summaryrefslogtreecommitdiff
path: root/loader/lib/parser.py
diff options
context:
space:
mode:
authorMagnus Hagander2012-07-07 14:23:52 +0000
committerMagnus Hagander2012-07-07 14:23:52 +0000
commitc1af12eb480511a884864218216840b28735adaf (patch)
treebcf63e294c087f93533f42d0fa8ee06ba40de8ed /loader/lib/parser.py
parentbcfb2f71306dd993517f7e700eaf83e4a68704ca (diff)
Attempt to get rid of the postgresql specific mail footer on all lists
Diffstat (limited to 'loader/lib/parser.py')
-rw-r--r--loader/lib/parser.py8
1 files changed, 8 insertions, 0 deletions
diff --git a/loader/lib/parser.py b/loader/lib/parser.py
index 6886723..c077413 100644
--- a/loader/lib/parser.py
+++ b/loader/lib/parser.py
@@ -145,6 +145,9 @@ class ArchivesParser(object):
# Return None or empty string, depending on what we got back
return b
+ # Regular expression matching the PostgreSQL custom mail footer that
+ # is appended to all emails.
+ _re_footer = re.compile('(.*)--\s+\nSent via [^\s]+ mailing list \([^\)]+\)\nTo make changes to your subscription:\nhttp://www\.postgresql\.org/mailpref/[^\s]+\s*$', re.DOTALL)
def get_body(self):
b = self._get_body()
if b:
@@ -152,6 +155,11 @@ class ArchivesParser(object):
# later reject..
if b.find(u'\udbff\n\udef8'):
b = b.replace(u'\udbff\n\udef8', '')
+
+ # Remove postgres specific mail footer - if it's there
+ m = _re_footer.match(b)
+ if m:
+ b = m.group(1)
return b
def _get_body(self):