How a trie works was explained in the last recipe, but how we fill it and how we query it looks a bit strange here. Let's have a closer look at the code snippet that fills the empty trie with the content of the text database file:

fstream infile {"db.txt"};
for (string line; getline(infile, line);) {
    istringstream iss {line};
    t.insert(istream_iterator<string>{iss}, {});
}

The loop fills the string line with the content of the text file, line by line. Then, we copy the string into an istringstream object. From such an input stream object, we can create an istream_iterator, which is useful because our trie does not only accept a container instance for looking up subtries but also primarily iterators. This way, we do not need to construct a vector or a list of words and can directly consume the string. The last piece of unnecessary memory allocations could be avoided by moving the content of line into iss. Unfortunately, std::istringstream does not provide a constructor that accepts std::string values to be moved. It will copy its input string, nevertheless.

When reading the user's input to look it up in the trie, we use exactly the same strategy but we do not use an input file stream. We use std::cin, instead. This works completely identically for our use case because trie::subtrie works with iterators just as trie::insert does.