.TH tac 1 "01 April 2018" "tac" "IOCCC 2018"
.SH NAME
tac \- tokenize and count C-like program source code
.SH SYNOPSIS
\fBtac\fP [\-t] [\-c] [\-k] [\-r] [\-s] [\-i] [file]
.SH DESCRIPTION
\fBtac\fP reads C or C-like source code from \fBstdin\fP or file, breaking it into tokens (lexemes),
counting these according to the rules of the IOCCC, and outputting the tokens,
counts, or source code as directed by the command line options.
The default output is the source code with comments removed on \fBstdout\fP,
and the counts on \fBstderr\fP.
The count format is a set of integer values in the following order:
.RS
.TP
.B lines
number of newlines
.TQ
.B words
number of words
.TQ
.B chars
number of characters
.TQ
.B ioccc
number of IOCCC countable tokens; the IOCCC rule 2 value
.TQ
.B saved
number of characters not counted due to rules and keywords
.TQ
.B kwcnt
number of keywords
.TQ
.B kwlen
number of characters saved by counting keyword length as 1
.RE
.PP
For the following code:
.sp
.RS 4
.nf
#include <stdio.h>
int main() {
    /* example for loop */
    for (int ac=1; ac<10; ++ac)
	    printf("%d\n", ac);
    return 0;
}
.fi
.RE
\fBtac\fP prints the code to \fBstdout\fP, and to \fBstderr\fP: '6 15 96 71 25 5 0'
which means 6 lines, 15 words, 96 chars, 71 IOCCC counts, 25 chars saved,
5 \fBkwcnt\fP, and no characters saved from \fBkwlen\fP. (By default, comments are not
counted.)
.SH OPTIONS
The following options are supported:
.TP
.B \-t
Output just the token stream.
.TP
.B \-c
Correct the accounting bugs in \fBiocccsize(1)\fP, output the token id and the token.
.TP
.B \-k
Keep the comments.
.TP
.B \-r
Reserved words count as size 1.
.TP
.B \-s
Suppress the source code, summarize counts only.
.TP
.B \-i
Report only the official IOCCC rule 2 size: number of octets (implies \fB\-r\fP and \fB\-s\fP).
.PP
Counts are impacted by the \fB\-k\fP, \fB\-r\fP and \fB\-c\fP options.
By default, the counts do not include comments, words are counted as in \fBwc(1)\fP, and reserved words count in full.
Using the example code above, the options change the counts as follows:
.PP
.TP
\fB\-r\fP
6 15 96 53 25 5 18
.TQ
\fB\-k\fP
7 20 120 89 31 5 0
.TQ
\fB\-c\fP
7 36 98 71 27 5 0
.PP
The \fB\-c\fP option counts tokens instead of words, and is therefore more accurate for source code.
Note the aberration for line count with \fB\-r\fP; it is as\-if the comment never existed.
.PP
The \fB\-c\fP and \fB\-t\fP options change the output, instead writing a stream of tokens to \fBstdout\fP, one token per line.
The \fB\-t\fP option suppresses both the counts and the token ids, writing only the tokens.
.SH EXAMPLES
.PP
Summarize the source code counts:
.sp
.RS
.ft B
tac -rs tac.c
.ft R
.RE
.PP
Count the 10 most frequently used unique tokens in the source code:
.sp
.RS
.ft B
tac -t tac.c 2>/dev/null | sort | uniq -c | sort -k1nr | sed 10q
.ft R
.RE
.PP
Count the unique constants in the source:
.sp
.RS
.ft B
tac -c tac.c 2>/dev/null | grep ^300 | sort | uniq -c | sort -k1nr | awk '{print $1,$3}'
.ft R
.RE
.SH BUGS
\fBtac\fP does not properly count preprocessor tokens with embedded space or comments as reserved words:
.sp
.RS
.ft B
echo '#/*comment*/include/*comment*/<stdio.h>' | tac -i
.br
echo '# include<stdio.h>' | tac -i
.br
echo '#include<stdio.h>' | tac -i
.ft R
.RE
.sp
report no reserved words on the first two inputs, and thus 17 as the IOCCC size, although 10 is the correct answer (as in the final example).
.PP
The original meaning of the \fB\-c\fP option was "become compatible with \fBiocccsize(1)\fP".
Its meaning was flipped so the default behavior is now compatible.
.SH SEE ALSO
\fBiocccsize(1)\fP, \fBwc(1)\fP
.br
\<https://ioccc.org/2015/rules.txt\>