How to filter pair of braces

Question

I have a .toc (table of contents file) from my .tex document.

It contains a lot of lines and some of them have the form

\contentsline {part}{Some title here\hfil }{5}
\contentsline {chapter}{\numberline {}Person name here}{5}

I know how to grep for part and for chapter. But I'd like to filter for those lines and have the output in a csv file like this:

{Some title here},{Person name here},{5}

or with no braces

Some title here,Person name here,5

1. For sure the number (page number) in the last pair {} is the same for both two lines, so we can filter only the second one.

2. Note that some empty pair {} could happens or also could contain another pair {}. For example, it could be

\contentsline {part}{Title with math $\frac{a}{b}$\hfil }{15}

which should be filtered as

Title with math $\frac{a}{b}$

edit 1: I was able to obtain the numbers without braces at end of line using

grep '{part}' file.toc | awk -F '[{}]' '{print $(NF-1)}'

edit 2: I was able to filter the chapter lines and remove the garbage with

grep '{chapter}' file.toc | sed 's/\\numberline//' | sed 's/\\contentsline//' | sed 's/{chapter}//' | sed 's/{}//' | sed 's/^ {/{/'

and the output without blank spaces was

    {Person name here}{5}

edit 3: I was able to filter for part and clean the output with

    \contentsline {chapter}{\numberline {}Person name here}{5}

which returns

{Title with math $\frac{a}{b}$}{15}

Answer 1

This is using GNU awk, using POSIX awk would be very troublesome (lack of gensub, which I use more than once).

#!/usr/bin/env gawk

function join(array, result, i)
{
    result = array[0];
    end = length(array) - 1;
    for (i = 1; i <= end; i++)
        result = result "," array[i];
    return result;
}
function push(arr, elem)
{
    arr[length(arr)] = elem;
}

# split("", arr) is a horribly unreadable way to clear an array
BEGIN { split("", arr); }

/{part}|{chapter}/ {
    l = gensub(".*{(.+)}{(.+)}{([0-9]+)}$", "\\1,\\3,\\2", "g");
    if ("part" == substr(l, 0, 4)) {
        if (length(arr) > 0) { print join(arr); }
        split("", arr);
        push(arr, gensub("^(.*),(.*),(.*)$", "\\2,\\3","g", l));
    } else {
        push(arr, gensub("^(.*),(.*),(.*)$", "\\3","g", l));
    }
}

END { print join(arr); }

This uses the fact that regexes are greedy, so the matches will get the full line each time. It took more effort than I though at first.

With the following input:

\contentsline {part}{Some title here\hfil }{5}
\contentsline {chapter}{\numberline {}Person name here}{5}
blah blah
\contentsline {chapter}{\numberline {}Person name here}{5}
blah blah
blah blah
\contentsline {chapter}{\numberline {}Person name here}{5}
\contentsline {chapter}{\numberline {}Person name here}{5}
blah blah
blah blah
\contentsline {chapter}{\numberline {}Person name here}{5}
\contentsline {chapter}{\numberline {}Person name here}{5}
\contentsline {part}{Some title here\hfil }{7}
\contentsline {chapter}{\numberline {}Person name here}{7}
blah blah
blah blah
\contentsline {chapter}{\numberline {}Person name here}{7}
blah blah
\contentsline {part}{Some title here\hfil }{9}
blah blah
blah blah
\contentsline {chapter}{\numberline {}Person name here}{9}

We produce with cat input | awk -f the_above_script.awk:

5,Some title here\hfil ,\numberline {}Person name here,\numberline {}Person name here,\numberline {}Person name here,\numberline {}Person name here,\numberline {}Person name here,\numberline {}Person name here
7,Some title here\hfil ,\numberline {}Person name here,\numberline {}Person name here
9,Some title here\hfil ,\numberline {}Person name here

The page number is taken from the {part} then any {chapter} that happens after the {part} is included. This allows for several chapters inside parts of a book.

Answer 2

With the Perl Text::Balanced module the top-level {} can have their contents extracted thusly:

#!/usr/bin/env perl
use strict;
use warnings;
use Text::Balanced qw(extract_bracketed);

# this will of course fail if the input is one multiple lines, as this
# is only a line-by-line parser of standard input or the filenames
# passed to this script
while ( my $line = readline ) {
    if ( $line =~ m/\\contentsline / ) {
        my @parts = extract_contents($line);
        # emit as CSV (though ideally instead use Text::CSV module)
        print join( ",", @parts ), "\n";
    } else {
        #print "NO MATCH ON $line";
    }
}

sub extract_contents {
    my $line = shift;
    my @parts;
    # while we can get a {} bit out of the input line, anywhere in the
    # input line
    while ( my $part = extract_bracketed( $line, '{}', qr/[^{]*/ ) ) {
        # trim off the delimiters
        $part = substr $part, 1, length($part) - 2;
        push @parts, $part;
    }
    return @parts;
}

With some input:

% < input 
not content line
\contentsline {chapter}{\numberline {}Person name here}{5}
\contentsline {part}{Title with math $\frac{a}{b}$\hfil }{15}
also not content line
% perl parser input
chapter,\numberline {}Person name here,5
part,Title with math $\frac{a}{b}$\hfil ,15
% 

Answer 3

In TXR

@(repeat)
\contentsline {part}{@title\hfil }{@page}
@  (trailer)
@  (skip)
\contentsline {chapter}{\numberline {}@author}{@page}
@  (do (put-line `@title,@author,@page`))
@(end)

Sample data:

\lorem{ipsum}
\contentsline {part}{The Art of The Meringue\hfil }{5}
a
b
c
j
\contentsline {chapter}{\numberline {}Doug LeMonjello}{5}


\contentsline {part}{Parachuting Primer\hfil }{16}

\contentsline {chapter}{\numberline {}Hugo Phirst}{16}

\contentsline {part}{Making Sense of $\frac{a}{b}$\hfil }{19}

\contentsline {part}{War and Peace\hfil }{27}

\contentsline {chapter}{\numberline {}D. Vide}{19}

\contentsline {part}{War and Peace\hfil }{19}

Run:

$ txr title-auth.txr data
The Art of The Meringue,Doug LeMonjello,5
Parachuting Primer,Hugo Phirst,16
Making Sense of $\frac{a}{b}$,D. Vide,19

Notes:

• Because @(trailer) is used, the lines which give the author do not have to strictly follow their part. The data could introduce several \contentsline {part} elements which are then followed by the chapter lines that match on page number.
• @(skip) implies a search through the entire remaining data. The performance can be improved by limiting the range by adding a numeric argument. If it can be assumed that a matching {chapter} is always found within 50 lines after {part}, we can use @(skip 50).