Replace \input{fileX} by the content of fileX automatically

Question

Suppose I have a document with multiple include or input statements

\input{fileA}
\input{fileB}

etc.

Is there an easy way to generate a single .tex file where \input{fileA} is replaced by the actual content of fileA etc. without copying it manually?

Answer 1

You can use the following tools to do this. All of them are on CTAN but not all of them are part of either TeXLive or MikTeX, so you may need to manually install them. They need either Perl or a C compiler installed. Both should not be a problem with Linux but might be one under Windows or Mac. However IIRC TeXLive installs its own Perl interpreter.

---

latexpand Perl script:

Latexpand is a Perl script that simply replaces \input and \include commands with the content of the file input/included. The script does not deal with \includeonly commands.

Installation:

Simply download it from http://mirrors.ctan.org/support/latexpand/latexpand and run it. You need Perl installed however.

Usage:

    perl latexpand mainfile.tex > newfile.tex

---

flatex

A C program to flatten a LaTeX file into a single file, by explicitly including the files included by \include and \input commands. Also, if BibTeX is being used, then includes the bbl file into the resulting file. The result is therefore a stand-alone LaTeX file that can be emailed to a collaborator.

Installation:

Get the single C file and compile it, e.g. with cc flatex.c -o flatex.

Usage:

    flatex mainfile.tex > newfile.tex

---

flatten

A program to flatten a LaTeX root file by copying \input and \include files into the root file.

Installation:

Get the ZIP file from CTAN, unpack it and run make followed by make install. You need a C compiler and maybe flex for this. However it seems quite old and you might run in trouble because of it. I failed compiling it!

Usage:

    flatten mainfile.tex newfile.tex

Answer 2

You need an installed Perl. Save it as buildFile.pl and make it executable or run it with perl buildFile.pl

eval '(exit $?0)' && eval 'exec perl -S $0 ${1+"$@"}' && eval 'exec perl -S $0 $argv:q'
  if 0;
#use strict;
# 
# Usage:
# buildFile.pl < main-file.tex > main-file.tot
#     
sub p_inc {
  $DateiName = shift;
    if ( open (my $datei, "$DateiName.tex") ) {
      print "%%%---------- open: ", $DateiName, "\n";
      while (<$datei>) {
        if (/^\s*\\include{\s+(\S+)/i) {
          my $include = $1;
          chomp($include);chop($include);
          print "%%%%%%%%% Springe nach ", $include, "\n";
          p_inc($include);
        } else { print unless /^\s*(#|$)/; }
      }
      print "%%%---------- close: ", $DateiName, "\n";
      close $datei;
    } else { print "%%%<===== Datei existiert nicht\n"; }
}
#
@zeilen = (<>);
for $zeile (@zeilen) {
  next if $zeile =~ /^\s*(%)/;
  if ($zeile =~ /^\s*\\include{\s*(\S+)/i) {
    my $include = $1;
    chomp($include);
    chop($include);
    print "%%%%%%%%%%% Springe nach ", $include, "\n";
    p_inc($include);
  } else { print $zeile; }

Answer 3

I've recently discovered TexSoup, a Python module to parse Latex files inspired by BeatifulSoup. It makes really easy to work with .tex files in a very straightforward way. For instance it allows to target specific Latex commands such as \input{} in just few lines of codes.

Suppose you have a main.tex file in your folder which reads as follows:

\documentclass[12pt]{article}
\begin{document}

% ..other commands here..

\section{First part}
\input{file1}

\section{Second part}
\input{file2}

% ..other commands here..

\end{document}

If the files file1.tex and file2.tex are in the same folder just save the script at the bottom in a file named flatten.py and execute the command:

python flatten.py main.tex expanded.tex

to get the expanded file.

#!/usr/bin/env python3
import sys
from TexSoup import TexSoup

def soupify(file):
    """
    Load file in TexSoup object.
    """
    with open(file, 'r') as file_to_read:
        content = file_to_read.read()
    soup = TexSoup(content) 
    return soup

def flatten(soup):
    """
    Expand \input{} commands in a tex file.
    """
    while soup.input:
        soup.input.replace(flatten(soupify(soup.input.args[0]+'.tex')))
    return soup

if __name__ == "__main__":    
    file = sys.argv[1]
    expanded_tex = flatten(soupify(file))

    if len(sys.argv) > 2:
        output_file = sys.argv[2]
    else:
        output_file = 'output.tex'

    with open(output_file, 'w') as output:
        output.write(repr(expanded_tex))

    pass

Notice: this work for Python 3 only and you have to install TexSoup before (pip install TexSoup). If you have both Python 2 and 3 installed on the same machine you may need to launch the command using python3 instead of python as first word.

Answer 4

I developed FLaP, a python script which replaces \input, \include and other inclusion directives by the file they refer to while supporting \includeonly. In addition it move graphic files around and adjusts graphic inclusion directive (\includegraphics, includesvg and \overpic) accordingly. It supports the \graphicspath directive and SVG, EPS and PDF images.

Answer 5

I tried all the options mentioned in this post, but the original flatex works best for me. There appears to be a newer version, but I was unable to compile it as I am not familiar with multi-file C sources. If anyone could try it and let me know about the outcome that would be great.

The original flatex has an issue as it only creates a .flt (the flattened document), even when you specify a different .tex as the output. But that's not really an issue. I have created a batch file that sorts that out and allows integration with Subversion as well (see this post on my website http://www.jwe.cc/2012/02/workflow-with-subversion-and-latex/).

Answer 6

The question asks for a single TeX file, but the existing answers appear to address the case of \include and \input statements only. If those statements are generated dynamically by LaTeX commands, then static replacement by a script will miss inclusion of such files.

A solution that captures all files used is:

• add \RequirePackage{snapshot} at the top of the document, to use the snapshot package
• run bundledoc --localonly on the resulting *.dep file

The resulting archive can then be used with one of the methods described in the other answers (e.g., latexpand, flatex, flatten, or FLaP) in order to generate a single LaTeX file, along the image and other non-LaTeX files included by that file.

(Thanks to this answer.)

Answer 7

I've extended a Python script to handle \include, \input, and \import hierarchically: master2single.py

USAGE: master2single.py masterfile.tex flattenfile.tex [-v] [-c]
        masterfile.tex      : Input file
        flattenfile.tex     : Flattened file
        -v                  : Verbose mode
        -c                  : Remove all comments

Answer 8

There is also texdirflatten which is a perl script.

This worked best for me. YMMV. There is also the consideration that perl is not installed by default on Windows machines.

Answer 9

An alternative is to use Pandoc. Not Pandoc's intended usage, but works more or less:

pandoc -s file_nested.tex -o file_flat.tex

This will convert all \input as well as tables etc to a flat file. When I tried, it did not convert my preamble correctly, so I had to copy paste that part. But all the text in the body of the file got flattened.

Answer 10

Late to the party, but here's another python script (let's call it merge-tex.py) for the same thing. It could be shorter if there were no diagnostics, but it's not recursive, and doesn't do \include.

Use as python merge_tex.py main.tex merge.tex.

import sys
import re
from pathlib import Path

if len(sys.argv) < 3:
    raise ValueError("Expected two arguments")

in_file = Path(sys.argv[1])
out_file = Path(sys.argv[2])

# https://regex101.com/r/azkAmW/1
regex = re.compile(r"^([^%\\]*)(.*)\\input{(.*)}(.*)$")

with open(out_file, "w") as fout:
    with open(in_file, "r") as fin:
        print(f"Reading: {in_file}")
        for line in fin:
            match = regex.findall(line)
            if not match:
                fout.write(line)
            else:
                # regex groups
                start, comment, input_file, end = match[0]
                if comment:  # nothing to do
                    fout.write(line)
                else:
                    if start:
                        fout.write(start)
                    print(f"Merging: {input_file}.tex")
                    with open(Path(input_file + ".tex"), "r") as finput:
                        [fout.write(inp) for inp in finput]
                    if end:
                        fout.write(end)

print(f"Finished. Merged to: {out_file}")