[tor-commits] [ooni-probe/master] Moved parsing logic for top-1m asset to Top1MAsset class in DNS test

[art at torproject.org]

commit bf3a0362a7cf6d4b22a600a17c7ea906b3ec48b8
Author: Isis Lovecruft <isis at patternsinthevoid.net>
Date:   Wed Mar 14 23:26:24 2012 -0700

    Moved parsing logic for top-1m asset to Top1MAsset class in DNS test
---
 assets/fix-top-1m.py |   34 ----------------------------------
 tests/dnstamper.py   |   12 +++++++++++-
 2 files changed, 11 insertions(+), 35 deletions(-)

diff --git a/assets/fix-top-1m.py b/assets/fix-top-1m.py
deleted file mode 100755
index 9e6b681..0000000
--- a/assets/fix-top-1m.py
+++ /dev/null
@@ -1,34 +0,0 @@
-#!/usr/bin/env python
-###############
-# fix-top-1m.py
-###############
-# Removes the 'N,' from the default top-1m.txt.
-# ::author:: Isis Lovecruft
-
-from __future__ import with_statement
-from contextlib import nested
-import os
-import re
-import tempfile
-
-def remove_cruft():
-    print 'Opening list of top webservers for parsing...'
-    in_file = os.path.abspath('top-1m.txt')
-    print 'Creating temporary file for parsing purposes...'
-    outfile = os.path.abspath('top-1m.txt~')
-    tempfile.NamedTemporaryFile('w+b', 1000, '.txt~', 'top-1m',
-                                os.getcwd())
-    with open(in_file, 'r+') as in_, open('top-1m.txt~', 'w+') as out_:
-        lines = in_.readlines()
-        print 'Replacing cruft with nothingness...'
-        for line in lines:
-            out_.write(re.sub('^(\d+),', '', line))
-        print 'Removing write access to temp file...'
-    out_.close()
-    in_.close()
-    print 'Renaming parsed temp file as orginal file...'
-    os.rename('top-1m.txt~', 'top-1m.txt')
-    print 'List of webservers now ready for use!'
-
-if __name__=="__main__":
-    remove_cruft()
diff --git a/tests/dnstamper.py b/tests/dnstamper.py
index 23318b6..1e59d26 100644
--- a/tests/dnstamper.py
+++ b/tests/dnstamper.py
@@ -7,11 +7,21 @@ import os
 import plugoo
 from plugoo.assets import Asset
 from plugoo.tests import Test
-
+from __future__ import with_statement
 
 __plugoo__ = "DNST"
 __desc__ = "DNS censorship detection test"
 
+class Top1MAsset(Asset):
+    def __init__(self, file=None):
+        self = Asset.__init__(self, file)
+    
+    def next_asset(self):
+        with self.fh as asset:
+            lines = asset.readlines()
+            for line in lines:
+                yield line.split(',')[1]
+
 class DNSTAsset(Asset):
     def __init__(self, file=None):
         self = Asset.__init__(self, file)