Limit overlapping segments to just one through weighted randomness

src/routes/getSkipSegmentsByHash.js

@@ -12,24 +12,64 @@ const getIP = require('../utils/getIP.js');
  * @property {string} UUID Unique identifier for the specific segment
  */
 
+/**
+ * @typedef {Object} Row
+ * @property {string} videoID
+ * @property {number} startTime
+ * @property {number} endTime
+ * @property {number} votes
+ * @property {string} UUID
+ * @property {string} category
+ * @property {number} shadowHidden
+ */
+
+/**
+ * Input an array of database records and get only one back, weighed on votes.
+ * The logic is taken from getWeightedRandomChoice, just simplified input and output to not work on indices only.
+ * 
+ * @param {Row[]} rows
+ * @returns {?Row}
+ */
+function pickWeightedRandomRow(rows) {
+    if (rows.length === 0) {
+        return null;
+    } else if (rows.length === 1) {
+        return rows[0];
+    }
+
+    const sqrtWeightsList = [];
+    let totalSqrtWeights = 0;
+    for (const row of rows) {
+        let sqrtVote = Math.sqrt((row.votes + 3) * 10);
+        sqrtWeightsList.push(sqrtVote);
+        totalSqrtWeights += sqrtVote;
+    }
+
+    const randomNumber = Math.random();
+    let currentVoteNumber = 0;
+    for (let i = 0; i < sqrtWeightsList.length; i++) {
+        if (randomNumber > currentVoteNumber / totalSqrtWeights && randomNumber < (currentVoteNumber + sqrtWeightsList[i]) / totalSqrtWeights) {
+            return rows[i];
+        }
+        currentVoteNumber += sqrtWeightsList[i];
+    }
+}
 /**
  * @param {string} prefix Lowercased hexadecimal hash prefix
  * @param {string} hashedIP Custom hash of the visitor’s IP address
- * @returns {Segment[]}
+ * @returns {Object.<string, Segment[]>}
  */
 function getSkipSegmentsByHash(prefix, hashedIP) {
-    /**
+    /** @type Row[] */
-     * @constant
+    const rows = db.prepare('SELECT videoID, startTime, endTime, votes, UUID, category, shadowHidden FROM sponsorTimes WHERE votes >= -1 AND hashedVideoID LIKE ? ORDER BY videoID, startTime')
-     * @type {Segment[]}
-     * @default
-     */
-    const segments = [];
-
-    const rows = db.prepare('SELECT videoID, startTime, endTime, UUID, category, shadowHidden FROM sponsorTimes WHERE votes >= -1 AND hashedVideoID LIKE ? ORDER BY startTime')
             .all(prefix + '%');
-
+    /** @type {string[]} */
     const onlyForCurrentUser = privateDB.prepare('SELECT videoID FROM sponsorTimes WHERE hashedIP = ?').all(hashedIP).map(row => row.videoID);
+    /** @type {Object.<string, Segment[][]>} */
+    const rowGroupsPerVideo = {};
 
+    let previousVideoID = null;
+    let previousEndTime = null;
     for (const row of rows) {
         /** @TODO check if this logic does what is expected. */
         if (row.shadowHidden === 1 && onlyForCurrentUser.indexOf(row.videoID) === -1) {
@@ -37,16 +77,30 @@ function getSkipSegmentsByHash(prefix, hashedIP) {
             // Do not send shadowHidden segments to them.
             continue;
         }
-
+        // Split up the rows per video and group overlapping segments together.
-        segments.push({
+        if (!(row.videoID in rowGroupsPerVideo)) {
-            videoID: row.videoID,
+            rowGroupsPerVideo[row.videoID] = [];
-            segment: [row.startTime, row.endTime],
+        }
-            category: row.category,
+        if (previousVideoID === row.videoID && row.startTime <= previousEndTime) {
-            UUID: row.UUID
+            rowGroupsPerVideo[row.videoID][rowGroupsPerVideo[row.videoID].length - 1].push(row);
-        });
+            previousEndTime = Math.max(previousEndTime, row.endTime);
+        } else {
+            rowGroupsPerVideo[row.videoID].push([row]);
+            previousVideoID = row.videoID;
+            previousEndTime = row.endTime;
+        }
     }
 
-    return segments;
+    /** @type {Object.<string, Segment[]>} */
+    const output = {};
+    for (const videoID in rowGroupsPerVideo) {
+        const pickedVideosForVideoID = [];
+        for (const group of rowGroupsPerVideo[videoID]) {
+            pickedVideosForVideoID.push(pickWeightedRandomRow(group));
+        }
+        output[videoID] = pickedVideosForVideoID.map(row => ({ videoID: row.videoID, segment: [row.startTime, row.endTime], category: row.category, UUID: row.UUID }));
+    }
+    return output;
 }
 
 const minimumPrefix = config.minimumPrefix || '3';