Lines as records

What you will be able to do

After this lesson you can

• read a file in which each line is a record,
• turn the texts of a line into the right types,
• change a record and write the file again, and
• keep going when a line is bad.

A line is a record

A file is just text, but its lines often mean something. A teacher keeps a file in which each line is a name, a colon and a mark: Asha:72. The program reads the file and gets back a dictionary.

Predict first. What is the average? Who is the best student? Then run it.

The steps are always the same. Learn them as a pattern.

1. Make the file. (In a real program, the file already exists. We write it here so that the program runs alone.)
2. Read each line: for line in f:.
3. Clean it: line.strip() removes the new line.
4. Cut it: .split(":") gives ["Asha", "72"]. We unpack it into name, text.
5. Change the type: int(text). Everything in a file is text. A mark must be made a number before you add it.
6. Keep it: marks[name] = ....

Then the usual work: the average is 77.75, and the best is Kabir with 90.

To change a record, change the dictionary, and write the whole file again with "w". There is no way to change one line in the middle of a text file. We write the names in sorted order, one line each, with a \n. The new file holds Asha:72, Kabir:90, Meera:64 and Ravi:88, one on each line.

Counting what is in a file

The same loop counts lines, words and different words.

int(text) is the step that turns the file's text into a number. Without it, the comparison with the pass mark would fail.

Bad lines

Real files are messy. A teacher may leave an empty line, forget a mark, type eighty, or add a stray colon.

The loop catches the ValueError, records the line number, and goes on. The result: two good records and three bad lines (2, 3 and 4). enumerate(f, start=1) gives the numbers that a person would use. A report that says "line 2 is bad" is far more useful than a crash, or than silence.

The three-step rule for a file that comes from outside: expect trouble, keep the good, and report the bad.

Common mistakes

• Forgetting that everything is text. "72" + "85" is "7285".
• Forgetting strip(). The last field has a \n stuck to it, and int("72\n") works, but a name with \n does not match.
• Splitting on the wrong separator, or on a separator that also appears inside a value.
• Reading the whole file into a variable when you only need to go through it once.
• Changing a record and forgetting to write the file. The change is lost when the program ends.
• Writing the file again with "a". The old records stay, and you get duplicates. Use "w".

Practice

Start with two quick questions. Each New numbers gives you a fresh one, so you can practise as long as you like.

Then the labs and the problems:

1. Warm-up. Open the lab Read the marks file. You write a function that turns the lines of a file into a dictionary, and the checker gives it good and bad lines.
2. Build. Open the problem Count the words in a file and the problem The longest line.
3. Stretch. Open the problem Update a record and change one mark in a file without losing the others.

Remember these

• A line can be a record. Strip it, split it, and change the types.
• Everything that you read from a file is text.
• To change a record, change the data and write the whole file again with "w".
• Expect bad lines. Keep the good, report the bad, with their numbers.

Check yourself

Try each one on your own first. If you are stuck, open a hint. Then check with Show the answer.

1. What does "Asha:72".split(":") give?

2. Why must you write int(text) after you read a mark?

3. What should a program do with a bad line?