Unit 3: Finding Things

Counting without reading

wc counts. On its own it gives three numbers: lines, words, characters, in that order.

A flag picks just one. -l for lines is the one you will use nine times out of ten.

Terminal

kid@alguni:~$ wc story.txt
2 6 23 story.txt
kid@alguni:~$ wc -l story.txt
2 story.txt
kid@alguni:~$ wc -w story.txt
6 story.txt

How big is that folder?

ls -l is ls in long form: one line per thing, with a lot more on it.

The number in the middle is the size in characters. There is plenty more on that line — you will read the rest of it in unit 5, which is entirely about the first ten letters.

Terminal

kid@alguni:~$ ls -l
total 12
-rw-r--r-- 1 kid kid  292 Jan  1 09:00 big.txt
drwxr-xr-x 2 kid kid 4096 Jan  1 09:00 folder
-rw-r--r-- 1 kid kid    6 Jan  1 09:00 small.txt

The best command there is

grep prints every line that has your word in it, and ignores the rest.

Give it the word first, then the file. On a file of ten thousand lines it takes no longer than on ten.

Terminal

kid@alguni:~$ grep Turing people.txt
Alan Turing 1912
kid@alguni:~$ grep 19 people.txt
Alan Turing 1912
Grace Hopper 1906
Katherine Johnson 1918

Its odd name

grep is not a word. It comes from an instruction in an ancient editor: g/re/p, meaning "globally search for a pattern and print it".

Nobody would name it that today. Everybody uses it anyway, and now you know why it looks like a cat walked over the keyboard.

Nothing found

When nothing matches, grep says nothing at all. No "0 results", no message.

That silence *is* the answer, and it takes a bit of getting used to. Unix commands that worked and had nothing to say keep quiet.

Terminal

kid@alguni:~$ grep pineapple fruit.txt
kid@alguni:~$ grep apple fruit.txt
apple

Four flags worth knowing

-i — ignore capitals, so cat finds Cat
-n — show which line number each one was
-v — turn it inside out: the lines that do not match
-c — just count them

You can stack them: grep -in does both.

Terminal

kid@alguni:~$ grep cat pets.txt
cat food
kid@alguni:~$ grep -i cat pets.txt
Cat
cat food
kid@alguni:~$ grep -n -i cat pets.txt
1:Cat
3:cat food
kid@alguni:~$ grep -c -i cat pets.txt
2

Everything except

-v is the odd one and it is enormously useful. It gives you every line that does not match — which is how you throw away the boring lines and keep the rest.

Terminal

kid@alguni:~$ grep -v "#" config.txt
real line
another real line

Standing for anything

A * in a name means "anything at all, however long".

*.txt means every name ending in .txt. photo* means every name starting with photo. And ? is the same idea for exactly one character.

Terminal

kid@alguni:~$ ls
a.txt
b.txt
notes.txt
photo1.jpg
photo2.jpg
song.mp3
kid@alguni:~$ ls *.txt
a.txt
b.txt
notes.txt
kid@alguni:~$ ls photo*
photo1.jpg
photo2.jpg
kid@alguni:~$ ls ?.txt
a.txt
b.txt

The command never sees the star

Here is the part that surprises everybody. ls knows nothing about *.

The shell looks at *.txt, works out which files match, and hands ls the finished list. By the time ls wakes up it has been given three plain names.

Watch echo prove it: echo cannot search for files, and yet it prints them all.

Terminal

kid@alguni:~$ echo *.txt
a.txt b.txt c.txt
kid@alguni:~$ echo "*.txt"
*.txt

When nothing matches

If no file matches, the shell gives up and hands the command the pattern exactly as you typed it.

So ls *.zip in a folder with no zip files complains about a file literally called *.zip. That message confuses everybody once. Now it will not confuse you.

Terminal

kid@alguni:~$ ls *.zip
ls: cannot access '*.zip': No such file or directory

Looking everywhere at once

A wildcard only looks in one folder. find goes all the way down, into every folder inside every folder.

Say where to start, then what you are looking for. find . -name "*.txt" means "starting here, everything ending in .txt".

Terminal

kid@alguni:~$ find school
school
school/art
school/art/paint.jpg
school/maths
school/maths/sums.txt
school/timetable.txt
kid@alguni:~$ find school -name "*.txt"
school/maths/sums.txt
school/timetable.txt

Files or folders

-type f says files only. -type d says directories only.

And notice the quotes round "*.txt". Without them the shell would expand the star itself, before find ever saw it — which is exactly what the last lesson warned about.

Terminal

kid@alguni:~$ find a -type f
a/b/two.txt
a/one.txt
kid@alguni:~$ find a -type d
a
a/b

A real search

You wrote something about volcanoes. You cannot remember what you called it or where you put it.

You know two things about it: it is a .txt file, and the word volcano is in it. That is enough.