IB Computer Science Topic B2.5.1: Constructing Code to Perform File-processing Operations Notes

File Processing: Keeping Data After the Program Stops

1. A variable lives in RAM, so every value a program holds disappears the moment it stops running.
2. Writing that data to a file puts it in secondary storage, where the next run can read it back.
3. B2.5.1 covers text files, which hold characters and are processed as a sequence of lines.
4. Text files are sequential, so a program works through them from the start toward the end.
   1. A file position marker moves forward as each line is read.
   2. Reading a line a second time means closing the file and opening it again.

Opening a File: The Mode Decides What Happens to Existing Data

1. A program cannot read or write a file until it opens it, which links the file on disk to an object in the code.
2. The mode given at that moment fixes what the program may do and what happens to data already in the file.
   1. Read mode ("r") opens an existing file for input and raises an error if the file is not there.
   2. Write mode ("w") creates the file if it is missing and empties it if it already exists.
   3. Append mode ("a") creates the file if needed and adds new data at the end, leaving the existing contents intact.
3. Python passes the mode as the second argument to open().
4. Java has no mode letter, so the class you choose is the mode.
   1. Scanner and BufferedReader open a file for reading.
   2. FileWriter opens a file for writing, and passing true as its second argument switches it to append.

Opening a File in Python and Java

Reading a File: Three Ways to Pull Text Out

1. read() in Python returns the whole file as one string, which suits a small file you want to search or print in one go.
2. readline() in Python returns the next single line and moves the position marker past it.
3. Looping over the file object reads one line at a time without holding the whole file in memory, which is the usual choice.
4. Java reads with a loop and a test for the end of the file.
   1. Scanner pairs hasNextLine() with nextLine(), so the loop stops when no next line remains.
   2. BufferedReader uses readLine(), which returns null once the end of the file is reached.
5. Python keeps the newline character on the end of each line it reads, so strip() removes it before the value is used.
6. Java's nextLine() and readLine() return the line without its newline character.
7. Every line arrives as a string, so a number must be converted with int() in Python or Integer.parseInt() in Java before any arithmetic.

Reading Line by Line in Both Languages

Writing and Appending: Text Only, and Newlines Are Yours to Add

1. write() sends a string to the file and adds nothing of its own.
   1. A line break appears only if the code writes the newline character itself.
   2. A number must be converted first, with str(total) in Python or String.valueOf(total) in Java.
2. Write mode starts from an empty file, so a program that saves a whole updated list each run uses "w".
3. Append mode keeps what is already there, so a program that adds one record per run uses "a".
   1. A log file, a high-score table and an attendance register are all append jobs.

Writing and Appending in Both Languages

Closing a File: The Step That Actually Saves the Data

1. Written data sits in a buffer in memory until the file is closed, so an unclosed file can leave the copy on disk short or empty.
2. Closing also releases the file handle, an operating system resource that comes in limited numbers.
3. Python's with block closes the file when the block ends, even if an error is raised inside it.
4. Python's close() does the same job explicitly when with is not used.
5. Java closes nothing on its own, so close() must be called on every Scanner, BufferedReader and FileWriter.
   1. Java's try-with-resources form closes the file for you when the try block ends.

Two Ways to Guarantee a Close

File Errors: A File Can Fail For Reasons the Code Cannot See

1. A file is an external resource, so it fails in ways ordinary input does not: the file is missing, the path is wrong, or the operating system refuses permission.
2. Python raises FileNotFoundError when read mode cannot find the file, and IOError for wider input and output failures.
3. Java throws FileNotFoundException and IOException, which are checked exceptions.
   1. A checked exception stops the program compiling until it is caught in a catch block or declared with throws.
   2. This is why Java file code is almost always wrapped in try and catch while arithmetic is not.
4. The try/except, try/catch and finally constructs themselves belong to B2.1.3 Exception Handling, so this section covers only what files add to them.

Guarding a Read That May Fail

A Full Program: Read One File, Filter It, Write Another

1. Paper 2 file questions usually move data between two files rather than printing it, so this pattern is worth knowing by heart.
2. The shape stays the same whatever the task.
   1. Open the input file for reading and the output file for writing.
   2. Loop through the input one line at a time.
   3. Split each line into its fields and test the condition.
   4. Write the records that pass to the output file.
   5. Close both files.
3. In the example below scores.txt holds one record per line in the form name,mark, and every record with a mark of 80 or more is copied into top.txt.

scores.txt to top.txt in Both Languages

Answering a Paper 2 File Question

1. File code is marked step by step, so an answer that opens the files, loops, tests and writes still collects most of the marks with one line wrong.
2. The specimen mark scheme withholds marks when files are opened without indicating the access mode, so always write the "r", "w" or "a", or the true in a Java append.
3. Some questions prohibit specific built-in functions such as sort, pop, len, max and min, and the paper says so on the question itself.
4. A Java method that touches a file must either catch IOException or declare throws IOException, and an answer that does neither will not compile.

Where the Marks Sit in a File Question

Adapting the Two-File Pattern to Other Tasks

1. Counting rather than copying replaces the write with a counter that rises each time the condition is true.
2. Totalling rather than copying adds the converted number to a running total, which is printed once the loop ends.
3. Updating one record means reading the whole file into a list, changing the entry in memory, then writing the list back in "w" mode.
   1. Append mode cannot do this, because a sequential file offers no way to overwrite a line in place.