bool LexicsCutter::_ReadLetterAnalogs(const std::string& fileName)
{
char szLine[1024];
FILE* file = fopen(fileName.c_str(), "rb");
if (!file)
return false;
while (!feof(file))
{
szLine[0] = 0x0;
fgets(szLine, 1020, file);
std::string line;
if (!_ProcessLine(szLine, line))
continue;
std::string ch;
unsigned int pos = 0;
if (ReadUTF8(line, ch, pos))
{
// Create analogs vector
std::string analog;
LC_AnalogVector av;
while (ReadUTF8(line, analog, pos))
av.push_back(analog);
// Store vector in hash map
_analogMap[ch] = av;
}
}
fclose(file);
return true;
}
bool LexicsCutter::Read_Letter_Analogs(std::string& FileName)
{
FILE *ma_file;
char line[1024];
unsigned int pos;
std::string line_s;
std::string lchar;
std::string lanalog;
ma_file = fopen(FileName.c_str(), "rb");
while (!feof(ma_file))
{
line[0] = 0x0;
fgets(line, 1020, ma_file);
// check for UTF8 prefix and comments
if (strlen(line) >= 3)
{
if (line[0] == '\xEF' && line[1] == '\xBB' && line[2] == '\xBF')
{
strncpy(&line[0], &line[3], strlen(line) - 3);
}
}
if (strlen(line) >= 2)
{
if (line[0] == '/' && line[1] == '/') continue;
}
// check for empty string
line_s = line;
line_s = trim(line_s, "\x0A\x0D\x20");
if (line_s == "") continue;
// process line without CR/LF
line_s = line;
line_s = trim(line_s, "\x0A\x0D");
pos = 0;
if (ReadUTF8(line_s, lchar, pos))
{
// create analogs vector
LC_AnalogVector av;
while (ReadUTF8(line_s, lanalog, pos))
{
av.push_back(lanalog);
}
// store vector in hash map
AnalogMap[lchar] = av;
}
}
fclose(ma_file);
return true;
}
