Merge pull request #81 from Block-Guard/feat/#76/fraud-type-news-api

[Refactor] 뉴스 api 정확도 개선

Author: chaeyuuu

src/main/java/com/blockguard/server/domain/news/domain/enums/Category.java

@@ -51,23 +51,4 @@ public static Category from(String input) {
         return Optional.ofNullable(MAPPING.get(input.trim()))
                 .orElse(ETC);
     }
-        /*return switch (input.trim()) {
-            case "보이스피싱", "보이스 피싱" -> VOICE_PHISHING;
-            case "스미싱" -> SMISHING;
-            case "메신저 피싱", "메신저피싱" -> MESSAGE_VOICE_PHISHING;
-
-            case "기관 사칭형" -> INSTITUTION_IMPERSONATION;
-            case "대출 사기형" -> LOAN_FRAUD;
-            case "카드사 사칭형" -> CARD_IMPERSONATION;
-            case "가족/지인 사칭형" -> FAMILY_IMPERSONATION;
-            case "경조사 사칭형" -> EVENT_IMPERSONATION;
-            case "공공기관 사칭형" -> PUBLIC_IMPERSONATION;
-            case "알바/부업 사기형" -> PART_TIME_SCAM;
-            case "정부지원금 위장형" -> GOVERNMENT_GRANT_SCAM;
-            case "택배 사기형" -> DELIVERY_SCAM;
-            case "투자 사기형" -> INVESTMENT_SCAM;
-            case "허위결제 사기형" -> FALSE_PAYMENT_SCAM;
-            default -> ETC;
-        };*/
-
 }

src/main/java/com/blockguard/server/infra/crawler/DaumNewsCrawler.java

@@ -20,6 +20,7 @@
 import java.time.LocalDateTime;
 import java.time.format.DateTimeFormatter;
 import java.util.Optional;
+import java.util.regex.Pattern;
 
 @Service
 @RequiredArgsConstructor
@@ -29,6 +30,16 @@ public class DaumNewsCrawler {
     private static final int ARTICLE_RETENTION_DAYS = 365;
     private static final long CRAWL_DELAY_MS = 500L;
 
+    private static final Pattern NON_KO_ALNUM = Pattern.compile("[^가-힣a-zA-Z0-9]");
+    private String norm(String s) {
+        if (s == null) return "";
+        return NON_KO_ALNUM.matcher(s).replaceAll("").toLowerCase();
+    }
+
+    private boolean titleContains(String title, String keyword) {
+        return norm(title).contains(norm(keyword));
+    }
+
     public void fetchNewsFromDaum(String keyword, Category forceCategory) {
         Category category = (forceCategory != null) ? forceCategory : Category.from(keyword);
         int savedCount = 0;
@@ -53,6 +64,7 @@ public void fetchNewsFromDaum(String keyword, Category forceCategory) {
                     String title = titleEl.text();
                     String url = titleEl.attr("href");
 
+                    if (!titleContains(title, keyword)) continue;
                     if (newsRepository.existsByUrl(url)) continue;
 
                     Element imageEl = item.selectFirst("a.thumb_bf img");