feat: add article page

data/article.go

@@ -0,0 +1,64 @@
+package data
+
+import (
+	"encoding/json"
+	"time"
+
+	"github.com/PuerkitoBio/goquery"
+	"github.com/rramiachraf/dumb/utils"
+)
+
+type Article struct {
+	Title       string
+	Subtitle    string
+	HTML        string
+	Authors     []Author
+	PublishedAt time.Time
+	Image       string
+}
+
+type Author struct {
+	Name  string
+	Role  string `json:"human_readable_role_for_display"`
+	About string `json:"about_me_summary"`
+}
+
+type articleResponse struct {
+	Article struct {
+		Title    string
+		Subtitle string `json:"dek"`
+		Authors  []Author
+		Body     struct {
+			HTML string
+		}
+		PublishedAt int64  `json:"published_at"`
+		Image       string `json:"preview_image"`
+	}
+}
+
+func (a *Article) parseArticleData(doc *goquery.Document) error {
+	pageMetadata, exists := doc.Find("meta[itemprop='page_data']").Attr("content")
+	if !exists {
+		return nil
+	}
+
+	var articleData articleResponse
+	if err := json.Unmarshal([]byte(pageMetadata), &articleData); err != nil {
+		return err
+	}
+	data := articleData.Article
+
+	a.Title = data.Title
+	a.Subtitle = data.Subtitle
+
+	a.HTML = utils.CleanBody(data.Body.HTML)
+	a.Authors = data.Authors
+	a.PublishedAt = time.Unix(data.PublishedAt, 0)
+	a.Image = ExtractImageURL(data.Image)
+
+	return nil
+}
+
+func (a *Article) Parse(doc *goquery.Document) error {
+	return a.parseArticleData(doc)
+}

handlers/annotations.go

@@ -6,10 +6,7 @@ import (
 	"encoding/json"
 	"fmt"
 	"net/http"
-	"regexp"
-	"strings"
 
-	"github.com/PuerkitoBio/goquery"
 	"github.com/gorilla/mux"
 	"github.com/rramiachraf/dumb/data"
 	"github.com/rramiachraf/dumb/utils"
@@ -67,7 +64,7 @@ func annotations(l *utils.Logger) http.HandlerFunc {
 		}
 
 		body := data.Response.Referent.Annotations[0].Body
-		body.HTML = cleanBody(body.HTML)
+		body.HTML = utils.CleanBody(body.HTML)
 
 		w.Header().Set("content-type", "application/json")
 		encoder := json.NewEncoder(w)
@@ -82,31 +79,3 @@ func annotations(l *utils.Logger) http.HandlerFunc {
 		}
 	}
 }
-
-func cleanBody(body string) string {
-	if doc, err := goquery.NewDocumentFromReader(strings.NewReader(body)); err == nil {
-		doc.Find("iframe").Each(func(i int, s *goquery.Selection) {
-			src, exists := s.Attr("src")
-			if exists {
-				html := fmt.Sprintf(`<a id="iframed-link" href="%s">Link</a>`, src)
-				s.ReplaceWithHtml(html)
-			}
-		})
-
-		doc.Find("img").Each(func(i int, s *goquery.Selection) {
-			src, exists := s.Attr("src")
-			if exists {
-				re := regexp.MustCompile(`(?i)https:\/\/images\.(rapgenius|genius)\.com\/`)
-				pSrc := re.ReplaceAllString(src, "/images/")
-				s.SetAttr("src", pSrc)
-			}
-		})
-
-		if source, err := doc.Html(); err == nil {
-			body = source
-		}
-	}
-
-	re := regexp.MustCompile(`https?:\/\/[a-z]*.?genius.com`)
-	return re.ReplaceAllString(body, "")
-}

handlers/article.go

@@ -0,0 +1,68 @@
+package handlers
+
+import (
+	"context"
+	"fmt"
+	"net/http"
+
+	"github.com/PuerkitoBio/goquery"
+	"github.com/gorilla/mux"
+	"github.com/rramiachraf/dumb/data"
+	"github.com/rramiachraf/dumb/utils"
+	"github.com/rramiachraf/dumb/views"
+)
+
+func article(l *utils.Logger) http.HandlerFunc {
+	return func(w http.ResponseWriter, r *http.Request) {
+		articleSlug := mux.Vars(r)["article"]
+
+		if a, err := getCache[data.Article](articleSlug); err == nil {
+			views.ArticlePage(a).Render(context.Background(), w)
+			return
+		}
+
+		url := fmt.Sprintf("https://genius.com/a/%s", articleSlug)
+
+		resp, err := utils.SendRequest(url)
+		if err != nil {
+			l.Error(err.Error())
+			w.WriteHeader(http.StatusInternalServerError)
+			views.ErrorPage(500, "cannot reach Genius servers").Render(context.Background(), w)
+			return
+		}
+
+		defer resp.Body.Close()
+
+		if resp.StatusCode == http.StatusNotFound {
+			w.WriteHeader(http.StatusNotFound)
+			views.ErrorPage(404, "page not found").Render(context.Background(), w)
+			return
+		}
+
+		doc, err := goquery.NewDocumentFromReader(resp.Body)
+		if err != nil {
+			l.Error(err.Error())
+			w.WriteHeader(http.StatusInternalServerError)
+			views.ErrorPage(500, "something went wrong").Render(context.Background(), w)
+			return
+		}
+
+		cf := doc.Find(".cloudflare_content").Length()
+		if cf > 0 {
+			l.Error("cloudflare got in the way")
+			views.ErrorPage(500, "cloudflare is detected").Render(context.Background(), w)
+			return
+		}
+
+		var a data.Article
+		if err = a.Parse(doc); err != nil {
+			l.Error(err.Error())
+		}
+
+		views.ArticlePage(a).Render(context.Background(), w)
+
+		if err = setCache(articleSlug, a); err != nil {
+			l.Error(err.Error())
+		}
+	}
+}

handlers/article_test.go

@@ -0,0 +1,55 @@
+package handlers
+
+import (
+	"net/http"
+	"net/http/httptest"
+	"os"
+	"testing"
+
+	"github.com/PuerkitoBio/goquery"
+
+	"github.com/rramiachraf/dumb/utils"
+)
+
+func TestArticle(t *testing.T) {
+	url := "/a/genius-celebrates-hip-hops-50th-anniversary-with-a-look-back-at-the-music-thats-defined-this-site"
+	title := "Genius Celebrates Hip-Hop’s 50th Anniversary With A Look Back At The Music That’s Defined This Site"
+	subtitle := "The first post in a yearlong look at the genre’s storied history."
+
+	r, err := http.NewRequest(http.MethodGet, url, nil)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	rr := httptest.NewRecorder()
+	l := utils.NewLogger(os.Stdout)
+	m := New(l, &assets{})
+
+	m.ServeHTTP(rr, r)
+
+	defer rr.Result().Body.Close()
+
+	if rr.Result().StatusCode != http.StatusOK {
+		t.Fatalf("expected %d, got %d\n", http.StatusOK, rr.Result().StatusCode)
+	}
+
+	doc, err := goquery.NewDocumentFromReader(rr.Result().Body)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	articleTitle := doc.Find("#article-title").First().Text()
+	if articleTitle != title {
+		t.Fatalf("expected %q, got %q\n", title, articleTitle)
+	}
+
+	articleSubtitle := doc.Find("#article-subtitle").First().Text()
+	if articleSubtitle != subtitle {
+		t.Fatalf("expected %q, got %q\n", subtitle, articleSubtitle)
+	}
+
+	articleBody := doc.Find("#article-body").First().Text()
+	if len(articleBody) == 0 {
+		t.Fatal("missing article body\n")
+	}
+}

handlers/cache.go

@@ -10,7 +10,7 @@ import (
 )
 
 type cachable interface {
-	data.Album | data.Song | data.Annotation | data.Artist | []byte
+	data.Album | data.Song | data.Annotation | data.Artist | data.Article | []byte
 }
 
 var c, _ = bigcache.New(context.Background(), bigcache.DefaultConfig(time.Hour*24))

handlers/handler.go

@@ -29,6 +29,7 @@ func New(logger *utils.Logger, staticFiles static) *mux.Router {
 		{Path: "/robots.txt", Handler: robotsHandler},
 		{Path: "/albums/{artist}/{albumName}", Handler: album},
 		{Path: "/artists/{artist}", Handler: artist},
+		{Path: "/a/{article}", Handler: article},
 		{Path: "/images/{filename}.{ext}", Handler: imageProxy},
 		{Path: "/search", Handler: search},
 		{Path: "/{annotation-id}/{artist-song}/{verse}/annotations", Handler: annotations},

style/annotation.css

@@ -1,4 +1,4 @@
-.annotation #iframed-link {
+#iframed-link {
 	font-weight: 500;
 	background-color: #ffcd38;
 	padding: 2px 6px;

style/article.css

@@ -0,0 +1,19 @@
+#article-body {
+	line-height: 1.75;
+}
+
+#metadata, #article-subtitle, #article-date {
+	color: #333;
+}
+
+.dark #metadata, .dark #article-subtitle, .dark #article-date {
+	color: #ccc;
+}
+
+.dark #article-title {
+	color: #eee;
+}
+
+.dark #article-body {
+	color: #eee;
+}

utils/clean_body.go

@@ -0,0 +1,37 @@
+package utils
+
+import (
+	"fmt"
+	"regexp"
+	"strings"
+
+	"github.com/PuerkitoBio/goquery"
+)
+
+func CleanBody(body string) string {
+	if doc, err := goquery.NewDocumentFromReader(strings.NewReader(body)); err == nil {
+		doc.Find("iframe").Each(func(i int, s *goquery.Selection) {
+			src, exists := s.Attr("src")
+			if exists {
+				html := fmt.Sprintf(`<a id="iframed-link" href="%s">Link</a>`, src)
+				s.ReplaceWithHtml(html)
+			}
+		})
+
+		doc.Find("img").Each(func(i int, s *goquery.Selection) {
+			src, exists := s.Attr("src")
+			if exists {
+				re := regexp.MustCompile(`(?i)https:\/\/images\.(rapgenius|genius)\.com\/(images\/)?`)
+				pSrc := re.ReplaceAllString(src, "/images/")
+				s.SetAttr("src", pSrc)
+			}
+		})
+
+		if source, err := doc.Html(); err == nil {
+			body = source
+		}
+	}
+
+	re := regexp.MustCompile(`https?:\/\/[a-z]*.?genius.com`)
+	return re.ReplaceAllString(body, "")
+}

views/article.templ

@@ -0,0 +1,32 @@
+package views
+
+import (
+	"time"
+
+	"github.com/rramiachraf/dumb/data"
+)
+
+templ ArticlePage(a data.Article) {
+	@layout(a.Title) {
+		<div id="container" class="duo-split">
+			<div id="metadata">
+				<img src={ a.Image } alt="Article image"/>
+				<h3>Authored By</h3>
+				for _, author := range a.Authors {
+					<details>
+						<summary>{ author.Name } - { author.Role }</summary>
+						{ author.About }
+					</details>
+				}
+			</div>
+			<div>
+				<h1 id="article-title">{ a.Title }</h1>
+				<time datetime={ a.PublishedAt.Format(time.RFC3339) } id="article-date">{ a.PublishedAt.Format(time.DateOnly) }</time>
+				<h2 id="article-subtitle">{ a.Subtitle }</h2>
+				<div id="article-body">
+					@templ.Raw(a.HTML)
+				</div>
+			</div>
+		</div>
+	}
+}