From d0f449549d92b952df04abee318c52592adadae1 Mon Sep 17 00:00:00 2001
From: Defozo <defozo@gmail.com>
Date: Sat, 6 Apr 2019 23:15:03 +0200
Subject: [PATCH] Added LanguageDetector.js module with
 getAverageConfidenceScoresOfDetectedLanguages(messages) function.

---
 package.json                    |  1 +
 src/modules/LanguageDetector.js | 30 ++++++++++++++++++++++++++++++
 2 files changed, 31 insertions(+)
 create mode 100644 src/modules/LanguageDetector.js

diff --git a/package.json b/package.json
index e6cb1f0..bd03d19 100644
--- a/package.json
+++ b/package.json
@@ -11,6 +11,7 @@
     "core-js": "^2.6.5",
     "iconv-lite": "^0.4.24",
     "jszip": "^3.2.1",
+    "languagedetect": "^1.2.0",
     "sentiment-polish": "^1.0.0",
     "vue": "^2.6.6",
     "vue-router": "^3.0.1",
diff --git a/src/modules/LanguageDetector.js b/src/modules/LanguageDetector.js
new file mode 100644
index 0000000..d8d4355
--- /dev/null
+++ b/src/modules/LanguageDetector.js
@@ -0,0 +1,30 @@
+/**
+ * @author Michał Kiełtyka
+ * @type {module:languagedetect}
+ */
+
+const LanguageDetect = require('languagedetect');
+const lngDetector = new LanguageDetect();
+
+/**
+ * Takes every message into LanguageDetect().detect(message) which returns languages with confidence scores.
+ * Then, sums confidences and divide them by number of messages.
+ *
+ * Note: Not very reliable way to detect languages but may give some better results when tested on bigger dataset.
+ *
+ * @param {string[]} messages Messages to analyze.
+ * @returns {Object} Object with keys as languages and values as averaged confidence scores (between messages).
+ */
+export function getAverageConfidenceScoresOfDetectedLanguages(messages) {
+    let languagesWithConfidencesMap = {};
+    for (let i = 0; i < messages.length; i++) {
+        let languagesWithConfidencesArray = lngDetector.detect(messages[i]);
+        languagesWithConfidencesArray.forEach(function(entry) {
+            languagesWithConfidencesMap[entry[0]] = (languagesWithConfidencesMap[entry[0]] || 0) + entry[1];
+        });
+    }
+    Object.keys(languagesWithConfidencesMap).forEach(function (key) {
+        languagesWithConfidencesMap[key] = languagesWithConfidencesMap[key] / messages.length;
+    });
+    return languagesWithConfidencesMap;
+}