Greedy tokens and tokenizer order.md

Greedy tokens and tokenizer order

Tokenizers are greedy. The order that tokenizers are passed into the matches(from: AnyTokenizer...) will effect how substrings are matched.

Here's an example using the CharacterSet.decimalDigits tokenizer and the custom tokenizer DateTokenizer that matches dates in the format MM/dd/yy ([see example for implementation](Template tokenizer.md)).

import Mustard

let numbers = "36 03/29/17"
let tokens = numbers.tokens(matchedWith: CharacterSet.decimalDigits.anyTokenizer, DateTokenizer.defaultTokenizer)
// tokens.count -> 4
//
// tokens[0] -> CharacterSetToken
// tokens[0].set -> CharacterSet.decimalDigits
// tokens[0].text -> "36"
//
// tokens[1] -> CharacterSetToken
// tokens[1].set -> CharacterSet.decimalDigits
// tokens[1].text -> "03"
//
// tokens[2] -> CharacterSetToken
// tokens[2].set -> CharacterSet.decimalDigits
// tokens[2].text -> "29"
//
// tokens[3] -> CharacterSetToken
// tokens[3].set -> CharacterSet.decimalDigits
// tokens[3].text -> "17"

To get expected behavior, of matching the dates, the more specific DateTokenizer should come before the more general CharacterSet tokenizer:

import Mustard

let numbers = "36 03/29/17"
let tokens = numbers.tokens(matchedWith: DateTokenizer.defaultTokenizer, CharacterSet.decimalDigits.anyTokenizer)
// tokens.count -> 2
//
// tokens[0] -> CharacterSetToken
// tokens[0].set -> CharacterSet.decimalDigits
// tokens[0].text -> "36"

// tokens[1] -> DateToken
// tokens[1].text -> "03/29/17"
// tokens[1].date -> Date(2017-03-29 05:00:00 +0000)

Then if the more specific tokenizer DateTokenizer fails to match a token, the CharacterSet tokenizer will still have a chance to perform matches:

import Mustard

let numbers = "99/88/77 36"
let tokens = numbers.tokens(matchedWith: DateTokenizer.defaultTokenizer, CharacterSet.decimalDigits.anyTokenizer)
// tokens.count -> 4
//
// tokens[0] -> CharacterSetToken
// tokens[0].text -> "99"
//
// tokens[1].tokenizer -> CharacterSetToken
// tokens[1].text -> "88"
//
// tokens[2].tokenizer -> CharacterSetToken
// tokens[2].text -> "77"
//
// tokens[3].tokenizer -> CharacterSetToken
// tokens[3].text -> "36"