I want to make a pdf book on Linux manual pages for C functions using LaTeX.

For example man malloc looks as bellow -

man ls output I want this type of style in my pdf book. Specific color is not mandatory, but the color must be matched with terminal color.

  • 9
    why use latex rather than groff since the man pages are all marked up for the (g)roff typesetting system? (Although convertors exist eg google for man2latex) Sep 10, 2017 at 14:06
  • 2
    Oh and also: the final “style” depends on your pager and terminal; it's not entirely under the control of man. For example, try PAGER=cat man malloc, PAGER=more man malloc, and PAGER=most man malloc (if you have most installed). You can also vary the LESS_TERMCAP_md (etc.) termcap settings, to change your terminals interpretation of what to do when it's asked to display something “bold” (whether to show it in a particular colour, etc). Sep 10, 2017 at 15:30
  • 6
    I just want to say that you have to be very careful with the licenses of the man pages because your book will have to respect every license of used man page and sometimes (if you will not have only one license) the licenses are not compatible.
    – koleygr
    Sep 10, 2017 at 18:35
  • A lot of GNU utilities have info pages, which are done through texinfo with more detailed documentation.
    – Batman
    Sep 10, 2017 at 23:24
  • @ShreevatsaR: It's also quite convenient (or at least I've found it so) to use none of those pagers, but instead write a macro that formats the page and dumps it into your editor-of-choice.
    – jamesqf
    Sep 11, 2017 at 4:46

3 Answers 3


If you did not know, man pages are already written in a typesetting system that is even older than TeX. This is the *roff family: at different points in history, related programs have gone by names like RUNOFF, runoff, roff, nroff, troff, ditroff, groff, etc. If you're using a Unix-like system (a good guess as you're asking about man pages), you probably have groff installed on your system already, and possibly under names like troff and nroff.

In fact, man pages are written in a macro package for the *roff typesetter. If you type man -d malloc, you get some debugging information: on my computer the last line shows what command it would have run, to (typeset and) display the malloc man page:

cd '/usr/share/man' && (echo ".ll 18.3i"; echo ".nr LL 18.3i"; /bin/cat '/usr/share/man/man3/malloc.3') | /usr/bin/tbl | /usr/bin/groff -Wall -mtty-char -Tascii -mandoc -c | (/usr/bin/less -is || true)

This shows that the file /usr/share/man/man3/malloc.3 is passed through first the tbl preprocessor (which deals with tables), and then is formatted by groff for display on screen. The “input” /usr/share/man/man3/malloc.3 file itself has instructions like this:

.Fn malloc
function allocates
.Fa size
bytes of memory and returns a pointer to the allocated memory.

This is analogous to writing (in some hypothetical LaTeX package) something like

The \functionname{malloc} function allocates \functionarg{size} bytes of memory and returns a pointer to the allocated memory.

This is “typeset” by the preprocessors ending with /usr/bin/groff -Wall -mtty-char -Tascii -mandoc -c (definitions of these .Fn and .Fa *roff macros, and the fact that function names should be typeset in bold and arguments should be underlined, are in the manpage-related macro package) and this is why it ends up on screen with appropriate bold and underlines as:

line from malloc man page

Therefore: if you want to generate PDF instead, you just have to change the output format. This you can do in multiple ways (unfortunately there are some differences in the output so you may want to try each of them and pick the one you like the most):

man -t malloc > malloc.ps
ps2pdf malloc.ps

or (yes there exist programs other than TeX that can generate DVI files!)

groff -T dvi -m mandoc '/usr/share/man/man3/malloc.3' > malloc.dvi
dvipdfmx malloc.dvi


groff -T ps -m mandoc '/usr/share/man/man3/malloc.3' > malloc.ps
ps2pdf malloc.ps

or on Linux, you may need:

zcat '/usr/share/man/man3/malloc.3.gz' | tbl | groff -T ps -m mandoc > malloc.ps
ps2pdf malloc.ps

or variants where you use a different converter from PS to PDF or from DVI to PDF. Then you can include the PDF directly into your LaTeX document; you can search on this site for many ways of doing that. If you don't like the page margins, line lengths etc., there are ways you can specify them to groff.

Another alternative is to use the mandoc program, which understands the source format of man files:

zcat '/usr/share/man/man3/malloc.3.gz' | mandoc -T pdf > malloc.pdf


zcat '/usr/share/man/man3/malloc.3.gz' | mandoc -T html > malloc.html
# convert from html to pdf or to latex in your preferred way

Note that a conversion to html opens up various possibilities for converting into LaTeX. For example, you can use pandoc. Here is an example that matches some aspects of the display in your screenshot:

  • all bold text to displayed in red (as your terminal evidently does)
  • the background is not white
  • instead of italics, underlines are used (your terminal does that because it does not have an italic font; you may consider whether you want to match that in PDF: underlining is usually considered poor typography)

Create mancolours.tex containing:


% Set background colour (of the page)

% Make bold text appear in a particular colour

% Use underlines instead of emphasis (ugh)

% % Use fixed-width font by default
% \renewcommand*\familydefault{\ttdefault}

and then:

zcat '/usr/share/man/man3/malloc.3.gz' | mandoc -T html > malloc.html
pandoc -s -o malloc.tex --include-in-header=mancolours.tex malloc.html
pdflatex malloc.tex

This produces stuff like:

matching the screenshot closely

Finally, if none of these are satisfactory, you can look at the source of the man page and write your own tool for translating *roff macros into whatever LaTeX macros you'd like as equivalents. There aren't too many of those, so this should be reasonably doable. (There are some scripts online where people have written similar translators, but I tried a couple and neither worked well enough. So it would be better to write your own.) You may also consider operating on the output from mandoc -Thtml or mandoc -Ttree, if you find those easier.

Yet another option, if you want to match formatted terminal output exactly, is dumping that to a file along with the formatting. When you run man malloc, the pager invoked is most likely something like less. If you dump to a file everything that is displayed, and open the file in a decent editor, you'll see how the terminal does it:

 The m^Hma^Hal^Hll^Hlo^Hoc^Hc() function allocates _^Hs_^Hi_^Hz_^He bytes of memory and…

(the actual character in the file is byte 8, I have changed it to the two characters ^H above so that you can see it). So: to make a character bold, it prints the character, then ^H, and then the character again. To make something underlined, it prints a _, then ^H and then the character. (These make sense if you imagine that ^H acted like moving backwards, and overprinting a character on itself made it bold — this is actually how things worked at some point historically.) On top of that, your terminal preferences get applied, for how it displays such bold and underlined characters.

So, now that you have this file, you can extract the formatting in it, into a format suitable for LaTeX. For example, with the following Python script I turn those into \bold{...} and \underline{...} respectively (man malloc happens to contain no backslashes, but if it did you'd probably want to replace those too):

import re
import sys

def parseFormatting(text):
    """Detects 'bold' and 'underlined' characters in the output to a terminal."""
    chars = [(c, '') for c in text]
    while True:  # Detect bold characters
        m = re.search('(.)\x08\\1', ''.join(c[0] for c in chars))
        if not m: break
        s = m.start()
        chars[s : s + 3] = [(chars[s + 2][0], 'bold')]
    while True:  # Detect underlined characters
        m = re.search('_\x08.', ''.join(c[0] for c in chars))
        if not m: break
        s = m.start()
        chars[s : s + 3] = [(chars[s + 2][0], 'underline')]
    i = 0
    while i < len(chars):  # Collapse runs of identical formatting (for efficiency later)
        j = i
        while j < len(chars) and chars[j][1] == chars[i][1]: j += 1
        chars[i : j] = [(''.join(chars[k][0] for k in range(i, j)), chars[i][1])]
        i += 1
    return chars

def parseFileReplaceFormatting(filename):
    text = open(filename, 'rb').read().decode('utf-8').split('\n')
    newtext = ''
    for line in text:
        for c in parseFormatting(line):
            if c[1] == '':
                newtext += c[0]
            elif c[1] == 'bold':
                newtext += '\\bold{%s}' % c[0]
            elif c[1] == 'underline':
                newtext += '\\underline{%s}' % c[0]
            else: assert False, ('Unknown formatting', c[1], 'for', c[0])
        newtext += '\n'
    return newtext

if __name__ == '__main__':
    infilename = sys.argv[1]
    outfilename = sys.argv[2]
    updated = parseFileReplaceFormatting(infilename)
    with open(outfilename, 'wb') as f:

So after running the above script with something like:

python malloc.man.less malloc.man.less.py2

you can process (\input) the resulting file with TeX. If you wish, you can even preserve line-breaks and whatever crude hyphenation-and-justification your terminal did! (Of course by doing this you lose all the benefits of TeX's beautiful line-breaking algorithm, but you get to match the terminal output exactly.) You just have to make sure that the width of your pages and your terminal are roughly compatible:


\usepackage[paperwidth=11in, textwidth=10in, textheight=4in, paperheight=5in]{geometry}


% Set background colour (of the page)
% Make bold text appear in a particular colour

% Foreground colour

\def\nextline{\null\par} % \null so that a blank line in input (two consecutive newlines) becomes an empty paragraph.
{\catcode`\^^M=\active \def^^M{\nextline} \catcode`#=12 \catcode`_=12 \catcode32=12\relax\input{malloc.man.less.py2}}


page 1 of output generated from above TeX document

(You can tell that the above output was generated by TeX because of the black page number at the bottom!)

  • /usr/share/man/man3/malloc.3 file not found. /usr/share/man/man3$ ls | grep malloc results malloc.3.gz malloc_get_state.3.gz __malloc_hook.3.gz malloc_hook.3.gz malloc_info.3.gz __malloc_initialize_hook.3.gz malloc_set_state.3.gz malloc_stats.3.gz malloc_trim.3.gz malloc_usable_size.3.gz
    – alhelal
    Sep 10, 2017 at 16:20
  • 2
    @BandaMuhammadAlHelal That's not a question about TeX/LaTeX :-) But if I had to guess, I'd say your terminal's pager settings have changed; try restarting your terminal or computer or whatever. Beyond that I don't know, sorry. Sep 18, 2017 at 17:24
  • 1
    @alhelal It's just U+2002 (en space); you can just write a definition for it, or replace the character in the .tex file. Or just compile with xelatex or lualatex (instead of pdflatex). Any of the standard solutions for dealing with Unicode characters will work; I don't think there's any special problem here. Dec 28, 2017 at 8:43
  • 1
    @alhelal I have given you some basic ideas that answer the question originally asked; it is somewhat unreasonable to provide “support” here for every single thing that you may encounter during your project. E.g. Unicode has nothing particularly to do with man pages; it's a general question about TeX. Similarly, nothing in this answer inserts a “colophon” section; if you get such a section, it's there because it's there in your manpages. Finally, about underlines, I already mentioned in the answer that it's “considered poor typography”—just shown how you can achieve them if you really want. Dec 28, 2017 at 17:25
  • 1
    @alhelal If you think this workflow (for generating PDF from man pages, in terminal style) will be useful to others (or even yourself in future), how about writing a blog post (or your own answer to this question), in which you describe all the changes you made, for all the assorted things that came up, etc.? You could keep editing it as you change or mind or encounter new concerns. Then there would be a better reference for you and for anyone who wanted to do the same thing. You could copy from that reference (your answer or blog post) instead of copying from this (old) answer. Dec 28, 2017 at 17:41

May be man2html plus pandoc could be a simple good start:

$ zcat '/usr/share/man/man3/malloc.3.gz' | man2html > malloc.html
$ pandoc -s -f html -t latex  malloc.html -o malloc.tex

But if you do not need modify the LaTeX source, you can export directly to PDF:

$ pandoc -f html -t latex malloc.html -o malloc.pdf


Then, with a new preamble (and removing some first lines):


\settextfraction {1}
\setlength{\parskip}{6pt plus 2pt minus 1pt}
\setlength{\emergencystretch}{3em}  % prevent overfull lines
\def\textbf#1{\textcolor{texto}{\oldbfseries #1}}
\title{Man page of MALLOC}



malloc, free, calloc, realloc - allocate and free dynamic memory

... % remaining text is not changed 


man -w man returns the path to the file that stores the man page of the command man.

On my machine , man -w man returns /usr/share/man/fr/man1/man.1.gz

so you can easily combine this command with others

man -l -Tdvi $(man -w man) > man.dvi && dvipdf man.dvi && rm -f man.dvi && xpdf man.pdf

explanation: 1. man -l -Tdvi $(man -w man) > man.dvi Ask man to tell groof to produce a dvi file of the manpage man

2.dvipdf man.dvi We tell dvipdf to create man.pdf for the man.dvi file

  1. rm -f man.dvi We remove the tempory man.dvi file

  2. xpdf man.pdf We use xpdf to display the file. I've used it here cause I guess most systems will have it by default.

you can use okular on kde system okular man.pdf

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.