From 429cc96d2ad22214f3589056f130c51717636be9 Mon Sep 17 00:00:00 2001
From: = <mokaddem.sami@gmail.com>
Date: Tue, 22 Aug 2017 15:17:49 +0200
Subject: [PATCH] Improvement of import_dir.py. Added feed name parameter, send
 the correct path (without potential ../) and gzipEncode it if needed.

---
 bin/import_dir.py | 73 ++++++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 69 insertions(+), 4 deletions(-)

diff --git a/bin/import_dir.py b/bin/import_dir.py
index 7977b303..1300cb43 100755
--- a/bin/import_dir.py
+++ b/bin/import_dir.py
@@ -3,15 +3,57 @@
 
 import zmq
 import base64
+import StringIO
+import gzip
 import argparse
 import os
-import time
+import time, datetime
+import mimetypes
+
+'''
+'
+'   Import content/pastes into redis.
+'   If content is not compressed yet, compress it.
+'
+'   /!\ WARNING /!\
+        Content to be imported must be placed in a directory tree of the form
+        root/
+        |
+        +-- Year/
+            |
+            +-- Month/
+                |
+                +-- Day/
+                    |
+                    +-- Content
+    e.g.:
+    ~/to_import/2017/08/22/paste1.gz
+'
+'''
+
+import StringIO
+import gzip
+def is_hierachy_valid(path):
+    var = path.split('/')
+    try:
+        newDate = datetime.datetime(int(var[-4]), int(var[-3]), int(var[-2]))
+        correctDate = True
+    except ValueError:
+        correctDate = False
+    except IndexError:
+        correctDate = False
+    except:
+        correctDate = False
+    return correctDate
+
 
 if __name__ == "__main__":
     parser = argparse.ArgumentParser(description='Take files from a directory and push them into a 0MQ feed.')
     parser.add_argument('-d', '--directory', type=str, required=True, help='Root directory to import')
     parser.add_argument('-p', '--port', type=int, default=5556, help='Zero MQ port')
     parser.add_argument('-c', '--channel', type=str, default='102', help='Zero MQ channel')
+    parser.add_argument('-n', '--name', type=str, default='import_dir', help='Name of the feeder')
+    parser.add_argument('--hierarchy', type=int, default=1, help='Number of parent directory forming the name')
 
     args = parser.parse_args()
 
@@ -22,7 +64,30 @@ if __name__ == "__main__":
 
     for dirname, dirnames, filenames in os.walk(args.directory):
         for filename in filenames:
-            messagedata = open(os.path.join(dirname, filename)).read()
-            print(os.path.join(dirname, filename))
-            socket.send('{} {} {}'.format(args.channel, os.path.join(dirname, filename), base64.b64encode(messagedata)))
+            complete_path = os.path.join(dirname, filename)
+
+            #take wanted path of the file
+            wanted_path = os.path.realpath(complete_path)
+            wanted_path = wanted_path.split('/')
+            wanted_path = '/'.join(wanted_path[-(4+args.hierarchy):])
+
+            messagedata = open(complete_path).read()
+
+            #if paste do not have a 'date hierarchy' ignore it
+            if not is_hierachy_valid(complete_path):
+                print('/!\ hierarchy not valid, should have the format yyyy/mm/dd/paste.gz /!\ ')
+                print(complete_path)
+                break
+
+            #verify that the data is gzipEncoded. if not compress it
+            if 'text' in str(mimetypes.guess_type(complete_path)[0]):
+                out = StringIO.StringIO()
+                with gzip.GzipFile(fileobj=out, mode="w") as f:
+                    f.write(messagedata)
+                messagedata = out.getvalue()
+                wanted_path += '.gz'
+
+            print(args.name+'>'+wanted_path)
+            path_to_send = args.name + '>' + wanted_path
+            socket.send('{} {} {}'.format(args.channel, path_to_send, base64.b64encode(messagedata)))
             time.sleep(.2)