@hdelva/termennetwerk_client - npm Package Compare versions

Comparing version 1.0.7 to 2.0.0

lib/index.js

@@ -7,4 +7,22 @@ "use strict";
 exports.default = void 0;
-var AutoComplete_1 = require("./AutoComplete");
-Object.defineProperty(exports, "default", { enumerable: true, get: function () { return __importDefault(AutoComplete_1).default; } });
+const AutoComplete_1 = __importDefault(require("./AutoComplete"));
+var AutoComplete_2 = require("./AutoComplete");
+Object.defineProperty(exports, "default", { enumerable: true, get: function () { return __importDefault(AutoComplete_2).default; } });
+const client = new AutoComplete_1.default([
+    "https://termen.opoi.org/nta",
+    "https://termen.opoi.org/vtmk",
+    "https://termen.opoi.org/cht",
+    "https://termen.opoi.org/rkdartists"
+], 10);
+client.on("data", (data) => {
+    //const results = [];
+    console.log(data.object.value, client.resolveSubject(data.subject.value).length);
+    for (const quad of client.resolveSubject(data.subject.value)) {
+        //console.log(quad.object.value);
+    }
+});
+client.on("reset", (data) => {
+    console.log("");
+});
+client.query("anne frank");
 //# sourceMappingURL=index.js.map

package.json

 {
   "name": "@hdelva/termennetwerk_client",
-  "version": "1.0.7",
-  "description": "Autocomplete over TREEs",
+  "version": "2.0.0",
+  "description": "Proof of concept of client-side autocompletion",
   "main": "lib/index.js",
@@ -9,3 +9,3 @@ "author": "Harm Delva",
   "dependencies": {
-    "@types/node": "^12.12.14",
+    "@types/node": "^12.19.3",
     "@types/typescript": "^2.0.0",
@@ -20,3 +20,3 @@ "collections": "^5.1.12",
   "devDependencies": {
-    "ts-loader": "^8.0.5"
+    "ts-loader": "^8.0.7"
   },
@@ -23,0 +23,0 @@ "directories": {

src/index.ts

@@ -1,1 +0,39 @@
-export { default } from "./AutoComplete";
+import FuzzyAutoComplete from "./examples/FuzzyAutoComplete";
+import StrictAutoComplete from "./examples/StrictAutoComplete";
+import QueryAgent from "./QueryAgent";
+import QueryAggregator from "./QueryAggregator";
+import ResultEmitter from "./ResultEmitter";
+import ResultRanking from "./ResultRanking";
+import ResultStore from "./ResultStore";
+import ResultUniqueFilter from "./ResultUniqueFilter";
+import asymmetricDiceCoefficient from "./similarity/asymmetricDiceCoefficient";
+import commonPrefixSimilarity from "./similarity/commonPrefix";
+import fuzzyIndexSimilarity from "./similarity/fuzzyIndex";
+import fuzzyPrefixSimilarity from "./similarity/fuzzyPrefix";
+import SimilarityConfiguration from "./similarity/SimilarityConfiguration";
+import strictPrefixSimilarity from "./similarity/strictPrefix";
+import tokenwiseCompare from "./similarity/tokenwise";
+
+export default {
+    examples: {
+        StrictAutoComplete,
+        FuzzyAutoComplete,
+    },
+    components: {
+        QueryAgent,
+        QueryAggregator,
+        ResultEmitter,
+        ResultRanking,
+        ResultStore,
+        ResultUniqueFilter,
+        SimilarityConfiguration,
+    },
+    similarityFunctions: {
+        asymmetricDiceCoefficient,
+        commonPrefixSimilarity,
+        fuzzyIndexSimilarity,
+        fuzzyPrefixSimilarity,
+        strictPrefixSimilarity,
+        tokenwiseCompare,
+    }
+};

src/normalizers/NFKD.ts

@@ -7,3 +7,7 @@ import INormalizer from "./INormalizer";
     constructor() {
-        this.regex = /[^\p{L}\p{N}\s]/gu;
+        // see https://javascript.info/regexp-unicode
+        // \p{L} is all unicode letters
+        // \p{N} is all numbers
+        // \p{Z} is all separators (e.g., whitespace)
+        this.regex = /[^\p{L}\p{N}\p{Z}]/gu;
     }
@@ -10,0 +14,0 @@

src/QueryAgent.ts

@@ -5,4 +5,4 @@ import LDFetch from "ldfetch";
 
-import IQueryEmitter from "./IQueryEmitter";
-import { SimilarityConfiguration } from "./similarity/SimilarityConfiguration";
+import ResultEmitter from "./ResultEmitter";
+import SimilarityConfiguration from "./similarity/SimilarityConfiguration";
 
@@ -35,8 +35,11 @@ class RankedRelation {
 
-export default class QueryAgent extends IQueryEmitter {
-    protected source: string;
-    protected fetcher: LDFetch;
-    protected activeQueries: Set<string>;
-    protected similarityConfigurations: SimilarityConfiguration[];
-    protected knownRelations: Map<string, string>; // URI -> value
+/*
+ * Traverses a single data source for the requested query string
+ */
+export default class QueryAgent extends ResultEmitter {
+    protected source: string; // access URI of the data source
+    protected fetcher: LDFetch; // object that fetches and parses the RDF for us
+    protected activeQueries: Set<string>; // link traversal is async, we may want to terminate query early
+    protected similarityConfigurations: SimilarityConfiguration[]; // functions used to prioritize discovered relations
+    protected knownRelations: Map<string, string>; // URI -> tree value
 
@@ -49,8 +52,11 @@ constructor(source: string, similarityConfigurations: SimilarityConfiguration[]) {
         this.knownRelations = new Map();
-        this.similarityConfigurations = similarityConfigurations; // || strictPrefixSimilarity;
+        this.similarityConfigurations = similarityConfigurations;
 
+        // todo, maybe make this optional
         this.prefetch();
     }
 
-    private async prefetch() {
+    // fetch the root node, and memorize all discovered relations
+    // useful if the root node is particularly large
+    public async prefetch() {
         const data = await this.fetcher.get(this.source);
@@ -95,3 +101,7 @@
 
+        // kickstart the link traversal
+        // start from the already discovered nodes
         for (const [uri, value] of this.knownRelations.entries()) {
+            // iteratively build the similarity vector
+            // stop as soon as this relations is certainly useless
             const similarityScores: number[] = [];
@@ -101,2 +111,3 @@ for (const conf of this.similarityConfigurations) {
                 if (!isNaN(similarity)) {
+                    // NaN similarities are not worth following
                     similarityScores.push(similarity);
@@ -109,2 +120,3 @@ } else {
             if (similarityScores.length === this.similarityConfigurations.length) {
+                // the entire vector was evaluated, so it's worth following
                 queue.push(new RankedRelation(uri, similarityScores));
@@ -118,2 +130,3 @@ }
             if (!this.activeQueries.has(input)) {
+                // we're no longer waiting for this query's results
                 break;
@@ -129,2 +142,3 @@ }
             if (visited.has(page)) {
+                // avoid hitting the same page multiple times
                 continue;
@@ -136,4 +150,4 @@ }
 
-            const nodes = {};
-            const nodeValues = {};
+            const nodes = {}; // URI -> URI
+            const nodeValues = {}; // URI -> list of tree values
 
@@ -146,3 +160,3 @@ for (const untyped_quad of data.triples) {
                 } else if (quad.predicate.value == "https://w3id.org/tree#value") {
-                    // be prepared for multiple values
+                    // a relation may contain several tree values
                     if (!nodeValues[quad.subject.value]) {
@@ -157,7 +171,10 @@ nodeValues[quad.subject.value] = [];
 
+            // this page has been processed; schedule the next useful pages
             for (const [key, node] of Object.entries(nodes)) {
                 if (!visited.has(node)) {
                     const value = nodeValues[key];
-                    this.knownRelations.set(node as string, value);
+                    this.knownRelations.set(node as string, value); // memorize this relation
                     if (value) {
+                        // iteratively build the similarity vector
+                        // stop as soon as this relations is certainly useless
                         const similarityScores: number[] = [];
@@ -167,2 +184,3 @@ for (const conf of this.similarityConfigurations) {
                             if (!isNaN(similarity)) {
+                                // NaN similarities are not worth following
                                 similarityScores.push(similarity);
@@ -175,2 +193,3 @@ } else {
                         if (similarityScores.length === this.similarityConfigurations.length) {
+                            // the entire vector was evaluated, so it's worth following
                             queue.push(new RankedRelation(node, similarityScores));
@@ -183,4 +202,4 @@ }
 
+        // signal that we're done following links
         this.activeQueries.delete(input);
-
         this.emit("end", input);
@@ -187,0 +206,0 @@ }

src/QueryAggregator.ts

@@ -1,18 +0,20 @@
-import IQueryEmitter from "./IQueryEmitter";
+import ResultEmitter from "./ResultEmitter";
 import { Quad } from "rdf-js";
 
-export default class QueryAggregator extends IQueryEmitter {
-    protected singleSourceAgents: IQueryEmitter[];
+/*
+ * Merges results from multiple other emitters
+ */
+export default class QueryAggregator extends ResultEmitter {
+    protected subEmitters: ResultEmitter[];
     protected finished: Map<string, number>;
 
-    constructor(sourceAgents: IQueryEmitter[]) {
+    constructor(subEmitters: ResultEmitter[]) {
         super();
-        this.singleSourceAgents = sourceAgents;
+        this.subEmitters = subEmitters;
         this.finished = new Map();
 
         const self = this;
-        for (const source of this.singleSourceAgents) {
+        for (const source of this.subEmitters) {
             source.on("data", (q) => self.emit("data", q));
             source.on("end", (q) => self.processEnd(q));
-            //source.on("reset", () => self.emit("reset"));
         }
@@ -23,3 +25,3 @@ }
         this.emit("reset");
-        for (const source of this.singleSourceAgents) {
+        for (const source of this.subEmitters) {
             source.query(input);
@@ -31,3 +33,3 @@ }
         let result: Quad[] = [];
-        for (const source of this.singleSourceAgents) {
+        for (const source of this.subEmitters) {
             result = result.concat(source.resolveSubject(uri));
@@ -42,3 +44,3 @@ }
 
-        if (count === this.singleSourceAgents.length) {
+        if (count === this.subEmitters.length) {
             this.finished.delete(query);
@@ -45,0 +47,0 @@ this.emit("end", query);

src/ResultRanking.ts

@@ -5,8 +5,13 @@ const SortedArray = require("collections/sorted-array");
 
-import IQueryEmitter from "./IQueryEmitter";
+import ResultEmitter from "./ResultEmitter";
 import INormalizer from "./normalizers/INormalizer";
-import { SimilarityConfiguration } from "./similarity/SimilarityConfiguration";
+import SimilarityConfiguration from "./similarity/SimilarityConfiguration";
 
-export default class ResultRanking extends IQueryEmitter {
-    protected subEmitter: IQueryEmitter;
+/*
+ * Emits only the most relevant results from the subEmitter
+ * Emits "reset" events each time the top N results changes, 
+ * followed by "data" events for each result in the top N 
+ */
+export default class ResultRanking extends ResultEmitter {
+    protected subEmitter: ResultEmitter;
     protected activeQuery: string;
@@ -21,3 +26,3 @@ protected size: number;
         size: number,
-        subEmitter: IQueryEmitter,
+        subEmitter: ResultEmitter,
         normalizer: INormalizer,
@@ -24,0 +29,0 @@ similarityConfigurations: SimilarityConfiguration[],

src/ResultStore.ts

 import { Quad } from "rdf-js";
 
-import IQueryEmitter from "./IQueryEmitter";
+import ResultEmitter from "./ResultEmitter";
 import N3 = require('n3');
 
-export default class ResultStore extends IQueryEmitter {
-    protected subEmitter: IQueryEmitter;
+/*
+ * Intercepts all data from the subEmitter, and only reemits the literal quads
+ * Clients can then request everything we know about a given subject using the resolveSubject method
+ */ 
+export default class ResultStore extends ResultEmitter {
+    protected subEmitter: ResultEmitter;
     protected store: N3.Store;
 
-    constructor(sourceAgents: IQueryEmitter) {
+    constructor(sourceAgents: ResultEmitter) {
         super();
@@ -12,0 +16,0 @@ this.subEmitter = sourceAgents;

src/ResultUniqueFilter.ts

 import { Quad } from "rdf-js";
 
-import IQueryEmitter from "./IQueryEmitter";
+import ResultEmitter from "./ResultEmitter";
 
-export default class ResultUniqueFilter extends IQueryEmitter {
-    protected subEmitter: IQueryEmitter;
+/*
+ * The same value can be in multiple pages across multiple datasets
+ * Avoid duplicate results by just slapping a unique filter in there
+ */
+export default class ResultUniqueFilter extends ResultEmitter {
+    protected subEmitter: ResultEmitter;
     protected known: Set<string>;
 
-    constructor(sourceAgents: IQueryEmitter) {
+    constructor(sourceAgents: ResultEmitter) {
         super();
@@ -11,0 +15,0 @@ this.subEmitter = sourceAgents;

src/similarity/asymmetricDiceCoefficient.ts

@@ -0,1 +1,3 @@
+// like the dice coefficient, but only looks for bigrams out of `expected` in `found`
+// this doesn't penalize longer results if we know the `expected` string is just incomplete
 export default function asymmetricDiceCoefficient(expected: string, found: string) {
@@ -2,0 +4,0 @@ let firstBigrams = new Map();

src/similarity/commonPrefix.ts

 /*
  * Length of the common prefix
  */
-export function commonPrefixSimilarity(expected: string, found: string) {
+export default function commonPrefixSimilarity(expected: string, found: string) {
     const minLength = Math.min(expected.length, found.length);
@@ -6,0 +6,0 @@

src/similarity/fuzzyPrefix.ts

@@ -1,2 +0,2 @@
-import { commonPrefixSimilarity } from "./commonPrefix";
+import commonPrefixSimilarity from "./commonPrefix";
 import fuzzyIndexSimilarity from "./fuzzyIndex";
@@ -7,4 +7,4 @@
  */
-export function fuzzyPrefixSimilarity(expected: string, found: string) {
+export default function fuzzyPrefixSimilarity(expected: string, found: string) {
     return commonPrefixSimilarity(expected, found) + fuzzyIndexSimilarity(expected, found);
 }

src/similarity/SimilarityConfiguration.ts

 import { FilterFunction, SimilarityFunction } from "./SimilarityFunction";
 
-export class SimilarityConfiguration {
+export default class SimilarityConfiguration {
     private similarityFunction: SimilarityFunction;
@@ -12,3 +12,5 @@ private filterFunction?: FilterFunction;
 
-    public evaluate(expected: string, found: string) {
+    public evaluate(expected: string, found: string): number {
+        // higher is better
+        // NaN indicates an ineligible result
         const similarity = this.similarityFunction(expected, found);
@@ -15,0 +17,0 @@

src/similarity/SimilarityFunction.ts

@@ -1,3 +0,5 @@
-// semantics of return is bigger = better
+// bigger is better, NaN is ineligible
 export type SimilarityFunction = (expected: string, found: string) => number;
+
+// return true if this similarity is good enough, false if not
 export type FilterFunction = (expected: string, found: string, similarity: number) => boolean;

No alert changes

Improved metrics

Total package byte prevSize

8069956

increased by0.05%

Lines of code

7999

increased by1.69%

Worsened metrics

Number of package files

81

decreased by-3.57%

Dependency changes

• Updated@types/node@^12.19.3