apertium-tools/apertium-utils/morph-indux/XMLParseUtil.cpp

   1 #include <XMLParseUtil.h>
   2
   3 #include <cstdlib>
   4 #include <iostream>
   5
   6 using namespace std;
   7
   8 wstring
   9 XMLParseUtil::attrib(xmlTextReaderPtr reader, wstring const &name)
  10 {
  11   string mystr = "";
  12   for(int i = 0, limit = name.size(); i != limit; i++)
  13   {
  14     mystr += static_cast<char>(name[i]);
  15   }
  16
  17   xmlChar *attrname = xmlCharStrdup(mystr.c_str());
  18   xmlChar *myattr = xmlTextReaderGetAttribute(reader, attrname);
  19   wstring result = towstring(myattr);
  20   xmlFree(myattr);
  21   xmlFree(attrname);
  22   return result;
  23 }
  24
  25 string
  26 XMLParseUtil::latin1(xmlChar const *input)
  27 {
  28  if(input == NULL)
  29   {
  30     return "";
  31   }
  32
  33   int outputlen = xmlStrlen(input) + 1;
  34   int inputlen = xmlStrlen(input);
  35
  36   unsigned char output[outputlen];
  37
  38   if(UTF8Toisolat1(output, &outputlen, input, &inputlen) != 0)
  39   {
  40   }
  41
  42   output[outputlen] = 0;
  43   string result = reinterpret_cast<char *>(output);
  44   return result;
  45 }
  46
  47 wstring
  48 XMLParseUtil::towstring(xmlChar const * input)
  49 {
  50   wstring result = L"";
  51
  52   for(int i = 0, limit = xmlStrlen(input); i != limit; i++)
  53   {
  54     int val = 0;
  55     if(((unsigned char) input[i] & 0x80) == 0x0)
  56     {
  57       val = static_cast<wchar_t>(input[i]);
  58     }
  59     else if(((unsigned char) input[i] & 0xE0) == 0xC0)
  60     {
  61       val = (input[i] & 0x1F) << 6;
  62       i++;
  63       val += input[i] & 0x7F;
  64     }
  65     else if(((unsigned char) input[i] & 0xF0) == 0xE0)
  66     {
  67       val = (input[i] & 0x0F) << 6;
  68       i++;
  69       val += input[i] & 0x7F;
  70       val = val << 6;
  71       i++;
  72       val += input[i] & 0x7F;
  73     }
  74     else if(((unsigned char) input[i] & 0xF8) == 0xF0)
  75     {
  76       val = (input[i] & 0x07) << 6;
  77       i++;
  78       val += input[i] & 0x7F;
  79       val = val << 6;
  80       i++;
  81       val += input[i] & 0x7F;
  82       val = val << 6;
  83       i++;
  84       val += input[i] & 0x7F;
  85     }
  86     else
  87     {
  88       wcerr << L"UTF-8 invalid string" << endl;
  89       exit(EXIT_FAILURE);
  90     }
  91
  92     result += static_cast<wchar_t>(val);
  93   }
  94   return result;
  95 }
  96
  97 wstring
  98 XMLParseUtil::stows(string const &str)
  99 {
 100   wchar_t result[str.size()];
 101   mbstowcs(result, str.c_str(), str.size());
 102   wstring result2 = result;
 103   return result2;
 104 }