export

archive/archive.py

@@ -267,6 +267,29 @@ class Archive:
 		finally:
 			cursor.close()
 
+	def select_all(self):			
+
+		if self.db_con is None:
+			logging.warning("Not connection to database...")
+			return
+
+		try:
+			cursor = self.db_con.cursor(buffered=True)
+
+			cursor.execute(archive.sql.SELECT_ALL.format(self.archive_name))
+
+			# print(cursor.rowcount)
+			results = []
+			for (from_, author_name_, to_, subject_, date_, content_type_, content_, url_) in cursor:
+				results.append((from_, author_name_, to_, subject_, date_, content_type_, content_, url_))
+				# print("{} {} {}".format(from_, str(date_), url_))
+			return results
+
+		except mariadb.Error as error:
+			logging.error("Error: {}".format(error))
+		finally:
+			cursor.close()
+
 	# analysis
 	def longest_field(self, field, thread, max_length=0):
 		import archive.util

archive/sql.py

@@ -33,4 +33,7 @@ FROM_QUERY_NL = ("SELECT from_, author_name_, subject_, date_, url_ FROM {} "
 
 SHOW_TABLE = "show tables"
 
+SELECT_ALL = ("SELECT * FROM {} "
+	"ORDER BY date_")
+
 # SELECT from_, author_name_, subject_, date_, url_ FROM nettime_l WHERE MATCH(content_) AGAINST('%s' IN BOOLEAN MODE)

export.py

@@ -0,0 +1,46 @@
+import argparse, mailbox, email
+import archive.archive as archive
+import config
+
+if __name__ == "__main__":
+
+    p = argparse.ArgumentParser(description='Export mailinglists archives')
+    p.add_argument('name', metavar="name", help="name of the archive to export")
+    p.add_argument('--out', '-o', help="mbox output")
+
+    args = p.parse_args()
+
+    with archive.Archive(args.name, config=config.db) as a:
+
+        r = a.select_all()
+
+        if args.out:
+
+            mbox = mailbox.mbox(args.out)
+
+            for (from_, author_name_, to_, subject_, date_, content_type_, content_, url_) in r:
+
+                m = mailbox.mboxMessage()
+                m['To'] = to_
+                m['Subject'] = subject_
+                addr = from_.replace('{at}', '@')
+                if author_name_:
+                    m['From'] = email.utils.formataddr((author_name_, addr))
+                else:
+                    m['From'] = addr
+                m['Date'] = email.utils.format_datetime(date_)
+                m['Content-Type'] = content_type_
+                m.set_payload(content_)
+
+                mbox.add(m)      
+
+            mbox.flush()     
+
+
+        else:
+
+            for (from_, author_name_, to_, subject_, date_, content_type_, content_, url_) in r:
+                print(date_)
+
+
+

search.py

@@ -28,7 +28,7 @@ def run(args):
     for l in args.list:
 
         if l not in all_lists:
-            logging.warning(l + "is not a valud list... continuing")
+            logging.warning(l + "is not a valid list... continuing")
             continue
 
         k_arg = args.keyword