//transform lib function
//void StringParse(ifstream &ifs, string &in_file_name, vector<string> vec_str) {
void StringParse(std::string &in_file_name) {
//cut words sector
EncodingConverter trans;
std::vector<std::string> word_vec; //store word into vector
std::vector<std::string>::iterator iter_word_vec;
std::string line; //读入行
std::string content; //一次性读入全部
std::string out_file_name = in_file_name; //生成新的文件名,就是替换替换txt为utf8
std::ifstream ifs; //文本输入流
std::ofstream ofs; //文本输出流
out_file_name.erase(out_file_name.size() - 3, out_file_name.npos);
out_file_name += "utf8";
OpenInFile(ifs, in_file_name);
OpenOutFile(ofs, out_file_name);
while (getline(ifs, line)) {
content += line + "\n";
}
ofs << trans.gbk_to_utf8(content) << std::endl;
log_file << "tranform complete " << out_file_name << std::endl;
ofs.close();
ifs.close();
}
/* 通过输入DocId搜索页面，借助索引可以快速找到docid */
void SearchIndex(std::string &index_file_name, std::string &lib_file_name) {
std::ifstream ifs_index;
std::vector<std::string> index;
std::string line;
OpenInFile(ifs_index, index_file_name);
while (getline(ifs_index, line)) {
index.push_back(line);
}
ifs_index.close();
int num;
while (std::cin >> num) {
if (num > (int)index.size() || num < 1) {
std::cout << "out of index range" << std::endl;
continue;
}
int doc_id;
int start;
int size_;
std::istringstream iss(index[num - 1].c_str());
iss >> doc_id >> start >> size_;
//std::cout << doc_id << start << size_;
std::ifstream ifs_lib;
OpenInFile(ifs_lib, lib_file_name);
ifs_lib.seekg(start);
std::string content;
while (getline(ifs_lib, line) && ifs_lib.tellg() <= (start + size_)) {
content += line + "\n";
}
//std::cout << content << std::endl;
ifs_lib.close();
}
}
void BuildPageLib(std::string &in_file_name, std::vector<std::string> &lib_vec) {
std::string line; //读入行
std::string content; //一次性读入全部
std::string title; //title
std::ofstream ofs;
std::ifstream ifs; //文本输入流
OpenInFile(ifs, in_file_name);
log_file << " processing file name: " << in_file_name << std::endl;
getline(ifs, title, '\r'); //先取第一行作为标题
TrimR(title);
content += title;
while (getline(ifs, line, '\r')) {
if (strncmp(line.c_str(), "\n【 标 题 】", 17) == 0) {
//title.clear();
//title = std::string(text_title);
title.clear();
title = line;
std::cout << "titile: ~~~~~~" << line << std::endl;
}
/* 每行后面添加换行符 */
if (TrimR(line)) {
content += line + "\n";
} else {
content += line;
}
/* 每行后面不添加换行符, 整篇为一行 */
//content +=line;
//char text_title[MAXSIZE];
//bzero(text_title, MAXSIZE);
//strcpy(text_title, line.c_str());
//if (strncmp(text_title, "【 标 题 】", 16) == 0) {
}
std::string doc;
char id[8];
bzero(id, 8);
sprintf(id, "%d", (int)(lib_vec.size() + 1));
doc = "<doc><docid>" + std::string(id) + "</docid><url>" + in_file_name + "</url><title>" + title + "</title><content>" + content + "</content></doc>\n";
lib_vec.push_back(doc);
ifs.close();
}
void CDTASingleInput::LoadAll(string strPath, vector<CSpectrum> & S)
{
try
{
OpenInFile(string(strPath));
}
catch(exception & e)
{
CErrInfo info("CDTASingleInput", "LoadAll", "OpenInFile: Cannot open " + strPath + ".");
throw runtime_error(info.Get(e).c_str());
}
catch(...)
{
CErrInfo info("CDTASingleInput", "LoadAll", "caught an unknown exception from OpenInFile: Cannot open " + strPath + ".");
throw runtime_error(info.Get().c_str());
}
CSpectrum spec;
try
{
ReadMHAndCharge(spec);
ReadMZAndItensity(spec);
}
catch(exception & e)
{
CErrInfo info("CDTASingleInput", "LoadAll", "ReadMHAndCharge & ReadMZandIntensity: Cannot read " + spec.m_strFilePath + ".");
throw runtime_error(info.Get(e).c_str());
}
catch(...)
{
CErrInfo info("CDTASingleInput", "LoadAll", "caught an unknown exception from ReadMHAndCharge & ReadMZandIntensity: Cannot read " + spec.m_strFilePath + ".");
throw runtime_error(info.Get().c_str());
}
S.clear();
S.push_back(spec);
CloseInFile();
}
static int Load_Dump(void) //Загрузка дампа
{
HANDLE inFile;
DWORD nb;
if(GetName_InFile(NameFDump) < 0) return -1; //Ввод имени обрабатываемого файла
if(OpenInFile(NameFDump, &inFile) < 0) return -1; //Открытие входного файла
inBuf = (char *)MyAllocMem(inSize);
if(inBuf == NULL) return -1;
if(ReadFile(inFile, inBuf, inSize, &nb, NULL) == FALSE || nb != inSize)
return ErrorSys2("Ошибка при чтении исходного файла", NameFDump);
CloseFile(&inFile); //Закрыли входной файл
AdrEnd = inBuf + inSize;
Adr = inBuf;
for(;;) //До конца файла
{ char *NewAdr = strchr(Adr, '\n');
if(NewAdr > AdrEnd || NewAdr == NULL) break; //Данные кончились и Больше нет ни одной полной строки
*NewAdr = 0; //Ограничили строку
Adr = NewAdr + 1;
}
return 0;
}
void CDTASingleInput::LoadNext(CSpectrum & spec, int & idx)
{
try
{
OpenInFile(string(m_strInPath));
}
catch(exception & e)
{
CErrInfo info("CDTASingleInput", "LoadNext", "OpenInFile: Cannot open " + m_strInPath + ".");
throw runtime_error(info.Get(e).c_str());
}
catch(...)
{
CErrInfo info("CDTASingleInput", "LoadNext", "caught an unknown exception from OpenInFile: Cannot open " + m_strInPath + ".");
throw runtime_error(info.Get().c_str());
}
ReadMHAndCharge(spec);
ReadMZAndItensity(spec);
CloseInFile();
}
void CDTASingleInput::StartLoad(string strPath, int & nTotal)
{
try
{
OpenInFile(strPath);
}
catch(exception & e)
{
CErrInfo info("CDTASingleInput", "StartWrite", "OpenInFile: " + strPath + " failed.");
throw runtime_error(info.Get(e).c_str());
}
catch(...)
{
CErrInfo info("CDTASingleInput", "StartWrite", "caught an unknown exception from OpenInFile: " + strPath + " failed.");
info.Append("strPath=" + strPath);
throw runtime_error(info.Get().c_str());
}
m_strInPath = strPath;
nTotal = 1;
CloseInFile();
// OpenInFile(strPath);
}
/* convert file ipInFN to DOS format text and write to file ipOutFN
* RetVal: 0 if success
* -1 otherwise
*/
int ConvertUnixToDosNewFile(char *ipInFN, char *ipOutFN, CFlag *ipFlag)
{
int RetVal = 0;
FILE *InF = NULL;
FILE *TempF = NULL;
char *TempPath;
struct stat StatBuf;
struct utimbuf UTimeBuf;
#ifndef NO_FCHMOD
mode_t mask;
#endif
#ifdef NO_MKSTEMP
FILE* fd;
#else
int fd;
#endif
if ((ipFlag->Force == 0) && regfile(ipInFN))
{
ipFlag->status |= NO_REGFILE ;
return -1;
}
else
ipFlag->status = 0 ;
/* retrieve ipInFN file date stamp */
if (stat(ipInFN, &StatBuf))
RetVal = -1;
#ifdef NO_MKSTEMP
if((fd = MakeTempFileFrom(ipOutFN, &TempPath))==NULL) {
#else
if((fd = MakeTempFileFrom (ipOutFN, &TempPath)) < 0) {
#endif
perror(_("unix2dos: Failed to open temporary output file"));
RetVal = -1;
}
#ifdef DEBUG
fprintf(stderr, _("unix2dos: using %s as temporary file\n"), TempPath);
#endif
/* can open in file? */
if ((!RetVal) && ((InF=OpenInFile(ipInFN)) == NULL))
RetVal = -1;
/* can open output file? */
#ifdef NO_MKSTEMP
if ((!RetVal) && (InF) && ((TempF=fd) == NULL))
#else
if ((!RetVal) && (InF) && ((TempF=OpenOutFile(fd)) == NULL))
#endif
{
fclose (InF);
InF = NULL;
RetVal = -1;
}
#ifndef NO_FCHMOD
/* preserve original mode as modified by umask */
mask = umask(0);
umask(mask);
if (!RetVal && fchmod(fd, StatBuf.st_mode & ~mask))
RetVal = -1;
#endif
/* conversion sucessful? */
if ((!RetVal) && (ConvertUnixToDos(InF, TempF, ipFlag)))
RetVal = -1;
/* can close in file? */
if ((InF) && (fclose(InF) == EOF))
RetVal = -1;
/* can close output file? */
if ((TempF) && (fclose(TempF) == EOF))
RetVal = -1;
#ifdef NO_MKSTEMP
if(fd!=NULL)
fclose(fd);
#else
if(fd>=0)
close(fd);
#endif
if ((!RetVal) && (ipFlag->KeepDate))
{
UTimeBuf.actime = StatBuf.st_atime;
UTimeBuf.modtime = StatBuf.st_mtime;
/* can change output file time to in file time? */
if (utime(TempPath, &UTimeBuf) == -1)
RetVal = -1;
}
/* any error? */
if ((RetVal) && (unlink(TempPath)))
RetVal = -1;
/* can rename temporary file to output file? */
if (!RetVal)
{
#ifdef NEED_REMOVE
remove(ipOutFN);
#endif
if ((rename(TempPath, ipOutFN) == -1) && (!ipFlag->Quiet))
{
fprintf(stderr, _("unix2dos: problems renaming '%s' to '%s'\n"), TempPath, ipOutFN);
fprintf(stderr, _(" output file remains in '%s'\n"), TempPath);
RetVal = -1;
}
}
free(TempPath);
return RetVal;
}
/* convert file ipInFN to DOS format text
* RetVal: 0 if success
* -1 otherwise
*/
int ConvertUnixToDosOldFile(char* ipInFN, CFlag *ipFlag)
{
int RetVal = 0;
FILE *InF = NULL;
FILE *TempF = NULL;
char *TempPath;
struct stat StatBuf;
struct utimbuf UTimeBuf;
#ifndef NO_FCHMOD
mode_t mode = S_IRUSR | S_IWUSR;
#endif
#ifdef NO_MKSTEMP
FILE* fd;
#else
int fd;
#endif
if ((ipFlag->Force == 0) && regfile(ipInFN))
{
ipFlag->status |= NO_REGFILE ;
return -1;
}
else
ipFlag->status = 0 ;
/* retrieve ipInFN file date stamp */
if (stat(ipInFN, &StatBuf))
RetVal = -1;
#ifndef NO_FCHMOD
else
mode = StatBuf.st_mode;
#endif
#ifdef NO_MKSTEMP
if((fd = MakeTempFileFrom(ipInFN, &TempPath))==NULL) {
#else
if((fd = MakeTempFileFrom(ipInFN, &TempPath)) < 0) {
#endif
perror(_("unix2dos: Failed to open temporary output file"));
RetVal = -1;
}
#ifndef NO_FCHMOD
if (!RetVal && fchmod (fd, mode) && fchmod (fd, S_IRUSR | S_IWUSR))
RetVal = -1;
#endif
#ifdef DEBUG
fprintf(stderr, _("unix2dos: using %s as temporary file\n"), TempPath);
#endif
/* can open in file? */
if ((!RetVal) && ((InF=OpenInFile(ipInFN)) == NULL))
RetVal = -1;
/* can open output file? */
#ifdef NO_MKSTEMP
if ((!RetVal) && (InF) && ((TempF=fd) == NULL))
#else
if ((!RetVal) && (InF) && ((TempF=OpenOutFile(fd)) == NULL))
#endif
{
fclose (InF);
InF = NULL;
RetVal = -1;
}
/* conversion sucessful? */
if ((!RetVal) && (ConvertUnixToDos(InF, TempF, ipFlag)))
RetVal = -1;
/* can close in file? */
if ((InF) && (fclose(InF) == EOF))
RetVal = -1;
/* can close output file? */
if ((TempF) && (fclose(TempF) == EOF))
RetVal = -1;
#ifdef NO_MKSTEMP
if(fd!=NULL)
fclose(fd);
#else
if(fd>=0)
close(fd);
#endif
if ((!RetVal) && (ipFlag->KeepDate))
{
UTimeBuf.actime = StatBuf.st_atime;
UTimeBuf.modtime = StatBuf.st_mtime;
/* can change output file time to in file time? */
if (utime(TempPath, &UTimeBuf) == -1)
RetVal = -1;
}
/* any error? */
if ((RetVal) && (unlink(TempPath)))
RetVal = -1;
#ifdef NEED_REMOVE
if (!RetVal)
remove(ipInFN);
#endif
/* can rename output file to in file? */
if ((!RetVal) && (rename(TempPath, ipInFN) == -1))
{
if (!ipFlag->Quiet)
{
fprintf(stderr, _("unix2dos: problems renaming '%s' to '%s'\n"), TempPath, ipInFN);
fprintf(stderr, _(" output file remains in '%s'\n"), TempPath);
}
RetVal = -1;
}
free(TempPath);
return RetVal;
}
/* convert stdin to DOS format text and write to stdout
* RetVal: 0 if success
* -1 otherwise
*/
int ConvertUnixToDosStdio(CFlag *ipFlag)
{
ipFlag->NewFile = 1;
ipFlag->Quiet = 1;
ipFlag->KeepDate = 0;
ipFlag->Force = 1;
#ifdef WIN32
/* stdin and stdout are by default text streams. We need
* to set them to binary mode. Otherwise an LF will
* automatically be converted to CR-LF on DOS/Windows.
* Erwin */
/* 'setmode' was deprecated by MicroSoft
* since Visual C++ 2005. Use '_setmode' instead. */
_setmode(fileno(stdout), O_BINARY);
_setmode(fileno(stdin), O_BINARY);
return (ConvertUnixToDos(stdin, stdout, ipFlag));
#elif defined(MSDOS) || defined(__OS2__)
setmode(fileno(stdout), O_BINARY);
setmode(fileno(stdin), O_BINARY);
return (ConvertUnixToDos(stdin, stdout, ipFlag));
#else
return (ConvertUnixToDos(stdin, stdout, ipFlag));
#endif
}
