From bfcb0e6c789a0af78f092be97b48769daaecc3ea Mon Sep 17 00:00:00 2001
From: Henrik Levkowetz <henrik@levkowetz.com>
Date: Mon, 11 Jun 2007 23:52:51 +0000
Subject: [PATCH] Two soup2text tweaks. - Legacy-Id: 324
---
ietf/utils/soup2text.py | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/ietf/utils/soup2text.py b/ietf/utils/soup2text.py
index 9e9a2c1d1..6dde124eb 100755
--- a/ietf/utils/soup2text.py
+++ b/ietf/utils/soup2text.py
@@ -97,7 +97,7 @@ class TextSoup(BeautifulSoup):
node = render(self, encoding)
str = node.text
str = re.sub("[ \t]+", " ", str)
- str = re.sub("\n\n+", "\n\n", str)
+ str = re.sub("\n\n+ *", "\n\n", str)
return str
def soup2text(html):
@@ -105,6 +105,7 @@ def soup2text(html):
html = html.replace("\r\n", "\n").replace("\r", "\n")
# some preprocessing to handle common pathological cases
html = re.sub("<br */?>[ \t\n]*(<br */?>)+", "<p/>", html)
+ html = re.sub("<br */?>([^\n])", r"<br />\n\1", html)
soup = TextSoup(html)
return str(soup)