Last month, I explained how you can use the grep and cut tools to extract specific data fields from complex XML-based files, especially in the Safari web browser's bookmarks file on Mac OS X (yeah, I know, Mac OS X isn't Linux, but the two systems share a lot of common Unix underpinnings and, well, the Mac is a prettier laptop. Roll with me, the scripting environment is basically identical anyway).

This time I want to switch to the similar, though not identical, task of producing a Web-friendly version of the /etc/passwd file, showing you how you can extract data and translate it into specific HTML so that the data file can be transformed into an easily-shared Web page in just a few lines.

Extracting the Specific Data
One powerful capability of shell scripts is that you can change the field separator from a space to any other character you might need. Working with the password file, the obvious switch is to use a ":" as is obvious when you look at a representative line from this data file:

leah:*:1008:1008:leah999:/home/leah:/bin/sh

The fields here are account name, encrypted password (the "*" is a placeholder because the encrypted password is stored in a separate shadow file), user ID, group ID, optional comment field, home directory, and, finally, login shell.

The trick is to be able to tease out all this data in a shell script, and to do that, I'll change the internal field separator value, or IFS, to a colon, and then use a while loop in a way you might not have seen before:

pwf="etc.passwd"

IFS=":"

while read name pw uid gid comment home shell
do
  echo "read in name = $name home = $home and shell = $shell"
  done < $pwf

exit 0

Running this rather interesting-looking script produces:

$ ./login-webpage.sh | head -5
read in name = root home = /root and shell = /bin/csh
read in name = toor home = /root and shell = 
read in name = daemon home = /root and shell = /sbin/nologin
read in name = operator home = / and shell = /sbin/nologin
read in name = bin home = / and shell = /sbin/nologin

Perfect.

Getting things HTMLized
Now, we need to take the nicely named data fields and turn them into some attractive HTML-based data. To do that, I'll pour them into a table in which the rows are denoted by <tr> and </tr> and the individual data cells by <td> and </td>. The script changes like this:

echo "<table border=1>"

while read name pw uid gid comment home shell
do
  echo "<tr><td>$name</td><td>$uid</td>"
  echo "<td>$comment</td><td>$home</td>"
  echo "<td>$shell</td></tr>"
done < $pwf

echo "</table>"

Now you can see that once the fields are identified, it's pretty darn easy to change the echo statements to produce what we wanted. Note the addition of the <table> at the beginning and the </table> at the end of the loop.

This isn't really what I wanted, though, because I only wanted to see accounts with valid login shells, so it's time to add a conditional test in the while loop to see if the shell's legit:

while read name pw uid gid comment home shell
do
  if [ "$shell" != "/sbin/nologin" ] ; then
    echo "<tr><td>$name</td><td>$uid</td>"
    echo "<td>$comment</td><td>$home</td>"
    echo "<td>$shell</td></tr>"
  fi
done < $pwf

Notice that you need to ensure that it's not the nologin shell rather than test to see if it's a legit login shell because accounts with no specified shell are given a default login shell of /bin/sh (that is, they're still valid login shells).

Putting It All Together
Put this all together and here's how we can use this script:

$ ./login-webpage.sh > account-info.html

and then the password file information can be viewed in a Web browser, as shown in Figure 1.

I'd also like to invite you, dear reader, to e-mail me with some ideas for shell scripts that you'd like to see me build and/or disassemble and explain too. Please send your ideas to me at d1taylor@gmail.com