karelklima · karelklima · Dec 19, 2023 · Dec 17, 2023 · Dec 18, 2023 · Dec 19, 2023
diff --git a/docs/table-of-contents.json b/docs/table-of-contents.json
@@ -23,6 +23,7 @@
   "v2": {
     "title": "Version 2 (not yet released)",
     "pages": [
+      ["filtering", "Filtering"],
       ["pagination", "Pagination"],
       ["working-with-arrays", "Working with arrays"],
       ["inverse-properties", "Inverse properties"]

diff --git a/docs/v2/filtering.md b/docs/v2/filtering.md
@@ -0,0 +1,88 @@
+# Filtering
+
+LDkit comes with powerful search and filtering capabilities, enabling users to
+narrow data results and explore large datasets. The feature integrates
+seamlessly with the existing [Lens](../components/lens) `find` method, allowing
+for controlled data retrieval.
+
+LDkit allows various search and filtering operations like `$equals`, `$not`,
+`$contains`, `$strStarts`, `$strEnds`, `$gt`, `$lt`, `$gte`, `$lte`, `$regex`,
+`$langMatches`, and `$filter`. Each is illustrated below with examples.
+
+### Simple example
+
+```ts
+import { createLens } from "ldkit";
+import { schema, xsd } from "ldkit/namespaces";
+
+// Create a schema
+const PersonSchema = {
+  "@type": schema.Person,
+  name: schema.name,
+  birthDate: {
+    "@id": schema.birthDate,
+    "@type": xsd.date,
+  },
+} as const;
+
+// Create a lens instance
+const Persons = createLens(PersonSchema);
+
+await Persons.find({
+  where: {
+    name: "Ada Lovelace",
+  },
+}); // Returns list of all persons named "Ada Lovelace"
+```
+
+### Comparison operators
+
+```typescript
+await Persons.find({
+  where: {
+    name: {
+      $equals: "Ada Lovelace", // FILTER (?value = "Ada Lovelace")
+      $not: "Alan Turing", // FILTER (?value != "Alan Turing")
+    },
+    birthDate: {
+      $lt: new Date("01-01-1900"), // FILTER (?value < "01-01-1900"@xsd:date)
+      $lte: new Date("01-01-1900"), // FILTER (?value <= "01-01-1900"@xsd:date)
+      $gt: new Date("01-01-1900"), // FILTER (?value > "01-01-1900"@xsd:date)
+      $gte: new Date("01-01-1900"), // FILTER (?value >= "01-01-1900"@xsd:date)
+    },
+  },
+});
+```
+
+### String functions
+
+```typescript
+await Persons.find({
+  where: {
+    name: {
+      $contains: "Ada", // FILTER CONTAINS(?value, "Ada")
+      $strStarts: "Ada", // FILTER STRSTARTS(?value, "Ada")
+      $strEnds: "Lovelace", // FILTER STRENDS(?value, "Lovelace")
+      $langMatches: "fr", // FILTER LANGMATCHES(LANG(?value), "fr")
+      $regex: "^A(.*)e$", // FILTER REGEX(?value, "^A(.*)e$")
+    },
+  },
+});
+```
+
+### Custom filtering
+
+On top of the above, it is possible to specify a custom filter function using
+the SPARQL filtering syntax. There is a special placeholder `?value` used for
+the value of the property to be filtered. This placeholder is replaced by actual
+variable name during runtime.
+
+```typescript
+await Persons.find({
+  where: {
+    name: {
+      $filter: "STRLEN(?value) > 10", // FILTER (STRLEN(?value) > 10)
+    },
+  },
+});
+```
diff --git a/library/engine/query_engine_proxy.ts b/library/engine/query_engine_proxy.ts
@@ -1,4 +1,10 @@
-import { type Context, IQueryEngine, quadsToGraph, type RDF } from "../rdf.ts";
+import {
+  type Context,
+  IQueryEngine,
+  N3,
+  quadsToGraph,
+  type RDF,
+} from "../rdf.ts";
 import { resolveContext, resolveEngine } from "../global.ts";
 import { type AsyncIterator } from "../asynciterator.ts";
 
@@ -29,7 +35,8 @@ export class QueryEngineProxy {
       this.context,
     ) as unknown as AsyncIterator<RDF.Quad>;
     const quads = await (quadStream.toArray());
-    return quadsToGraph(quads);
+    const store = new N3.Store(quads);
+    return quadsToGraph(store);
   }
 
   queryVoid(query: string) {

diff --git a/library/lens/lens.ts b/library/lens/lens.ts
@@ -6,6 +6,7 @@ import {
   type SchemaInterface,
   type SchemaInterfaceIdentity,
   type SchemaPrototype,
+  type SchemaSearchInterface,
   type SchemaUpdateInterface,
 } from "../schema/mod.ts";
 import { decode } from "../decoder.ts";
@@ -44,6 +45,7 @@ export class Lens<
   S extends SchemaPrototype,
   I = SchemaInterface<S>,
   U = SchemaUpdateInterface<S>,
+  X = SchemaSearchInterface<S>,
 > {
   private readonly schema: Schema;
   private readonly context: Context;
@@ -72,19 +74,25 @@ export class Lens<
 
   async query(sparqlConstructQuery: string) {
     const graph = await this.engine.queryGraph(sparqlConstructQuery);
+    console.log("GRAPH", graph);
     return this.decode(graph);
   }
 
   async find(
-    options: { where?: string | RDF.Quad[]; take?: number; skip?: number } = {},
+    options: { where?: X | string | RDF.Quad[]; take?: number; skip?: number } =
+      {},
   ) {
     const { where, take, skip } = {
       take: 1000,
       skip: 0,
       ...options,
     };
-    const q = this.queryBuilder.getQuery(where, take, skip);
-    // TODO: console.log(q);
+    const isRegularQuery = typeof where === "string" ||
+      typeof where === "undefined" || Array.isArray(where);
+
+    const q = isRegularQuery
+      ? this.queryBuilder.getQuery(where, take, skip)
+      : this.queryBuilder.getSearchQuery(where ?? {}, take, skip);
     const graph = await this.engine.queryGraph(q);
     return this.decode(graph);
   }

diff --git a/library/lens/query_builder.ts b/library/lens/query_builder.ts
@@ -1,11 +1,16 @@
-import { type Schema } from "../schema/mod.ts";
-import { getSchemaProperties } from "../schema/mod.ts";
+import {
+  getSchemaProperties,
+  type Property,
+  type Schema,
+  type SearchSchema,
+} from "../schema/mod.ts";
 import {
   CONSTRUCT,
   DELETE,
   INSERT,
   SELECT,
   sparql as $,
+  type SparqlValue,
 } from "../sparql/mod.ts";
 import { type Context, DataFactory, type Iri, type RDF } from "../rdf.ts";
 import ldkit from "../namespaces/ldkit.ts";
@@ -15,6 +20,7 @@ import { encode } from "../encoder.ts";
 
 import { type Entity } from "./types.ts";
 import { UpdateHelper } from "./update_helpe
1E0A
r.ts";
+import { SearchHelper } from "./search_helper.ts";
 
 enum Flags {
   None = 0,
@@ -52,16 +58,33 @@ export class QueryBuilder {
     return ([] as RDF.Quad[]).concat(...quadArrays);
   }
 
-  private getShape(flags: Flags) {
+  private getShape(flags: Flags, searchSchema?: SearchSchema) {
     const includeOptional = (flags & Flags.ExcludeOptional) === 0;
     const wrapOptional = (flags & Flags.UnwrapOptional) === 0;
     const omitRootTypes = (flags & Flags.IncludeTypes) === 0;
     const ignoreInverse = (flags & Flags.IgnoreInverse) === Flags.IgnoreInverse;
 
     const mainVar = "iri";
-    const conditions: (RDF.Quad | ReturnType<typeof $>)[] = [];
+    const conditions: SparqlValue[] = [];
+
+    const populateSearchConditions = (
+      property: Property,
+      varName: string,
+      search?: SearchSchema,
+    ) => {
+      if (search === undefined) {
+        return;
+      }
+      const helper = new SearchHelper(property, varName, search);
+      helper.process();
+      conditions.push(helper.sparqlValues);
+    };
 
-    const populateConditionsRecursive = (s: Schema, varPrefix: string) => {
+    const populateConditionsRecursive = (
+      s: Schema,
+      varPrefix: string,
+      search?: SearchSchema,
+    ) => {
       const rdfType = s["@type"];
       const properties = getSchemaProperties(s);
 
@@ -80,7 +103,8 @@ export class QueryBuilder {
       Object.keys(properties).forEach((prop, index) => {
         const property = properties[prop];
         const isOptional = property["@optional"];
-        if (!includeOptional && isOptional) {
+        const propertySchema = search?.[prop] as SearchSchema | undefined;
+        if (!includeOptional && isOptional && propertySchema === undefined) {
           return;
         }
         if (wrapOptional && isOptional) {
@@ -95,6 +119,11 @@ export class QueryBuilder {
               this.df.variable!(`${varPrefix}_${index}`),
             ),
           );
+          populateSearchConditions(
+            property,
+            `${varPrefix}_${index}`,
+            propertySchema,
+          );
         } else {
           conditions.push(
             this.df.quad(
@@ -108,6 +137,7 @@ export class QueryBuilder {
           populateConditionsRecursive(
             property["@context"] as Schema,
             `${varPrefix}_${index}`,
+            propertySchema,
           );
         }
         if (wrapOptional && isOptional) {
@@ -116,7 +146,7 @@ export class QueryBuilder {
       });
     };
 
-    populateConditionsRecursive(this.schema, mainVar);
+    populateConditionsRecursive(this.schema, mainVar, searchSchema);
     return conditions;
   }
 
@@ -146,6 +176,26 @@ export class QueryBuilder {
     return query;
   }
 
+  getSearchQuery(where: SearchSchema, limit: number, offset: number) {
+    const selectSubQuery = SELECT.DISTINCT`
+      ${this.df.variable!("iri")}
+    `.WHERE`
+      ${this.getShape(Flags.ExcludeOptional | Flags.IncludeTypes, where)} 
+    `.LIMIT(limit).OFFSET(offset).build();
+
+    const query = CONSTRUCT`
+      ${this.getResourceSignature()}
+      ${this.getShape(Flags.UnwrapOptional | Flags.IgnoreInverse)}
+    `.WHERE`
+      {
+        ${selectSubQuery}
+      }
+      ${this.getShape(Flags.None, where)}
+    `.build();
+
+    return query;
+  }
+
   getByIrisQuery(iris: Iri[]) {
     const query = CONSTRUCT`
       ${this.getResourceSignature()}