Merge pull request #111 from studentinovisad/as/fix/search-relevance

fix(search): include identifier (index) in order by relevance
studentinovisad · Dec 29, 2024 · af9046f · af9046f
2 parents f77081b + 26b8ceb
commit af9046f
Show file tree

Hide file tree

Showing 3 changed files with 59 additions and 12 deletions.
diff --git a/src/lib/server/db/employee.ts b/src/lib/server/db/employee.ts
@@ -59,8 +59,9 @@ export async function getEmployees(
 							entryBuilding: employeeEntry.building,
 							exitTimestamp: max(employeeExit.timestamp),
 							leastDistance: sqlLeast([
-								sqlLevenshteinDistance(sqlConcat([employee.fname], ' '), nonEmptySearchQuery),
-								sqlLevenshteinDistance(sqlConcat([employee.lname], ' '), nonEmptySearchQuery),
+								sqlLevenshteinDistance(sqlConcat([employee.identifier]), nonEmptySearchQuery),
+								sqlLevenshteinDistance(sqlConcat([employee.fname]), nonEmptySearchQuery),
+								sqlLevenshteinDistance(sqlConcat([employee.lname]), nonEmptySearchQuery),
 								sqlLevenshteinDistance(
 									sqlConcat([employee.fname, employee.lname], ' '),
 									nonEmptySearchQuery
@@ -69,7 +70,11 @@ export async function getEmployees(
 									sqlConcat([employee.lname, employee.fname], ' '),
 									nonEmptySearchQuery
 								)
-							]).as('least_distance')
+							]).as('least_distance'),
+							leastDistanceIdentifier: sqlLevenshteinDistance(
+								sqlConcat([employee.identifier]),
+								nonEmptySearchQuery
+							).as('least_distance_identifier')
 						})
 						.from(employee)
 						.leftJoin(maxEntrySubquery, eq(maxEntrySubquery.employeeId, employee.id))
@@ -107,7 +112,11 @@ export async function getEmployees(
 							maxEntrySubquery.maxEntryTimestamp,
 							employeeEntry.building
 						)
-						.orderBy(({ leastDistance, identifier }) => [leastDistance, identifier])
+						.orderBy(({ leastDistance, leastDistanceIdentifier, identifier }) => [
+							leastDistance,
+							leastDistanceIdentifier,
+							identifier
+						])
 						.limit(limit)
 						.offset(offset)
 				: await db

diff --git a/src/lib/server/db/fuzzysearch.ts b/src/lib/server/db/fuzzysearch.ts
@@ -5,6 +5,18 @@ type FuzzySearchFiltersOptions = {
 	substr?: boolean;
 };
 
+type LevenshteinOptions = {
+	insertCost: number;
+	deleteCost: number;
+	substitutionCost: number;
+};
+
+const defaultLevenshteinOptions: LevenshteinOptions = {
+	insertCost: 1,
+	deleteCost: 3,
+	substitutionCost: 2
+};
+
 export function fuzzySearchFilters(
 	dbFields: Column[],
 	searchQuery: string,
@@ -45,6 +57,14 @@ export function fuzzySearchFilters(
  * Returns the sql for concatenating multiple columns with a separator using CONCAT_WS
  */
 export function sqlConcat(cols: Column[], separator?: string): SQL<Column> {
+	if (cols.length === 0) {
+		throw new Error('Passed columns length is 0');
+	}
+
+	if (cols.length === 1) {
+		return sql<Column>`${cols[0]}`;
+	}
+
 	const sqlCols = cols
 		.map((col) => sql<Column>`${col}`)
 		.reduce((prev, curr) => sql`${prev}, ${curr}`);
@@ -64,13 +84,22 @@ export function sqlLeast(cols: SQL<number>[]): SQL<number> {
 /*
  * Returns the sql for determining if the levenshtein distance is less than or equal to the passed distance
  */
-export function sqlLevenshtein(col: SQL<Column>, input: string, distance: number): SQL<boolean> {
-	return sql<boolean>`LEVENSHTEIN(LOWER(${col}), LOWER(${input})) <= ${distance}`;
+export function sqlLevenshtein(
+	col: SQL<Column>,
+	input: string,
+	distance: number,
+	opts: LevenshteinOptions = defaultLevenshteinOptions
+): SQL<boolean> {
+	return sql<boolean>`LEVENSHTEIN(LOWER(${input}), LOWER(${col}), ${opts.insertCost}, ${opts.deleteCost}, ${opts.substitutionCost}) <= ${distance}`;
 }
 
 /*
  * Returns the sql for getting the levenshtein distance
  */
-export function sqlLevenshteinDistance(col: SQL<Column>, input: string): SQL<number> {
-	return sql<number>`LEVENSHTEIN(LOWER(${col}), LOWER(${input}))`;
+export function sqlLevenshteinDistance(
+	col: SQL<Column>,
+	input: string,
+	opts: LevenshteinOptions = defaultLevenshteinOptions
+): SQL<number> {
+	return sql<number>`LEVENSHTEIN(LOWER(${input}), LOWER(${col}), ${opts.insertCost}, ${opts.deleteCost}, ${opts.substitutionCost})`;
 }
diff --git a/src/lib/server/db/student.ts b/src/lib/server/db/student.ts
@@ -59,8 +59,9 @@ export async function getStudents(
 							entryBuilding: studentEntry.building,
 							exitTimestamp: max(studentExit.timestamp),
 							leastDistance: sqlLeast([
-								sqlLevenshteinDistance(sqlConcat([student.fname], ' '), nonEmptySearchQuery),
-								sqlLevenshteinDistance(sqlConcat([student.lname], ' '), nonEmptySearchQuery),
+								sqlLevenshteinDistance(sqlConcat([student.index]), nonEmptySearchQuery),
+								sqlLevenshteinDistance(sqlConcat([student.fname]), nonEmptySearchQuery),
+								sqlLevenshteinDistance(sqlConcat([student.lname]), nonEmptySearchQuery),
 								sqlLevenshteinDistance(
 									sqlConcat([student.fname, student.lname], ' '),
 									nonEmptySearchQuery
@@ -69,7 +70,11 @@ export async function getStudents(
 									sqlConcat([student.lname, student.fname], ' '),
 									nonEmptySearchQuery
 								)
-							]).as('least_distance')
+							]).as('least_distance'),
+							leastDistanceIdentifier: sqlLevenshteinDistance(
+								sqlConcat([student.index]),
+								nonEmptySearchQuery
+							).as('least_distance_identifier')
 						})
 						.from(student)
 						.leftJoin(maxEntrySubquery, eq(maxEntrySubquery.studentId, student.id))
@@ -105,7 +110,11 @@ export async function getStudents(
 							maxEntrySubquery.maxEntryTimestamp,
 							studentEntry.building
 						)
-						.orderBy(({ leastDistance, index }) => [leastDistance, index])
+						.orderBy(({ leastDistance, leastDistanceIdentifier, index }) => [
+							leastDistance,
+							leastDistanceIdentifier,
+							index
+						])
 						.limit(limit)
 						.offset(offset)
 				: await db