From 6fe95443a0aaccc576388e3ca8b347e2165106b4 Mon Sep 17 00:00:00 2001
From: Marvin Blum <marvin.blum@emvi.com>
Date: Sun, 14 Jun 2020 15:16:44 +0200
Subject: [PATCH] Download blog file attachments when caching articles.

---
 .gitignore            |  1 +
 blog/blog.go          | 58 +++++++++++++++++++++++++++++++++++++++----
 template/about.html   |  1 +
 template/article.html |  2 +-
 template/blog.html    |  1 +
 tpl/template.go       |  4 +++
 6 files changed, 61 insertions(+), 6 deletions(-)

diff --git a/.gitignore b/.gitignore
index 9f11b75..df45cc1 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1 +1,2 @@
 .idea/
+static/blog/
diff --git a/blog/blog.go b/blog/blog.go
index 19fc72a..0630747 100644
--- a/blog/blog.go
+++ b/blog/blog.go
@@ -1,28 +1,34 @@
 package blog
 
 import (
+	"fmt"
 	"github.com/Kugelschieber/marvinblum.de/tpl"
 	emvi "github.com/emvi/api-go"
 	"github.com/emvi/logbuch"
 	"github.com/gorilla/mux"
 	"html/template"
+	"io/ioutil"
 	"net/http"
 	"os"
+	"path/filepath"
+	"regexp"
 	"strings"
 	"time"
 )
 
 const (
 	blogCacheTime     = time.Hour
+	blogFileCache     = "static/blog"
 	maxLatestArticles = 3
 )
 
 var (
-	blog Blog
+	blog               Blog
+	linkRegex          = regexp.MustCompile(`(?iU)href="/read/([^"]+)"`)
+	attachmentRegex    = regexp.MustCompile(`(?iU)(href|src)="([^"]+)/api/v1/content/([^"]+)"`)
+	attachmentURLRegex = regexp.MustCompile(`(?iU)(href|src)="([^"]+/api/v1/content/)([^"]+)"`)
 )
 
-// TODO cache files
-// TODO fix mentions
 type Blog struct {
 	client       *emvi.Client
 	articles     map[string]emvi.Article // id -> article
@@ -37,6 +43,11 @@ func InitBlog() {
 		os.Getenv("MB_EMVI_ORGA"),
 		nil)
 	blog.nextUpdate = time.Now().Add(blogCacheTime)
+
+	if err := os.MkdirAll(blogFileCache, 0755); err != nil {
+		logbuch.Error("Error creating blog file cache directory", logbuch.Fields{"err": err})
+	}
+
 	blog.loadArticles()
 }
 
@@ -87,10 +98,47 @@ func (blog *Blog) loadArticle(article emvi.Article) *emvi.ArticleContent {
 		return nil
 	}
 
+	blog.downloadAttachments(article.Id, content.Content)
+	content.Content = linkRegex.ReplaceAllString(content.Content, `href="/blog/$1"`)
+	content.Content = attachmentRegex.ReplaceAllString(content.Content, fmt.Sprintf(`$1="/static/blog/%s/$3"`, article.Id))
 	logbuch.Debug("Article loaded", logbuch.Fields{"id": article.Id})
 	return content
 }
 
+func (blog *Blog) downloadAttachments(id, content string) {
+	if _, err := os.Stat(filepath.Join(blogFileCache, id)); os.IsNotExist(err) {
+		if err := os.MkdirAll(filepath.Join(blogFileCache, id), 0755); err != nil {
+			logbuch.Error("Error creating article file cache directory", logbuch.Fields{"err": err, "id": id})
+			return
+		}
+	}
+
+	results := attachmentURLRegex.FindAllStringSubmatch(content, -1)
+
+	for _, attachment := range results {
+		if len(attachment) == 4 {
+			resp, err := http.Get(attachment[2] + attachment[3])
+
+			if err != nil {
+				logbuch.Error("Error downloading blog attachment", logbuch.Fields{"err": err, "id": id})
+				continue
+			}
+
+			defer resp.Body.Close()
+			data, err := ioutil.ReadAll(resp.Body)
+
+			if err != nil {
+				logbuch.Error("Error reading blog attachment body", logbuch.Fields{"err": err, "id": id})
+				continue
+			}
+
+			if err := ioutil.WriteFile(filepath.Join(blogFileCache, id, attachment[3]), data, 0755); err != nil {
+				logbuch.Error("Error saving blog attachment on disk", logbuch.Fields{"err": err, "id": id})
+			}
+		}
+	}
+}
+
 func (blog *Blog) setArticles(articles map[string]emvi.Article) {
 	blog.articles = articles
 	blog.articlesYear = make(map[int][]emvi.Article)
@@ -176,11 +224,11 @@ func ServeBlogArticle() http.HandlerFunc {
 		data := struct {
 			Title     string
 			Content   template.HTML
-			Published string
+			Published time.Time
 		}{
 			article.LatestArticleContent.Title,
 			template.HTML(article.LatestArticleContent.Content),
-			article.Published.Format("2. January 2006"),
+			article.Published,
 		}
 
 		if err := tpl.Get().ExecuteTemplate(w, "article.html", data); err != nil {
diff --git a/template/about.html b/template/about.html
index 9f9b943..42cce9e 100644
--- a/template/about.html
+++ b/template/about.html
@@ -22,6 +22,7 @@
     {{range $article := .Articles}}
         <p>
             <a href="/blog/{{slug $article.LatestArticleContent.Title}}-{{$article.Id}}">{{$article.LatestArticleContent.Title}}</a>
+            <small>{{format $article.Published "2. January 2006"}}</small>
         </p>
     {{else}}
         <p>There are no blog posts yet...</p>
diff --git a/template/article.html b/template/article.html
index 4c49f22..405113d 100644
--- a/template/article.html
+++ b/template/article.html
@@ -3,7 +3,7 @@
 
 <section>
     <h1>{{.Title}}</h1>
-    <small>Published on {{.Published}}</small>
+    <small>Published on {{format .Published "2. January 2006"}}</small>
     {{.Content}}
 </section>
 
diff --git a/template/blog.html b/template/blog.html
index 1da03c3..1550da0 100644
--- a/template/blog.html
+++ b/template/blog.html
@@ -9,6 +9,7 @@
         {{range $article := $articles}}
             <p>
                 <a href="/blog/{{slug $article.LatestArticleContent.Title}}-{{$article.Id}}">{{$article.LatestArticleContent.Title}}</a>
+                <small>{{format $article.Published "2. January 2006"}}</small>
             </p>
         {{end}}
     {{else}}
diff --git a/tpl/template.go b/tpl/template.go
index d37f20c..ecea8dc 100644
--- a/tpl/template.go
+++ b/tpl/template.go
@@ -7,6 +7,7 @@ import (
 	"html/template"
 	"net/http"
 	"os"
+	"time"
 )
 
 const (
@@ -21,6 +22,9 @@ var (
 
 var funcMap = template.FuncMap{
 	"slug": slug.Make,
+	"format": func(t time.Time, layout string) string {
+		return t.Format(layout)
+	},
 }
 
 func LoadTemplate() {