/*  Driver for Paice/Husk Stemmer program */
/* 06/26/2003 - A. Zamora */

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

int readrules(FILE *fp,struct rule ttable[26], int debugsw);
struct rule stem1(char *word,struct rule tbl[26], int debugsw, char *trace);

int main (int argc, char * argv[]) {
  /* There are up to 3 arguments: rule file name, 
                                  input file (optional), 
                                  output file (optional) */
  FILE *outfile1;
  int i, j, k, len;
  int wl, rl;
  int rc;
#define MAXLINE 500
  char line[MAXLINE];  
#define MAXWORD 60
  char word[MAXWORD+1];
  char *result;
#define  DNLEN 64
  char rulefile[DNLEN];  /* rule file name */
  char infile[DNLEN];
  char outfile[DNLEN];
  FILE *infile1;
  FILE *rulp;
  int  infilsw = 0;
  int  outfilsw = 0;  
  char *p1;
struct rule ttable[26];
struct rule trule;
   int initialized = 0;
   int debug = 0;
#define TRACELEN 240
   char trace[TRACELEN]; /* trace output */

  /* Set default rule file name */
  strcpy(rulefile,"azrules.txt");

  if (argc < 2) {
    printf("Parameters: \n rule_file, input_file (optional), output_file (optional)\n\n");
    printf("Using default rule_file: azrules.txt\n\n");
  }

  if (argc >= 2 ) {
    strcpy(rulefile, argv[1]);     /* rule file name */
  }

  if (argc >= 3 ) {
    p1 = &infile[0];
    for (i=0; i< sizeof(infile)-1; i++ ) {
     *p1++ = argv[2][i];
      if (argv[2][i] == '\0') break;
    }
    *p1 = '\0';  /* add null terminator */
    /* strcpy(infile, argv[2]);       input file name */
    infilsw = 1;
  }


  outfile[0] = '\0';              /* null output file name */
  if (argc >= 4 ) {
    p1 = &outfile[0];
    for (i=0; i< sizeof(outfile)-1; i++ ) {
      *p1++ = argv[3][i];
      if (argv[3][i] == '\0') break;
    }
    *p1 = '\0';  /* add null terminator */
    /* strcpy(outfile, argv[3]);      output file name */
    rc = remove(outfile);         /* erase file, if present */
    outfilsw = 1;
    outfile1 = fopen(outfile,"a");  /* open in append mode */
    if (outfile1 == NULL)
      return 4;
  }
      

 /* interactive loop */
   memset(line, 0, MAXLINE);
   if (infilsw == 0) {  /* Interactive input */
     printf("\nInput line of text to be stemmed:\n");
     result = fgets(line,MAXLINE-1,stdin);     /* read first line */
   }
   else {  /* batch input */
     infile1 = fopen(infile, "r");
     if (infile1 == NULL) 
       return 4;
     result = fgets(line,MAXLINE-1,infile1);     /* read first line */
   }

   while (result != NULL) {
    /* remove trailing CR, LF, and blank */
    i = strlen(line)-1;
    while (line[i] == '\n' || line[i] == '\r' || line[i] == ' ')
      line[i--] = '\0';
      len = strlen(line);

    /* Don't do anything for empty lines */
    if (len == 0)
      goto next1;

    /* convert tabs to blanks */
    for (i=0; i < len; i++) {
      if (line[i] == '\t') 
        line[i] = ' ';
    }

    i = 0;   /* initialize scan position */

    while (i < len) {  /* process a line */
      /* Tokenize input line */
      while (!isalpha(line[i]) && i < len) { /* find word start */
        i++;
      }
      if (i >= len) break;  /* exit while (i < len) */

      j = i;
      k = 0;
      /* find word end and convert to lower case */
      while (isalpha(line[j]) && (j < len) && (k < MAXWORD)) { 
          word[k++] = tolower(line[j++]);
      }
      word[k] = '\0';  /* add null terminator */
      
      if (strcmp(word,"debugon") == 0) 
        debug = 1;
      if (strcmp(word,"debugoff") == 0) 
        debug = 0;
      if (strcmp(word,"printrules") == 0) 
        debug = 1;

      if (initialized == 0) {
        initialized = 1;
        /* Initialization is within the loop to be able to print 
            the stemming rules based on user input.   */
        /* If file of stemmer rules is not opened correctly exit. */ 
        if((rulp = fopen(rulefile,"r")) == NULL){
          printf("\n*** ERROR *** - Cannot open rule file.\n");
          exit(1); 
        }
        /* Read in a rule set from the file */
        rc = readrules(rulp,ttable,debug); 
        fclose(rulp); /* And then close the file */    
        if (rc > 0)  {
          printf("\n*** ERROR *** - Rule file contains %d error(s).\n",rc);
          exit (1);
        }

        if (strcmp(word,"printrules") == 0) {
          debug = 0;
          goto next1;
        }
      }

      /* We have isolated a word, now stem it */
      trace[0] = '\0';  /* initialize trace to null */
      wl = strlen(word);        /* word length */
      if (wl > MAXWDSZ) {  /* truncate long words */
        strncpy(trule.text, word, MAXSTEMSIZE);
        trule.text[MAXSTEMSIZE] = '\0';  /* terminate string */
      }
      else {
        trule = stem1(word,ttable,debug,trace); /* Stem the word.. */
      }

      /* print trace to SYSOUT*/
      if (trace[0] != '\0') {
        printf("%s",trace);
      }
      /* print in format:  "root ing"  with space after the root  */
      rl = strlen(trule.text);  /* root length */
      if (outfilsw == 0) { /* Interactive input */
         /*  printf("%s %s %d\n",word, trule.text,trule.rulenum); */
        if (wl <= rl)  /* root == word */
          printf("%s\n",trule.text);
        else {
          if (strncmp(trule.text,word,rl) == 0) {
            /* stem is a substring of the word */
            p1 = word + rl;  /* address of ending */
            printf("%s %s %d\n",trule.text, p1, trule.rulenum);
          } 
          else {  /* stem has characters that are not in original word */
            printf("%s %d\n",trule.text, trule.rulenum);
          }
        }
      }
      else { /* batch input */
        if (wl <= rl)  /* root == word */
          fprintf(outfile1,"%s\n",word);
        else {
          if (strncmp(trule.text,word,rl) == 0) {
            /* stem is a substring of the word */
            p1 = word + rl;  /* address of ending */
            fprintf(outfile1,"%s %s %d\n",trule.text, p1, trule.rulenum);
          } 
          else {  /* stem has characters that are not in original word */
            fprintf(outfile1,"%s\n",trule.text);
          }
        }
      }

      i = j;
    }  /* end while (i < len) */


next1: ;  /* we get here for empty lines */
    memset(line, 0, MAXLINE);
    if (infilsw == 0) {  /* Interactive input */
      printf("\nInput line of text to be stemmed:\n");
      result = fgets(line,MAXLINE-1,stdin);     /* read line */
    }
    else {  /* batch input */
      result = fgets(line,MAXLINE-1,infile1);     /* read line */
    }


   } /* end while (result != NULL)*/

   if (infilsw != 0) { /* batch input */
     fclose(infile1);
   }
   if (outfilsw != 0) {
    fclose(outfile1);
   }

   return 0;

} /* end main */