kurious/internal/infrastructure/interfaceadapters/courses/sravni/client.go

package sravni

import (
	"context"
	"encoding/json"
	"fmt"
	"io"
	"log/slog"
	"strings"

	"git.loyso.art/frx/kurious/internal/domain"

	"github.com/go-resty/resty/v2"
	"golang.org/x/net/html"
	"golang.org/x/net/html/atom"
)

const (
	baseURL = "https://www.sravni.ru/kursy"
)

func NewClient(ctx context.Context, log *slog.Logger, debug bool) (c *client, err error) {
	c = &client{
		log: log.With(slog.String("client", "sravni")),
		http: resty.New().
			SetBaseURL(baseURL).
			SetDebug(debug),
	}

	c.cachedMainPageInfo, err = c.getMainPageState(ctx)
	if err != nil {
		return nil, err
	}

	return c, nil
}

type client struct {
	log  *slog.Logger
	http *resty.Client

	cachedMainPageInfo *PageState
}

type PageStateRuntimeConfig struct {
	BrandingURL      string `json:"brandingUrl"`
	Release          string `json:"release"`
	Environment      string `json:"environment"`
	Gateway          string `json:"gatewayUrl"`
	APIGatewayURL    string `json:"apiGatewayUrl"`
	EducationURL     string `json:"educationUrl"`
	PhoneVerifierURL string `json:"phoneVerifierUrl"`
	WebPath          string `json:"webPath"`
	ServiceName      string `json:"serviceName"`
	OrgnazationURL   string `json:"organizationsUrl"`
}

type Link struct {
	URL   string `json:"url"`
	Title string `json:"title"`
}

type ReduxStatePrefooterItem struct {
	Title string `json:"title"`
	Links []Link `json:"links"`
}

type ReduxMetadata struct {
	Data struct {
		Prefooter []ReduxStatePrefooterItem `json:"prefooter"`
	} `json:"data"`
}

type InitialReduxState struct {
	Metadata   ReduxMetadata `json:"metadata"`
	Categories struct {
		Data map[string]int `json:"data"`
	} `json:"categories"`
}

type PageStateProperties struct {
	InitialReduxState InitialReduxState `json:"initialReduxState"`
}

type PageState struct {
	Page          string                 `json:"page"`
	Query         map[string]string      `json:"query"`
	BuildID       string                 `json:"buildId"`
	RuntimeConfig PageStateRuntimeConfig `json:"runtimeConfig"`
	Props         PageStateProperties    `json:"props"`
}

func (p *PageState) Clone() *PageState {
	copiedState := *p
	copiedState.Query = make(map[string]string, len(p.Query))
	for k, v := range p.Query {
		copiedState.Query[k] = v
	}

	data := p.Props.InitialReduxState.Categories.Data
	copiedData := make(map[string]int, len(data))
	for k, v := range data {
		copiedData[k] = v
	}
	copiedState.Props.InitialReduxState.Categories.Data = copiedData

	return &copiedState
}

func (c *client) GetMainPageState() *PageState {
	return c.cachedMainPageInfo.Clone()
}

func (c *client) getMainPageState(ctx context.Context) (*PageState, error) {
	ctxLogger := restyCtxLogger{
		ctx: ctx,
		log: c.log,
	}

	req := c.http.R().
		SetContext(ctx).
		SetLogger(ctxLogger).
		SetDoNotParseResponse(true).
		EnableTrace()

	resp, err := req.Get("/")
	if err != nil {
		return nil, fmt.Errorf("getting request: %w", err)
	}

	if resp.IsError() {
		c.log.ErrorContext(ctx, "unable to proceed request", slog.String("body", string(resp.Body())))
		return nil, fmt.Errorf("got %d, but expected success: %w", resp.StatusCode(), domain.UnexpectedStatusError)
	}

	traceInfo := resp.Request.TraceInfo()
	c.log.InfoContext(ctx, "request proceeded", slog.Any("trace", traceInfo))

	return c.parsePageState(ctx, resp.RawBody())
}

func (c *client) parsePageState(ctx context.Context, body io.Reader) (*PageState, error) {
	page, err := html.Parse(body)
	if err != nil {
		return nil, fmt.Errorf("parsing body: %w", err)
	}

	c.log.DebugContext(ctx, "finding page state")

	var (
		foundHtml bool
		foundBody bool
	)
	findFunc := func(node *html.Node) (found, deeper bool) {
		if node == nil {
			c.log.DebugContext(ctx, "node is null, skipping")
			return false, false
		}

		if !foundHtml && node.Type == html.ElementNode {
			c.log.DebugContext(ctx, "found html node")
			foundHtml = true
			return false, true
		}

		if foundHtml && !foundBody && node.DataAtom == atom.Body {
			c.log.DebugContext(ctx, "found body node")
			foundBody = true
			return false, true
		}

		if foundHtml && foundBody && node.DataAtom == atom.Script {
			for _, attr := range node.Attr {
				if attr.Key == "id" && attr.Val == "__NEXT_DATA__" {
					c.log.DebugContext(ctx, "found script node with next_data")
					return true, false
				}
			}
		}

		return false, false
	}

	nextData := findNode(page, findFunc)
	if nextData == nil {
		return nil, nil
	}

	var out PageState
	dataReader := strings.NewReader(nextData.FirstChild.Data)
	err = json.NewDecoder(dataReader).Decode(&out)
	if err != nil {
		return nil, fmt.Errorf("decoding html data: %w", err)
	}
	return &out, nil
}

func findNode(parent *html.Node, eq func(*html.Node) (found, deeper bool)) *html.Node {
	for child := parent.FirstChild; child != nil; child = child.NextSibling {
		found, deeper := eq(child)
		if found {
			return child
		}
		if deeper {
			deeperChild := findNode(child, eq)
			if deeperChild != nil {
				return deeperChild
			}
		}
	}

	return nil
}