Lecture 4 – Sep 13th, 2022

Setup

Log on to clyde: ssh user@clyde.cs.oberlin.edu.
Create a directory with whatever name you like and cd into it.

Task

The command $ ls /bin /usr/bin will print out a list of all the files in /bin and /usr/bin. Each of these is a program you can run, but there are likely too many to display in your terminal. So run the command again, but redirect the output to a file called binaries.
Run $ wc -l binaries to see how many lines are in the file.
Open the file with your favorite command line editor (e.g., nvim, vim, emacs, nano) and take a look at the output. Is the number you printed out in step 2 the actual number of binaries? Hint: Look closely at the first line.
One thing to notice is the output from ls is different when directed to a file than it is when it goes to the terminal. It turns out that programs can detect if their stdout is connected to a terminal or not. Some programs use this ability to change their output, some do not. Run $ man isatty and take a look at the manual page for the isatty() function. This is a C function and we’ll get to programming in C soon enough. For now, it’s enough to know that stdout is file descriptor 1 so the function call isatty(1) tells the program if stdout is a terminal or not.
Let’s figure out how many binaries start with the letter z by using grep. Recall grep reads files (or stdin) and prints lines matching a pattern. Run $ grep z binaries. Notice that it prints out all lines containing a z. That’s not what we want, but we’re close.
Grep interprets the ^ character to mean “match the beginning of the line” so $ grep ^z binaries should do what we want. Run that and look at the output.
Let’s figure out how many of them there are. Rerun that grep command and redirect stdout to a file zfiles. Run the appropriate wc command to figure out how many there are.

To recap what we’ve done so far, we created a file containing a list of all of the binaries in /bin and /usr/bin. We used grep to extract just those that start with z and stored that in a file. Finally, we ran wc on that to figure out how many there are. Pretty simple but we had to create two files for it that we don’t really care about if our goal was just to figure out that final number.
We can skip the creation of the first file by piping the output of ls to grep and storing the result to a file. Explicitly,
```
$ ls /bin /usr/bin | grep '^z' >zfiles1
```
Notice that we don’t specific a file name for grep. When we omit it, it reads from stdin.
Let’s check that the two files we just created are identical by using the diff program to print out the differences between two files. Run
```
$ diff -u zfiles zfiles1
```
Nothing should be printed when you do that because the files should be the same.
Append a line of text (anything you want) to zfiles1 using echo with an append redirection (>>).
Rerun the diff command from step 9. You should see some output. (Diffing files is extremely useful. We’ll see more of this later and you’ll see a lot of it as you program more.)
Enough diff for now. In step 8, we skipped the creation of one file by using a pipe but we still created a second file. Construct a new command piping the output of ls to grep and piping that output to wc. You should get exactly the output you got in step 7.
We actually did more work than we needed to. Remember that the shell will expand wildcards that match file names. For example, /bin/z* will be expanded to the paths of all of the files that match that pattern. Run $ echo /bin/z* to see the result.
Using ls rather than echo (because echo prints a single line and we want multiple lines) and the appropriate patterns, construct a pipeline to count all of the binaries starting with z again. That is, fill in the missing bits in
```
$ ls [...] | wc -l
```
That’s enough for today. You can delete the directory you created.