Pipeline Stages

Search is a fluent builder. Every configuration method returns this (or a widened this, in register’s case), so a pipeline reads top to bottom as one chain ending in .search(query).

new Search()
	.register(NaiveSearch())
	.register(KMP(), { stopOnMatch: true })
	.register(Levenshtein())
	.defineDataset(allModels)
	.runOrder(["naive", "kmp", "levenshtein"])
	.defineBehavior("find-all")
	.cullMatchedRows()
	.search("MacBook Pr 16");

None of these calls execute a search. Nothing runs until .search(). Up to that point you are only building configuration.

.register(algorithm, options?)

Adds an algorithm to the pipeline under its own name. Names must be unique within a pipeline; registering the same name twice throws a SearchConfigError. The second argument configures the algorithm’s behavior, threshold or match limit for this pipeline, separately from whatever options its factory was constructed with. See Match Behavior and Thresholds and Confidence for what those options do, and Run Order and Type Safety for what register does to the instance’s type.

.defineDataset(strings)

Sets the array of strings every algorithm searches. It must be an array of strings: a non-array or an array containing a non-string entry throws a SearchConfigError naming the offending index.

.runOrder(names)

Restricts and orders which registered algorithms actually run. A subset is fine; anything left out of the list never runs. Skip this call entirely and algorithms run in registration order. Its parameter type is the union of names registered so far. See Run Order and Type Safety.

.defineBehavior(behavior)

Sets the outcome behavior for the whole pipeline: "find-all" (the default) or "stop-on-match". See Match Behavior.

.cullMatchedRows(enabled?)

Controls whether a row one algorithm matches is removed from the pool later algorithms search through. true by default. See Culling Matched Rows.

.search(query)

Runs the pipeline and returns a SearchResult. This is the only method that executes anything: everything before it is configuration. It throws a SearchConfigError if no dataset was defined or nothing was registered. See Working with Results for the shape it returns.