@sc-voice/tools 3.0.0 → 3.2.0

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@sc-voice/tools",
-  "version": "3.0.0",
+  "version": "3.2.0",
   "description": "Utilities for SC-Voice",
   "main": "index.mjs",
   "files": [

package/src/defines.mjs

@@ -9,6 +9,6 @@ export const DBG = {
   DEEPL_TEST_API: 0, // test with live DeepL API ($$$)
   DEEPL_XLT: 0, // test live translation
   L7C_FETCH_LEGACY: 0,
-  L7C_FETCH_LEGACY_SC: 0, // ignore test cache and use SC 
+  L7C_FETCH_LEGACY_SC: 0, // ignore test cache and use SC
   WORD_MAP_TRANFORMER: 0,
 };

package/src/text/legacy-doc.mjs

@@ -37,7 +37,7 @@ export class LegacyDoc {
     return true;
   }
 
-  static legacyUrl(opts={}) {
+  static legacyUrl(opts = {}) {
     let {
       endPoint = 'https://suttacentral.net/api/suttas',
       sutta_uid,
@@ -51,17 +51,14 @@ export class LegacyDoc {
   static async fetchLegacy(opts = {}) {
     const msg = 'L7c.fetchLegacy:';
     const dbg = DBG.L7C_FETCH_LEGACY;
-    let {
-      maxBuffer = 10 * 1024 * 1024,
-      cache,
-    } = opts;
+    let { maxBuffer = 10 * 1024 * 1024, cache } = opts;
     let url = LegacyDoc.legacyUrl(opts);
     let res;
     if (cache) {
       res = cache(url);
       dbg && console.log(msg, '[1]cached', res.ok);
     } else {
-      res = await fetch(url, {maxBuffer});
+      res = await fetch(url, { maxBuffer });
       dbg && console.log(msg, '[2]scapi', res.ok);
     }
     if (!res.ok) {

package/src/text/tfidf-space.mjs

@@ -52,7 +52,7 @@ export class TfidfSpace {
     return s.replace(/<[^>]*>/gi, '');
   }
 
-  static removeNonWords(s, opts={}) {
+  static removeNonWords(s, opts = {}) {
     const RE_RESERVED = /[_-]/g; // allowed in bow words
     const RE_LQUOTE = /[“‘«]/g;
     const RE_PUNCT = /[.,:;$"'“”‘’!?«»\[\]]/g;
@@ -67,11 +67,11 @@ export class TfidfSpace {
       .trim();
   }
 
-  static normalizeEN(s, opts={}) {
+  static normalizeEN(s, opts = {}) {
     return TfidfSpace.removeNonWords(s.toLowerCase(), opts);
   }
 
-  static normalizeFR(s, opts={}) {
+  static normalizeFR(s, opts = {}) {
     let sAbbr = s
       .toLowerCase()
       .replace(/\bd[’']/gi, 'de ')

package/src/text/word-vector.mjs

@@ -49,9 +49,10 @@ export class WordVector extends Object {
     let sv = entries.reduce((a, e) => {
       let [k, v] = e;
       if (minValue <= v) {
-        let vf = v.toFixed(precision)
+        let vf = v
+          .toFixed(precision)
           .replace(/\.0*$/, '')
-          .replace(/0\./,'.');
+          .replace(/0\./, '.');
         a.push(`${k}:${vf}`);
       }
       return a;
@@ -59,7 +60,8 @@ export class WordVector extends Object {
     return sv.join(',');
   }
 
-  norm() { // L2 norm
+  norm() {
+    // L2 norm
     let keys = Object.keys(this);
     if (keys.length === 0) {
       return 0;
@@ -74,9 +76,10 @@ export class WordVector extends Object {
   add(vec2) {
     let keys = Object.keys(vec2);
     return keys.reduce((a, k) => {
+      let v1 = a[k] || 0;
       let v2 = vec2[k];
       if (v2) {
-        a[k] = (a[k] || 0) + v2;
+        a[k] = v1 + v2;
       }
       return a;
     }, new WordVector(this));
@@ -85,14 +88,28 @@ export class WordVector extends Object {
   increment(vec2) {
     let keys = Object.keys(vec2);
     return keys.reduce((a, k) => {
+      let v1 = a[k] || 0;
       let v2 = vec2[k];
       if (v2) {
-        a[k] = (a[k] || 0) + v2;
+        a[k] = v1 + v2;
       }
       return a;
     }, this);
   }
 
+  multiply(vec2) {
+    const msg = 'w8r.multiply:';
+    let keys = Object.keys(vec2);
+    return keys.reduce((a, k) => {
+      let v1 = this[k];
+      let v2 = vec2[k];
+      if (v1 && v2) {
+        a[k] = v1 * v2;
+      }
+      return a;
+    }, new WordVector({}));
+  }
+
   dot(vec2) {
     const msg = 'w8r.dot:';
     if (vec2 == null) {
@@ -114,7 +131,7 @@ export class WordVector extends Object {
     }, this);
   }
 
-  hadamardL1(vec2 = {}) { 
+  hadamardL1(vec2 = {}) {
     // L1-norm of Hadamard product shows how
     // the cosine similarity score is apportioned
     let keys = Object.keys(this);
@@ -134,7 +151,7 @@ export class WordVector extends Object {
       return hadamard; // empty vector
     }
     let n12 = this.norm() * vec2.norm();
-    return hadamard.scale(1/n12);
+    return hadamard.scale(1 / n12);
   }
 
   similar(vec2) {
@@ -157,4 +174,28 @@ export class WordVector extends Object {
       return a;
     }, new WordVector());
   }
+
+  andOneHot(vec2) {
+    return Object.keys(this).reduce((a, k) => {
+      if (this[k] && vec2[k]) {
+        a[k] = 1;
+      }
+      return a;
+    }, new WordVector());
+  }
+
+  orOneHot(vec2) {
+    let result = Object.keys(this).reduce((a, k) => {
+      if (this[k] || vec2[k]) {
+        a[k] = 1;
+      }
+      return a;
+    }, new WordVector());
+    return Object.keys(vec2).reduce((a, k) => {
+      if (this[k] || vec2[k]) {
+        a[k] = 1;
+      }
+      return a;
+    }, result);
+  }
 } // WordVector