[IDLE-531] 배치 작업 정상화 및 최적화

idle-batch/src/main/kotlin/com/swm/idle/batch/common/scheduler/CrawlingJobScheduler.kt → idle-batch/src/main/kotlin/com/swm/idle/batch/common/launcher/CrawlingJobLauncher.kt

@@ -1,16 +1,18 @@
-package com.swm.idle.batch.common.scheduler
+package com.swm.idle.batch.common.launcher
 
-import com.swm.idle.batch.job.CrawlingJobConfig
+import com.swm.idle.batch.job.JobConfig
 import org.springframework.batch.core.JobParameters
 import org.springframework.batch.core.JobParametersBuilder
+import org.springframework.batch.core.configuration.JobRegistry
 import org.springframework.batch.core.launch.JobLauncher
 import org.springframework.scheduling.annotation.Scheduled
 import org.springframework.stereotype.Component
 
 @Component
-class CrawlingJobScheduler(
+class CrawlingJobLauncher(
     private val jobLauncher: JobLauncher,
-    private val crawlingJobConfig: CrawlingJobConfig,
+    private val jobRegistry: JobRegistry,
+    private val crawlingJobConfig: JobConfig,
 ) {
 
     @Scheduled(cron = "0 0 23 * * *")
@@ -22,4 +24,11 @@ class CrawlingJobScheduler(
         jobLauncher.run(crawlingJobConfig.crawlingJob(), jobParameters)
     }
 
+    fun jobStart() {
+        val jobParameters: JobParameters = JobParametersBuilder()
+            .addLong("timestamp", System.currentTimeMillis())
+            .toJobParameters()
+
+        jobLauncher.run(jobRegistry.getJob("crawlingJob"), jobParameters)
+    }
 }

idle-batch/src/main/kotlin/com/swm/idle/batch/crawler/CrawlerConsts.kt

@@ -0,0 +1,62 @@
+package com.swm.idle.batch.crawler
+
+enum class CrawlerConsts(val location: String, val value: String) {
+    CRAWLING_TARGET_URL_FORMAT("CRAWLING_TARGET_URL_FORMAT","https://www.work24.go.kr/wk/a/b/1200/retriveDtlEmpSrchList.do?basicSetupYn=&careerTo=&keywordJobCd=&occupation=&seqNo=&cloDateEndtParam=&payGbn=&templateInfo=&rot2WorkYn=&shsyWorkSecd=&srcKeywordParam=%EC%9A%94%EC%96%91%EB%B3%B4%ED%98%B8%EC%82%AC&resultCnt=50&keywordJobCont=&cert=&moreButtonYn=Y&minPay=&codeDepth2Info=11000&currentPageNo=1&eventNo=&mode=&major=&resrDutyExcYn=&eodwYn=&sortField=DATE&staArea=&sortOrderBy=DESC&keyword=%EC%9A%94%EC%96%91%EB%B3%B4%ED%98%B8%EC%82%AC&termSearchGbn=all&carrEssYns=&benefitSrchAndOr=O&disableEmpHopeGbn=&actServExcYn=&keywordStaAreaNm=&maxPay=&emailApplyYn=&codeDepth1Info=11000&keywordEtcYn=&regDateStdtParam={yesterday}&publDutyExcYn=&keywordJobCdSeqNo=&viewType=&exJobsCd=&templateDepthNmInfo=&region=&employGbn=&empTpGbcd=&computerPreferential=&infaYn=&cloDateStdtParam=&siteClcd=WORK&searchMode=Y&birthFromYY=&indArea=&careerTypes=&subEmpHopeYn=&tlmgYn=&academicGbn=&templateDepthNoInfo=&foriegn=&entryRoute=&mealOfferClcd=&basicSetupYnChk=&station=&holidayGbn=&srcKeyword=%EC%9A%94%EC%96%91%EB%B3%B4%ED%98%B8%EC%82%AC&academicGbnoEdu=noEdu&enterPriseGbn=all&cloTermSearchGbn=all&birthToYY=&keywordWantedTitle=&stationNm=&benefitGbn=&notSrcKeywordParam=&keywordFlag=&notSrcKeyword=&essCertChk=&depth2SelCode=&keywordBusiNm=&preferentialGbn=&rot3WorkYn=&regDateEndtParam={yesterday}&pfMatterPreferential=&pageIndex={pageIndex}&termContractMmcnt=&careerFrom=&laborHrShortYn=#scrollLoc"),
+    JOB_POSTING_COUNT_PER_PAGE("JOB_POSTING_COUNT_PER_PAGE","50"),
+    JOB_POSTING_COUNT("JOB_POSTING_COUNT","//*[@id=\"mForm\"]/div[2]/div/div[1]/div[1]/span/span"),
+
+    //공고 정보
+    TITLE("TITLE", "//*[@id=\"contents\"]/div/div/div/div[1]/div[3]/div[1]/div[1]/strong"),
+    CONTENT("CONTENT", "//*[@id=\"tab-panel01\"]/div[1]/div"),
+
+    //근무 정보
+    PAY_INFO("PAY_INFO", "//*[@id=\"tab-panel02\"]/div/table/tbody/tr[1]/td[2]"),
+    WORK_TIME("WORK_TIME","//*[@id=\"tab-panel02\"]/div/table/tbody/tr[2]/td"),
+    WORK_SCHEDULE("WORK_SCHEDULE","//*[@id=\"tab-panel02\"]/div/table/tbody/tr[3]/td[2]"),
+
+    //모집 정보
+    RECRUITMENT_PROCESS("RECRUITMENT_PROCESS","//*[@id=\"tab-panel05\"]/div[2]/div/div[2]/p[1]"),
+    REQUIRED_DOCUMENT("REQUIRED_DOCUMENT","//*[@id=\"tab-panel05\"]/div[2]/div/div[2]/p[2]"),
+    APPLY_METHOD("APPLY_METHOD","//*[@id=\"tab-panel05\"]/div[2]/div/div[2]/p[1]"),
+    APPLY_DEADLINE("APPLY_DEADLINE","//*[@id=\"tab-panel05\"]/div[2]/div/div[1]/div[1]/p"),
+    CREATED_AT("CREATED_AT","//*[@id=\"contents\"]/div/div/div/div[1]/div[5]/div[11]/div[2]/table/tbody/tr[1]/td[1]"),
+
+    //센터 정보
+    CENTER_NAME("CENTER_NAME","//*[@id=\"contents\"]/div/div/div/div[1]/div[3]/div[1]/div[1]/p/strong"),
+    CENTER_ADDRESS1("CENTER_ADDRESS1","//*[@id=\"tab-panel02\"]/div/table/tbody/tr[5]/td/div[1]/p"),
+    CENTER_ADDRESS2("CENTER_ADDRESs2","//*[@id=\"tab-panel02\"]/div/table/tbody/tr[5]/td/div[1]/p"),
+    CENTER_ADDRESS3("CENTER_ADDRESS3","//*[@id=\"tab-panel02\"]/div/table/tbody/tr[5]/td/div[1]/p"),
+
+    //노인 주소
+    CLIENT_ADDRESS1("CLIENT_ADDRESS1","//*[@id=\"tab-panel02\"]/div/table/tbody/tr[5]/td/div[1]/p"),
+    CLIENT_ADDRESS2("CLIENT_ADDRESS2","//*[@id=\"tab-panel02\"]/div/table/tbody/tr[5]/td/div[1]/p"),
+
+    //ChromDriver-Options
+    HEADLESS("HEADLESS","--headless"),
+    NO_SANDBOX("NO_SANDBOX","--no-sandbox"),
+    DISABLE_DEV_SHM_USAGE("DISABLE_DEV_SHM_USAGE","--disable-dev-shm-usage"),
+    DISABLE_GPU("DISABLE_GPU","--disable-gpu"),
+    WINDOW_SIZE("WINDOW_SIZE","window-size=1920x1080"),
+    DISABLE_SOFTWARE_RASTERIZER("DISABLE_SOFTWARE_RASTERIZER","--disable-software-rasterizer"),
+    IGNORE_SSL_ERRORS("IGNORE_SSL_ERRORS","--ignore-ssl-errors=yes"),
+    IGNORE_CERTIFICATE_ERRORS("IGNORE_CERTIFICATE_ERRORS","--ignore-certificate-errors");
+
+    companion object {
+        fun getChromOptions(): Array<String> {
+            return arrayOf(
+                HEADLESS.value,
+                NO_SANDBOX.value,
+                DISABLE_DEV_SHM_USAGE.value,
+                DISABLE_GPU.value,
+                WINDOW_SIZE.value,
+                DISABLE_SOFTWARE_RASTERIZER.value,
+                IGNORE_SSL_ERRORS.value,
+                IGNORE_CERTIFICATE_ERRORS.value
+            )
+        }
+    }
+
+    fun getIntValue(): Int {
+        return value.toInt()
+    }
+}

idle-batch/src/main/kotlin/com/swm/idle/batch/crawler/DriverInitializer.kt

@@ -0,0 +1,29 @@
+package com.swm.idle.batch.crawler
+
+import io.github.oshai.kotlinlogging.KotlinLogging
+import org.openqa.selenium.chrome.ChromeDriver
+import org.openqa.selenium.chrome.ChromeDriverService
+import org.openqa.selenium.chrome.ChromeOptions
+import java.io.File
+
+object DriverInitializer {
+    private val logger = KotlinLogging.logger { }
+
+    fun init(): ChromeDriver {
+        return runCatching {
+            ChromeDriver(
+                ChromeDriverService.Builder()
+                    .usingDriverExecutable(File(System.getenv("CHROMEDRIVER_BIN")))
+                    .build()
+                    .also { logger.info { System.getenv("CHROMEDRIVER_BIN") } },
+                ChromeOptions().apply {
+                    addArguments(*CrawlerConsts.getChromOptions())
+                    setBinary(System.getenv("CHROME_BIN"))
+                }.also { logger.info { System.getenv("CHROME_BIN")} }
+            )
+        }.getOrElse {
+            logger.error { "ChromeDriver initialization failed: ${it.message}" }
+            throw RuntimeException("ChromeDriver initialization failed, application will exit.") // 이후 코드가 실행되지 않도록 예외 던짐
+        }
+    }
+}

idle-batch/src/main/kotlin/com/swm/idle/batch/crawler/WorknetPageCrawler.kt

@@ -0,0 +1,44 @@
+package com.swm.idle.batch.crawler
+
+import com.swm.idle.batch.step.PostingReader
+import org.openqa.selenium.By
+import org.openqa.selenium.WebDriver
+import org.openqa.selenium.support.ui.ExpectedConditions
+import org.openqa.selenium.support.ui.WebDriverWait
+import java.time.Duration
+import java.time.LocalDate
+import java.time.format.DateTimeFormatter
+
+class WorknetPageCrawler {
+    private var driver: WebDriver = DriverInitializer.init()
+
+    fun initCounts(reader: PostingReader) {
+        reader.crawlingUrl = CrawlerConsts.CRAWLING_TARGET_URL_FORMAT.value
+                .replace("{yesterday}", LocalDate.now().format(DateTimeFormatter.ofPattern("yyyyMMdd")))
+                .replace("{pageIndex}", "1")
+
+        moveToPage(reader)
+
+        reader.postingCount = driver
+            .findElement(By.xpath(CrawlerConsts.JOB_POSTING_COUNT.value))
+            .text.toInt()
+            .takeIf { it > 0 }
+            ?: run {
+                driver.quit()
+                throw Exception("크롤링 할 공고가 없습니다.")
+            }
+
+        reader.pageCount = (reader.postingCount + CrawlerConsts.JOB_POSTING_COUNT_PER_PAGE.getIntValue() - 1) /
+                CrawlerConsts.JOB_POSTING_COUNT_PER_PAGE.getIntValue()
+        reader.lastPageJobPostingCount = reader.postingCount % CrawlerConsts.JOB_POSTING_COUNT_PER_PAGE.getIntValue()
+        driver.quit()
+    }
+
+    private fun moveToPage(reader: PostingReader) {
+        driver.get(reader.crawlingUrl)
+        WebDriverWait(driver, Duration.ofSeconds(10))
+            .also {
+                it.until(ExpectedConditions.visibilityOfElementLocated(By.xpath(CrawlerConsts.JOB_POSTING_COUNT.value)))
+            }
+    }
+}

idle-batch/src/main/kotlin/com/swm/idle/batch/crawler/WorknetPostCrawler.kt

@@ -0,0 +1,142 @@
+package com.swm.idle.batch.crawler
+
+import com.swm.idle.batch.common.dto.CrawledJobPostingDto
+import io.github.oshai.kotlinlogging.KotlinLogging
+import org.openqa.selenium.By
+import org.openqa.selenium.WebDriver
+import org.openqa.selenium.support.ui.ExpectedConditions
+import org.openqa.selenium.support.ui.WebDriverWait
+import java.time.Duration
+import java.time.LocalDate
+import java.time.format.DateTimeFormatter
+import org.openqa.selenium.WebElement
+
+class WorknetPostCrawler {
+    private val logger = KotlinLogging.logger { }
+    private var driver: WebDriver = DriverInitializer.init()
+    private var errorCountMap: MutableMap<String, Int> = mutableMapOf()
+
+    fun crawlPosts(end: Int, url: String): List<CrawledJobPostingDto> {
+        moveToPage(url)
+
+        val crawledPostings = mutableListOf<CrawledJobPostingDto>()
+        repeat(end) { i ->
+            val originalWindow = driver.windowHandle
+            val titleElement = findElementSafe(By.xpath("//*[@id=\"list${i+1}\"]/td[1]/div/div[2]/a")) ?:  return@repeat
+
+            moveToPostDetailWindow(titleElement, originalWindow)
+
+            try {
+                val post: CrawledJobPostingDto = createPost()
+                crawledPostings.add(post)
+            } catch (e: Exception) {
+                logger.warn { "실패" }
+            }
+
+            backWindow(originalWindow)
+        }
+        errorCountMap.asSequence().forEach { (key, value) -> println("$key -> $value") }
+        driver.quit()
+        return crawledPostings
+    }
+
+    private fun moveToPage(url: String) {
+        driver.get(url)
+        WebDriverWait(driver, Duration.ofSeconds(10))
+            .until(
+                ExpectedConditions.visibilityOfElementLocated(By.cssSelector("#list1"))
+            )
+    }
+
+    private fun createPost(): CrawledJobPostingDto {
+        return CrawledJobPostingDto(
+            title = extractText(CrawlerConsts.TITLE),
+            content = extractText(CrawlerConsts.CONTENT),
+            createdAt = extractText(CrawlerConsts.CREATED_AT),
+            payInfo = extractText(CrawlerConsts.PAY_INFO),
+            workSchedule = extractText(CrawlerConsts.WORK_SCHEDULE),
+            recruitmentProcess = extractText(CrawlerConsts.RECRUITMENT_PROCESS),
+            applyMethod = extractText(CrawlerConsts.APPLY_METHOD),
+            requiredDocument = extractText(CrawlerConsts.REQUIRED_DOCUMENT),
+            centerName = extractText(CrawlerConsts.CENTER_NAME),
+            applyDeadline = extractApplyDeadline(CrawlerConsts.APPLY_DEADLINE),
+            workTime = extractWorkTime(CrawlerConsts.WORK_TIME),
+            centerAddress = extractAddress(
+                CrawlerConsts.CLIENT_ADDRESS1,
+                CrawlerConsts.CLIENT_ADDRESS2
+            ),
+            clientAddress = extractAddress(
+                CrawlerConsts.CENTER_ADDRESS1,
+                CrawlerConsts.CENTER_ADDRESS2,
+                CrawlerConsts.CENTER_ADDRESS3
+            ),
+            directUrl = driver.currentUrl
+        )
+    }
+
+
+    private inline fun <T> errorRecord(location: String, action: () -> T): T {
+        return runCatching { action() }
+            .getOrElse { e ->
+                logError(location)
+                throw e
+            }
+    }
+
+    private fun findElementSafe(by: By): WebElement? {
+        return runCatching { driver.findElement(by) }.getOrNull()
+    }
+
+    private fun moveToPostDetailWindow(titleElement: WebElement, originalWindow: String) {
+        titleElement.click()
+        WebDriverWait(driver, Duration.ofSeconds(10))
+            .until(ExpectedConditions.numberOfWindowsToBe(2))
+        driver.switchTo().window(driver.windowHandles.first { it != originalWindow })
+    }
+
+    private fun extractText(con: CrawlerConsts): String {
+        return errorRecord(con.location) { driver.findElement(By.xpath(con.value)).text }
+    }
+
+    private fun extractApplyDeadline(con: CrawlerConsts): String {
+        return errorRecord(con.location) {
+            driver.findElement(By.xpath(con.value)).text.let {
+                if (it.contains("채용시까지"))
+                    LocalDate.now().plusDays(15).format(DateTimeFormatter.ofPattern("yyyyMMdd"))
+                else
+                    it
+            }
+        }
+    }
+
+    private fun extractAddress(vararg cons: CrawlerConsts): String {
+        for (con in cons) {
+            runCatching {
+                val address = driver.findElement(By.xpath(con.value)).text
+                return address.replace("지도보기", "").trim().replace(Regex("\\(\\d{5}\\)"), "").trim()
+            } .getOrElse { e ->
+                logError(con.location)
+                throw e
+            }
+        }
+        throw NoSuchElementException("Center address not found using any of the provided XPaths")
+    }
+
+    private fun extractWorkTime(con: CrawlerConsts): String {
+        return errorRecord(con.location) {
+            driver.findElement(By.xpath(con.value)).text
+                .replace("도움말", "")
+                .replace("(근무시간)", "")
+                .replace("\n", "")
+        }
+    }
+
+    private fun logError(location: String) {
+        errorCountMap[location] = errorCountMap.getOrDefault(location, 0) + 1
+    }
+
+    private fun backWindow(originalWindow: String?) {
+        driver.close()
+        driver.switchTo().window(originalWindow)
+    }
+}