File comparison can include regex scrubbing to ignore certain parts.

1 files changed, 16 insertions, 2 deletions

diff --git a/test/test_farm.py b/test/test_farm.py
index 01934df..a20a9a2 100644
--- a/test/test_farm.py
+++ b/test/test_farm.py
@@ -1,6 +1,6 @@
 """Run tests in the farm subdirectory.  Designed for nose."""
 
-import difflib, filecmp, fnmatch, glob, os, shutil, sys
+import difflib, filecmp, fnmatch, glob, os, re, shutil, sys
 
 sys.path.insert(0, os.path.split(__file__)[0]) # Force relative import for Py3k
 from backtest import run_command, execfile # pylint: disable-msg=W0622
@@ -158,7 +158,7 @@ class FarmTestCase(object):
             self.cd(cwd)
 
     def compare(self, dir1, dir2, filepattern=None, size_within=0,
-            left_extra=False, right_extra=False
+            left_extra=False, right_extra=False, scrubs=None
             ):
         """Compare files matching `filepattern` in `dir1` and `dir2`.
         
@@ -177,6 +177,9 @@ class FarmTestCase(object):
         without triggering an assertion.  `right_extra` means the right
         directory can.
         
+        `scrubs` is a list of pairs, regex find and replace patterns to use to
+        scrub the files of unimportant differences.
+        
         An assertion will be raised if the directories fail one of their
         matches.
         
@@ -220,6 +223,9 @@ class FarmTestCase(object):
             for f in diff_files:
                 left = open(os.path.join(dir1, f), "r").readlines()
                 right = open(os.path.join(dir2, f), "r").readlines()
+                if scrubs:
+                    left = self._scrub(left, scrubs)
+                    right = self._scrub(right, scrubs)
                 if left != right:
                     text_diff.append(f)
                     print("".join(list(difflib.Differ().compare(left, right))))
@@ -230,6 +236,14 @@ class FarmTestCase(object):
         if not right_extra:
             assert not right_only, "Files in %s only: %s" % (dir2, right_only)
 
+    def _scrub(self, strlist, scrubs):
+        scrubbed = []
+        for s in strlist:
+            for rgx_find, rgx_replace in scrubs:
+                s = re.sub(rgx_find, rgx_replace, s)
+            scrubbed.append(s)
+        return scrubbed
+
     def contains(self, filename, *strlist):
         """Check that the file contains all of a list of strings.