct OCROpts
    {
        bool        _bPrettyOutput;
        bool        _bSinglePageSave;
        int         _numPagesInError;
        int         _nThreadCount;
        int         _nStartPage;
        int         _nEndPage;
        std::string _inFile;
        std::string _tempFolder;
        std::string _singlePageFolder;

        OCROpts() : _bPrettyOutput(false), _bSinglePageSave(false), _numPagesInError(0), _nThreadCount(DEFAULT_THREAD_COUNT), _nStartPage(0), _nEndPage(0), _tempFolder(DEFAULT_TMP_DIR), _singlePageFolder("") {}
    };

	/**
	* A base class for an OCR engine.
	*/
	class COCRBase
	{
	public:
		COCRBase() {}
		COCRBase(std::string &inFile);
		virtual ~COCRBase() {}

		/**
		* Recognize/OCR something.
		*/
        int Recognize(const OCROpts &opts);

		/**
		* Validates that the OCR engine is configured correctly.
		*/
		bool Validate();

	protected:
#pragma region Members

		std::string getXMLName(const std::string inFile);


		void copyFile(const std::string& startFile, const std::string& destFile);

		/**
		* The number of threads the program will utalize.
		*/
		int threadCount;

		/**
		* Specified weather or not the image is a tif
		*/
		bool isTIF;

		/**
		* Specified weather or not the image is a pdf
		*/
		bool isPDF;

		/**
		* The name of the origional image file
		*/
		std::string imgFile;

		/**
		* The name of the fixed image file
		*/
		std::string fixedImgFile;

		std::string fixedImgPageFile;

		/**
		* The number of pages allowed to be in error, before bailing out.
		*/
		int nPagesinErrToAllow;

		/**
		* The name of the temporary fixed image file
		*/
		std::string tmpImgFile;
#pragma endregion Members to be inherited by the base classes.

		/**
		* Prints a message to the console in a "thread safe" way.
		*/
		void printMessage(std::string msg);
		/**
		* OCR a single file.
		*/
        int OCRFile(const OCROpts &opts);
		/**
		* OCR all files in a given directory.
		*/
        int OCRFiles(const OCROpts &opts);

		/*
		* The one, and only, interface. For now...
		*/
        virtual int OCR(const OCROpts &opts, const std::string &outFile) = 0;
		
		virtual int OCRMT(const std::string &inFile, int threadCount = 2) = 0;

		/**
		* Checks to see if the engine is properly set up
		*/
		virtual bool verify() = 0;

		/**
		* Get a file type string, based on it's extension.
		*/
		std::string getTypeByFileExt(const std::string &fileName);

		/**
		* Get the value of a binary number in base 10.
		*/
		long fromBin(long n);

		/**
		* Converts the 'error' nuance gives to a confidence value between 1-9.
		*/
		int convertConfidence(long confidence);

		/**
		* Gets the bit from a byte at the specified position.
		*/
		bool getBit(unsigned char byte, int position);

		/**
		* Gets the maximum value between start and stop in the array vals.
		*/
		int getMax(int start, int stop, int* vals);

		/**
		* Gets the minimum value between start and stop in the array vals.
		*/
		int getMin(int start, int stop, int* vals);

		/**
		* Gets the number of words (seperated by spaces) in the given array.
		*/
		int getNumWords(char chars[], int numChars);

		/**
		* Returns true if the given fileName is a tif image.
		*/
		bool isTIFImg(const std::string &fileName);

		/**
		* Returns true if the given fileName is a PDF image.
		*/
		bool isPDFImg(const std::string &fileName);

		/**
		* Gets the name for the "fixed" image.
		*/
		std::string getFixedImgName(const std::string inFile);

		std::string getFixedImgName(const std::string inFile, int startPage, int endPage);

		/**
		* Gets the temporary name for the "fixed" image.
		*/
		std::string getTmpImgName(const std::string &tmpDir, const  std::string &inFile);

		//---------------------FOR TESTING---------------------------------------------------------------------------
		std::string getFixedImgName(const std::string inFile, int pageNum);

		std::string getJSONName(const std::string &inFile, int pageNum);

	private:
		/**
		* Get all files in a given directory. Returns the number of files found.
		*/
		size_t getFilesInDirectory(const std::string &dirName, std::vector<std::string> &files);

		/**
		* The mutex to prevent messages causing a race condition.
		*/
		Mutex messageLock;
	};
};

#endif