potato-count.c

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include <sys/types.h>
#include <sys/stat.h>
#include <dirent.h>
#include <limits.h>
#include <ctype.h>

#define TSZ 18199

struct wordcount {
  char *word;
  size_t count;
  struct wordcount *next;
} table[TSZ];

int table_entries = 0;

static int wcompare(const void *a, const void *b) { 
  const struct wordcount *wa = a, *wb = b;
  if(wb->count < wa->count) { 
    return -1;
  } 

  return(wb->count != wa->count);
} 

static unsigned hash(char *s) {
  unsigned h = 5381;
  int c; 
  while((c = (*s++))) { 
    h = ((h << 5) + h) + c;
  }
  return h;
}

static void count(char *word) { 
  struct wordcount *w = &table[hash(word) % TSZ];
  while(w->word) { 
    if(!strcmp(word, w->word)) { 
      w->count += 1;
      return;
    }

    if(!w->next) 
      w->next = calloc(1, sizeof(*w));

    w = w->next;
  }

  table_entries += 1;
  w->count = 1;
  w->word = strdup(word);   
} 

static void count_file(char *path) { 
  FILE *fp = fopen(path, "r"); 
  if(!fp) { 
    return;
  } 

  struct stat sb;
  if(fstat(fileno(fp), &sb) < 0 || !(sb.st_mode & S_IFREG)) { 
    fclose(fp);
    return;
  }

  char buf[1024];
  while(fgets(buf, 1024, fp)) { 
    // XXX check long line; potato code
    char *bp = buf;
    for(char *cp = strsep(&bp, " \t\n"); cp; cp = strsep(&bp, " \t\n")) { 
      if(*cp && cp[1] && cp[2] && strchr("abcdefghijklmnopqrstuvwxyz", tolower(*cp))) { 
        count(cp);
      }
    }
  }

  fclose(fp);
}

void walk(char *path) { 
  struct dirent ent, *ep = NULL; 
  
  DIR *d = opendir(path);
  if(!d) { 
    return;
  } 

  while(readdir_r(d, &ent, &ep) == 0 && ep != NULL) { 
    if(!strcmp(ep->d_name, ".") || !strcmp(ep->d_name, "..")) { 
      continue;
    }

    char pathbuf[PATH_MAX];
    snprintf(pathbuf, PATH_MAX, "%s/%s", path, ep->d_name);

    if(ep->d_type == DT_DIR) { 
      walk(pathbuf);
    } else if(ep->d_type == DT_REG) { 
      size_t len = strlen(ep->d_name);
      if(len > 3 && !strcmp(&ep->d_name[len-4], ".txt")) {
        count_file(pathbuf);
      }
    }
  }

  closedir(d);
}

int main(int argc, char **argv) { 
  walk(argv[1] ? argv[1] : ".");
 
  size_t cents = 0;
  struct wordcount *counts = calloc(table_entries, sizeof(struct wordcount));
   
  for(int i = 0; i < TSZ; i++) { 
    struct wordcount *w = &table[i];
    while(w && w->word) { 
      counts[cents++] = *w;
      w = w->next;
    }
  } 

  if(cents) { 
    qsort(counts, cents, sizeof(*counts), wcompare);

    for(int i = 0; i <= 10 && i < cents; i++) 
      printf("%d. %s / %lu\n", i, counts[i].word, counts[i].count);
  }
 
  return 0;
}