dummy graph data generator (exection: go run gen_graph_data.go)

gen_graph_data.go

package main

import (
	"bufio"
	"fmt"
	"log"
	"math/rand"
	"os"
	"strconv"
)

const USER_NUM_BASE = 100 * 10000 // 1M
const AVERAGE_FOLLOW = 100

func makeUserList() []*string {
	dupMap := make(map[string]bool)
	userList := make([]*string, USER_NUM_BASE*2)
	num := 0
	randGenerator := rand.New(rand.NewSource(2024))
	for num < USER_NUM_BASE*2 {
		userID := randGenerator.Intn(USER_NUM_BASE * 10)
		if _, ok := dupMap[string(userID)]; ok {
			// duplicated
			continue
		} else {
			dupMap[string(userID)] = true
			tmpStr1 := strconv.FormatInt(int64(userID), 16) + "X"
			userList[num] = &tmpStr1
			tmpStr2 := strconv.FormatInt(int64(userID), 16) + "Y"
			userList[num+1] = &tmpStr2
			num += 2
		}
	}

	return userList
}

func main() {
	file, err := os.Create("./social_graph.csv")
	writer := bufio.NewWriter(file)
	defer file.Close()
	if err != nil {
		log.Fatal(err)
	}

	userList := makeUserList()

	randGenerator := rand.New(rand.NewSource(2023))
	for ii := 0; ii < USER_NUM_BASE*2; ii++ {
		if ii%10000 == 0 {
			fmt.Println(ii)
		}
		dupMap := make(map[int]bool)
		if ii%2 == 0 {
			_, err = writer.WriteString(*userList[ii] + "," + *userList[ii+1] + "\n")
			if err != nil {
				log.Fatal(err)
			}
			dupMap[ii+1] = true
			_, err = writer.WriteString(*userList[ii+1] + "," + *userList[ii] + "\n")
			if err != nil {
				log.Fatal(err)
			}
		} else {
			dupMap[ii-1] = true
		}

		for jj := 0; jj < AVERAGE_FOLLOW; jj++ {
			followIdx := randGenerator.Intn(USER_NUM_BASE * 2)
			if _, ok := dupMap[followIdx]; ok {
				// duplicated
				continue
			} else {
				dupMap[followIdx] = true
				_, err = writer.WriteString(*userList[ii] + "," + *userList[followIdx] + "\n")
				if err != nil {
					log.Fatal(err)
				}
			}
		}
	}
}