diff options
Diffstat (limited to 'loader')
| -rw-r--r-- | loader/lib/parser.py | 1 | ||||
| -rw-r--r-- | loader/lib/storage.py | 3 | ||||
| -rw-r--r-- | loader/sql/schema.sql | 3 |
3 files changed, 5 insertions, 2 deletions
diff --git a/loader/lib/parser.py b/loader/lib/parser.py index 258cc2b..670a70c 100644 --- a/loader/lib/parser.py +++ b/loader/lib/parser.py @@ -18,6 +18,7 @@ class ArchivesParser(object): def parse(self, stream): self.msg = self.parser.parse(stream) + self.rawtxt = unicode(self.msg) def is_msgid(self, msgid): # Look for a specific messageid. This means we might parse it twice, diff --git a/loader/lib/storage.py b/loader/lib/storage.py index ab25793..1a34b1b 100644 --- a/loader/lib/storage.py +++ b/loader/lib/storage.py @@ -131,7 +131,7 @@ class ArchivesParserStorage(ArchivesParser): if len(curs.fetchall()): log.status("Tagged thread %s with listid %s" % (self.threadid, listid)) - curs.execute("INSERT INTO messages (parentid, threadid, _from, _to, cc, subject, date, has_attachment, messageid, bodytxt) VALUES (%(parentid)s, %(threadid)s, %(from)s, %(to)s, %(cc)s, %(subject)s, %(date)s, %(has_attachment)s, %(messageid)s, %(bodytxt)s) RETURNING id", { + curs.execute("INSERT INTO messages (parentid, threadid, _from, _to, cc, subject, date, has_attachment, messageid, bodytxt, rawtxt) VALUES (%(parentid)s, %(threadid)s, %(from)s, %(to)s, %(cc)s, %(subject)s, %(date)s, %(has_attachment)s, %(messageid)s, %(bodytxt)s, %(rawtxt)s) RETURNING id", { 'parentid': self.parentid, 'threadid': self.threadid, 'from': self._from, @@ -142,6 +142,7 @@ class ArchivesParserStorage(ArchivesParser): 'has_attachment': len(self.attachments) > 0, 'messageid': self.msgid, 'bodytxt': self.bodytxt, + 'rawtxt': self.rawtxt, }) id = curs.fetchall()[0][0] if len(self.attachments): diff --git a/loader/sql/schema.sql b/loader/sql/schema.sql index ff8e5d3..b6cf49f 100644 --- a/loader/sql/schema.sql +++ b/loader/sql/schema.sql @@ -11,7 +11,8 @@ CREATE TABLE messages ( date timestamptz NOT NULL, has_attachment boolean NOT NULL, messageid text NOT NULL, - bodytxt text NOT NULL + bodytxt text NOT NULL, + rawtxt text NOT NULL ); CREATE INDEX idx_messages_threadid ON messages(threadid); CREATE UNIQUE INDEX idx_messages_msgid ON messages(messageid); |