gotchi-battler-game-logic 2.0.7 → 2.0.8

package/scripts/balancing/processSims.js

@@ -1,230 +1,230 @@
-require('dotenv').config()
-const fs = require('fs')
-const path = require('path')
-const { Storage } = require('@google-cloud/storage')
-const storage = new Storage()
-
-if (!process.env.CLOUD_RUN_JOB) {
-    // Use key file locally
-    process.env.GOOGLE_APPLICATION_CREDENTIALS = path.join(__dirname, '../../keyfile.json')
-}
-
-const downloadAndCombineResults = async (executionId, numOfTasks) => {
-    const results = []
-
-    // Download all the files in batches of 100
-    for (let i = 0; i < numOfTasks; i += 100) {
-        const batch = []
-
-        for (let j = 0; j < 100; j++) {
-            if (i + j < numOfTasks) {
-                batch.push(storage.bucket(process.env.SIMS_BUCKET).file(`${executionId}_${i + j}.json`).download())
-            }
-        }
-
-        const batchResults = await Promise.all(batch)
-        results.push(...batchResults)
-
-        console.log(`Downloaded ${i + batchResults.length} files`)
-    }
-
-    const combinedResults = []
-
-    for (const result of results) {
-        combinedResults.push(JSON.parse(result.toString()))
-    }
-
-    return combinedResults
-}
-
-const classes = ['Ninja','Enlightened','Cleaver','Tank','Cursed','Healer', 'Mage', 'Troll']
-
-const calculateKpis = (data) => {
-    // Extract wins values
-    const wins = data.map(item => item.wins);
-
-    // Calculate the mean of wins
-    const winsMean = wins.reduce((sum, value) => sum + value, 0) / wins.length;
-
-    // Calculate the wins per slot
-    const winsPerSlot = winsMean / 5
-
-    // Calculate the standard deviation of wins
-    const winsVariance = wins.reduce((sum, value) => sum + Math.pow(value - winsMean, 2), 0) / (wins.length - 1);
-    const winsStdev = Math.sqrt(winsVariance)
-
-    // Calculate Interquartile Range (IQR)
-    wins.sort((a, b) => a - b)
-    const q1 = wins[Math.floor(wins.length / 4)]
-    const q3 = wins[Math.floor(wins.length * 3 / 4)]
-    const iqr = q3 - q1
-
-    // Analyse top and bottom quartiles
-    const topQuartile = data.filter(item => item.wins >= q3)
-    const bottomQuartile = data.filter(item => item.wins <= q1)
-    const topQuartileWins = topQuartile.map(item => item.wins)
-    const bottomQuartileWins = bottomQuartile.map(item => item.wins)
-    const topQuartileWinsMean = topQuartileWins.reduce((sum, value) => sum + value, 0) / topQuartileWins.length
-    const bottomQuartileWinsMean = bottomQuartileWins.reduce((sum, value) => sum + value, 0) / bottomQuartileWins.length
-
-    // Non leader global KPIs
-     
-
-    const results = {
-        winsMean: winsMean.toFixed(2),
-        winsPerSlot: winsPerSlot.toFixed(2),
-        winsStdev: winsStdev.toFixed(2),
-        iqr: iqr.toFixed(2),
-        topQuartileWinsMean: topQuartileWinsMean.toFixed(2),
-        bottomQuartileWinsMean: bottomQuartileWinsMean.toFixed(2),
-        classes: {}
-    }
-
-    
-    classes.forEach((className, i) => {
-        /**
-         * Get leader KPIs
-         * Leaders are in slot 1
-         */
-
-        // Names are in format "G|avg|1_B" 
-        // G is power level, avg is trait combo, 1 is class number, B is formation position
-        const leaderWins = data.filter(item => item.slot1.includes(`${i + 1}`)).map(item => item.wins)
-
-        const leaderWinsMean = leaderWins.reduce((sum, value) => sum + value, 0) / leaderWins.length;
-        const leaderWinsVariance = leaderWins.reduce((sum, value) => sum + Math.pow(value - leaderWinsMean, 2), 0) / (leaderWins.length - 1);
-        const leaderWinsStdev = Math.sqrt(leaderWinsVariance)
-
-        // Calculate Interquartile Range (IQR)
-        leaderWins.sort((a, b) => a - b)
-        const leaderQ1 = leaderWins[Math.floor(leaderWins.length / 4)]
-        const leaderQ3 = leaderWins[Math.floor(leaderWins.length * 3 / 4)]
-        const leaderIqr = leaderQ3 - leaderQ1
-
-        // Analyse top and bottom quartiles
-        const leaderTopQuartile = data.filter(item => item.slot1.includes(`${i + 1}`) && item.wins >= leaderQ3)
-        const leaderBottomQuartile = data.filter(item => item.slot1.includes(`${i + 1}`) && item.wins <= leaderQ1)
-        const leaderTopQuartileWins = leaderTopQuartile.map(item => item.wins)
-        const leaderBottomQuartileWins = leaderBottomQuartile.map(item => item.wins)
-        const leaderTopQuartileWinsMean = leaderTopQuartileWins.reduce((sum, value) => sum + value, 0) / leaderTopQuartileWins.length
-        const leaderBottomQuartileWinsMean = leaderBottomQuartileWins.reduce((sum, value) => sum + value, 0) / leaderBottomQuartileWins.length
-
-        /**
-         * Get non-leader KPIs
-         * Non leaders are in slots 2, 3, 4, 5 
-         */
-
-        // Get the teams where the class is in the non-leader slots
-        const nonLeaderTeams = data.filter(item => item.slot2.includes(`${i + 1}`) || item.slot3.includes(`${i + 1}`) || item.slot4.includes(`${i + 1}`) || item.slot5.includes(`${i + 1}`))
-
-        const totalNonLeaderWins = nonLeaderTeams.reduce((sum, item) => {
-            const team = [item.slot2, item.slot3, item.slot4, item.slot5]
-            const occurrences = team.filter(slot => slot.includes(`${i + 1}`)).length
-            return sum + item.wins / 5 * occurrences
-        }, 0)
-
-        // Get total non-leader occurrences
-        const totalOccurrences = nonLeaderTeams.reduce((sum, item) => {
-            const team = [item.slot2, item.slot3, item.slot4, item.slot5]
-            return sum + team.filter(slot => slot.includes(`${i + 1}`)).length
-        }, 0)
-
-        // Calulate win per non-leader occurance
-        const winsPerOccurance = totalNonLeaderWins / totalOccurrences
-
-
-        results.classes[className] = {
-            leader: {
-                winsMean: leaderWinsMean.toFixed(2),
-                winsMeanVsGlobal: (leaderWinsMean - winsMean).toFixed(2),
-                winsMeanVsGlobalPercentage: ((leaderWinsMean - winsMean) / winsMean * 100).toFixed(2),
-                winsStdev: leaderWinsStdev.toFixed(2),
-                winsStdevVsGlobal: (leaderWinsStdev - winsStdev).toFixed(2),
-                winsStdevVsGlobalPercentage: ((leaderWinsStdev - winsStdev) / winsStdev * 100).toFixed(2),
-                iqr: leaderIqr.toFixed(2),
-                iqrVsGlobal: (leaderIqr - iqr).toFixed(2),
-                iqrVsGlobalPercentage: ((leaderIqr - iqr) / iqr * 100).toFixed(2),
-                topQuartileWinsMean: leaderTopQuartileWinsMean.toFixed(2),
-                topQuartileWinsMeanVsGlobal: (leaderTopQuartileWinsMean - topQuartileWinsMean).toFixed(2),
-                topQuartileWinsMeanVsGlobalPercentage: ((leaderTopQuartileWinsMean - topQuartileWinsMean) / topQuartileWinsMean * 100).toFixed(2),
-                bottomQuartileWinsMean: leaderBottomQuartileWinsMean.toFixed(2),
-                bottomQuartileWinsMeanVsGlobal: (leaderBottomQuartileWinsMean - bottomQuartileWinsMean).toFixed(2),
-                bottomQuartileWinsMeanVsGlobalPercentage: ((leaderBottomQuartileWinsMean - bottomQuartileWinsMean) / bottomQuartileWinsMean * 100).toFixed(2)
-            },
-            nonLeader: {
-                winsPerOccurance: winsPerOccurance.toFixed(2),
-                winsPerOccuranceVsGlobal: (winsPerOccurance - winsPerSlot).toFixed(2),
-                winsPerOccuranceVsGlobalPercentage: ((winsPerOccurance - winsPerSlot) / winsPerSlot * 100).toFixed(2)
-            }
-        }
-    })
-
-    return results
-}
-
-const main = async (executionId, numOfTasks, gameLogicVersion) => {
-    const gameLogicConstants = require(`../../game-logic/${gameLogicVersion}/constants.js`)
-
-    const data = await downloadAndCombineResults(executionId, numOfTasks)
-
-    const availablePowerLevels = {
-        'G': 'Godlike',
-        'M': 'Mythical',
-        'L': 'Legendary',
-        'R': 'Rare',
-        'U': 'Uncommon',
-        'C': 'Common'
-    }
-
-    // Find power levels from data
-    // slot1 is in the format "G|avg|1_B" where the first character is the power level
-    // Find the unique first characters in slot1
-    const powerLevels = Array.from(new Set(data.map(item => item.slot1[0])))
-
-    // Calculate KPIs for each power level
-    const results = {
-        metadata: {
-            gameLogicVersion: gameLogicVersion,
-            gameLogicConstants: gameLogicConstants,
-            powerLevels: powerLevels.map(powerLevel => availablePowerLevels[powerLevel])
-        },
-        results: {}
-    }
-    powerLevels.forEach((powerLevel) => {
-        const powerLevelData = data.filter(item => item.slot1[0] === powerLevel)
-        results.results[availablePowerLevels[powerLevel]] = calculateKpis(powerLevelData)
-    })
-
-    if(process.env.CLOUD_RUN_JOB) {
-        const bucket = storage.bucket('gotchi-battler-sims-v1-7-1')
-        
-        // Get number of files in the bucket
-        const [files] = await bucket.getFiles()
-        const numOfFiles = files.length
-
-        // Write the results to a file
-        await bucket.file(`${numOfFiles + 1}.json`).save(JSON.stringify(results, null, 2))
-    } else {
-        // Write to /scripts/balancing/output/<executionId>.json
-        fs.writeFileSync(path.join(__dirname, '/output/', `${executionId}.json`), JSON.stringify(results, null, 2))
-    }
-}
-
-module.exports = main
-
-// node scripts/balancing/processSims.js avg-sims-99hzc 7920 v1.7
-if (require.main === module) {
-    const executionId = process.env.EXECUTION_ID || process.argv[2]
-    const numOfTasks = parseInt(process.env.NUM_OF_TASKS) || parseInt(process.argv[3])
-    const gameLogicVersion = process.env.GAME_LOGIC_VERSION || process.argv[4]
-
-    main(executionId, numOfTasks, gameLogicVersion)
-        .then(() => {
-            console.log('Done')
-            process.exit(0)
-        })
-        .catch((error) => {
-            console.error(error)
-            process.exit(1)
-        })
+require('dotenv').config()
+const fs = require('fs')
+const path = require('path')
+const { Storage } = require('@google-cloud/storage')
+const storage = new Storage()
+
+if (!process.env.CLOUD_RUN_JOB) {
+    // Use key file locally
+    process.env.GOOGLE_APPLICATION_CREDENTIALS = path.join(__dirname, '../../keyfile.json')
+}
+
+const downloadAndCombineResults = async (executionId, numOfTasks) => {
+    const results = []
+
+    // Download all the files in batches of 100
+    for (let i = 0; i < numOfTasks; i += 100) {
+        const batch = []
+
+        for (let j = 0; j < 100; j++) {
+            if (i + j < numOfTasks) {
+                batch.push(storage.bucket(process.env.SIMS_BUCKET).file(`${executionId}_${i + j}.json`).download())
+            }
+        }
+
+        const batchResults = await Promise.all(batch)
+        results.push(...batchResults)
+
+        console.log(`Downloaded ${i + batchResults.length} files`)
+    }
+
+    const combinedResults = []
+
+    for (const result of results) {
+        combinedResults.push(JSON.parse(result.toString()))
+    }
+
+    return combinedResults
+}
+
+const classes = ['Ninja','Enlightened','Cleaver','Tank','Cursed','Healer', 'Mage', 'Troll']
+
+const calculateKpis = (data) => {
+    // Extract wins values
+    const wins = data.map(item => item.wins);
+
+    // Calculate the mean of wins
+    const winsMean = wins.reduce((sum, value) => sum + value, 0) / wins.length;
+
+    // Calculate the wins per slot
+    const winsPerSlot = winsMean / 5
+
+    // Calculate the standard deviation of wins
+    const winsVariance = wins.reduce((sum, value) => sum + Math.pow(value - winsMean, 2), 0) / (wins.length - 1);
+    const winsStdev = Math.sqrt(winsVariance)
+
+    // Calculate Interquartile Range (IQR)
+    wins.sort((a, b) => a - b)
+    const q1 = wins[Math.floor(wins.length / 4)]
+    const q3 = wins[Math.floor(wins.length * 3 / 4)]
+    const iqr = q3 - q1
+
+    // Analyse top and bottom quartiles
+    const topQuartile = data.filter(item => item.wins >= q3)
+    const bottomQuartile = data.filter(item => item.wins <= q1)
+    const topQuartileWins = topQuartile.map(item => item.wins)
+    const bottomQuartileWins = bottomQuartile.map(item => item.wins)
+    const topQuartileWinsMean = topQuartileWins.reduce((sum, value) => sum + value, 0) / topQuartileWins.length
+    const bottomQuartileWinsMean = bottomQuartileWins.reduce((sum, value) => sum + value, 0) / bottomQuartileWins.length
+
+    // Non leader global KPIs
+     
+
+    const results = {
+        winsMean: winsMean.toFixed(2),
+        winsPerSlot: winsPerSlot.toFixed(2),
+        winsStdev: winsStdev.toFixed(2),
+        iqr: iqr.toFixed(2),
+        topQuartileWinsMean: topQuartileWinsMean.toFixed(2),
+        bottomQuartileWinsMean: bottomQuartileWinsMean.toFixed(2),
+        classes: {}
+    }
+
+    
+    classes.forEach((className, i) => {
+        /**
+         * Get leader KPIs
+         * Leaders are in slot 1
+         */
+
+        // Names are in format "G|avg|1_B" 
+        // G is power level, avg is trait combo, 1 is class number, B is formation position
+        const leaderWins = data.filter(item => item.slot1.includes(`${i + 1}`)).map(item => item.wins)
+
+        const leaderWinsMean = leaderWins.reduce((sum, value) => sum + value, 0) / leaderWins.length;
+        const leaderWinsVariance = leaderWins.reduce((sum, value) => sum + Math.pow(value - leaderWinsMean, 2), 0) / (leaderWins.length - 1);
+        const leaderWinsStdev = Math.sqrt(leaderWinsVariance)
+
+        // Calculate Interquartile Range (IQR)
+        leaderWins.sort((a, b) => a - b)
+        const leaderQ1 = leaderWins[Math.floor(leaderWins.length / 4)]
+        const leaderQ3 = leaderWins[Math.floor(leaderWins.length * 3 / 4)]
+        const leaderIqr = leaderQ3 - leaderQ1
+
+        // Analyse top and bottom quartiles
+        const leaderTopQuartile = data.filter(item => item.slot1.includes(`${i + 1}`) && item.wins >= leaderQ3)
+        const leaderBottomQuartile = data.filter(item => item.slot1.includes(`${i + 1}`) && item.wins <= leaderQ1)
+        const leaderTopQuartileWins = leaderTopQuartile.map(item => item.wins)
+        const leaderBottomQuartileWins = leaderBottomQuartile.map(item => item.wins)
+        const leaderTopQuartileWinsMean = leaderTopQuartileWins.reduce((sum, value) => sum + value, 0) / leaderTopQuartileWins.length
+        const leaderBottomQuartileWinsMean = leaderBottomQuartileWins.reduce((sum, value) => sum + value, 0) / leaderBottomQuartileWins.length
+
+        /**
+         * Get non-leader KPIs
+         * Non leaders are in slots 2, 3, 4, 5 
+         */
+
+        // Get the teams where the class is in the non-leader slots
+        const nonLeaderTeams = data.filter(item => item.slot2.includes(`${i + 1}`) || item.slot3.includes(`${i + 1}`) || item.slot4.includes(`${i + 1}`) || item.slot5.includes(`${i + 1}`))
+
+        const totalNonLeaderWins = nonLeaderTeams.reduce((sum, item) => {
+            const team = [item.slot2, item.slot3, item.slot4, item.slot5]
+            const occurrences = team.filter(slot => slot.includes(`${i + 1}`)).length
+            return sum + item.wins / 5 * occurrences
+        }, 0)
+
+        // Get total non-leader occurrences
+        const totalOccurrences = nonLeaderTeams.reduce((sum, item) => {
+            const team = [item.slot2, item.slot3, item.slot4, item.slot5]
+            return sum + team.filter(slot => slot.includes(`${i + 1}`)).length
+        }, 0)
+
+        // Calulate win per non-leader occurance
+        const winsPerOccurance = totalNonLeaderWins / totalOccurrences
+
+
+        results.classes[className] = {
+            leader: {
+                winsMean: leaderWinsMean.toFixed(2),
+                winsMeanVsGlobal: (leaderWinsMean - winsMean).toFixed(2),
+                winsMeanVsGlobalPercentage: ((leaderWinsMean - winsMean) / winsMean * 100).toFixed(2),
+                winsStdev: leaderWinsStdev.toFixed(2),
+                winsStdevVsGlobal: (leaderWinsStdev - winsStdev).toFixed(2),
+                winsStdevVsGlobalPercentage: ((leaderWinsStdev - winsStdev) / winsStdev * 100).toFixed(2),
+                iqr: leaderIqr.toFixed(2),
+                iqrVsGlobal: (leaderIqr - iqr).toFixed(2),
+                iqrVsGlobalPercentage: ((leaderIqr - iqr) / iqr * 100).toFixed(2),
+                topQuartileWinsMean: leaderTopQuartileWinsMean.toFixed(2),
+                topQuartileWinsMeanVsGlobal: (leaderTopQuartileWinsMean - topQuartileWinsMean).toFixed(2),
+                topQuartileWinsMeanVsGlobalPercentage: ((leaderTopQuartileWinsMean - topQuartileWinsMean) / topQuartileWinsMean * 100).toFixed(2),
+                bottomQuartileWinsMean: leaderBottomQuartileWinsMean.toFixed(2),
+                bottomQuartileWinsMeanVsGlobal: (leaderBottomQuartileWinsMean - bottomQuartileWinsMean).toFixed(2),
+                bottomQuartileWinsMeanVsGlobalPercentage: ((leaderBottomQuartileWinsMean - bottomQuartileWinsMean) / bottomQuartileWinsMean * 100).toFixed(2)
+            },
+            nonLeader: {
+                winsPerOccurance: winsPerOccurance.toFixed(2),
+                winsPerOccuranceVsGlobal: (winsPerOccurance - winsPerSlot).toFixed(2),
+                winsPerOccuranceVsGlobalPercentage: ((winsPerOccurance - winsPerSlot) / winsPerSlot * 100).toFixed(2)
+            }
+        }
+    })
+
+    return results
+}
+
+const main = async (executionId, numOfTasks, gameLogicVersion) => {
+    const gameLogicConstants = require(`../../game-logic/${gameLogicVersion}/constants.js`)
+
+    const data = await downloadAndCombineResults(executionId, numOfTasks)
+
+    const availablePowerLevels = {
+        'G': 'Godlike',
+        'M': 'Mythical',
+        'L': 'Legendary',
+        'R': 'Rare',
+        'U': 'Uncommon',
+        'C': 'Common'
+    }
+
+    // Find power levels from data
+    // slot1 is in the format "G|avg|1_B" where the first character is the power level
+    // Find the unique first characters in slot1
+    const powerLevels = Array.from(new Set(data.map(item => item.slot1[0])))
+
+    // Calculate KPIs for each power level
+    const results = {
+        metadata: {
+            gameLogicVersion: gameLogicVersion,
+            gameLogicConstants: gameLogicConstants,
+            powerLevels: powerLevels.map(powerLevel => availablePowerLevels[powerLevel])
+        },
+        results: {}
+    }
+    powerLevels.forEach((powerLevel) => {
+        const powerLevelData = data.filter(item => item.slot1[0] === powerLevel)
+        results.results[availablePowerLevels[powerLevel]] = calculateKpis(powerLevelData)
+    })
+
+    if(process.env.CLOUD_RUN_JOB) {
+        const bucket = storage.bucket('gotchi-battler-sims-v1-7-2')
+        
+        // Get number of files in the bucket
+        const [files] = await bucket.getFiles()
+        const numOfFiles = files.length
+
+        // Write the results to a file
+        await bucket.file(`${numOfFiles + 1}.json`).save(JSON.stringify(results, null, 2))
+    } else {
+        // Write to /scripts/balancing/output/<executionId>.json
+        fs.writeFileSync(path.join(__dirname, '/output/', `${executionId}.json`), JSON.stringify(results, null, 2))
+    }
+}
+
+module.exports = main
+
+// node scripts/balancing/processSims.js avg-sims-99hzc 7920 v1.7
+if (require.main === module) {
+    const executionId = process.env.EXECUTION_ID || process.argv[2]
+    const numOfTasks = parseInt(process.env.NUM_OF_TASKS) || parseInt(process.argv[3])
+    const gameLogicVersion = process.env.GAME_LOGIC_VERSION || process.argv[4]
+
+    main(executionId, numOfTasks, gameLogicVersion)
+        .then(() => {
+            console.log('Done')
+            process.exit(0)
+        })
+        .catch((error) => {
+            console.error(error)
+            process.exit(1)
+        })
 }