1 files changed, 31 insertions, 0 deletions

diff --git a/convert.sed b/convert.sed
new file mode 100755
index 0000000..2449d1f
--- /dev/null
+++ b/convert.sed
@@ -0,0 +1,31 @@
+#!/bin/sed -f
+
+/data-href/s/&/&/g
+
+# xsltproc --html doesnt understand html5
+s/section/div/g
+s/header/h1/g
+s/<nav/<div/g
+s:</nav:</div:g
+s/footer/div/g
+s/article/div/g
+
+# Fix incorrect < and > inside p tags, that is by allowing only
+# known tag be surrounded by < and >.
+ta
+:a
+s/<p>\(.*\)<\/p>/\1/
+tfix
+b
+
+:fix
+s/</\&lt;/g
+s/>/\&gt;/g
+
+s/&lt;span\([^;]*\)&gt;/<span\1>/g
+s/&lt;\/span&gt;/<\/span>/g
+
+s/\&lt;em\&gt;/<em>/g
+s/\&lt;\/em\&gt;/<\/em>/g
+
+s:.*:<p>&</p>: