From c0b4982e33106794683ce54490e7357296811346 Mon Sep 17 00:00:00 2001
From: SouthFox <master@southfox.me>
Date: Sat, 26 Aug 2023 10:42:59 +0800
Subject: [PATCH] [back/feat] clean styles, links and show images

---
 src/main/backend/handlers.clj | 32 ++++++++++++++++++++++++--------
 1 file changed, 24 insertions(+), 8 deletions(-)

diff --git a/src/main/backend/handlers.clj b/src/main/backend/handlers.clj
index 37fbda2..d28353b 100644
--- a/src/main/backend/handlers.clj
+++ b/src/main/backend/handlers.clj
@@ -1,6 +1,7 @@
 (ns backend.handlers
   (:require [clojure.string :as str]
             [babashka.http-client :as client]
+            [ring.util.codec :refer [url-decode]]
             [cheshire.core :refer [generate-string]])
   (:import [org.jsoup Jsoup]))
 
@@ -9,16 +10,31 @@
   [content]
   (generate-string {:content content}))
 
+(defn clean-html
+  [docs]
+  (-> (.select docs "style[data-emotion-css~=^[a-z0-9]*$]")
+      (.remove))
+  (-> (.select docs "figure > img")
+      (.remove))
+  (-> (.select docs "figure > noscript")
+      (.tagName "div"))
+  (vec
+   (for [a (.select docs "a")]
+     (.attr a "href"
+            (url-decode
+             (str/replace
+              (.attr a "href")
+              "https://link.zhihu.com/?target=" ""))))))
 
-(defn fetch-hu-post [request]
+
+(defn fetch-hu-post
+  [request]
   (let [id (-> request :path-params :id)
-        post-url (str/join ["https://zhuanlan.zhihu.com/p/" id])]
-
+        post-url (str/join ["https://zhuanlan.zhihu.com/p/" id])
+        docs (-> (client/get post-url) :body Jsoup/parse
+                 (.getElementsByClass "Post-RichTextContainer"))]
+    (clean-html docs)
     {:status 200
      :headers {"Content-Type" "application/json; charset=utf-8"}
-     :body (-> (client/get post-url)
-               :body
-               Jsoup/parse
-               (.getElementsByClass "Post-RichTextContainer")
-               (.toString)
+     :body (-> (.toString docs)
                (wrap-json))}))