From 385ef0db111ad842be4817171f9a19276d312fb5 Mon Sep 17 00:00:00 2001
From: Alan Pearce
Date: Mon, 22 Sep 2014 20:36:26 +0100
Subject: Trim extra spaces from HTML content

---
 src/microformats/parser.clj | 9 +++++++++
 1 file changed, 9 insertions(+)

(limited to 'src/microformats')

diff --git a/src/microformats/parser.clj b/src/microformats/parser.clj
index eade3dc..cd57693 100644
--- a/src/microformats/parser.clj
+++ b/src/microformats/parser.clj
@@ -101,6 +101,15 @@
           %
           (apply str (persistent! (html/emit-tag % (transient []))))) el))
 
+(defn- node-to-text
+  "Turn a node into a text string"
+  [content]
+  (->> content
+       html/texts
+       (map #(str/replace % #"\s+" " "))
+       (apply str)
+       str/trim))
+
 (defn get-e-value
   "Get the e-x propery value of an element"
   [el]
-- 
cgit 1.4.1