refactor: 기존 인구 데이터 처리용 테스트 클래스 통합 및 외부 호출을 위해 함수로 변경

This commit is contained in:
이상진 2025-09-27 15:06:02 +09:00
parent 3e5bdc3e05
commit 6752421fda
3 changed files with 236 additions and 229 deletions

View File

@ -1,99 +0,0 @@
package roomescape.data
import io.kotest.core.spec.style.StringSpec
import org.apache.poi.xssf.usermodel.XSSFWorkbook
import java.io.File
const val BASE_DIR = "data"
const val PARSED_REGION_POPULATION_FILE = "$BASE_DIR/region_population.txt"
const val REGION_SQL_FILE = "data/region.sql"
const val MIN_POPULATION_FOR_PER_STORE = 200_000
/**
* 행안부 202508 인구 동향 데이터 사용( /data/population.xlsx )
*/
class PopulationDataSqlParser() : StringSpec({
val regionCodePattern = Regex("^[0-9]{10}$")
"인구 데이터를 이용하여 지역 정보 SQL 파일로 변환하고, 추가로 $MIN_POPULATION_FOR_PER_STORE 이상의 시/군/구는 매장 데이터 생성을 위해 따로 분류한다.".config(
enabled = false
) {
val populationXlsx = XSSFWorkbook(File("data/population.xlsx"))
val sheet = populationXlsx.getSheetAt(0)
val allRegion = mutableListOf<List<String>>()
val regionsMoreThanMinPopulation = mutableListOf<List<String?>>()
sheet.rowIterator().forEach { row ->
val regionCode = row.getCell(0)?.stringCellValue ?: return@forEach
if (regionCodePattern.matches(regionCode).not()) {
return@forEach
}
val sidoCode = regionCode.substring(0, 2)
val sigunguCode = regionCode.substring(2, 5)
val population = row.getCell(2).stringCellValue.replace(",", "")
if (Regex("^[0-9]+$").matches(population).not()) {
return@forEach
}
val regionName = row.getCell(1).stringCellValue
if (!regionName.trim().contains(" ")) {
return@forEach
}
val parts = regionName.split(" ")
if (parts.size < 2) {
return@forEach
}
val sidoName = parts[0].trim()
val sigunguName = parts[1].trim()
val populationInt = population.toInt()
if (populationInt <= 0) {
return@forEach
}
if (populationInt >= MIN_POPULATION_FOR_PER_STORE) {
regionsMoreThanMinPopulation.add(
listOf(
regionCode,
sidoCode,
sigunguCode,
sidoName,
sigunguName,
population
)
)
}
allRegion.add(listOf(regionCode, sidoCode, sigunguCode, sidoName, sigunguName))
}
regionsMoreThanMinPopulation.filter {
val sidoName = it[3]
val sigunguName = it[4]
val sameSigungu = allRegion.filter { r -> r[3] == sidoName && r[4] == sigunguName }
!((sameSigungu.size > 1) && sameSigungu.minByOrNull { r -> r[2].toInt() }!![2] != it[2])
}.mapIndexed { idx, values ->
"${values[0]}, ${values[1]}, ${values[2]}, ${values[3]}, ${values[4]}, ${values[5]}"
}.joinToString(separator = "\n").also {
File(PARSED_REGION_POPULATION_FILE).writeText(it)
}
allRegion.distinctBy { it[1] to it[2] }.filter {
val sidoName = it[3]
val sigunguName = it[4]
val sameSigungu = allRegion.filter { r -> r[3] == sidoName && r[4] == sigunguName }
!((sameSigungu.size > 1) && sameSigungu.minByOrNull { r -> r[2].toInt() }!![2] != it[2])
}.joinToString(
prefix = "INSERT INTO region(code, sido_code, sigungu_code, sido_name, sigungu_name) VALUES ",
separator = ",\n"
) { region ->
"('${region[0]}', '${region[1]}', '${region[2]}', '${region[3]}', '${region[4]}')"
}.also {
File(REGION_SQL_FILE).writeText("${it};")
}
}
})

View File

@ -0,0 +1,236 @@
package roomescape.data
import org.apache.poi.xssf.usermodel.XSSFWorkbook
import roomescape.common.config.next
import roomescape.store.infrastructure.persistence.StoreStatus
import roomescape.supports.randomPhoneNumber
import roomescape.supports.tsidFactory
import java.io.File
import java.time.LocalDateTime
import java.time.ZoneId
import java.time.format.DateTimeFormatter
import kotlin.random.Random
const val BASE_DIR = "data"
const val PARSED_REGION_POPULATION_FILE = "$BASE_DIR/region_population.txt"
const val REGION_SQL_FILE = "data/region.sql"
const val MIN_POPULATION_FOR_PER_STORE = 200_000
/**
* 행안부 202508 인구 동향 데이터 사용( /data/population.xlsx )
*/
class PopulationDataSqlParser() {
val regionCodePattern = Regex("^[0-9]{10}$")
// 인구 데이터를 이용하여 지역 정보 SQL 파일로 변환하고, 추가로 $MIN_POPULATION_FOR_PER_STORE 이상의 시/군/구는 매장 데이터 생성을 위해 따로 분류한다.
fun createParsedRegionPopulationFiles() {
val populationXlsx = XSSFWorkbook(File("data/population.xlsx"))
val sheet = populationXlsx.getSheetAt(0)
val allRegion = mutableListOf<List<String>>()
val regionsMoreThanMinPopulation = mutableListOf<List<String?>>()
sheet.rowIterator().forEach { row ->
val regionCode = row.getCell(0)?.stringCellValue ?: return@forEach
if (regionCodePattern.matches(regionCode).not()) {
return@forEach
}
val sidoCode = regionCode.substring(0, 2)
val sigunguCode = regionCode.substring(2, 5)
val population = row.getCell(2).stringCellValue.replace(",", "")
if (Regex("^[0-9]+$").matches(population).not()) {
return@forEach
}
val regionName = row.getCell(1).stringCellValue
if (!regionName.trim().contains(" ")) {
return@forEach
}
val parts = regionName.split(" ")
if (parts.size < 2) {
return@forEach
}
val sidoName = parts[0].trim()
val sigunguName = parts[1].trim()
val populationInt = population.toInt()
if (populationInt <= 0) {
return@forEach
}
if (populationInt >= MIN_POPULATION_FOR_PER_STORE) {
regionsMoreThanMinPopulation.add(
listOf(
regionCode,
sidoCode,
sigunguCode,
sidoName,
sigunguName,
population
)
)
}
allRegion.add(listOf(regionCode, sidoCode, sigunguCode, sidoName, sigunguName))
}
regionsMoreThanMinPopulation.filter {
val sidoName = it[3]
val sigunguName = it[4]
val sameSigungu = allRegion.filter { r -> r[3] == sidoName && r[4] == sigunguName }
!((sameSigungu.size > 1) && sameSigungu.minByOrNull { r -> r[2].toInt() }!![2] != it[2])
}.mapIndexed { idx, values ->
"${values[0]}, ${values[1]}, ${values[2]}, ${values[3]}, ${values[4]}, ${values[5]}"
}.joinToString(separator = "\n").also {
File(PARSED_REGION_POPULATION_FILE).writeText(it)
}
allRegion.distinctBy { it[1] to it[2] }.filter {
val sidoName = it[3]
val sigunguName = it[4]
val sameSigungu = allRegion.filter { r -> r[3] == sidoName && r[4] == sigunguName }
!((sameSigungu.size > 1) && sameSigungu.minByOrNull { r -> r[2].toInt() }!![2] != it[2])
}.joinToString(
prefix = "INSERT INTO region(code, sido_code, sigungu_code, sido_name, sigungu_name) VALUES ",
separator = ",\n"
) { region ->
"('${region[0]}', '${region[1]}', '${region[2]}', '${region[3]}', '${region[4]}')"
}.also {
File(REGION_SQL_FILE).writeText("${it};")
}
}
}
/**
* PopulationDataSqlParser에서 전처리된 지역 + 인구 정보를 이용하여 Store 초기 데이터 SQL 생성
*/
class StoreDataInitializer {
val positiveWords = listOf(
"사랑", "행복", "희망", "감사", "기쁨", "미소", "축복", "선물", "평화",
"열정", "미래", "자유", "도전", "지혜", "행운"
)
fun createStoreDataSqlFile(creatableAdminIds: List<String>): File {
val regions = initializeRegionWithStoreCount()
val usedStoreName = mutableListOf<String>()
val usedBusinessRegNums = mutableListOf<String>()
val storeSqlRows = mutableListOf<String>()
val storeDataRows = mutableListOf<String>()
val storeIds = mutableListOf<Long>()
regions.forEachIndexed { idx, region ->
for (i in 0..region.storeCount) {
var address: String
var storeName: String
do {
val randomPositiveWord = positiveWords.random()
storeName = "${parseSigunguName(region.sigunguName)}${randomPositiveWord}"
address =
"${region.sidoName} ${region.sigunguName} ${randomPositiveWord}${Random.nextInt(1, 10)}${Random.nextInt(1, 100)}"
} while (usedStoreName.contains(storeName))
usedStoreName.add(storeName)
val contact = randomPhoneNumber()
var businessRegNum: String
do {
businessRegNum = generateBusinessRegNum()
} while (usedBusinessRegNums.contains(businessRegNum))
usedBusinessRegNums.add(businessRegNum)
val createdAt = randomLocalDateTime()
val updatedAt = createdAt
val id: Long = tsidFactory.next().also { storeIds.add(it) }
val createdBy = creatableAdminIds.random()
storeSqlRows.add(
"(${id}, '$storeName', '$address', '$contact', '$businessRegNum', '${region.regionCode}', '${StoreStatus.ACTIVE.name}', '$createdAt', '${createdBy}', '$updatedAt', '${createdBy}')"
)
storeDataRows.add(
"$id, $storeName, $address, $contact, $businessRegNum, ${region.regionCode}, ${StoreStatus.ACTIVE.name}, $createdAt, $createdBy, $updatedAt, $createdBy"
)
}
}
File("$BASE_DIR/store_data.txt").also {
if (it.exists()) { it.delete() }
}.writeText(
storeDataRows.joinToString("\n")
)
return File("$BASE_DIR/store_data.sql").also {
if (it.exists()) { it.delete() }
StringBuilder("INSERT INTO store (id, name, address, contact, business_reg_num, region_code, status, created_at, created_by, updated_at, updated_by) VALUES ")
.append(storeSqlRows.joinToString(",\n"))
.append(";")
.toString()
.also { sql -> it.writeText(sql) }
}
}
}
private fun parseSigunguName(sigunguName: String): String {
if (sigunguName.length == 2) {
return sigunguName
}
return sigunguName.substring(0, sigunguName.length - 1)
}
private fun randomLocalDateTime(): String {
val year = Random.nextInt(2020, 2024)
val month = Random.nextInt(1, 13)
val day = when (month) {
2 -> Random.nextInt(1, 29)
else -> Random.nextInt(1, 31)
}
val hour = Random.nextInt(9, 19)
val minute = Random.nextInt(0, 60)
val second = Random.nextInt(0, 60)
return LocalDateTime.of(year, month, day, hour, minute, second)
.atZone(ZoneId.systemDefault())
.toOffsetDateTime()
.format(DateTimeFormatter.ofPattern("yyyy-MM-dd'T'HH:mm:ssXXX"))
}
private fun generateBusinessRegNum(): String {
val part1 = Random.nextInt(100, 1000)
val part2 = Random.nextInt(10, 100)
val part3 = Random.nextInt(10000, 100000)
return "$part1-$part2-$part3"
}
private fun initializeRegionWithStoreCount(): List<RegionWithStoreCount> {
return File(PARSED_REGION_POPULATION_FILE).also {
if (it.exists().not()) {
PopulationDataSqlParser().createParsedRegionPopulationFiles()
}
}.readText().lines().map {
val parts = it.split(", ")
val regionCode = parts[0]
val sidoName = parts[3]
val sigunguName = parts[4]
val storeCount: Int = (parts[5].toInt() / MIN_POPULATION_FOR_PER_STORE)
RegionWithStoreCount(
regionCode = regionCode,
sidoName = sidoName,
sigunguName = sigunguName,
storeCount = storeCount
)
}
}
data class RegionWithStoreCount(
val regionCode: String,
val sidoName: String,
val sigunguName: String,
var storeCount: Int
)

View File

@ -1,130 +0,0 @@
package roomescape.data
import io.kotest.core.spec.style.StringSpec
import roomescape.common.config.next
import roomescape.supports.randomPhoneNumber
import roomescape.supports.tsidFactory
import java.io.File
import java.time.LocalDateTime
import java.time.ZoneId
import java.time.format.DateTimeFormatter
import kotlin.random.Random
/**
* DataParser에서 전처리된 지역 + 인구 정보를 이용하여 Store 초기 데이터 생성
*/
class StoreDataInitializer : StringSpec({
val positiveWords = listOf(
"사랑", "행복", "희망", "감사", "기쁨", "미소", "축복", "선물", "평화",
"열정", "미래", "자유", "도전", "지혜", "행운"
)
"초기 매장 데이터를 준비한다.".config(enabled = false) {
val regions = initializeRegionWithStoreCount()
val usedStoreName = mutableListOf<String>()
val usedBusinessRegNums = mutableListOf<String>()
val storeSqlRows = mutableListOf<String>()
val storeDataRows = mutableListOf<String>()
val storeIds = mutableListOf<Long>()
regions.forEachIndexed { idx, region ->
for (i in 0..region.storeCount) {
var address: String
var storeName: String
do {
val randomPositiveWord = positiveWords.random()
storeName = "${parseSigunguName(region.sigunguName)}${randomPositiveWord}"
address =
"${region.sidoName} ${region.sigunguName} ${randomPositiveWord}${Random.nextInt(1, 10)}${Random.nextInt(1, 100)}"
} while (usedStoreName.contains(storeName))
usedStoreName.add(storeName)
val contact = randomPhoneNumber()
var businessRegNum: String
do {
businessRegNum = generateBusinessRegNum()
} while (usedBusinessRegNums.contains(businessRegNum))
usedBusinessRegNums.add(businessRegNum)
val createdAt = randomLocalDateTime()
val updatedAt = createdAt
val id: Long = tsidFactory.next().also { storeIds.add(it) }
storeSqlRows.add(
"(${id}, '$storeName', '$address', '$contact', '$businessRegNum', '${region.regionCode}', '$createdAt', '$updatedAt')"
)
storeDataRows.add(
"$id, $storeName, $address, $contact, $businessRegNum, ${region.regionCode}, $createdAt, $updatedAt"
)
}
}
StringBuilder("INSERT INTO store (id, name, address, contact, business_reg_num, region_code, created_at, updated_at) VALUES ")
.append(storeSqlRows.joinToString(",\n"))
.append(";")
.toString()
.also { File("$BASE_DIR/store_data.sql").writeText(it) }
File("$BASE_DIR/store_data.txt").writeText(
storeDataRows.joinToString("\n")
)
}
})
private fun parseSigunguName(sigunguName: String): String {
if (sigunguName.length == 2) {
return sigunguName
}
return sigunguName.substring(0, sigunguName.length - 1)
}
private fun randomLocalDateTime(): String {
val year = Random.nextInt(2020, 2024)
val month = Random.nextInt(1, 13)
val day = when (month) {
2 -> Random.nextInt(1, 29)
else -> Random.nextInt(1, 31)
}
val hour = Random.nextInt(9, 19)
val minute = Random.nextInt(0, 60)
val second = Random.nextInt(0, 60)
return LocalDateTime.of(year, month, day, hour, minute, second)
.atZone(ZoneId.systemDefault())
.toOffsetDateTime()
.format(DateTimeFormatter.ofPattern("yyyy-MM-dd'T'HH:mm:ssXXX"))
}
private fun generateBusinessRegNum(): String {
val part1 = Random.nextInt(100, 1000)
val part2 = Random.nextInt(10, 100)
val part3 = Random.nextInt(10000, 100000)
return "$part1-$part2-$part3"
}
private fun initializeRegionWithStoreCount(): List<RegionWithStoreCount> {
return File(PARSED_REGION_POPULATION_FILE).readText().lines().map {
val parts = it.split(", ")
val regionCode = parts[0]
val sidoName = parts[3]
val sigunguName = parts[4]
val storeCount: Int = (parts[5].toInt() / MIN_POPULATION_FOR_PER_STORE)
RegionWithStoreCount(
regionCode = regionCode,
sidoName = sidoName,
sigunguName = sigunguName,
storeCount = storeCount
)
}
}
data class RegionWithStoreCount(
val regionCode: String,
val sidoName: String,
val sigunguName: String,
var storeCount: Int
)