suggester 0.9.7

Autocompletion engine with highly configurable partial matching. Suitable for web or mobile.

An autocompetion engine for Dart.

Supports highly configurable partial matching.

Fast and memory efficient.

Suitable for both web and mobile devices.

Test It Live !

Play with a Live instance of Suggester.

Test different combinations of data and parameters tofind what suits your application.

Usage

Import

import 'package:suggester/suggester.dart';

Choose Term Mapping

Suggester works by tokenising both suggestions and input using a Term Mapping and finding partial matches between them via prefix searching.

For example a TermMapping that groups inputs and suggestions into sequences of letters and numbers:

print(AlphaOrNumeric().map('[e-mail]', false, false));

{armani, klein, 3, veda, io}

TermMapping that groups inputs and suggestions into NGrams of size 3, i.e. Trigrams.

print(Ngrams(3).map('[e-mail]', false, false));

{arm, rma, man, ani, ni., i.k, .kl, kle, lei, ein, in3, n3@, 3@v, @ve, ved, eda, da., a.i, .io}

Construct Suggester and Load Entries

A small list of entries. Suggester comfortably handles any number.

final entries = [
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]',
  '[e-mail]'
];

Construct 2 example Suggester objects, each using a different Term Mapping. Load each with the above list of entries.

final suggesterAlphaOrNumeric = Suggester(AlphaOrNumeric())
  ..addAll(entries);

final suggesterTrigram = Suggester(Ngrams(3))..addAll(entries);

Suggest!

Find all suggestions featuring search term 'wen', note the difference between TermMappings.

print(suggesterAlphaOrNumeric.suggest('wen').map((e) => e.entry.value));

([e-mail])

print(suggesterTrigram.suggest('wen').map((e) => e.entry.value));

([e-mail], [e-mail], [e-mail])

Mark Suggestion Search Terms

User feedback is helped by marking a suggestion with the search terms. Pass 2 custom functions to be applied to marked and unmarked terms. Terms can be mapped to any type, here it is mapped to html tags but could as well return an iterable of UI elements.

print(suggesterTrigram.suggest('wen').map((e) => e
    .mapTermsAndNonTerms((term) => '<mark>$term</mark>', (nonTerm) => nonTerm)
    .join()));

(Ro<mark>wen</mark>[e-mail], banjo.ber<mark>wen</mark>@newell.ca, Curtis_McGlynn@<mark>wen</mark>dell.me)

[e-mail], [e-mail], [e-mail]

Expand Search With Edit Distance

An edit distance may also be specified during search allowing terms to be matched using a Hamming edit distance. This extends the range of a TermMapping. Using the original AlphaOrNumeric example with a maxEditDistance of 1:

print(suggesterAlphaOrNumeric
    .suggest('wen', maxEditDistance: 1)
    .map((e) => e.entry.value));

([e-mail], [e-mail])

... and with a maxEditDistance of 2

for (final suggestion
    in suggesterAlphaOrNumeric.suggest('wen', maxEditDistance: 2)) {
  print(suggestion.entry.value);
}

[e-mail]
[e-mail]
[e-mail]
[e-mail]
[e-mail]
[e-mail]
[e-mail]
[e-mail]
[e-mail]
[e-mail]
[e-mail]
[e-mail]
[e-mail]
[e-mail]
[e-mail]
[e-mail]

Accept Suggestions

Once a selection is accepted this can be recorded via the accept method:

final result = suggesterAlphaOrNumeric.suggest('lu');

print(result.map((final suggestion) => suggestion.entry.value));

Note ordering of search suggestions.

([e-mail], [e-mail])

Accept last suggestion in ordering:

result.last.entry.accept();

print(result.map((final suggestion) => suggestion.entry.value));

Last suggestion now becomes first to reflect prior acceptance.

([e-mail], [e-mail])

By default when a suggestion is accepted it results in a subScore value incrementing but any int value may specified including a timestamp so that last used suggestions are always favoured. For example:

result.last.entry.accept(subScore: DateTime.now().millisecondsSinceEpoch);

print(result.first.entry);

{value: [e-mail], subScore: 1609579026000}

Check Current Entries

Iterate through all current entries

print(suggesterAlphaOrNumeric.entries);

Iterate all entries ordered by subScore. Note that the last accepted entry is first with timestamp subScore.

print(suggesterAlphaOrNumeric.entriesBySubScore());

[{value: [e-mail], subScore: 1609569378812},
  {value: [e-mail], subScore: 1},
  {value: [e-mail], subScore: 0},
  {value: [e-mail], subScore: 0},
  {value: [e-mail], subScore: 0},
  {value: [e-mail], subScore: 0},
  {value: [e-mail], subScore: 0},
  {value: [e-mail], subScore: 0},
  {value: [e-mail], subScore: 0},
  {value: [e-mail], subScore: 0},
  ...]

Serialise

Suggester is serialisable so suggestions and approval history are easily stored and reused.

final json = suggesterAlphaOrNumeric.toJson();

final rehydrated = Suggester.fromJson(AlphaOrNumeric(), json);

if (SuggesterEquality().equals(rehydrated, suggesterAlphaOrNumeric)) {
  print('Rehydrated successfully!');
}

Rehydrated successfully!

Features and bugs

Please file feature requests and bugs at the issue tracker.