pdfproject/MFCApplication1/AssignWords.cpp

#include "pch.h"
#include "AssignWords.h"
#include "CvxText.h"
#include <iostream>
#include <sstream>
#include <fstream>
#include <io.h>
#include "BaseUtility.h"
/*
string& replace_str(string& str, const string& to_replaced, const string& newchars)
{
	for (string::size_type pos(0); pos != string::npos; pos += newchars.length())
	{
		pos = str.find(to_replaced, pos);
		if (pos != string::npos)
			str.replace(pos, to_replaced.length(), newchars);
		else
			break;
	}
	return   str;
}
*/


int preNUm(unsigned char byte) {
	unsigned char mask = 0x80;
	int num = 0;
	for (int i = 0; i < 8; i++) {
		if ((byte & mask) == mask) {
			mask = mask >> 1;
			num++;
		}
		else {
			break;
		}
	}
	return num;
}


bool isUtf8(unsigned char* data, int len) {
	int num = 0;
	int i = 0;
	while (i < len) {
		if ((data[i] & 0x80) == 0x00) {
			// 0XXX_XXXX
			i++;
			continue;
		}
		else if ((num = preNUm(data[i])) > 2) {
			// 110X_XXXX 10XX_XXXX
			// 1110_XXXX 10XX_XXXX 10XX_XXXX
			// 1111_0XXX 10XX_XXXX 10XX_XXXX 10XX_XXXX
			// 1111_10XX 10XX_XXXX 10XX_XXXX 10XX_XXXX 10XX_XXXX
			// 1111_110X 10XX_XXXX 10XX_XXXX 10XX_XXXX 10XX_XXXX 10XX_XXXX
			// preNUm() 返回首个字节8个bits中首 ?0bit前面1bit的个数，该数量也是该字符所使用的字节数        
			i++;
			for (int j = 0; j < num - 1; j++) {
				//判断后面num - 1 个字节是不是都是10开
				if ((data[i] & 0xc0) != 0x80) {
					return false;
				}
				i++;
			}
		}
		else {
			//其他情况说明不是utf-8
			return false;
		}
	}
	return true;
}

void split(const std::string& s, const std::string& delim, std::vector< std::string >* ret)
{
	size_t last = 0;
	size_t index = s.find_first_of(delim, last);
	while (index != std::string::npos)
	{
		ret->push_back(s.substr(last, index - last));
		last = index + 1;
		index = s.find_first_of(delim, last);
	}
	if (index - last > 0)
	{
		ret->push_back(s.substr(last, index - last));
	}
}

void deleteFlagWord(std::string& str, const std::string& to_replaced)
{
	std::string::size_type pos = str.find(to_replaced, 0);
	if (pos == 0)
		str.replace(pos, to_replaced.length(), "");
}

void ReplaceFlagWord(std::vector<std::string> &list, std::string &strRet)
{
	for (std::vector<std::string>::iterator it = list.begin(); it != list.end(); it++)
	{
		deleteFlagWord(strRet, (*it));
	}
}


void DivideEquallyStr(const char *buff, int len, char left[], char right[])
{
	std::string strtemp = buff;
	std::size_t keyindex1 = std::string::npos, keyindex2 = std::string::npos, keyindex;
	int i = 0;
	for (i = 0; i < len / 2; i++)
	{
		if (!(buff[i] < 0x80 && buff[i] >= 0x0))
		{
			i++;
		}
	}
	strncpy(left, buff, i);
	strncpy(right, buff + i, len - i);
	return;
	/* 直接中间分开，不用特殊处理*/
	std::string strKey1 = "；", strKey2 = " ";
	std::size_t found = strtemp.find(strKey1);
	bool flag1 = false, flag2 = false;
	while (found != std::string::npos && found < strtemp.length() / 2)
	{
		keyindex1 = found;
		found = strtemp.find(strKey1, found + 1);
		flag1 = true;
	}
	int w = 1;
	found = strtemp.find(strKey2);
	while (found != std::string::npos && found < strtemp.length() / 2)
	{
		keyindex2 = found;
		found = strtemp.find(strKey2, found + 1);
		flag2 = true;
	}
	if (flag1 && flag2)
	{
		if (keyindex1 > keyindex2)
		{
			keyindex = keyindex1;
			w = 2;
		}
		else
		{
			keyindex = keyindex2;
		}
	}
	else if (flag1)
	{
		keyindex = keyindex1;
		w = 2;
	}
	else if (flag2)
	{
		keyindex = keyindex2;
	}
	else 
	{
		int i = 0;
		for (i = 0; i < len / 2; i++)
		{
			if (!(buff[i] < 0x80 && buff[i] >= 0x0))
			{
				i++;
			}
		}
		strncpy(left, buff, i);
		strncpy(right, buff + i, len - i);
		return;
	}
	std::string ttm = strtemp.substr(0, keyindex + w);
	std::string ttm2 = strtemp.substr(keyindex + w, strtemp.size());
	strncpy(left, ttm.c_str(), ttm.length());
	strncpy(right, ttm2.c_str(), ttm2.length());
}

void RandReplaceStr(std::vector<std::string> list, std::string &ret)
{
	std::string to_replaced = "；";
	srand((unsigned)time(NULL));
	std::string newchars;
	int index = rand() % list.size();
	newchars = list[index];
	for (string::size_type pos(0); pos != string::npos; )
	{
		pos = ret.find(to_replaced, pos);
		if (pos != string::npos)
			ret.replace(pos, to_replaced.length(), newchars);
		else
			break;
		pos += newchars.length();
		index = rand() % list.size();
		newchars = list[index];
	}
}

void AssignWordsFromString(std::string strContent, int maxTwoWith, int maxTreeWith, int maxFourWith, std::vector<std::string> &rePlaceList,
	std::vector<std::string> &twoList, std::vector<std::string> &threeList, std::vector<std::string> &fourList)
{
	cv::Size WordsWith = GetTextSize(strContent.c_str(), font_size, font_family);
	if (WordsWith.width < maxFourWith)
	{
		if (rePlaceList.size() > 0)
		{
			RandReplaceStr(rePlaceList, strContent);
		}
		fourList.push_back(strContent);
	}
	else if (WordsWith.width < maxTreeWith)
	{
		if (rePlaceList.size() > 0)
		{
			RandReplaceStr(rePlaceList, strContent);
		}
		threeList.push_back(strContent);
	}
	else if (WordsWith.width < maxTwoWith)
	{
		if (rePlaceList.size() > 0)
		{
			RandReplaceStr(rePlaceList, strContent);
		}
		twoList.push_back(strContent);
	}
	else
	{
		char left[2048] = { 0 }, right[2048] = { 0 };
		DivideEquallyStr(strContent.c_str(), strContent.length(), left, right);
		AssignWordsFromString(left, maxTwoWith, maxTreeWith, maxFourWith, rePlaceList, twoList, threeList, fourList);
		AssignWordsFromString(right, maxTwoWith, maxTreeWith, maxFourWith, rePlaceList, twoList, threeList, fourList);
	}
}



int AssignWordsFromTest(std::string pathName, std::vector<std::string> &twoList, std::vector<std::string> &threeList, std::vector<std::string> &fourList)
{
	// 词性处理
	HMODULE module = GetModuleHandle(0);
	TCHAR pFileName[MAX_PATH + 2] = { 0 };
	GetModuleFileName(module, pFileName, MAX_PATH);

	CString csFullPath(pFileName);
	int nPos = csFullPath.ReverseFind(_T('\\'));
	CString filepath = csFullPath.Left(nPos);
	CString FilePath1 = filepath + L"\\config.ini";
	CString FilePath2 = filepath + L"\\words.txt";
	WCHAR wflagWord[10],wFourColScaleParam[10],wThreeColScaleParam[10],wTwoColScaleParam[10];
	GetPrivateProfileString(L"USER", L"flagWord", L"1", wflagWord, 10, FilePath1);
	GetPrivateProfileString(L"USER", L"TwoColScaleParam", L"6", wTwoColScaleParam, 10, FilePath1);
	GetPrivateProfileString(L"USER", L"ThreeColScaleParam", L"9", wThreeColScaleParam, 10, FilePath1);
	GetPrivateProfileString(L"USER", L"FourColScaleParam", L"16", wFourColScaleParam, 10, FilePath1);
	int nFourColScaleParam = 0, nThreeColScaleParam = 0, nTwoColScaleParam = 0;
	nFourColScaleParam = _wtoi(wFourColScaleParam);
	nFourColScaleParam = nFourColScaleParam > 0 ? nFourColScaleParam : 16;
	nThreeColScaleParam = _wtoi(wThreeColScaleParam);
	nThreeColScaleParam = nThreeColScaleParam > 0 ? nThreeColScaleParam : 9;
	nTwoColScaleParam = _wtoi(wTwoColScaleParam);
	nTwoColScaleParam = nTwoColScaleParam > 0 ? nTwoColScaleParam : 6;
	CString lpszflagWord(wflagWord);
	std::vector<std::string> WordList;
	WordList.push_back("vt.");
	WordList.push_back("vi.");
	WordList.push_back("n.");
	WordList.push_back("pron.");
	WordList.push_back("adj.");
	WordList.push_back("num.");
	WordList.push_back("v.");
	WordList.push_back("adv.");
	WordList.push_back("art.");
	WordList.push_back("prep.");
	WordList.push_back("conj.");
	WordList.push_back("int.");

	// 随机替换规则
	std::vector<std::string> rePlaceList;
	std::string strFilePath2 = CT2A(FilePath2);
	if (_access(strFilePath2.c_str(), 00) != -1)
	{
		char line[100] = { 0 };
		std::ifstream finfile(strFilePath2.c_str(), std::ios::in);
		while (finfile.getline(line, sizeof(line)))
		{
			std::vector<std::string> retlist;
			std::string strTemp = line;
			UTF8toANSI(strTemp);
			split(strTemp, "@", &retlist);
			if (retlist.size() == 2)
			{
				int x = atoi(retlist[1].c_str());
				if (x > 0)
				{
					for (int i = 0; i < x; i++)
					{
						rePlaceList.push_back(retlist[0]);
					}
				}
			}
		}
		memset(line, 0, 100);
	}

	int cout = 0;
	if (_access(pathName.c_str(), 00) != -1)
	{
		int maxTwoWith = main_wdith / 2 - main_wdith / nTwoColScaleParam - chk_width;
		int maxTreeWith = main_wdith / 3 - main_wdith / nThreeColScaleParam - chk_width;
		int maxFourWith = main_wdith  / 4 - main_wdith/ nFourColScaleParam - chk_width;
		std::ifstream fin(pathName.c_str(), std::ios::in);
		char line[4096] = { 0 };
		while (fin.getline(line, sizeof(line)))
		{
			std::string strTemp = line;
			if(isUtf8((unsigned char*)strTemp.c_str(),strTemp.length()))
				UTF8toANSI(strTemp);
			if (lpszflagWord == "0")
			{
				ReplaceFlagWord(WordList, strTemp);
			}
			if (strTemp == "")
				continue;
			
			cv::Size WordsWith = GetTextSize(strTemp.c_str(), font_size, font_family);
			if (WordsWith.width < maxFourWith)
			{
				if (rePlaceList.size() > 0)
				{
					RandReplaceStr(rePlaceList, strTemp);
				}
				fourList.push_back(strTemp);
			}
			else if (WordsWith.width < maxTreeWith)
			{
				if (rePlaceList.size() > 0)
				{
					RandReplaceStr(rePlaceList, strTemp);
				}
				threeList.push_back(strTemp);
			}
			else if (WordsWith.width < maxTwoWith)
			{
				if (rePlaceList.size() > 0)
				{
					RandReplaceStr(rePlaceList, strTemp);
				}
				twoList.push_back(strTemp);
			}
			else
			{
				char left[2048] = { 0 }, right[2048] = { 0 };
				DivideEquallyStr(strTemp.c_str(), strTemp.length(), left,right);
				AssignWordsFromString(left, maxTwoWith, maxTreeWith, maxFourWith, rePlaceList, twoList, threeList, fourList);
				AssignWordsFromString(right, maxTwoWith, maxTreeWith, maxFourWith, rePlaceList, twoList, threeList, fourList);
			}
			memset(line, 0, 4096);
		}
		
	}
	return cout;
}