Is there any parsing when reading 'clean' text file using Scanner?

Question

I know that:

Parsing is the process of turning some kind of data into another kind of data.

But then I also came across this difference between Scanner and BufferedReader:

BufferedReader is faster than Scanner because BufferedReader does not need to parse the data.

So my question is how is using Scanner slower than using BufferedReader if I am reading just text file (plain characters) and I am not doing any parsing? Is there any parsing I am not aware of?

Or from following code perspective, how is here Scanner slower because of parsing than using BufferedReader?

//1
BufferedReader bufferedReader = new BufferedReader(new FileReader("xanadu.txt"));
System.out.println(bufferedReader.readLine());
    
//2
Scanner scanner = new Scanner(new FileReader("xanadu.txt"));
scanner.useDelimiter("\n");
System.out.println(scanner.next());

I don't understand quote how Scanner is slower because of parsing, when I am technically not parsing any data..

But yes, there's parsing to know when the end delimiter is found — OneCricketeer, Sep 06 '20 at 13:59
The only way you can determine this is by benchmarking. To be honest, I doubt the different in speed will matter until you get into the hundreds of thousands of lines. — NomadMaker, Sep 06 '20 at 13:59

score 1 · Accepted Answer · answered Sep 06 '20 at 19:42

1

Dividing an input stream into lines is a (very limited) form of parsing, but as you say BufferedReader can also do that. The difference, if there is one, will be that BufferedReader can use a highly-optimised procedure to implement a single use case (divide a stream into lines) while Scanner needs to be able to be considerably more flexible (divide a stream into tokens delimited by an arbitrary string or regular expression). Flexibility almost always comes at a price, although you won't know what that cost is without doing some benchmarking. (And it may be very small, since it is conceivable that Scanner has optimised algorithms for particular special cases which it can recognise.)

In short, "because parsing" is not a very good explanation for why one interface is slower than another one. But the more flexibly and precisely you parse an input, the more time it is expected to take.

answered Sep 06 '20 at 19:42

rici

234,347
28
237
341

Thanks for answer. I thought that term parsing is only used when converting. For example nextInt() would be parse method as it converts String into int. What does parsing mean here? You said dividing stream into lines is form of parsing. But it is technically not converting anything, right? – Stefan Sep 07 '20 at 01:44
1

@stefan: Parsing means to analyse a string into its constituent parts. Indeed, the similarity between the word "parse" and the word "parts" is not a coincidence; it is actually the etymology of "parse". So in a very primitive sense, you can parse an input into individual lines (or fields). Originally, parsing applied to sentences and meant figuring out which part of speech each word is (and that usage goes back centuries before computers). Conversion really has little to do with parsing; it's at best a side effect. – rici Sep 07 '20 at 01:59

Is there any parsing when reading 'clean' text file using Scanner?

1 Answers1