Operating on Collections

• We usually use an iterator to process a collection.

  • Suppose we have split a text file as a list of words.
  • We want to count the number of long words in the list.

  List<String> words = ....;
  long count = 0;
  for(String w : words){
  	if(w.length() > 10){
  		count++;
  	}
  }
  

• An iterator generates all elements from a collection as a sequence.

• Alternative Approach

  • Generate a stream of values from a collection.
  • Operations on the stream transforms input streams to output streams.
  • Terminate with a result.

  long count = words.stream().filter(w -> w.length() >10).count();
  

Why Streams?

• Stream processing is declarative.

  • Recall, declarative vs imperative.
  • Focus on what to compute, rather then how.

• Processing can be parallelized.

  long count = words.paralleStream().filter(w->w.length()>10).count();
  

• Lazy Evaluation is possible.

  • Suppose we want only the first 10 words.
  • Stop generating the stream once we find 10 such words.
  • Need not generate the entire stream in advance.

Working with Streams

• Create a stream.
• Pass through intermediate operations that transforms streams.
• Apply a terminal operation to get a result.

A stream does not actually stores its elements.

• Elements are stores in the underlying collection.
• Or is generated by a function, on demand.

Stream operations are non-destructive

• Input stream is untouched.

Creating Streams

• Apply stream() to a collection.

  • Part of the Collections interface.

  List<String> wordlist = .... ;
  Stream<String> wordstream = wordlist.stream();
  

• Use static method Stream.of() for arrays.

  String[] wordarr = .... ;
  Stream<String> wordstream = Stream.of(wordarr);
  

• Static method: Stream.generate() generates a stream from a function.

  Stream<String> echos = Stream.generate( () -> "Echo" ) ;
  

  Stream<Double> randomds = Stream.generate(Math::random)