/*
Program: TXTINFO.C
Author : Kim Moser
Date   : March 17, 1991
System : IBM PC / Borland Turbo C 2.0
Descrip: For any text file piped from stdin, tells:

	Number of lines
	Average line length
	Longest line
	Shortest line

	Number of words
	Average word length
	Longest word
	Shortest word

*/

#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <dir.h>
#include <dos.h>

typedef unsigned long ulong;

static ulong grandlinecount=0;

static int FILESPROCESSED=0;
/* How many files processed */

static void dofile(FILE *fp);
static void dofile(FILE *fp)
{
	int ch;
	int inword=0;

	ulong	linecount=0, longestlinelen=0, shortestlinelen=-1,
		totallinelen=0,
		wordcount=0, longestwordlen=0, shortestwordlen=-1,
		totalwordlen=0,
		minwordsonline=-1, maxwordsonline=0;
	ulong thislinelen=0, thiswordlen=0, wordsonthisline=0;

	/* File pointers: */
	long	longestline=-1, shortestline=-1, longestword=-1, shortestword=-1,
		linestart=0, wordstart=-1;

	FILESPROCESSED++;
	while (1) {
		ch = getc(fp);
		if ((ch == '\t') || (ch == '\n') || (ch == ' ') || (ch == EOF)) {
			if (inword) {
				wordsonthisline++;
			}
			if (ch == '\n' || ((ch==EOF) && thislinelen)) {
				linecount++;

				if (!(linecount % 100)) {
					fprintf(stderr, "%lu lines so far\r", linecount);
				}

				if ((thislinelen < shortestlinelen) || (shortestlinelen == -1)) {
					shortestlinelen = thislinelen;
					shortestline = linestart;
				}
				if (thislinelen > longestlinelen) {
					longestlinelen = thislinelen;
					longestline = linestart;
				}
				totallinelen += thislinelen;
				thislinelen = 0;

				if (wordsonthisline > maxwordsonline) maxwordsonline = wordsonthisline;
				if ((wordsonthisline < minwordsonline) || (minwordsonline==-1)) minwordsonline = wordsonthisline;

				wordsonthisline = 0;

				/* Remember where this line begins: */
				if ((linestart = ftell(fp)) == -1L) {
					fprintf(stderr, "ftell() failed for linestart.\n");
				}
			} else {
				thislinelen++;
			}

			if (inword) {
				wordcount++;
				if ((thiswordlen < shortestwordlen) || (shortestwordlen == -1)) {
					shortestwordlen = thiswordlen;
					shortestword = wordstart;
				}
				if (thiswordlen > longestwordlen) {
					longestwordlen = thiswordlen;
					longestword = wordstart;
				}
				totalwordlen += thiswordlen;
				thiswordlen = 0;

				inword = 0;
			}
		} else {
			if (!inword) {
				if ((wordstart = ftell(fp)) == -1L) {
					fprintf(stderr, "ftell() failed for wordstart.\n");
				}
				if (wordstart) wordstart--;
			}
			inword = 1;
			thislinelen++;
			thiswordlen++;
		}
		if (ch == EOF) break;
	}

	grandlinecount += linecount;

	fprintf(stderr, "%lu lines (min len=%lu, max len=%lu, avg len=%f)\n",
			linecount, shortestlinelen, longestlinelen, ((float) totallinelen) / ((float) linecount));
	if (shortestline != -1) {
		fprintf(stderr, "shortest line = '");
		if (fseek(fp, shortestline, SEEK_SET)) {
			fprintf(stderr, "fseek() failed for shortest line offset (%ld).\n", shortestline);
		} else {
			while ((ch = getc(fp)) != '\n') {
				if (ch == EOF) break;
				fputc(ch, stderr);
			}
			fprintf(stderr, "'\n");
		}
	}
	if (longestline != -1) {
		fprintf(stderr, "longest line = '");
		if (fseek(fp, longestline, SEEK_SET)) {
			fprintf(stderr, "fseek() failed for longest line offset (%ld).\n", longestline);
		} else {
			while ((ch = getc(fp)) != '\n') {
				if (ch == EOF) break;
				fputc(ch, stderr);
			}
			fprintf(stderr, "'\n");
		}
	}

	fprintf(stderr, "\n%lu words (min len=%lu, max len=%lu, avg len=%f)\n",
			wordcount, shortestwordlen, longestwordlen, ((float) totalwordlen)/((float) wordcount));
	if (shortestword != -1) {
		fprintf(stderr, "shortest word = '");
		if (fseek(fp, shortestword, SEEK_SET)) {
			fprintf(stderr, "fseek() failed for shortest word offset (%ld).\n", shortestword);
		} else {
			while ((ch = getc(fp)) != EOF) {
				if (ch == '\n' || ch == '\t' || ch == ' ') break;
				fputc(ch, stderr);
			}
			fprintf(stderr, "'\n");
		}
	}
	if (longestword != -1) {
		fprintf(stderr, "longest word = '");
		if (fseek(fp, longestword, SEEK_SET)) {
			fprintf(stderr, "fseek() failed for longest word offset (%ld).\n", longestword);
		} else {
			while ((ch = getc(fp)) != EOF) {
				if (ch == '\n' || ch == '\t' || ch == ' ') break;
				fputc(ch, stderr);
			}
			fprintf(stderr, "'\n");
		}
	}

	fprintf(stderr, "\nmin words on line: %lu\n", minwordsonline);
	fprintf(stderr, "max words on line: %lu\n", maxwordsonline);
	fprintf(stderr, "avg words on line: %f\n", ((float) wordcount) / ((float) linecount));
	fprintf(stderr, "\n");
}

void expand_and_do(char *s);
void expand_and_do(char *s)
{
	int first=1;
	FILE *fp;
	struct ffblk info;
	static char *fname = NULL;
	int i;

	while (first ? !findfirst(s, &info, ~0) : !findnext(&info)) {
		first = 0;
		/* If it's not (a directory named "." or "..") then print it: */
		if (!(info.ff_attrib & FA_DIREC) && !(info.ff_attrib & FA_LABEL)) {  /* It's not a directory or a label... */
			if (fname != NULL) free(fname);
			if ((fname = (char*) malloc(strlen(s)+12)) == NULL) {
				fprintf(stderr, "malloc() failed for filename.\n");
				exit(-1);
			}
			strcpy(fname, s);

			/* Add info.ff_name in appropriate spot: */
			for (i=strlen(fname)-1; i>=0; i--) {
				if (fname[i] == '\\' || fname[i] == ':') break;
			}
			strcpy(fname+i+1, info.ff_name);

			fprintf(stderr, "%s\n", fname);
			if ((fp = fopen(fname, "r")) == NULL) {
				fprintf(stderr, "fopen() failed for '%s'.\n", info.ff_name);
				exit(-1);
			}
			dofile(fp);
			if (fclose(fp)) {
				fprintf(stderr, "fclose() failed for '%s'.\n", info.ff_name);
				exit(-1);
			}
		}
	}
}

void main(int argc, char **argv)
{
	int i;

	if (argc == 1) {
		fputs("Reading from stdin...\n", stderr);
		dofile(stdin);
	} else {
		for (i=1; i<argc; i++) {
			expand_and_do(argv[i]);
		}
	}

	if (FILESPROCESSED) {
		fprintf(stderr, "\nTOTAL: %lu lines\n", grandlinecount);
	} else {
		fprintf(stderr, "Sorry, no matching files found.\n");
	}
}