Node *index_file(Node *head, char *fname, char **filenames) {
char line[MAXLINE];
char *marker, *token;
int countlines = 0;
FILE *fp;
if((fp = fopen(fname, "r")) == NULL) {
perror(fname);
exit(1);
}
while((fgets(line, MAXLINE, fp)) != NULL) {
countlines++;
if((countlines % 1000) == 0) {
printf("processed %d lines from %s (words%d)\n", countlines, fname, num_words);
}
line[strlen(line)-1] = '\0';
if(strlen(line) == 0) {
continue;
}
marker = line;
while((token = strsep(&marker, " \t")) != NULL) {
if(strlen(token) == 0) {
continue;
}
token = remove_punc(token);
if((strlen(token) <= 3) || isdigit(*token)) {
continue;
}
if(*token != '\0') {
head = add_word(head, filenames, token, fname);
}
}
}
return head;
}
int main(int argc, char *argv[])
{
const char* INPUT_PATH = argv[1];
const char* OUTPUT_PATH = argv[2];
const char* QUIET = argv[3];
/* Check arugments are present */
if (!INPUT_PATH || !OUTPUT_PATH) {
printf(ANSI_RED "WordCounter: Arguments Missing.\n" ANSI_RESET);
printf("Usage: WordCounter input-path output-path [-q]\n");
exit(EXIT_FAILURE);
}
/* Try and open file */
FILE *fp = fopen(argv[1], "r");
if (fp == NULL) {
printf(ANSI_RED "Failed to open file: %s\n" ANSI_RESET, argv[1]);
exit(EXIT_FAILURE);
}
/* Iterate text file adding words */
WordList *wl = WordList_new();
while(!feof(fp)){
char *word = malloc(sizeof(char));
fscanf(fp, "%s", word);
remove_punc(word);
if(!wl->increment(wl, word)){
wl->add(wl, word);
}
}
fclose(fp);
wl->sort(wl, true);
/* If not quiet then print the results */
if (!QUIET) {
wl->print(wl);
}
wl->save(wl, argv[2]);
/* Free memory */
for (int i=0; i<wl->used; i++) {
free(wl->words[i].chars);
}
free(wl->words);
free(wl);
printf(ANSI_GREEN "Completed, %d words saved to '%s'\n" ANSI_RESET, wl->used, argv[2]);
}
