Spring Boot整合Postgres實現輕量級全文搜索

有這樣一個帶有搜索功能的用戶界面需求：

搜索流程如下所示：

這個需求涉及兩個實體：

• “評分（Rating）、用戶名（Username）”數據與User實體相關
• “創建日期（create date）、觀看次數（number of views）、標題（title）、正文（body）”與Story實體相關

需要支持的功能對User實體中的評分（Rating）的頻繁修改以及下列搜索功能：

• 按User評分進行範圍搜索
• 按Story創建日期進行範圍搜索
• 按Story瀏覽量進行範圍搜索
• 按Story標題進行全文搜索
• 按Story正文進行全文搜索

Postgres中創建表結構和索引

創建users表和stories表以及對應搜索需求相關的索引，包括：

• 使用 btree 索引來支持按User評分搜索
• 使用 btree 索引來支持按Story創建日期、查看次數的搜索
• 使用 gin 索引來支持全文搜索內容（同時創建全文搜索列fulltext，類型使用tsvector以支持全文搜索）

具體創建腳本如下：

--Create Users table
CREATE TABLE IF NOT EXISTS users
(
  id bigserial NOT NULL,
  name character varying(100) NOT NULL,
rating integer,
PRIMARY KEY (id)
)
;
CREATE INDEX usr_rating_idx
ON users USING btree
(rating ASC NULLS LAST)
TABLESPACE pg_default
;

--Create Stories table
CREATE TABLE  IF NOT EXISTS stories
(
    id bigserial NOT NULL,
    create_date timestamp without time zone NOT NULL,
    num_views bigint NOT NULL,
    title text NOT NULL,
    body text NOT NULL,
    fulltext tsvector,
    user_id bigint,
    PRIMARY KEY (id),
CONSTRAINT user_id_fk FOREIGN KEY (user_id)
REFERENCES users (id) MATCH SIMPLE
ON UPDATE NO ACTION
ON DELETE NO ACTION
NOT VALID
)
;
CREATE INDEX str_bt_idx
ON stories USING btree
(create_date ASC NULLS LAST,
num_views ASC NULLS LAST, user_id ASC NULLS LAST)
;

CREATE INDEX fulltext_search_idx
ON stories USING gin
(fulltext)
;

創建Spring Boot應用

1. 項目依賴關係（這裏使用Gradle構建）：

plugins {
   id 'java'
   id 'org.springframework.boot' version '3.1.3'
   id 'io.spring.dependency-management' version '1.1.3'
}

group = 'com.example'
version = '0.0.1-SNAPSHOT'

java {
   sourceCompatibility = '17'
}

repositories {
   mavenCentral()
}

dependencies {
   implementation 'org.springframework.boot:spring-boot-starter-data-jdbc'
   implementation 'org.springframework.boot:spring-boot-starter-web'
   runtimeOnly 'org.postgresql:postgresql'
   testImplementation 'org.springframework.boot:spring-boot-starter-test'
}

tasks.named('test') {
   useJUnitPlatform()
}

2. application.yaml中配置數據庫連接信息

spring:
  datasource: 
    url: jdbc:postgresql://localhost:5432/postgres
    username: postgres
    password: postgres

3. 數據模型

定義需要用到的各種數據模型：

public record Period(String fieldName, LocalDateTime min, LocalDateTime max) {
}

public record Range(String fieldName, long min, long max) {
}

public record Search(List<Period> periods, List<Range> ranges, String fullText, long offset, long limit) {
}

public record UserStory(Long id, LocalDateTime createDate, Long numberOfViews,
                        String title, String body, Long userRating, String userName, Long userId) {
}

這裏使用Java 16推出的新特性record實現，所以代碼非常簡潔。如果您還不瞭解的話，可以前往程序猿DD的Java新特性專欄補全一下知識點。

4. 數據訪問（Repository）

@Repository
public class UserStoryRepository {

    private final JdbcTemplate jdbcTemplate;


    @Autowired
    public UserStoryRepository(JdbcTemplate jdbcTemplate) {
        this.jdbcTemplate = jdbcTemplate;
    }

    public List<UserStory> findByFilters(Search search) {
        return jdbcTemplate.query(
                """
                  SELECT s.id id, create_date, num_views, 
                         title, body, user_id, name user_name, 
                         rating user_rating 
                  FROM stories s INNER JOIN users u 
                      ON s.user_id = u.id
                  WHERE true
                """ + buildDynamicFiltersText(search)
                        + " order by create_date desc offset ? limit ?",
                (rs, rowNum) -> new UserStory(
                        rs.getLong("id"),
                        rs.getTimestamp("create_date").toLocalDateTime(),
                        rs.getLong("num_views"),
                        rs.getString("title"),
                        rs.getString("body"),
                        rs.getLong("user_rating"),
                        rs.getString("user_name"),
                        rs.getLong("user_id")
                ),
                buildDynamicFilters(search)
        );
    }

    public void save(UserStory userStory) {
        var keyHolder = new GeneratedKeyHolder();

        jdbcTemplate.update(connection -> {
            PreparedStatement ps = connection
                .prepareStatement(
                    """
                      INSERT INTO stories (create_date, num_views, title, body, user_id)
                          VALUES (?, ?, ?, ?, ?)
                    """,
                    Statement.RETURN_GENERATED_KEYS
            );
            ps.setTimestamp(1, Timestamp.valueOf(userStory.createDate()));
            ps.setLong(2, userStory.numberOfViews());
            ps.setString(3, userStory.title());
            ps.setString(4, userStory.body());
            ps.setLong(5, userStory.userId());

            return ps;
        }, keyHolder);

        var generatedId = (Long) keyHolder.getKeys().get("id");

        if (generatedId != null) {
            updateFullTextField(generatedId);
        }
    }

    private void updateFullTextField(Long generatedId) {
        jdbcTemplate.update(
            """
              UPDATE stories SET fulltext = to_tsvector(title || ' ' || body)
              where id = ?
            """,
            generatedId
        );
    }

    private Object[] buildDynamicFilters(Search search) {
        var filtersStream = search.ranges().stream()
                .flatMap(
                    range -> Stream.of((Object) range.min(), range.max())
                );

        var periodsStream = search.periods().stream()
                .flatMap(
                    range -> Stream.of((Object) Timestamp.valueOf(range.min()), Timestamp.valueOf(range.max()))
                );

        filtersStream = Stream.concat(filtersStream, periodsStream);

        if (!search.fullText().isBlank()) {
            filtersStream = Stream.concat(filtersStream, Stream.of(search.fullText()));
        }

        filtersStream = Stream.concat(filtersStream, Stream.of(search.offset(), search.limit()));

        return filtersStream.toArray();
    }

    private String buildDynamicFiltersText(Search search) {
        var rangesFilterString =
                Stream.concat(
                  search.ranges()
                        .stream()
                        .map(
                            range -> String.format(" and %s between ? and ? ", range.fieldName())
                        ),
                  search.periods()
                        .stream()
                        .map(
                            range -> String.format(" and %s between ? and ? ", range.fieldName())
                        )
                  )
                  .collect(Collectors.joining(" "));

        return rangesFilterString + buildFulltextFilterText(search.fullText());
    }

    private String buildFulltextFilterText(String fullText) {
        return fullText.isBlank() ? "" : " and fulltext @@ plainto_tsquery(?) ";
    }
}

5. Controller實現

@RestController
@RequestMapping("/user-stories")
public class UserStoryController {
    private final UserStoryRepository userStoryRepository;

    @Autowired
    public UserStoryController(UserStoryRepository userStoryRepository) {
        this.userStoryRepository = userStoryRepository;
    }

    @PostMapping
    public void save(@RequestBody UserStory userStory) {
        userStoryRepository.save(userStory);
    }

    @PostMapping("/search")
    public List<UserStory> search(@RequestBody Search search) {
        return userStoryRepository.findByFilters(search);
    }
}

小結

本文介紹瞭如何在Spring Boot中結合Postgres數據庫實現全文搜索的功能，該方法比起使用Elasticsearch更爲輕量級，非常適合一些小項目場景使用。希望本文內容對您有所幫助。如果您學習過程中如遇困難？可以加入我們超高質量的Spring技術交流羣，參與交流與討論，更好的學習與進步！更多Spring Boot教程可以點擊直達！，歡迎收藏與轉發支持！

參考資料

• Postgres full-text search Spring boot integration
• Java 16新特性：record

歡迎關注我的公衆號：程序猿DD。第一時間瞭解前沿行業消息、分享深度技術乾貨、獲取優質學習資源