How do I tell which command has the longest manual on my system?

Question

Is there an easy way to find out which command has the longest manual pages?

Answer 1

You can calculate it yourself for your system with simple command

$ find /usr/share/man/ -type f -exec ls -S {} + 2>/dev/null | head | while \
  read -r file; do printf "%-40s" "$file"; \
  man "$file" 2>/dev/null | wc -lwm; done | sort -nrk 4

which returns on my box

      (file)                             (lines) (words) (chars)
/usr/share/man/man1/zshall.1.bz2          27017  186394 1688174
/usr/share/man/man1/cmake.1.bz2           22477  106148 1004288
/usr/share/man/man1/cmake-gui.1.bz2       21362  100055  951110
/usr/share/man/man1/perltoc.1.bz2         18179   59783  780134
/usr/share/man/man1/cpack.1.bz2            9694   48264  458528
/usr/share/man/man1/cmakemodules.1.bz2    10637   42022  419127
/usr/share/man/man5/smb.conf.5.bz2         8306   49991  404190
/usr/share/man/man1/perlapi.1.bz2          8548   43873  387237
/usr/share/man/man1/perldiag.1.bz2         5662   37910  276778
/usr/share/man/e                           1518    5919   58630

where columns represent number of lines, words and characters respectively. Rows (commands) are sorted by last column.

We can do similar thing for info pages, but we have to bear in mind that it's content can span over many files. Thus let's use the benefits of zsh to keep above one-liner in compact form:

 $ for inf in ${(u)$(echo /usr/share/info/**/*(.:t:r:r))}; do \
   printf "%-40s" "$inf"; \
   info "$inf" 2>/dev/null | wc -lwm; done | sort -nrk 4

what gives

   (info title)                          (lines) (words) (chars)
elisp                                     72925  457537 3379403
libc                                      69813  411216 3066817
lispref                                   62753  374938 2806412
emacs                                     47507  322194 2291425
calc                                      33716  244394 1680763
internals                                 32221  219772 1549305
zsh                                       34932  206851 1544909
gsl-ref                                   32493  179954 1518248
gnus                                      31723  180613 1405064
gawk                                      27150  167135 1203395
xemacs                                    25734  170403 1184250

Info pages are huge mostly for gnu-related stuff what is understandable, but I find interesting that for example zsh has more lines and words but less characters than in man pages. It is interesting because at first glance the content is the same, just formatting is a little bit different.

---

_{Explanation of zsh tricks in the selection of the files for the loop: for inf in ${(u)$(echo /usr/share/info/**/*(.:t:r:r))}; do}

_{The goal is to create the list of unique file names from /usr/share/info directory and all subdirectories. Files should be stripped from dirname, extenstions and all numbers. The above snippet can be rewritten as ${(u)$(echo /usr/share/info/**/*(.)):t:r:r}, what gives the same result but uses probably more decent syntax, namely:}

• **/*: descent into all subdirectories and mark everything there
• (.): select only plain files
• :t: remove pathname components (works like basename)
• :r: remove extension (everything after last dot, including dot). It is applied twice to remove also unnecessary string and number (e.g. .info-6 from file zsh.info-6.bz2)
• (u): show only unique words (after previous operations there are many the same words - different files/chapters for the same info command)

Answer 2

Man pages are stored in /usr/share/man/manX where X is the section (described in man man). They're compressed in gzip format, so let's assume a larger compressed file means a bigger manpage.

By checking in /usr/share/man/man1 (section 1: Executable programs or shell commands) with the command gzip -l *.gz | sort -n -k2, I get this (which will probably vary from distro to distro, this was on an Arch Linux system with a bunch of packages) where the first column is the compressed file size and the second column is the uncompressed file size:

         161077              607106  73.5% ffprobe-all.1
         198943              757155  73.7% ffserver-all.1
         217339              792577  72.6% msp430-g++.1
         217339              792577  72.6% msp430-gcc.1
         209129              794118  73.7% ffmpeg-all.1
         261778              972719  73.1% avr-g++.1
         261778              972719  73.1% avr-gcc.1
         262154              975423  73.1% g++.1
         262154              975423  73.1% gcc.1
         227830             1123043  79.7% perltoc.1perl

Answer 3

An alternative to the methods provided by @Renan and @jimmij yields wireshark-filter the big winner on my system.

for i in {1..9}; do du -sh man"$i"/*.gz | grep -v "^..0K" | grep -v "^0\|^12K\|^16K\|^[0-9][0-9]K" ; done

Based on that I did a opened each of the largest entries with man and checked the number of lines at the end of the file with a :f and came up with:

wireshark-filter = 245016 lines
gcc              =   8341 lines
perlfunc         =   5132 lines

Answer 4

The example by @jimmij is interesting, but incorrect because it returns results from only one subdirectory. Running that script (on my Debian 7), I get this:

/usr/share/man/man3/DBI.3pm.gz             6182   35812  271206
/usr/share/man/man8/openvpn.8.gz           4021   24702  202032
/usr/share/man/pt/man1/nmap.1.gz           2563   21214  159284
/usr/share/man/man8/lsof.8.gz              2714   18670  142698
/usr/share/man/man3/pcrepattern.3.gz       2579   18631  131204
/usr/share/man/man3/pcreapi.3.gz           2382   16966  123349
/usr/share/man/man8/iptables.8.gz          2631   14844  114354
/usr/share/man/man8/ip6tables.8.gz         2465   13619  105283
/usr/share/man/man3/CPAN.3perl.gz          2142   12346   98823
/usr/share/man/man8/mount.8.gz             2136   12059   97407

Here's a quick revision which fixes that problem:

find /usr/share/man/ -type f -ls | \
        awk '{ printf "%s %s\n", $7, $11; }' | \
        sort -r -n 2>/dev/null | \
        awk '{ printf "%s\n", $2; }' 2>/dev/null | head -n 20 |
while \
        read -r file; do printf "%-40s" "$file"; \
        man "$file" 2>/dev/null | wc -lwm; done | sort -nrk 4

On my Debian 7 system that gives

/usr/share/man/man1/cmake.1.gz            19264   86499  803021
/usr/share/man/man1/perltoc.1.gz          18755   62738  657990
/usr/share/man/man1/fvwm2.1.gz             8942   57060  462861
/usr/share/man/man1/cmakemodules.1.gz      9843   37965  376314
/usr/share/man/man1/perlfunc.1.gz          7535   47684  370246
/usr/share/man/man5/smb.conf.5.gz          8303   49962  359329
/usr/share/man/man1/cpack.1.gz             8037   38214  342264
/usr/share/man/man1/bash.1.gz              5465   42031  320015
/usr/share/man/man1/perlapi.1.gz           7074   36791  309196
/usr/share/man/man1/xterm-dev.1.gz         6242   34747  297639
/usr/share/man/man3/DBI.3pm.gz             6182   35812  271206
/usr/share/man/man1/xterm.1.gz             5238   29434  253526
/usr/share/man/man1/zshcompsys.1.gz        4502   31336  244115
/usr/share/man/man1/perldiag.1.gz          4939   33200  237882
/usr/share/man/man1/tcsh.1.gz              4355   29640  226498
/usr/share/man/ru/man1/nmap.1.gz           3048   21396  187181
/usr/share/man/hu/man1/nmap.1.gz           3020   21519  186431
/usr/share/man/de/man1/nmap.1.gz           2965   21384  182926
/usr/share/man/man1/nmap.1.gz              3005   24785  179485
/usr/share/man/ru/man1/mc.1.gz             3571   19200  173292

For what it's worth, there are 10326 files under /usr/share/man on that machine (ymmv).