answer/pkg/extractor/extractor.go

package extractor

import (
	"context"
	"errors"

	"gitea.stevedudenhoeffer.com/steve/answer/pkg/cache"
)

var ErrFailedToExtract = errors.New("failed to extract")

type Article struct {
	URL   string
	Title string
	Body  string
}

// Extractor is an interface of systems that can extract the contents of
type Extractor interface {
	Extract(ctx context.Context, url string) (Article, error)
}

type multiExtractor struct {
	extractors []Extractor
}

var _ Extractor = multiExtractor{}

// Extract will try to extract the contents of a URL using all the extractors, and return the first successful result.
func (m multiExtractor) Extract(ctx context.Context, url string) (Article, error) {
	var errs []error
	for _, e := range m.extractors {
		article, err := e.Extract(ctx, url)
		if err == nil {
			return article, nil
		}

		if errors.Is(err, ErrFailedToExtract) {
			continue
		}

		errs = append(errs, err)
	}

	if len(errs) > 0 {
		return Article{}, errors.Join(errs...)
	}
	return Article{}, ErrFailedToExtract
}

func MultiExtractor(e ...Extractor) Extractor {
	return multiExtractor{extractors: e}
}

type CacheExtractor struct {
	Cache     cache.Cache
	Tag       string
	Extractor Extractor
}

var _ Extractor = CacheExtractor{}

func (c CacheExtractor) Extract(ctx context.Context, url string) (Article, error) {
	tag := c.Tag
	if tag == "" {
		tag = "defaultextractor:"
	}
	key := tag + ":" + url

	var article Article

	err := c.Cache.GetJSON(key, &article)
	if err == nil {
		return article, nil
	}

	article, err = c.Extractor.Extract(ctx, url)
	if err != nil {
		return Article{}, err
	}

	err = c.Cache.SetJSON(key, article)
	if err != nil {
		return Article{}, err
	}

	return article, nil
}