2 files changed, 99 insertions, 18 deletions

diff --git a/src/dompa/core.clj b/src/dompa/core.clj
index b97d5f2..d8fff41 100644
--- a/src/dompa/core.clj
+++ b/src/dompa/core.clj
@@ -4,10 +4,41 @@
     [dompa.nodes :refer [coordinates->nodes]]))
 
 (defn html->nodes [html]
-  (let [coordinates (html->coordinates html)
-        nodes (coordinates->nodes html coordinates)]
+  (->> html
+       html->coordinates
+       (coordinates->nodes html)))
+
+(defn nodes->html [nodes]
+  (reduce
+    (fn [html node]
+      (cond
+        (= (-> node :name) :dompa/text)
+        (str html (-> node :value))
+
+        :else
+        (let [node-name (-> node :name name)
+              node-child-html (nodes->html (-> node :children))]
+          (str html "<" node-name ">" node-child-html "</" node-name ">"))))
+    ""
     nodes))
 
+(defn traverse-nodes [nodes pred]
+  (reduce
+    (fn [updated-nodes node]
+      (if-let [updated-node (pred node)]
+        (let [children (traverse-nodes (-> updated-node :children) pred)]
+          (conj updated-nodes (assoc updated-node :children children)))
+        updated-nodes))
+    []
+    nodes))
+
+(defn traverse-html [html pred]
+  (-> (html->nodes html)
+      (traverse-nodes pred)
+      nodes->html))
+
 (comment
+  (traverse-html "<div>asdasd<span>hello</span></div>" #(when-not (= (-> % :name) :span)
+                                                          %))
   (html->coordinates "<div>hello<span>asd</span><strong>asdasdadad<img></strong></div>hello some text<div>another root element</div>")
-  (html->nodes "<div>hello<span>asd</span><strong>asdasdadad</strong></div>"))
\ No newline at end of file
+  (html->nodes "<div>hello<span><img src=\"test.jpg\" ckcche/>asd</span><strong>asdasdadad</strong>"))
\ No newline at end of file
diff --git a/src/dompa/nodes.clj b/src/dompa/nodes.clj
index 9b30be3..eb96027 100644
--- a/src/dompa/nodes.clj
+++ b/src/dompa/nodes.clj
@@ -1,19 +1,71 @@
 (ns dompa.nodes
-  (:require [clojure.string :as str]
-            [dompa.coordinates :as coordinates]))
+  (:require
+    [clojure.string :as str]
+    [dompa.coordinates :as coordinates]))
 
-(defn- html->node-name [html]
+(defn- html->node-name
+  "Parses a given HTML string of a node to get its name as
+  a keyword. A text node will return `:dompa/text`."
+  [html]
   (if (str/starts-with? html "<")
-    (-> html
-        (str/split #"[\s\>]")
-        first
-        (str/replace #"[\<\>\/]" "")
-        keyword)
-    :text-node))
+    (->> (subs html 1)
+         (take-while #(not (contains? #{\space \>} %)))
+         (reduce str)
+         keyword)
+    :dompa/text))
 
-(defn- html->node-attrs [html])
+(defn- attr->k-v
+  "Parses a given HTML node attribute string into a
+  key-value pair."
+  [attr]
+  (->> (partition-by #(= % \=) attr)
+       (filter #(not= (-> % first) \=))
+       (map #(reduce str %))))
 
+(defn- normalize-attr-str
+  "Normalizes a given HTML attribute string. If it
+  has surrounding quotes, removes them."
+  [attr-str]
+  (if (str/starts-with? attr-str "\"")
+    (->> (subs attr-str 1)
+         (take-while #(not= % \"))
+         (reduce str))
+    attr-str))
 
+(defn- parse-attr
+  "Parses a given HTML attribute into a normalized
+  key-value map. Attributes with no value part are
+  treated as boolean attributes, and are always `true`."
+  [attr]
+  (let [[k v] (attr->k-v attr)
+        k (keyword k)
+        v (if (nil? v) true (normalize-attr-str v))]
+    {k v}))
+
+(defn- html->node-attrs [html]
+  (when (str/starts-with? html "<")
+    (->> (subs html 1)
+         (take-while #(not (contains? #{\> \/} %)))
+         (partition-by #(= % \space))
+         (drop 1)
+         (filter #(not= (-> % first) \space))
+         (map parse-attr)
+         (into {}))))
+
+(comment
+  (html->node-attrs "<img src=\"test.jpg\" checked />"))
+
+(defn- construct-node
+  [node-html node-children]
+  (let [node-name (html->node-name node-html)]
+    (merge
+      {:name node-name}
+      (when (= node-name :dompa/text)
+        {:value node-html})
+      (when-let [attrs (html->node-attrs node-html)]
+        {:attrs attrs})
+      (when node-children
+        {:children node-children}))))
 
 (defn coordinates->nodes
   [html coordinates]
@@ -22,9 +74,7 @@
           [parent-from parent-to] (first sorted-coordinates)
           children (coordinates/children sorted-coordinates [parent-from parent-to])
           remaining (coordinates/without-children sorted-coordinates [parent-from parent-to])
-          node-html (subs html parent-from (inc parent-to))]
-      (cons {:value (subs html parent-from (inc parent-to))
-             :name (html->node-name node-html)
-             :attrs (html->node-attrs node-html)
-             :children (coordinates->nodes html children)}
+          node-html (subs html parent-from (inc parent-to))
+          node-children (coordinates->nodes html children)]
+      (cons (construct-node node-html node-children)
             (coordinates->nodes html remaining)))))