LaTeXTidy.pl

#!/usr/bin/perl

# LaTeXTidy (c) 2004 by Eric Hsu <drerichsu@gmail.com>.

# Little Perl script to neaten up the format of LaTeX files.
# This will be simple and naive. This takes STDIN .tex files and prints to STDOUT.
# Check your file! Backup! No guarantees!

# License
# -------
# This is released as Niceware, which is like the Perl Artistic License, except you have to be nice to me when you criticize the code.

# General Idea
# ------------

# Eat all single newlines. Add newlines after all "\\"
# Newlines before each \begin and \end. After each \end{}
# Each environment \begin adds a level of tab.
# Newlines before each \item.
# \n before each \[  and after each \]

my $in="";

while (<STDIN>) {
    $in .= $_;
}

my @keywords = qw(
    appendix
    author
    bibliography
    bigskip
    chapter
    date
    def
    document
    evensidemargin
    font
    headheight
    headsep
    include
    index
    make
    new
    noindent
    oddsidemargin
    page
    paragraph
    part
    ragged
    renew
    section
    subsection
    subsubsection
    subsubsubsection
    table
    textheight
    textwidth
    title
    topmargin
    use
    vfil
);

# Let's ignore all comments in the following way. We first find all \%(.*?)\n.
# Then we put a second \n at the end, and two leading \n
# to ensure that they all land in
# separate pieces. Then each piece that has a leading % is immediately passed.

$in =~ s/(?<!\\)\%(.*?)\n/\n\n\%$1\n\n/g;

my @pieces = split(/\n\s*\n/, $in);
my $string="";
my $keyword;

foreach (@pieces){

# Every comment is left as is.  But ignore % that are immediately preceded by \
    if (/^\s*(?<!\\)\%/) {
        $string .= $_ . "\n" ;
        next;
    }

#Eat all single newlines.

    s/\s+/ /g;

#Put @keywords on their own line.

    foreach $keyword (@keywords) {
        s/(\\$keyword)/\n$1/g;
    }

#Newlines before each \begin and \end. After each \end{}
#We want to ignore begin and end document, since those shouldn't
#induce additional indenting

    s/([^\\]\%)/\n$1/g;


    s/(\\begin\{)((?!document).*?)(\})/\n$1$2$3\n/g;
    s/(\\end\{)((?!document).*?)(\})/\n$1$2$3\n/g;
    s/(\\begin\{array\})\n(\{)(.*?)(\})/\n$1$2$3$4\n/g;

#Newlines before each \item.

    s/(\\item)(.*?)(\\item)/$1$2\n$3/g;
    s/(\\item)/\n$1/g;

#\n before each \[  and after each \]
#Add newlines after all "\\" and "\\[...]"

    s/[^\\](\\\[)/\n$1/g;
    s/(\\\])/$1\n/g;

    s/(\\\\)\s/$1\n/g;
    s/(\\\\\[)(.*?)(\])\s/$1$2$3\n/g;


# nuke accidentally added double newlines.

    s/\n\s*\n/\n/g;

# collect the cleaned string.

    s/^\n//;
    chomp;
    $string .= $_ . "\n\n";

}


# First let's collapse all multiple \n's into double \n.

$string =~ s/\n\s+\n/\n\n/g;

# We will soon mark the \end and \begin keywords, but we want to ignore ones
# found as comments. Hence we'll (awful kludge) wedge in a \{\n\n\n\} to commented
# \begin and \end to avoid their processing.
# We'll fix them right after the pieces are split.

$string =~ s/(\%[^\n]*)(\\)(end)/$1$2\{\n\n\n\}$3/g;
$string =~ s/(\%[^\n]*)(\\)(begin)/$1$2\{\n\n\n\}$3/g;


# Now let's put [triple \n] at the start of each \begin and the start of each
# \end. Then we'll split on them, since they are unique.
# Each of those pieces must be at the same indent level. Again, we need to
# ignore beginning and end of document.

#$string =~s/(\\end)/\[\n\n\n$\]$1/g;
#$string =~s/(\\begin)/\[\n\n\n$\]$1/g;
#
#@pieces = split(/\[\n\n\n$\]/, $string);

$string =~s/(\\end(?!\{document\}))/\[\n\n\n\]$1/g;
$string =~s/(\\begin(?!\{document\}))/\[\n\n\n\]$1/g;

@pieces = split(/\[\n\n\n\]/, $string);

my $indent =1;
my @lines;
my ($piece, $i);

# $string is now free for reuse.
$string="";

foreach $piece (@pieces) {
    #first, is this a begin block, or after an end block?

    $piece =~ s/\{\n\n\n\}//g; #get rid of awful kludge.
    $piece =~ /^\\(.*?)\{/;
    if (lc($1) eq "begin") {
        $indent++;
    } else {
        $indent--;
    }
    # each piece is split on \n. these pieces must begin with $indent tabs.
    # We need to avoid combining comment lines with others.

    @lines = split(/\n/, $piece);

    foreach (@lines) {
        s/^\s+//; #no leading whitespace
        if (/^\\begin/i) {
            for($i=1;$i<=$indent-1;$i++){
                $string .= "\t";
            }
        } else {
            for($i=1;$i<=$indent;$i++){
                $string .= "\t";
            }
        }
        $string .= $_ . "\n";
    }

}

    print $string;

#(0.1) First version works. It indents LaTeX more or less correctly.
#(0.2) Added a big list of LaTeX words to check. Squashed bug losing double lines. Handles comments. Handles sections, more or less. (5/3/02)
#(0.21) Pushed keywords out to its own array. Added some more keywords. (5/4/02)
#(0.22) Trying to ignore comments.
#(0.23) Trying to port it to BBEdit.
#(0.3) Now for TextMate.
#(0.31) Neating the code so strict mode doesn't complain as much.

# To Do.
# Not catching \usecommand!
#  Take all such \sections
#   and give the header a line of its own?