Add fuzzy-matching option for tag names

guardian · Jul 14, 2024 · bd1e54b · bd1e54b
1 parent 01b2597
commit bd1e54b
Show file tree

Hide file tree

Showing 5 changed files with 43 additions and 15 deletions.
diff --git a/...scala/deprecated/anotherschema/Edge.scala → ...m/gu/contentapi/porter/graphql/Edge.scala b/...scala/deprecated/anotherschema/Edge.scala → ...m/gu/contentapi/porter/graphql/Edge.scala
@@ -9,11 +9,11 @@ import java.nio.charset.StandardCharsets
 import java.util.Base64
 import scala.util.Try
 import io.circe.syntax._
-@deprecated("you should be using com.gu.contentapi.porter.graphql")
+
 case class Edge[T:io.circe.Decoder](totalCount:Long, endCursor:Option[String], hasNextPage:Boolean, nodes:Seq[T]) {
   def map[V:io.circe.Decoder](mapper:(T)=>V) = Edge[V](totalCount, endCursor, hasNextPage, nodes.map(mapper))
 }
-@deprecated("you should be using com.gu.contentapi.porter.graphql")
+
 object Edge {
   private val logger = LoggerFactory.getLogger(getClass)
   private val encoder = Base64.getEncoder

diff --git a/src/main/scala/com/gu/contentapi/porter/graphql/RootQuery.scala b/src/main/scala/com/gu/contentapi/porter/graphql/RootQuery.scala
@@ -74,6 +74,7 @@ object RootQuery {
         arguments = TagQueryParameters.AllTagQueryParameters,
         resolve = ctx =>
           ctx.ctx.repo.marshalledTags(ctx arg TagQueryParameters.QueryString,
+            ctx arg TagQueryParameters.Fuzziness,
             ctx arg TagQueryParameters.tagId,
             ctx arg TagQueryParameters.Section,
             ctx arg TagQueryParameters.TagType,

diff --git a/src/main/scala/com/gu/contentapi/porter/graphql/TagQueryParameters.scala b/src/main/scala/com/gu/contentapi/porter/graphql/TagQueryParameters.scala
@@ -96,13 +96,29 @@ object TagQueryParameters {
     )
   )
 
+  val FuzzinessOptions = EnumType(
+    "FuzzinessOptions",
+    Some("Valid options for making a fuzzy-match query"),
+    List(
+      EnumValue("AUTO",
+        value="AUTO",
+        description=Some("Generates an edit distance based on the length of the term. If the term is >5 chars, then 2 edits allowed; if <3 chars than no edits allowed")
+      ),
+      EnumValue("OFF",
+        value="OFF",
+        description=Some("Disable fuzzy-matching")
+      )
+    )
+  )
+
   val tagId = Argument("tagId", OptionInputType(StringType), description = "Retrieve this specific tag")
   val Section = Argument("section", OptionInputType(StringType), description = "Only return tags from this section")
   val TagType = Argument("type", OptionInputType(TagTypes), description = "Type of the tag to return")
-  val QueryString = Argument("q", OptionInputType(StringType), description = "Generic Lucene query string for finding tags")
+  val QueryString = Argument("q", OptionInputType(StringType), description = "Search for tags that match this public-facing name")
+  val Fuzziness = Argument("fuzzy", OptionInputType(FuzzinessOptions), description = "Perform a fuzzy-matching query (default). Set to `OFF` to disable fuzzy-matching.")
   val Category = Argument("category", OptionInputType(StringType), description = "A category to match against tags")
   val Reference = Argument("reference", OptionInputType(StringType), description = "A reference to match against tags")
-  val AllTagQueryParameters = QueryString :: tagId :: Section :: TagType :: Category ::
+  val AllTagQueryParameters = QueryString :: tagId :: Section :: TagType :: Fuzziness :: Category ::
     Reference :: Cursor :: OrderBy :: Limit :: Nil
 
   val NonPaginatedTagQueryParameters = Section :: TagType :: Nil

diff --git a/src/main/scala/datastore/DocumentRepo.scala b/src/main/scala/datastore/DocumentRepo.scala
@@ -21,6 +21,7 @@ trait DocumentRepo {
                      limit: Option[Int], cursor: Option[String]): Future[Edge[Content]]
 
   def marshalledTags(maybeQuery:Option[String],
+                     maybeFuzziness:Option[String],
                      maybeTagId:Option[String],
                      maybeSection: Option[String],
                      tagType:Option[String],

diff --git a/src/main/scala/datastore/ElasticsearchRepo.scala b/src/main/scala/datastore/ElasticsearchRepo.scala
@@ -8,7 +8,7 @@ import com.sksamuel.elastic4s.ElasticDsl._
 import scala.concurrent.ExecutionContext.Implicits.global
 import scala.concurrent.Future
 import com.sksamuel.elastic4s.requests.searches.SearchResponse
-import com.sksamuel.elastic4s.requests.searches.queries.{ExistsQuery, NestedQuery, Query, RangeQuery}
+import com.sksamuel.elastic4s.requests.searches.queries.{DisMaxQuery, ExistsQuery, Fuzzy, FuzzyQuery, NestedQuery, Query, RangeQuery}
 import com.sksamuel.elastic4s.requests.searches.queries.compound.BoolQuery
 import com.sksamuel.elastic4s.requests.searches.queries.matches.{FieldWithOptionalBoost, MatchAllQuery, MatchQuery, MultiMatchQuery}
 import com.sksamuel.elastic4s.requests.searches.sort.{FieldSort, ScoreSort, Sort, SortOrder}
@@ -185,8 +185,19 @@ class ElasticsearchRepo(endpoint:ElasticNodeEndpoint, val defaultPageSize:Int=20
 
   private def tagQueryParams(maybeTagId:Option[String], maybeSection:Option[String],
                              tagType:Option[String], maybeCategory:Option[String],
-                             maybeReferences: Option[String]):Seq[Query] = {
+                             maybeReferences: Option[String], queryString:Option[String], fuzziness:Option[String]):Seq[Query] = {
     Seq(
+      queryString.map(qs=>{
+        if(fuzziness.getOrElse("AUTO") != "OFF") {
+          //Why DisMax here? Because we want to include exact-matches as well, if they are relevant. E.g. FuzzyQuery on "politics" returns no results!
+          DisMaxQuery(Seq(
+            FuzzyQuery("webTitle", qs, fuzziness),
+            MatchQuery("webTitle", qs)
+          ))
+        } else {
+          MatchQuery("webTitle", qs)
+        }
+      }),
       maybeTagId.map(MatchQuery("id", _)),
       maybeSection.map(MatchQuery("sectionId", _)),
       tagType.map({
@@ -204,18 +215,16 @@ class ElasticsearchRepo(endpoint:ElasticNodeEndpoint, val defaultPageSize:Int=20
   private def buildTagQuery(maybeTagId:Option[String],
                             maybeSection:Option[String],
                             tagType:Option[String], maybeQuery:Option[String],
+                            maybeFuzziness:Option[String],
                             maybeCategory:Option[String], maybeReferences:Option[String]) = {
     val baseSearch = search("tag")
-    val searchWithQuery = maybeQuery match {
-      case Some(q)=>baseSearch.query(q)
-      case None=>baseSearch
-    }
-    val params = tagQueryParams(maybeTagId, maybeSection, tagType, maybeCategory, maybeReferences)
+
+    val params = tagQueryParams(maybeTagId, maybeSection, tagType, maybeCategory, maybeReferences, maybeQuery, maybeFuzziness)
 
     if(params.isEmpty) {
-      searchWithQuery
+      baseSearch
     } else {
-      searchWithQuery.query(BoolQuery(must=params))
+      baseSearch.query(BoolQuery(must=params))
     }
   }
 
@@ -230,6 +239,7 @@ class ElasticsearchRepo(endpoint:ElasticNodeEndpoint, val defaultPageSize:Int=20
   //FIXME: tagsForList / marshalledTags could be DRY'd out a bit
 
   override def marshalledTags(maybeQuery:Option[String],
+                              maybeFuzziness:Option[String],
                               maybeTagId:Option[String],
                               maybeSection: Option[String],
                               tagType:Option[String],
@@ -249,7 +259,7 @@ class ElasticsearchRepo(endpoint:ElasticNodeEndpoint, val defaultPageSize:Int=20
     Edge.decodeCursor(cursor) match {
       case Right(maybeCursor)=>
         client.execute {
-          buildTagQuery(maybeTagId, maybeSection, tagType, maybeQuery, maybeCategory, maybeReferences)
+          buildTagQuery(maybeTagId, maybeSection, tagType, maybeQuery, maybeFuzziness, maybeCategory, maybeReferences)
             .sortBy(sortParam)
             .limit(pageSize)
             .searchAfter(maybeCursor)
@@ -281,7 +291,7 @@ class ElasticsearchRepo(endpoint:ElasticNodeEndpoint, val defaultPageSize:Int=20
     val tagIdMatches = tagIdList.map(MatchQuery("id", _))
 
     val response = client.execute {
-      val restrictions = tagQueryParams(None, maybeSection, tagType, maybeCategory, maybeReferences)
+      val restrictions = tagQueryParams(None, maybeSection, tagType, maybeCategory, maybeReferences, None, None)
 
       if(restrictions.nonEmpty) {
         search("tag").query(