Merge pull request #22 from reedho/html-support

html-support (with hickory)
markus-wa · Oct 18, 2023 · 28889c3 · 28889c3
2 parents a852403 + efd9774
commit 28889c3
Show file tree

Hide file tree

Showing 3 changed files with 52 additions and 3 deletions.
diff --git a/deps.edn b/deps.edn
@@ -14,7 +14,9 @@
   com.cognitect/transit-clj {:mvn/version "1.0.324"}
   medley/medley {:mvn/version "1.3.0"}
   mvxcvi/puget {:mvn/version "1.3.1"}
-  tolitius/xml-in {:mvn/version "0.1.1"}}
+  tolitius/xml-in {:mvn/version "0.1.1"}
+  org.clj-commons/hickory {:mvn/version "0.7.3"}
+  org.jsoup/jsoup {:mvn/version "1.14.3"}}
  :aliases
  {:run
   {:main-opts ["-m" "cq.main"]}

diff --git a/src/cq/formats.clj b/src/cq/formats.clj
@@ -2,6 +2,8 @@
   (:require [clojure.data.csv :as csv]
             [clojure.data.json :as json]
             [clojure.data.xml :as xml]
+            [hickory.core :as html]
+            [hickory.render :refer [hickory-to-html]]
             [clojure.edn :as edn]
             [clojure.pprint :as ppt]
             [clojure.java.io :as io]
@@ -167,6 +169,18 @@
       (with-open [w (io/writer out)]
         (emit x w)))))
 
+(defn ->html-reader
+  [_]
+  (fn [in]
+    (html/as-hickory (html/parse (slurp (io/reader in))))))
+
+(defn ->html-writer
+  [_]
+  (fn [x out]
+    (binding [*out* (io/writer out)]
+      (print (hickory-to-html x))
+      (flush))))
+
 (def formats
   {"json"    {:->reader ->json-reader
               :->writer ->json-writer}
@@ -185,7 +199,9 @@
    "transit" {:->reader ->transit-reader
               :->writer ->transit-writer}
    "xml"     {:->reader ->xml-reader
-              :->writer ->xml-writer}})
+              :->writer ->xml-writer}
+   "html"    {:->reader ->html-reader
+              :->writer ->html-writer}})
 
 (defn format->reader
   [format in opts]

diff --git a/test/cq/formats_test.clj b/test/cq/formats_test.clj
@@ -2,7 +2,8 @@
   (:require [cq.formats :as sut]
             [clojure.test :refer :all]
             [clojure.java.io :as io]
-            [clojure.string :as str])
+            [clojure.string :as str]
+            [hickory.core :as html])
   (:import [java.io ByteArrayInputStream BufferedInputStream PrintStream ByteArrayOutputStream]))
 
 (defn- to-out-stream
@@ -152,3 +153,33 @@
   <a:article>Hello</a:article>
 </a:html>\n"
              (test-writer-str sut/->xml-writer {:pretty true} test-xml-data))))))
+
+(def test-html-str
+  "<p>hello</p>")
+
+(def test-html-data
+  {:type :document
+   :content
+   [{:type  :element
+     :attrs nil
+     :tag   :html
+     :content
+     [{:type  :element
+       :attrs nil
+       :tag   :head :content nil}
+      {:type  :element
+       :attrs nil
+       :tag   :body :content
+       [{:type    :element
+         :attrs   nil
+         :tag     :p
+         :content ["hello"]}]}]}]})
+
+(deftest html
+  (testing "reader"
+    (is (= test-html-data
+           (test-reader-str sut/->html-reader nil test-html-str))))
+
+  (testing "writer"
+    (is (= "<html><head></head><body><p>hello</p></body></html>"
+           (test-writer-str sut/->html-writer nil test-html-data)))))