kvoctrain/kvoctrain/tools/extractlang.cpp

   1 #include <qtextstream.h>
   2 #include <qstring.h>
   3
   4 ///////////////////////////////////////////////////////////
   5 //
   6 // (c) Kevin Krammer <kevin.krammer@gmx.at>
   7 //
   8 // 2001-12-08
   9 //
  10 // automatically extracts a structure from
  11 //   http://www.loc.gov/standards/iso639-2/englangn.html
  12 //
  13 // see also extractlang.README
  14 //
  15 ///////////////////////////////////////////////////////////
  16
  17 QString twoLangs;
  18 QString threeLangs;
  19
  20 void parseLanguage(QString name, QString three, QString two)
  21 {
  22   int star = three.find('*');
  23   if (star > 0)
  24     three.remove(star, 1);
  25
  26   QString three_1 = three;
  27
  28   int slash = three_1.find('/');
  29   if (slash > 0) {
  30     three_1.remove(slash, 1);
  31     three_1.insert(slash, ", ");
  32   }
  33
  34   if (!two.contains("&nbsp;")) {
  35       twoLangs.append("\t{{\"");
  36       twoLangs.append(two);
  37       twoLangs.append("\"}, {\"");
  38       twoLangs.append(three_1);
  39       twoLangs.append("\"}, {I18N_NOOP(\"");
  40       twoLangs.append(name);
  41       twoLangs.append("\")}},\n");
  42   }
  43   else
  44     two = "";
  45
  46   slash = three.find('/');
  47   if (slash > 0) {
  48     if (two.length() == 0)
  49       two = three.mid(slash+1);
  50     three.remove(slash, three.length());
  51   }
  52
  53   threeLangs.append("\t{{\"");
  54   threeLangs.append(two);
  55   threeLangs.append("\"}, {\"");
  56   threeLangs.append(three);
  57   threeLangs.append("\"}, {I18N_NOOP(\"");
  58   threeLangs.append(name);
  59   threeLangs.append("\")}},\n");
  60 }
  61
  62 QString parseTD(QString& line)
  63 {
  64   QCString cline = line.local8Bit();
  65   uint clineLength = cline.length();
  66   QCString result = cline;
  67   int pos = -1;
  68   bool inTag = false;
  69
  70   for (uint i = 0; i < clineLength; ++i)
  71     {
  72       if (cline[i] == '<')
  73         inTag = true;
  74       else if (cline[i] == '>')
  75         inTag = false;
  76       else if (!inTag)
  77         {
  78           pos++;
  79           result[pos] = cline[i];
  80         }
  81     }
  82
  83   QString retVal(result);
  84   retVal = retVal.left(pos+1);
  85
  86   return retVal.simplifyWhiteSpace();
  87 }
  88
  89 bool parseTR(QString& line, QTextIStream& in)
  90 {
  91   QString text[4];
  92
  93   line = in.readLine();
  94   if (line.isNull())
  95     return true;
  96
  97
  98   uint count = 0;
  99   while (line.contains("<td"))
 100     {
 101       if (count > 3)
 102         return true;
 103
 104       text[count] = parseTD(line);
 105       count++;
 106       line = in.readLine();
 107     }
 108
 109   if (count == 4)
 110     {
 111       parseLanguage(text[0], text[2], text[3]);
 112     }
 113
 114   if (line.contains("/tr"))
 115     return true;
 116
 117   return false;
 118 }
 119
 120 int main()
 121 {
 122   QString line;
 123   bool readNext = true;
 124
 125   QTextIStream in(stdin);
 126   QTextOStream out(stdout);
 127
 128   while (!in.atEnd())
 129     {
 130       if (readNext)
 131         line = in.readLine();
 132
 133       if (line.contains("<tr"))
 134         {
 135           readNext = parseTR(line, in);
 136         }
 137       else
 138         readNext = true;
 139     }
 140
 141   out << "struct KV_ISO639_Code {\n"
 142          "   const char *iso1code;\n"
 143          "   const char *iso2code;\n"
 144          "   const char *langname;\n"
 145          "};\n"
 146          "\n";
 147
 148   out << "KV_ISO639_Code kv_iso639_1[] = {\n"
 149       << twoLangs.left(twoLangs.length()-1) << endl
 150       << "\t{{0}, {0}, {0}}\n"
 151       << "};\n";
 152
 153   out << endl;
 154
 155   out << "KV_ISO639_Code kv_iso639_2[] = {\n"
 156       << threeLangs.left(threeLangs.length()-1) << endl
 157       <<"\t{{0}, {0}, {0}}\n"
 158       << "};\n";
 159
 160   return 0;
 161 }