Spring Boot 4 Complete Guide · 课时

分页、排序与切片流式处理

为大型数据集和无限滚动界面高效返回分页、排序和流式处理的结果集。

分页、排序与切片流式处理 是 CoddyKit 上的免费 Spring Boot 4 Complete Guide 课时。这是第 4 节课，共 4 节。你可以在下方免费阅读本课时的完整内容 — 然后在浏览器中使用内置代码编辑器和全天候 AI 导师进行实践。这是 Spring Boot 4 Complete Guide 学习路径的一部分，你的进度在网页和 CoddyKit 应用中同步。 Spring Boot 4 Complete Guide 课程共包含 4 节课。

本课时的部分内容尚未翻译，以英文显示。

Why Pagination Matters

Loading an entire table into memory is one of the fastest ways to crash a service. A findAll() on a million-row table builds a giant list, exhausts the heap, and blocks the request thread.

Pagination solves this by fetching results in small chunks (pages). Spring Data JPA gives you first-class support through the Pageable abstraction.

Page — knows the total element count and total page count.
Slice — only knows whether a next chunk exists (cheaper).
Stream — processes rows one-by-one without materializing a list.

In this lesson you will learn when to reach for each one.

The Pageable Parameter

Add a Pageable parameter to any repository method and Spring Data appends LIMIT and OFFSET (or the dialect equivalent) to the generated query automatically.

The method returns a Page<T>, which wraps the content list plus paging metadata.

public interface ProductRepository extends JpaRepository<Product, Long> {

    Page<Product> findByCategory(String category, Pageable pageable);
}

Building a PageRequest

The concrete implementation of Pageable is PageRequest. You create one with a zero-based page number and a page size.

PageRequest.of(0, 20) — first page, 20 items.
PageRequest.of(2, 20) — third page (rows 40-59).

In a controller you usually let Spring resolve it from the request automatically, but you can also build it by hand in a service.

@Service
public class ProductService {

    private final ProductRepository repository;

    public ProductService(ProductRepository repository) {
        this.repository = repository;
    }

    public Page<Product> firstPage(String category) {
        Pageable pageable = PageRequest.of(0, 20);
        return repository.findByCategory(category, pageable);
    }
}

Reading Page Metadata

A Page<T> exposes everything a UI needs to render pagination controls.

getContent() — the rows on this page.
getTotalElements() — total matching rows in the table.
getTotalPages() — total number of pages.
getNumber() — current zero-based page index.
hasNext() / hasPrevious() — navigation flags.

Important: computing getTotalElements() requires an extra COUNT(*) query on every call.

Page<Product> page = repository.findByCategory("books", PageRequest.of(0, 20));

List<Product> rows = page.getContent();
long total = page.getTotalElements();
int totalPages = page.getTotalPages();
boolean more = page.hasNext();

Sorting with Sort

Sorting is part of Pageable. Pass a Sort object into PageRequest.of(page, size, sort) and Spring Data adds an ORDER BY clause.

Sort.by("price") — ascending by price.
Sort.by("price").descending() — descending.
Chain multiple orders for tie-breaking.

Sort sort = Sort.by("price").descending()
                .and(Sort.by("name").ascending());

Pageable pageable = PageRequest.of(0, 20, sort);
Page<Product> page = repository.findByCategory("books", pageable);

Multi-Field Sort with Order

For finer control over null handling and direction per field, build the Sort from Sort.Order objects.

This is the clearest way to express something like "newest first, then alphabetical, nulls last".

Sort sort = Sort.by(
    Sort.Order.desc("createdAt"),
    Sort.Order.asc("name").nullsLast()
);

Pageable pageable = PageRequest.of(0, 25, sort);
Page<Product> page = repository.findAll(pageable);

Slice vs Page

A Page always runs an extra COUNT(*) to know the total. For an infinite-scroll UI you rarely need the total — you only need to know if there is a next chunk.

Return Slice<T> instead. Spring fetches size + 1 rows internally: if the extra row exists, hasNext() is true. No count query runs, so it is noticeably cheaper on large tables.

Slice.getContent() and Slice.hasNext() work just like Page.
Slice.getTotalElements() does not exist.

public interface FeedRepository extends JpaRepository<Post, Long> {

    Slice<Post> findByAuthorId(Long authorId, Pageable pageable);
}

Consuming a Slice for Infinite Scroll

On the client you keep incrementing the page number while hasNext() stays true. The service stays clean because the repository handles the size + 1 trick.

public Slice<Post> nextChunk(Long authorId, int pageNumber) {
    Pageable pageable = PageRequest.of(
        pageNumber, 15, Sort.by("createdAt").descending());

    Slice<Post> slice = feedRepository.findByAuthorId(authorId, pageable);

    if (slice.hasNext()) {
        // tell the UI to request pageNumber + 1
    }
    return slice;
}

The Offset Problem

Offset pagination has a hidden cost: to return page 10000 with size 20, the database must scan and discard 200000 rows before returning yours. Deep pages get slower and slower.

Keyset (cursor) pagination avoids this. Instead of OFFSET, you filter on the last seen value. The query stays fast at any depth because it can use an index seek.

public interface PostRepository extends JpaRepository<Post, Long> {

    @Query("SELECT p FROM Post p WHERE p.id < :lastId ORDER BY p.id DESC")
    Slice<Post> findOlderThan(@Param("lastId") Long lastId, Pageable pageable);
}

Streaming Large Result Sets

When you must process every row (export, batch job, report) but cannot hold them all in memory, return a Stream<T>. JPA reads rows lazily from a forward-only cursor.

Two rules are non-negotiable:

The method must run inside a transaction (@Transactional) so the cursor stays open.
You must close the stream — use try-with-resources — to release the DB cursor.

public interface OrderRepository extends JpaRepository<Order, Long> {

    @QueryHints(@QueryHint(name = HINT_FETCH_SIZE, value = "100"))
    @Query("SELECT o FROM Order o WHERE o.status = :status")
    Stream<Order> streamByStatus(@Param("status") String status);
}

Draining a Stream Safely

Wrap the stream in try-with-resources and keep the whole consumption inside one transactional method. Detach or clear entities periodically if you mutate them, so the persistence context does not grow unbounded.

@Service
public class OrderExporter {

    private final OrderRepository repository;

    public OrderExporter(OrderRepository repository) {
        this.repository = repository;
    }

    @Transactional(readOnly = true)
    public long exportPending(Consumer<Order> writer) {
        long count = 0;
        try (Stream<Order> stream = repository.streamByStatus("PENDING")) {
            for (Order order : (Iterable<Order>) stream::iterator) {
                writer.accept(order);
                count++;
            }
        }
        return count;
    }
}

Quick Check: Choosing the Right Return Type

You are building an infinite-scroll feed over a 5-million-row table. The UI never shows a total count, only a "load more" button. Which repository return type is the best fit?

Recap

You now have a toolbox for handling large result sets in Spring Data JPA:

Page<T> — content plus total counts; pays for a COUNT(*) query. Use when the UI shows total pages.
Slice<T> — content plus hasNext(); no count query. Best for infinite scroll.
Sort — combine with PageRequest.of(page, size, sort) for ordered, deterministic pages.
Keyset pagination — filter on the last seen value to keep deep pages fast.
Stream<T> — lazy, forward-only processing inside @Transactional with try-with-resources for huge batch jobs.

Match the return type to the access pattern, and your queries stay fast and memory-safe at any scale.

常见问题解答

「分页、排序与切片流式处理」课时是免费的吗？

是的 — 「分页、排序与切片流式处理」的完整文本可在网页上免费阅读。要进行交互式练习（内置代码编辑器和全天候 AI 导师）并解锁 Spring Boot 4 Complete Guide 课程的其余内容，请升级到 CoddyKit PRO。 Spring Boot 4 Complete Guide 课程共包含 4 节课。

「分页、排序与切片流式处理」这节课中我会学到什么？

为大型数据集和无限滚动界面高效返回分页、排序和流式处理的结果集。你通过在浏览器中直接运行的动手代码来练习 Spring Boot 4 Complete Guide，全天候 AI 导师会在你学习这节课的过程中回答你的问题。

学习 Spring Boot 4 Complete Guide 需要有经验吗？

无需任何先前经验。CoddyKit 上的 Spring Boot 4 Complete Guide 课程适合初学者到高级学习者，你可以从这里开始或从头开始，按照自己的节奏学习。这是第 4 节课，共 4 节。

「分页、排序与切片流式处理」课时需要多长时间？

大多数 CoddyKit 课程大约需要 5–10 分钟。每节课都很精短且互动，所以你能稳步进步，并在网页和应用中从离开的地方继续。

我能在这节 Spring Boot 4 Complete Guide 课中编写并运行代码吗？

能。每节 Spring Boot 4 Complete Guide 课都包含内置代码编辑器，你可以在浏览器中直接编写并运行真实代码，并获得即时 AI 反馈 — 无需本地设置。

此课程中的所有课时

← 返回 Spring Boot 4 Complete Guide