Feat : 학교 홈페이지 리뉴얼로인한 교직원 스크랩 및 스크랩간 직위 추가

src/main/java/com/kustacks/kuring/common/utils/converter/EmailSupporter.java

@@ -0,0 +1,68 @@
+package com.kustacks.kuring.common.utils.converter;
+
+import java.util.Arrays;
+import java.util.regex.Pattern;
+
+public class EmailSupporter {
+    private static final Pattern AT_PATTERN = Pattern.compile("\\s+at\\s+");
+    private static final Pattern DOT_PATTERN = Pattern.compile("\\s+dot\\s+");
+    private static final Pattern EMAIL_PATTERN = Pattern.compile("^[a-zA-Z0-9_!#$%&'\\*+/=?{|}~^.-]+@[a-zA-Z0-9.-]+$");
+
+    private static final String KONKUK_DOMAIN = "@konkuk.ac.kr";
+    private static final String EMPTY_EMAIL = "";
+
+    public static boolean isNullOrBlank(String email) {
+        return email == null || email.isBlank();
+    }
+
+    public static String convertValidEmail(String email) {
+        if (isNullOrBlank(email)) {
+            return EMPTY_EMAIL;
+        }
+
+        String[] emailGroups = splitEmails(email);
+        String[] normalizedEmails = normalizeEmails(emailGroups);
+
+        //여러 이메일 중 konkuk을 우선 선택, 없으면 첫번째 내용
+        return selectPreferredEmail(normalizedEmails);
+    }
+
+    private static String[] splitEmails(String email) {
+        return email.split("[/,]");
+    }
+
+    private static String[] normalizeEmails(String[] emailGroups) {
+        return Arrays.stream(emailGroups)
+                .map(EmailSupporter::normalizeEmail)
+                .toArray(String[]::new);
+    }
+
+    private static String normalizeEmail(String email) {
+        if (EMAIL_PATTERN.matcher(email).matches()) {
+            return email;
+        }
+
+        if (containsSubstitutePatterns(email)) {
+            return replaceSubstitutePatterns(email);
+        }
+
+        return EMPTY_EMAIL;
+    }
+
+    private static String replaceSubstitutePatterns(String email) {
+        return email.replaceAll(DOT_PATTERN.pattern(), ".")
+                .replaceAll(AT_PATTERN.pattern(), "@");
+    }
+
+    private static boolean containsSubstitutePatterns(String email) {
+        return DOT_PATTERN.matcher(email).find() && AT_PATTERN.matcher(email).find();
+    }
+
+    // Konkuk 도메인 우선 선택
+    private static String selectPreferredEmail(String[] emails) {
+        return Arrays.stream(emails)
+                .filter(email -> email.endsWith(KONKUK_DOMAIN))
+                .findFirst()
+                .orElseGet(() -> emails.length > 0 ? emails[0] : EMPTY_EMAIL);
+    }
+}

src/main/java/com/kustacks/kuring/common/utils/converter/PhoneNumberSupporter.java

@@ -0,0 +1,42 @@
+package com.kustacks.kuring.common.utils.converter;
+
+import java.util.regex.Pattern;
+
+public class PhoneNumberSupporter {
+
+    private static final Pattern LAST_FOUR_NUMBER_PATTERN = Pattern.compile("\\d{4}");
+    private static final Pattern FULL_NUMBER_PATTERN = Pattern.compile("02-\\d{3,4}-\\d{4}");
+    private static final Pattern FULL_NUMBER_WITH_PARENTHESES_PATTERN = Pattern.compile("02[)]\\d{3,4}-\\d{4}");
+
+    private static final String EMPTY_PHONE = "";
+
+    public static boolean isNullOrBlank(String number) {
+        return number == null || number.isBlank();
+    }
+
+    public static String convertFullExtensionNumber(String number) {
+        if (isNullOrBlank(number)) {
+            return EMPTY_PHONE;
+        }
+
+        if (FULL_NUMBER_PATTERN.matcher(number).matches()) {
+            return number;
+        }
+        if (containsLastFourNumber(number)) {
+            return "02-450-" + number;
+        }
+        if (containsParenthesesPattern(number)) {
+            return number.replace(")", "-");
+        }
+
+        return EMPTY_PHONE;
+    }
+
+    private static boolean containsLastFourNumber(String number) {
+        return LAST_FOUR_NUMBER_PATTERN.matcher(number).matches();
+    }
+
+    private static boolean containsParenthesesPattern(String number) {
+        return FULL_NUMBER_WITH_PARENTHESES_PATTERN.matcher(number).matches();
+    }
+}

src/main/java/com/kustacks/kuring/staff/domain/Email.java

@@ -30,7 +30,8 @@ public Email(String email) {
     }
 
     private boolean isValidEmail(String email) {
-        return !Objects.isNull(email) && patternMatches(email);
+        return Objects.nonNull(email) &&
+                (patternMatches(email) || Objects.equals(email,""));
     }
 
     private boolean patternMatches(String email) {

src/main/java/com/kustacks/kuring/staff/domain/Phone.java

@@ -22,13 +22,14 @@ public class Phone {
             = Pattern.compile("(\\d{3,4})[-\\s]*(\\d{4})");
     private static final String SEOUL_AREA_CODE = "02";
     private static final String DELIMITER = "-";
+    private static final String EMPTY_NUMBER = "";
 
     @Column(name = "phone", length = 64)
     private String value;
 
     public Phone(String phone) {
         if(isEmptyNumbers(phone)) {
-            this.value = DELIMITER;
+            this.value = EMPTY_NUMBER;
             return;
         }
 
@@ -71,7 +72,7 @@ private boolean isValidNumbersAndSet(String phone) {
     }
 
     private static boolean isEmptyNumbers(String phone) {
-        return phone == null || phone.isBlank() || phone.equals(DELIMITER);
+        return phone == null || phone.isBlank();
     }
 
     public boolean isSameValue(String phone) {

src/main/java/com/kustacks/kuring/staff/domain/Staff.java

@@ -29,6 +29,10 @@ public class Staff {
     @Column(name = "lab", length = 64)
     private String lab;
 
+    @Getter(AccessLevel.PUBLIC)
+    @Column(name = "position", length = 64)
+    private String position;
+
     @Embedded
     private Phone phone;
 
@@ -45,24 +49,26 @@ public class Staff {
     private College college;
 
     @Builder
-    private Staff(String name, String major, String lab, String phone, String email, String dept, String college) {
+    private Staff(String name, String major, String lab, String phone, String email, String dept, String college, String position) {
         this.name = new Name(name);
         this.major = major;
         this.lab = lab;
         this.phone = new Phone(phone);
         this.email = new Email(email);
         this.dept = dept;
         this.college = College.valueOf(college);
+        this.position = position;
     }
 
-    public void updateInformation(String name, String major, String lab, String phone, String email, String deptName, String college) {
+    public void updateInformation(String name, String major, String lab, String phone, String email, String deptName, String college, String position) {
         this.name = new Name(name);
         this.major = major;
         this.lab = lab;
         this.phone = new Phone(phone);
         this.email = new Email(email);
         this.dept = deptName;
         this.college = College.valueOf(college);
+        this.position = position;
     }
 
     public String getEmail() {
@@ -105,6 +111,14 @@ public boolean isSameCollege(String collegeName) {
         return this.college == College.valueOf(collegeName);
     }
 
+    public boolean isSamePosition(String position) {
+        return this.position.equals(position);
+    }
+
+    public String identifier() {
+        return String.join(",", getName(), position, dept);
+    }
+
     @Override
     public boolean equals(Object o) {
         if (this == o) return true;

src/main/java/com/kustacks/kuring/worker/parser/staff/EachDeptStaffHtmlParser.java

@@ -1,8 +1,6 @@
 package com.kustacks.kuring.worker.parser.staff;
 
 import com.kustacks.kuring.worker.scrap.deptinfo.DeptInfo;
-import com.kustacks.kuring.worker.scrap.deptinfo.art_design.CommunicationDesignDept;
-import com.kustacks.kuring.worker.scrap.deptinfo.art_design.LivingDesignDept;
 import com.kustacks.kuring.worker.scrap.deptinfo.real_estate.RealEstateDept;
 import lombok.NoArgsConstructor;
 import lombok.extern.slf4j.Slf4j;
@@ -18,33 +16,22 @@ public class EachDeptStaffHtmlParser extends StaffHtmlParserTemplate {
 
     @Override
     public boolean support(DeptInfo deptInfo) {
-        return !(deptInfo instanceof RealEstateDept) &&
-                !(deptInfo instanceof LivingDesignDept) &&
-                !(deptInfo instanceof CommunicationDesignDept);
+        return !(deptInfo instanceof RealEstateDept);
     }
 
     protected Elements selectStaffInfoRows(Document document) {
-        Element table = document.select(".photo_intro").get(0);
-        return table.getElementsByTag("dl");
+        return document.select(".row");
     }
 
     protected String[] extractStaffInfoFromRow(Element row) {
-        Elements infos = row.getElementsByTag("dd");
-
-        // 교수명, 직위, 세부전공, 연구실, 연락처, 이메일 순으로 파싱
-        // 연구실, 연락처 정보는 없는 경우가 종종 있으므로, childNode접근 전 인덱스 체크하는 로직을 넣었음
-        String name = infos.get(0).getElementsByTag("span").get(1).text();
-
-        String jobPosition = String.valueOf(infos.get(1).childNodeSize() < 2 ? "" : infos.get(1).childNode(1));
-        if (jobPosition.contains("명예") || jobPosition.contains("대우") || jobPosition.contains("휴직") || !jobPosition.contains("교수")) {
-            log.info("스크래핑 스킵 -> {} 교수", name);
-            return new String[]{};
-        }
-
-        String major = infos.get(2).childNodeSize() < 2 ? "" : String.valueOf(infos.get(2).childNode(1));
-        String lab = infos.get(3).childNodeSize() < 2 ? "" : String.valueOf(infos.get(3).childNode(1));
-        String phone = infos.get(4).childNodeSize() < 2 ? "" : String.valueOf(infos.get(4).childNode(1));
-        String email = infos.get(5).getElementsByTag("a").get(0).text();
-        return new String[]{name, major, lab, phone, email};
+        String name = row.select(".info .title .name").text();
+
+        Elements detailElement = row.select(".detail");
+        String jobPosition = detailElement.select(".ico1 dd").text().trim();
+        String major = detailElement.select(".ico2 dd").text().trim();
+        String lab = detailElement.select(".ico3 dd").text().trim();
+        String extensionNumber = detailElement.select(".ico4 dd").text().trim();
+        String email = detailElement.select(".ico5 dd").text().trim();
+        return new String[]{name, jobPosition, major, lab, extensionNumber, email};
     }
 }

src/main/java/com/kustacks/kuring/worker/parser/staff/RealEstateStaffHtmlParser.java

@@ -14,23 +14,20 @@ public class RealEstateStaffHtmlParser extends StaffHtmlParserTemplate {
     public boolean support(DeptInfo deptInfo) {
         return deptInfo instanceof RealEstateDept;
     }
-
     protected Elements selectStaffInfoRows(Document document) {
-        Element table = document.select(".sub0201_list").get(0).getElementsByTag("ul").get(0);
-        return table.getElementsByTag("li");
+        return document.select(".row");
     }
 
     protected String[] extractStaffInfoFromRow(Element row) {
-        Element content = row.select(".con").get(0);
-
-        String name = content.select("dl > dt > a > strong").get(0).text();
-        String major = String.valueOf(content.select("dl > dd").get(0).childNode(4)).replaceFirst("\\s", "").trim();
-
-        Element textMore = content.select(".text_more").get(0);
-
-        String lab = String.valueOf(textMore.childNode(4)).split(":")[1].replaceFirst("\\s", "").trim();
-        String phone = String.valueOf(textMore.childNode(6)).split(":")[1].replaceFirst("\\s", "").trim();
-        String email = textMore.getElementsByTag("a").get(0).text();
-        return new String[]{name, major, lab, phone, email};
+        String name = row.select(".info .title .name").text();
+
+        Elements detalTagElement = row.select(".detail");
+        String jobPosition = detalTagElement.select("dt:contains(직위) + dd").text();
+        String major = detalTagElement.select("dt:contains(연구분야) + dd").text().trim();
+        String lab = detalTagElement.select("dt:contains(연구실) + dd").text().trim();
+        String extensionNumber = detalTagElement.select("dt:contains(연락처) + dd").text().trim();
+        String email = detalTagElement.select("dt:contains(이메일) + dd").text().trim();
+        return new String[]{name, jobPosition, major, lab, extensionNumber, email};
     }
 }
+

src/main/java/com/kustacks/kuring/worker/scrap/StaffScraper.java

@@ -60,10 +60,11 @@ private static List<StaffDto> convertStaffDtos(DeptInfo deptInfo, List<String[]>
         return parseResult.stream()
                 .map(oneStaffInfo -> StaffDto.builder()
                         .name(oneStaffInfo[0])
-                        .major(oneStaffInfo[1])
-                        .lab(oneStaffInfo[2])
-                        .phone(oneStaffInfo[3])
-                        .email(oneStaffInfo[4])
+                        .position(oneStaffInfo[1])
+                        .major(oneStaffInfo[2])
+                        .lab(oneStaffInfo[3])
+                        .phone(oneStaffInfo[4])
+                        .email(oneStaffInfo[5])
                         .deptName(deptInfo.getDeptName())
                         .collegeName(deptInfo.getCollegeName()
                         ).build()