/*  File   : ll.c
    Author : Richard A. O'Keefe
    Updated: %G%
    Purpose: Print the longest line in each file.
    Usage  : ll file...
    TODO   : handle UTF-8.

    If there are no file names in the command line,
    the file name "-" is used.  As a file name, "-"
    means to use standard input, and should be used
    at most once, but this is not checked.

    For each file,
    - if the file cannot be read, an error message is printed
      and processing stops.
    - if an error occurs while processing a file, an error
      message is printed and processing stops.
    - otherwise one line is printed to standard input.
	* If the file is empty, the line is
	  <filename>:empty
	* If the longest line exceeds 1024 characters, the line is
	  <filename>:<linenum>: > 1024 <text truncated to 1024>
	* If the longest line has no more than 1024 characters, it's
	  <filename>:<linenum>:<width>:<text>
      where <linenum> is the 1-origin index of the first line to
      have the greatest number of columns.

    TABs are expanded according to the 40-year UNIX convention of
    tab stops being set every 8 columns.  CR characters are
    ignored.

    Currently this is set up for 8-bit character sets.
    All bytes other than TAB, CR, and LF are taken as representing
    one-column characters.  To handle UTF-8, we would need to
    + distinguish between buflen (in bytes) and bufcnt (in chars).
    + don't count bytes 10xxxxxx towards bufcnt
    + do this only in a UTF-8 locale.
*/
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#define MAX_LINE_LENGTH 1024

static char buffer[MAX_LINE_LENGTH+1] = "";
static int  buflen = 0;
static int  overflow = 0;

static int read_line(char const *file_name, FILE *in) {
    int i;  /* index into buffer */
    int c;  /* unsigned character U {EOF} */

    i = 0;
    for (;;) {
	c = getc_unlocked(in);
	if (c == '\n' || c < 0) break;
	if (c == '\r') continue;
	if (i != MAX_LINE_LENGTH) {
	    if (c == '\t') {
		do buffer[i++] = ' '; while (i % 8 != 0);
	    } else {
		buffer[i++] = (char)c;
	    }
	} else {
	    overflow = 1;
	}
    }
    if (c < 0) {
	if (ferror(in)) perror(file_name), exit(EXIT_FAILURE);
	if (i == 0) return 0;
    }
    buflen = i;
    buffer[i] = '\0';
    return 1;
}	    

static void process(char const *file_name) {
    char max_str[MAX_LINE_LENGTH + 1];
    int   max_len = -1;
    int   max_num = 0;
    int   line_no;
    FILE *in = 0 == strcmp("-", file_name) ? stdin : fopen(file_name, "r");

    if (in == 0) perror(file_name), exit(EXIT_FAILURE);
    overflow = 0;
    for (line_no = 1; read_line(file_name, in); line_no++) {
	if (buflen > max_len) {
	    (void)strcpy(max_str, buffer);
	    max_len = buflen;
	    max_num = line_no;
	}
    }
    (void)fclose(in);
    printf("%s:%d:\t", file_name, max_num);
    printf(overflow ? " >%5d " : "%7d ", max_len);
    printf("%s\n", max_str);
}

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

    if (argc < 2) {
	process("-");
    } else {
	for (i = 1; i < argc; i++) process(argv[i]);
    }
    return 0;
}