How come grep [sd] testfile1 won't find anything in testfile1 (even though the characters sd are there in great quantity), but grep '[sd]' testfile1 will find plenty?

Do the single quotes prevent the shell from interpreting the [ test command? So when not using single quotes, the grep [sd] testfile1 is interpreted as: grep *test whether or not characters sd exist* testfile1?

Or is the explanation much simpler? Thanks!

You are correct - you have to prevent the shell from interpreting the the '[sd]' regex.

Just for completeness, [sd] without the quotes is actually a shell wildcard, not a "test" invocation. It matches any file whose name is s or d.

Thank you!

Second question: Is there way to insert math into a grep, for example:

In a large chat log file I want to grep for every instance of an abbreviation of an ordinal number: 21st, 31st, 41st, 101st, 401st, 2nd, 22nd, 32nd etc.

So I would want to do egrep [x+10]st|y+10]nd|[z+10]rd] chatlog.txt or something like that. Is there any way to have the shell evaluate this math and append 'st' 'nd' 'rd' as necessary? Thanks!

Read up on regular expressions. The egrep manual page has a very brief summary; some systems also have a regex(5) manual page which is a bit more detailed. Any good Unix beginners book will have a chapter or five on the subject.

There's absolutely no math in there but it's a very powerful and versatile facility for matching ... stuff.

Note that this will also match '21st', '103245th', etc. The grep family of tools are strictly substring-based and line-oriented; any line containing a match anywhere in it will be printed.

Great explanation. It's much more clear now. I was just making things a bit too complicated. Thank you for the help.