/* Vide
   https://rudhar.com/lingtics/intrlnga/cgi-grep/fria/intro-ia.htm
   e specificamente
   https://rudhar.com/lingtics/intrlnga/cgi-grep/fria/intro-ia.htm#frhtmtxt
   Isto es un variante, que extrahe le parolas e expression in interlingua
   ex le dictionario francese-interlingua.

   Le 16 de februario 2020, © Ruud Harmsen, ma face con illo quecunque vos vole.
 */

#include <stdlib.h>
#include <stdio.h>
#include <string.h>
#include <ctype.h>

char *DT    = "<dt>";
char *DD    = "<dd>";
char *DL    = "<dl>";
char *SDL   = "</dl>";
char *CLN   = " : ";
char *ITAL  = "<i>";
char *SITAL = "</i>";

char *OutFile1 = "tmp03.txt";
char *OutFile2 = "tmp04.txt";

struct se_SubString
{
   char *sub;
   int len;
};
typedef struct se_SubString se_SubstrSpec;

static int TractaLinea (char *linea, FILE *fpo1, FILE *fpo2);
static int TrovaClasseDeParola (char *line, char *ClasseDeParola, size_t sizeClasseDeParola);
static int PlenaLista (char *linea, char *ClasseDeParola, FILE *fpo1);
static se_SubstrSpec *TrovaCommas (char *linea);

int main (int argc, char **argv)
{
   static char line1[4096];
   FILE *fpi = stdin, *fpo1, *fpo2;
   int inprepost = 1;
   long linecnt = 0, blanks = 0, dts = 0, dtss = 0, dds = 0;
   long prepost = 0, unrecog = 0, invalid = 0;
   char *s;

   if ((fpo1 = fopen(OutFile1, "wt")) == NULL)
   {
      fprintf(stderr, "%s: cannot open output file %s\n", argv[0], OutFile1);
      return 1;
   }
   if ((fpo2 = fopen(OutFile2, "wt")) == NULL)
   {
      fprintf(stderr, "%s: cannot open output file %s\n", argv[0], OutFile2);
      return 2;
   }

   while (fgets(line1, sizeof line1, fpi))
   {
      linecnt++;

      if (s = strstr(line1, DL), s && s == line1)
      {
         inprepost = 0;
         prepost++;
         continue;
      }
      else if (s = strstr(line1, SDL), s && s == line1)
      {
         inprepost = 1;
      }

      if (inprepost)
      {
         prepost++;
      }
      else
      {
         if (line1[0] == '\n' && line1[1] == '\0')
         {
            /* Skip empty lines */
            blanks++;
         }
         else
         {
            TractaLinea(line1, fpo1, fpo2);
         }
      }
   }

   if (fpo1) fclose(fpo1);
   if (fpo2) fclose(fpo2);

   return 0;
}

static int TractaLinea (char *linea, FILE *fpo1, FILE *fpo2)
{
   char *s, *n, *c, *l;
   char ClasseDeParola[20];

   c = strstr(linea, CLN);
   if (c)
   {
      c += strlen(CLN);
      l = c + strlen(c) - 1;
      if (*l == '\n')
         *l = '\0';
   }

   if (s = strstr(linea, DT), s && s == linea)
   {
      if (c)
      {
         TrovaClasseDeParola(linea, ClasseDeParola, sizeof ClasseDeParola);
         PlenaLista(c, ClasseDeParola, fpo1);
      }
   }
   else if (s = strstr(linea, DD), s && s == linea)
   {
      n = s + strlen(DD);
      if ((isdigit(n[0]) && n[1] == '.') ||
          (isdigit(n[0]) && isdigit(n[1]) && n[2] == '.'))
      {
         if (c)
         {
            TrovaClasseDeParola(linea, ClasseDeParola, sizeof ClasseDeParola);
            PlenaLista(c, ClasseDeParola, fpo1);
         }
      }
      else
      {
         if (c)
         {
            se_SubstrSpec *Places, *p;

            Places = TrovaCommas(c);

            for (p = Places; p->sub; p++)
            {
               fprintf(fpo2, "· %.*s\n", (int)p->len, p->sub);
            }
         }
      }
   }

   return 0;
}

static int TrovaClasseDeParola (char *line, char *ClasseDeParola, size_t sizeClasseDeParola)
{
   char *b, *e;

   memset(ClasseDeParola, 0, sizeClasseDeParola);
   if ((b = strstr(line, ITAL)) != NULL &&
       (e = strstr(b,   SITAL)) != NULL)
   {
      b += strlen(ITAL);
      if (e - b > sizeClasseDeParola - 1)
      {
         e = b + sizeClasseDeParola - 1;
      }
      if (e > b && e[-1] == '.')
         e--;
      strncpy(ClasseDeParola, b, e - b);
   }

   return 0;
}

static int PlenaLista (char *linea, char *ClasseDeParola, FILE *fpo1)
{
   se_SubstrSpec *Places, *p;

   Places = TrovaCommas(linea);

   for (p = Places; p->sub; p++)
   {
      fprintf(fpo1, "· %.*s · %s\n", (int)p->len, p->sub, ClasseDeParola);
   }

   return 0;
}

#define NUMPOINTERS 20
char *sep = ", ";

static se_SubstrSpec *TrovaCommas (char *linea)
{
   static se_SubstrSpec Places[NUMPOINTERS + 1];

   int i = 0;
   char *b, *bn, *e;

   memset(Places, '\0', sizeof Places);
   for (b = linea, i = 0; i < NUMPOINTERS && *b; i++)
   {
      e = strstr(b, sep);
      if (e)
      {
         bn = e + strlen(sep);
      }
      else
      {
         e = b + strlen(b);
         bn = e + 1;
      }

      Places[i].sub = b;
      Places[i].len = e - b;

      b = bn;
   }

   Places[i].sub = NULL;

   return Places;
}