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

#define NUM_OF_LETTERS 26
#define MAX_WORD_LEN 20

typedef struct wordcount {
  char *word;
  int count;
  struct wordcount *next;
} WordCount;

WordCount *ltable[NUM_OF_LETTERS];
char **stopwords;

int CreateSWArray(char *filename, int *pnumsw);
int ProcessInputFile(char *filename, int fileno);
int convertToLower(char *);
int isStopWord(char *word, int numstopwords);
int insertWord(char *word);
int insertIntoList(char *pw, WordCount **phead);
void Cleanup(int numstopwords);
void freestopwordarray(int numstopwords);
void freeltable();

int main(int argc , char *argv[])
{
  int numstopwords; /* for storing the number of stop words */

  /* Check the number of command line arguments */
  if (argc < 3) {
     fprintf(stderr, "Usage: a4 <stop words file> <input file>\n");
     exit(1);
  }

  /* Set up an array to hold stop words */
  if (!CreateSWArray(argv[1], &numstopwords)) {
     exit(1); 
  }

  if (!ProcessInputFile(argv[2], numstopwords)) {
     exit(1);
  } 
 
  if (!GenerateOutFile(argv[2])) {
     Cleanup(numstopwords);
     exit(1);
  }
     
  exit(0);
}

/* Create an array to hold all the stop words */
int CreateSWArray(char *filename, int *pnumsw)
{
  FILE *fstopwords;
  char line[MAX_WORD_LEN+2], c;
  int i;

  /* Check for the existence of the stopwords file */
  if ((fstopwords = fopen(filename, "r")) == NULL) {
     fprintf(stderr, "File %s does not exist\n", filename);
     return(0);
  } 

  /* Get the number of stop words and store it in an int variable */
  /* pointed by pnumsw                                            */
  while ((c = fgetc(fstopwords)) != EOF) {
     if (c == '\n')
        (*pnumsw)++;
  }
  
  /* Allocate memory for stopwords array */
  if ((stopwords = (char **)malloc(sizeof(char *)**pnumsw)) == NULL) {
     fprintf(stderr, "not enough memory\n");
     fclose(fstopwords);
     return(0);
  }
  
  /* Set the file position indicator to the beginning of the file */
  rewind(fstopwords); 

  /* Read stop words from file to memory pointed by the stopwords array */ 
  i=0;
  while (fgets(line, sizeof(line), fstopwords)) {
     if ((stopwords[i] = (char *)malloc(strlen(line))) == NULL) {
        fprintf(stderr, "not enough memory\n");
        fclose(fstopwords);
        freestopwordarray(i);
        return(0);
     }
     line[strlen(line)-1]='\0'; /* remove the newline character read from fgets() */
     strcpy(stopwords[i], line);
     i++;
  }
  
  fclose(fstopwords);
  return(1);
}

/* Extract words from the input file and remove stop words */ 
int ProcessInputFile(char *filename, int numstopwords) 
{
  FILE *fin;
  int i, insertstatus;
  char c, word[MAX_WORD_LEN+1];

  if ((fin = fopen(filename, "r")) == NULL) {
     fprintf(stderr, "File %s does not exist\n", filename);
     return(0);
  }
  
  i=0;
  while ((c = fgetc(fin)) != EOF) {
     if (isalpha(c))
        word[i++]=c;
     else {
       word[i]='\0';
       if (i!=0) {
          convertToLower(word);
          if (!isStopWord(word, numstopwords)) {
             insertstatus = insertWord(word);
             if (!insertstatus)
                return(0);
          }
       }
       i=0;
     }
  }
  fclose(fin);

  return(1);
}

/* Convert the word into lower cases */
int convertToLower(char * word)
{
  int i;
  for (i=0; i<strlen(word); i++)
     word[i] = tolower(word[i]);
} 

/* Check if word is a stop word */
int isStopWord(char *word, int numstopwords)
{
  int i;

  for (i=0; i<=numstopwords; i++)
     if (strcmp(stopwords[i], word) = 0)
        return(1);
  return(0);
}

/* Insert a word into ltable */
int insertWord(char *word)
{
  if (insertIntoList(word, &ltable[word[0]-'a']))
     return(1);
  else
     return(0);
}

/* Insert a word into a linked list */
int insertIntoList(char *word, WordCount **phead)
{
  WordCount *nwc;

  if (*phead == NULL) {
     if ((nwc = (WordCount *)malloc(sizeof(WordCount))) == NULL) {
        fprintf(stderr, "not enough memory\n");
        return(0);
     }

     if ((nwc->word = (char *)malloc(strlen(word)+1)) == NULL) {
        fprintf(stderr, "not enough memory\n");
        return(0);
     }

     strcpy(nwc->word, word);
     nwc->count = 1;
     nwc->next = NULL; 
     *phead = nwc;
  }
  else if (strcmp(word, (*phead)->word)<0) {
     if ((nwc = (WordCount *)malloc(sizeof(WordCount))) == NULL) {
        fprintf(stderr, "not enough memory\n");
        return(0);
     }
     if ((nwc->word = (char *)malloc(strlen(word)+1)) == NULL) {
        fprintf(stderr, "not enough memory\n");
        return(0);
     }

     strcpy(nwc->word, word);
     nwc->next = *phead;
     *phead = nwc;
  }
  else if (strcmp(word, (*phead)->word)==0) {
     (*phead)->count++;
  }
  else 
     insertIntoList(word, &((*phead)->next));
  return(1);
}

/* Generate outputfile */
int GenerateOutFile(char *infilename)
{
  FILE *outfile;
  WordCount *p;
  int i;
  char fn[20];

  
  if ((outfile = fopen("output", "w")) == NULL) {
     fprintf(stderr, "cannot create the output file\n");
     return(0);
  }
  for (i=0; i<NUM_OF_LETTERS; i++) {
     p = ltable[i];
     while (p != NULL) {
        fprintf(outfile, "%s %s %f\n", p->word, infilename, p->count);
        p = p->next;
     }
  }
  fclose(outfile);            
  return(1);
}

void Cleanup(int numstopwords)
{
  freestopwordarray(numstopwords);
  freeltable();
}

/* free the memory space dynamically allocated for the stopwords */
/* array, including the space for storing each stop word         */
void freestopwordarray(int numstopwords)
{
  int i;

  for (i=0; i<numstopwords; i++) 
     free(stopwords[i]);
  free(stopwords);
}

/* free the memory space dynamically allocated for the nodes in */
/* ltable, including the space for storing each word            */
void freeltable() 
{
  WordCount *p, *q;
  int i;

  for (i=0; i<NUM_OF_LETTERS; i++) {
     p = ltable[i];
     while (p != NULL) {
        q = p->next;
        free(p->word);
        free(p);
        p = q;
     }     
  } 
}

/* Check if word is a stop word */
int isStopWord(char *word, int numstopwords)
{
  int i;

  for (i=0; i<=numstopwords; i++)
    if (strcmp(stopwords[i], word) = 0)  //Pling Pling!
      return(1);
  return(0);
}