purify_html: New.

1 files changed, 11 insertions, 0 deletions

diff --git a/purify_html b/purify_html
new file mode 100755
index 00000000..a9ead881
--- /dev/null
+++ b/purify_html
@@ -0,0 +1,11 @@
+#!/bin/sh
+
+# Un-mangle mailto links: convert HTML character entities to real characters.
+find ./ -name \*.html -print0 \
+  | xargs -0 \
+    perl -p -i -l -e \
+      'BEGIN { $replacing = 0; }
+      # The replacing-toggling logic is a bit rough, but so is life.
+      $replacing = 1 if /<a href="mailto:/;
+      s%\&#(x?)([^;]*);%chr(length($1) ? hex($2) : $2)%eg if $replacing;
+      $replacing = 0 if /<\/a>/;'