#!/bin/sh # wordcount.sh 1.7 Copyright 2016/06/03 Michael John Downes # Following Michael's death, his family have requested that his TeX # related files are made publicly available under the LPPL license. # # This file is available under the LaTeX Project Public License version 1 # or (at your option) any later version. # https://latex-project.org/lppl/ # (LPPL maintenance status: unmaintained) # # This is a Unix Bourne shell script that counts the characters and # spaces in a LaTeX document by running it through LaTeX using # wordcount.tex, and then counting the lines in the resulting # wordcount.log file that match a special pattern. It assumes that you # have a program "grep" with a -c option meaning "Report the number of # matching lines". To use perl or some other tool instead of grep, # redefine $COUNT in some suitable way, below. # # The characters that are counted with this method are actual font # characters in the page body (excluding running heads): in math, each # subscript and superscript character is counted separately, even when # they appear one over the other; and so are the separate combining # pieces used to make large delimiters. Interword spaces are counted as # separate characters, and so are the spaces around mathrel symbols in # math formulas (equal signs and the like)---but not the spaces around # mathbin symbols (plus, minus, and the like). # # You can count words by dividing the character count by 5 (or # whatever), or by counting the occurrences of interword spaces. The # latter method (illustrated here) is highly accurate at counting the # actual number of words---until tables or math formulas enter the # picture. In the case of tables, the number of words will tend to be # overestimated because tabskip glue, counted as an interword space, # appears on both sides of each table cell. (Not counting the tabskip # glue at all would result in an underestimate, and the underestimate # would tend to be slightly less accurate than the overestimate that we # have chosen to do.) # # And of course, in the case of math formulas, how do you define what is # a "word"? SCRIPT=`basename $0` LOGFILE=wordcount.log COUNT="grep -c" LATEX=latex die () { echo "$@ was apparently unsuccessful." exit 1 } no_files_msg () { echo "$SCRIPT: At least one file name must be supplied" \ "on the command line." exit 1 } case X$1 in X) no_files_msg ;; *) : ;; esac for f in "$@" do # Since \? is not defined by LaTeX itself, this use via \csname causes # it to be defined as \relax, and it then serves as a filename # terminator. And within wordcount.tex if \? is found to be no longer # undefined, certain messages are suppressed. echo "$f" | $LATEX 'wordcount.tex\csname ?\endcsname' || die "LaTeX run" # Count interword spaces and end-of-paragraph spaces words=`$COUNT '3[.]0863[35]' $LOGFILE` || die "Log search" # Count nonspace characters and interword spaces chars=`$COUNT '3[.]0863[23]' $LOGFILE` || die "Log search" # If you want to see what the log file looks like, comment out the # following line: rm $LOGFILE echo "$f contains $chars characters and $words words." done