[back/refactor] fetch and build post

This commit is contained in:
SouthFox 2023-09-01 15:39:24 +08:00
parent 86e08d5478
commit 547ce09b45
2 changed files with 56 additions and 54 deletions

View file

@ -11,10 +11,7 @@
(generate-string content)) (generate-string content))
(defn clean-html (defn clean-html
[docs hugo] [docs replace-str]
(let [replace-str (if hugo
"/hp/"
"#/item/")]
(-> (.select docs "style[data-emotion-css~=^[a-z0-9]*$]") (-> (.select docs "style[data-emotion-css~=^[a-z0-9]*$]")
(.remove)) (.remove))
(doseq [a (.select docs "a")] (doseq [a (.select docs "a")]
@ -22,7 +19,7 @@
(-> (.attr a "href") (-> (.attr a "href")
(str/replace "https://link.zhihu.com/?target=" "") (str/replace "https://link.zhihu.com/?target=" "")
(str/replace "https://zhuanlan.zhihu.com/p/" replace-str) (str/replace "https://zhuanlan.zhihu.com/p/" replace-str)
(url-decode)))))) (url-decode)))))
(defn clean-images (defn clean-images
[docs] [docs]
@ -53,38 +50,39 @@
(apply str (mapv build-catalog-item catalog)))) (apply str (mapv build-catalog-item catalog))))
(defn process-hu-post (defn process-hu-post
[page hugo] [page {:keys [content-type replace-str wrap-fn]}]
(let [post-content (.getElementsByClass page "Post-RichTextContainer") (let [post-content (.getElementsByClass page "Post-RichTextContainer")
title (.getElementsByClass page "Post-Title") title (.getElementsByClass page "Post-Title")
post-time (.getElementsByClass page "ContentItem-time")] post-time (.getElementsByClass page "ContentItem-time")]
(clean-html post-content hugo) (clean-html post-content replace-str)
(clean-images post-content) (clean-images post-content)
(render-linkcard post-content) (render-linkcard post-content)
{:status 200 {:status 200
:body {:content (.toString post-content) :headers content-type
:body (wrap-fn
{:content (.toString post-content)
:title (.text title) :title (.text title)
:time (first (str/split (.text post-time) #"・")) :time (first (str/split (.text post-time) #"・"))
:catalog (build-catalog post-content)}})) :catalog (build-catalog post-content)})}))
(defn fetch-hu-post (defn fetch-hu-post
[post-id & {:keys [hugo]}] [post-id params]
(let [post-url (str/join ["https://zhuanlan.zhihu.com/p/" post-id]) (let [post-url (str/join ["https://zhuanlan.zhihu.com/p/" post-id])
page (-> (client/get post-url) :body Jsoup/parse) page (-> (client/get post-url) :body Jsoup/parse)
docs (.getElementsByClass page "Post-RichTextContainer")] docs (.getElementsByClass page "Post-RichTextContainer")]
(if (empty? docs) (if (empty? docs)
{:status 404 {:status 404
:headers (:content-type params)
:body {:content "Not Found" :body {:content "Not Found"
:title "Not Found"}} :title "Not Found"}}
(process-hu-post page hugo)))) (process-hu-post page params))))
(defn build-api-hu-post (defn build-api-hu-post
[request] [request]
(let [post-id (-> request :path-params :id) (let [post-id (-> request :path-params :id)]
content (fetch-hu-post post-id)] (fetch-hu-post post-id
{:status (:status content) {:content-type {"Content-Type" "application/json; charset=utf-8"}
:headers {"Content-Type" "application/json; charset=utf-8"} :replace-str "#/item/"
:body (wrap-json (:body content))})) :wrap-fn wrap-json})))

View file

@ -3,14 +3,9 @@
[hiccup.page :as page])) [hiccup.page :as page]))
(defn build-hugo-post (def hugo-post-template
[request] #(page/html5
(let [post-id (-> request :path-params :id) {:escape-strings? false}
result (handlers/fetch-hu-post post-id {:hugo true})
content (:body result)]
{:status (:status result)
:headers {"Content-Type" "text/html; charset=utf-8"}
:body (page/html5 {:escape-strings? false}
[:html [:html
[:head [:head
[:meta {:name "viewport" :content "width=device-width, initial-scale=1.0"}] [:meta {:name "viewport" :content "width=device-width, initial-scale=1.0"}]
@ -22,10 +17,19 @@
[:div {:class "flex flex-row flex-wrap py-4"} [:div {:class "flex flex-row flex-wrap py-4"}
[:div {:class "w-full sm:w-1/3 md:w-1/4 px-2"} [:div {:class "w-full sm:w-1/3 md:w-1/4 px-2"}
[:div {:class "sticky top-0 p-4 bg-slate-300 rounded-xl w-full"} [:div {:class "sticky top-0 p-4 bg-slate-300 rounded-xl w-full"}
[:div (:catalog content)]]] [:div (:catalog %)]]]
[:div {:class "w-full sm:w-2/3 md:w-3/4 pt-1 px-2"} [:div {:class "w-full sm:w-2/3 md:w-3/4 pt-1 px-2"}
[:h1 (:title content)] [:h1 (:title %)]
[:time (:time content)] [:time (:time %)]
[:hr] [:hr]
[:div [:div
(:content content)]]]]]]])})) (:content %)]]]]]]]))
(defn build-hugo-post
[request]
(let [post-id (-> request :path-params :id)]
(handlers/fetch-hu-post
post-id
{:content-type {"Content-Type" "text/html; charset=utf-8"}
:replace-str "/hp/"
:wrap-fn hugo-post-template})))