Modify HTTP logs in clients for safety and clarity

Refs grnet#32

Implement method escape_ctrl_chars
Escape control characters in HTTP data logs
Remove misleading separators from HTTP logs

kamaki/clients/__init__.py

@@ -43,6 +43,8 @@
 
 from kamaki.clients.utils import https
 
+from kamaki.clients import utils
+
 
 TIMEOUT = 60.0   # seconds
 HTTP_METHODS = ['GET', 'POST', 'PUT', 'HEAD', 'DELETE', 'COPY', 'MOVE']
@@ -148,7 +150,6 @@ def __init__(
 
     def dump_log(self):
         plog = ('\t[%s]' % self) if self.LOG_PID else ''
-        sendlog.info('- -  -   -     -        -             -')
         sendlog.info('%s %s://%s%s%s' % (
             self.method, self.scheme, self.netloc, self.path, plog))
         for key, val in self.headers.items():
@@ -158,8 +159,8 @@ def dump_log(self):
         if self.data:
             sendlog.info('data size: %s%s' % (len(self.data), plog))
             if self.LOG_DATA:
-                sendlog.info(self.data.replace(self._token, '...') if (
-                    self._token) else self.data)
+                sendlog.info(utils.escape_ctrl_chars(self.data.replace(
+                    self._token, '...') if self._token else self.data))
         else:
             sendlog.info('data size: 0%s' % plog)
 
@@ -274,8 +275,7 @@ def _get_response(self):
                     self._status_code, self._status = r.status, unquote(
                         r.reason)
                     recvlog.info(
-                        '%d %s%s' % (
-                            self.status_code, self.status, plog))
+                        '%d %s%s' % (self.status_code, self.status, plog))
                     self._headers = dict()
 
                     r_headers = r.getheaders()
@@ -291,8 +291,7 @@ def _get_response(self):
                         data = '%s%s' % (self._content, plog)
                         if self._token:
                             data = data.replace(self._token, '...')
-                        recvlog.info(data)
-                    recvlog.info('-             -        -     -   -  - -')
+                        recvlog.info(utils.escape_ctrl_chars(data))
                 break
             except Exception as err:
                 if isinstance(err, HTTPException):
@@ -382,11 +381,9 @@ def run(self):
         try:
             self._value = self.method(*(self.args), **(self.kwargs))
         except Exception as e:
+            estatus = e.status if isinstance(e, ClientError) else ''
             recvlog.debug('Thread %s got exception %s\n<%s %s' % (
-                self,
-                type(e),
-                e.status if isinstance(e, ClientError) else '',
-                e))
+                self, type(e), estatus, e))
             self._exception = e
 
 

kamaki/clients/utils/__init__.py

@@ -31,6 +31,7 @@
 # interpreted as representing official policies, either expressed
 # or implied, of GRNET S.A.
 
+import unicodedata
 
 def _matches(val1, val2, exactMath=True):
     """Case Insensitive match"""
@@ -105,3 +106,14 @@ def readall(openfile, size, retries=7):
                 continue
         return buf
     raise IOError('Failed to read %s bytes from file' % size)
+
+
+def escape_ctrl_chars(s):
+    """Escape control characters from unicode and string objects."""
+    if isinstance(s, unicode):
+        return "".join(ch.encode("unicode_escape") if (
+            unicodedata.category(ch)[0]) == "C" else ch for ch in s)
+    if isinstance(s, basestring):
+        return "".join(
+            [c if 31 < ord(c) < 127 else c.encode("string_escape") for c in s])
+        return s

kamaki/clients/utils/test.py

@@ -35,6 +35,7 @@
 
 from unittest import TestCase
 from tempfile import TemporaryFile
+from itertools import product
 
 from kamaki.clients import utils
 
@@ -140,6 +141,24 @@ def test_readall(self):
             self.assertEqual(utils.readall(f, 1), '')
             self.assertRaises(IOError, utils.readall, f, 1, 0)
 
+    def test_escape_ctrl_chars(self):
+        gr_synnefo = u'\u03c3\u03cd\u03bd\u03bd\u03b5\u03c6\u03bf'
+        gr_kamaki = u'\u03ba\u03b1\u03bc\u03ac\u03ba\u03b9'
+
+        char_pairs = (
+            ('\b', '\\x08'), ('\n', '\\n'), ('\a', '\\x07'), ('\f', '\\x0c'),
+            ('\t', '\\t'), ('\v', '\\x0b'), ('\r', '\\r'), ('\072', ':'),
+            ('\016', '\\x0e'), ('\\', '\\'), ('\\n', '\\n'), ("'", '\''),
+            ('"', '"'),
+        )
+
+        for orig_char, esc_char in char_pairs:
+            for word1, word2 in product(
+                    ('synnefo', gr_kamaki), ('kamaki', gr_synnefo)):
+                orig_str = word1 + orig_char + word2
+                esc_str = word1 + esc_char + word2
+                self.assertEqual(utils.escape_ctrl_chars(orig_str), esc_str)
+
 if __name__ == '__main__':
     from sys import argv
     from kamaki.clients.test import runTestCase