gqlgen go，通过添加一个解析器减少数据库调用

Question

我在解决导致性能下降的特定情况时遇到了一些麻烦。我很确定这是可以做到的，但我不知道该怎么做。

这是暴露问题的示例模式：

type Answer{

    answerId: String!

    text: String!

    topic: Topic!

}

type Topic {

    topicId: String!

    name: String!

    level: Int!

}

extend type Query {

    answer(answerId: String!): Answer!

    answers: [Answer!]! 

}

我遵循了文档，特别是这部分https://gqlgen.com/getting-started/#dont-eagerly-fetch-the-user 从我的架构中，它生成以下解析器：

func (r *queryResolver) Answer(ctx context.Context, answerId string) (*models.Answer, error) {

...

#Single Query which retrives single record of Answer from DB.

#Fills a model Answer with the Id and the text

#Proceeds by calling the Topic resolver

...

}

func (r *queryResolver) Answers(ctx context.Context) ([]*models.Answer, error) {

...

#Single Query which retrives list of Answers from DB

#Fills a list of model Answer with the Id and the text

-->#For each element of that list, it calls the Topic resolver

...

}

func (r *answerResolver) Topic(ctx context.Context, obj *models.Answer) (*models.Topic, error) {

...

#Single Query which retrives single record of Topic from DB

#Return a model Topic with id, name and level

...

}

当使用参数answer调用查询时，将触发解析器，它解析属性并调用解析器。解析器按预期工作，检索并将其合并到并返回。answerIdanswertextTopicTopicTopicAnswer

当answers查询在没有answerId参数的情况下被调用时，answer解析器被触发，它answers通过单个查询检索列表。然后，对于该列表的每个元素，它调用Topic解析器。Topic 检索 a并将其Topic合并到单 例中Answer并返回。

结果在这两种情况下都可以，但是answers如果我要求很多答案，查询就会成为性能问题。对于每个答案，Topic解析器都会被触发并执行查询以检索单个记录。

前任。如果我有 2 个答案 --> 1 个查询[Answer0, Answer1]，然后 1 个查询 Topic0和 1 个Topic1

前任。10 个答案 --> 1 个[Answer0, ..., Answer9]，然后每个 10 个TopicN

我想获得一些topic数组解析器，例如

func (r *answersResolver) Topics(ctx context.Context, obj *[]models.Answer) (*[]models.Topic, error) {

...

#Single Query which retrives list of Topics from DB

#Return a list of model Topic with id, name and level

...

}

我希望返回数组的每个元素都与数组的相应元素合并Answers。

有可能吗？我在哪里可以找到这种方法的例子？谢谢

Answer 1

可以使用数据加载器（文档）解决问题我必须为以下数据源实现Topics：package dataloaderimport (    "github.com/graph-gophers/dataloader")type ctxKey stringconst (    loadersKey = ctxKey("dataloaders"))type TopicReader struct {    conn *sql.DB}func (t *TopicReader) GetTopics(ctx context.Context, keys dataloader.Keys) []*dataloader.Result {    topicIDs := make([]string, len(keys))    for ix, key := range keys {        topicIDs[ix] = key.String()    }    res := u.db.Exec(        r.Conn,        "SELECT id, name, level        FROM topics        WHERE id IN (?" + strings.Repeat(",?", len(topicIDs-1)) + ")",        topicIDs...,    )    defer res.Close()    output := make([]*dataloader.Result, len(keys))    for index, _ := range keys {            output[index] = &dataloader.Result{Data: res[index], Error: nil}    }    return output}type Loaders struct {    TopicLoader *dataloader.Loader}func NewLoaders(conn *sql.DB) *Loaders {    topicReader := &TopicReader{conn: conn}    loaders := &Loaders{        TopicLoader: dataloader.NewBatchedLoader(t.GetTopics),    }    return loaders}func Middleware(loaders *Loaders, next http.Handler) http.Handler {    return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {        nextCtx := context.WithValue(r.Context(), loadersKey, loaders)        r = r.WithContext(nextCtx)        next.ServeHTTP(w, r)    })}func For(ctx context.Context) *Loaders {    return ctx.Value(loadersKey).(*Loaders)}func GetTopic(ctx context.Context, topicID string) (*model.Topic, error) {    loaders := For(ctx)    thunk := loaders.TopicLoader.Load(ctx, dataloader.StringKey(topicID))    result, err := thunk()    if err != nil {        return nil, err    }    return result.(*model.Topic), nil}