newsbot-api/internal/services/input/rss.go

package input

import (
	"strings"

	"git.jamestombleson.com/jtom38/newsbot-api/internal/entity"
	"github.com/mmcdole/gofeed"
)

type FeedInput interface {
	GetArticles() (entity.ArticleEntity, error)
}

type rssClient struct {
	SourceRecord entity.SourceEntity
}

func NewRssClient(sourceRecord entity.SourceEntity) rssClient {
	client := rssClient{
		SourceRecord: sourceRecord,
	}

	return client
}

func (rc rssClient) GetArticles() ([]entity.ArticleEntity, error) {
	parser := gofeed.NewParser()
	feed, err := parser.ParseURL(rc.SourceRecord.Url)
	if err != nil {
		return nil, err
	}

	sourceTags := strings.Split(rc.SourceRecord.Tags, ",")
	var articles []entity.ArticleEntity
	for _, post := range feed.Items {
		article := entity.ArticleEntity{
			SourceID:    rc.SourceRecord.ID,
			Title:       post.Title,
			Description: post.Content,
			Url:         post.Link,
			PubDate:     *post.PublishedParsed,
			//AuthorName:  post.Authors[0].Email,
		}

		if len(post.Authors) != 0 {
			article.AuthorName = post.Authors[0].Email
		}

		var postTags []string
		postTags = append(postTags, sourceTags...)
		postTags = append(postTags, post.Categories...)
		article.Tags = strings.Join(postTags, ",")

		/*
			pageContent, err := getHttpContent(article.Url)
			if err != nil {
				continue
			}

			htmlNode, err := html.Parse(bytes.NewReader(pageContent))
			if err != nil {
				continue
			}
			htmlNode.

			fmt.Println(htmlNode)
		*/

		if post.Image == nil {
			article.Thumbnail = ""
		}

		articles = append(articles, article)
	}

	return articles, nil
}