Automating a directory diff while ignoring some particular lines in files

Question

I need to compare two directories, and produce some sort of structured output (text file is fine) of the differences. That is, the output might looks something like this:

file1 exists only in directory2
file2 exists only in directory1
file3 is different between directory1 and directory2

I don't care about the format, so long as the information is there. The second requirement is that I need to be able to ignore certain character sequences when diffing two files. Araxis Merge has this ability: you can type in a Regex and any files whose only difference is in character sequences matching that Regex will be reported as identical.

That would make Araxis Merge a good candidate, BUT, as of yet I have found no way to produce a structured output of the diff. Even when launching consolecompare.exe with command-line argumetns, it just opens an Araxis GUI window showing the differences.

So, does either of the following exist?

• A way to get Araxis Merge to print a diff result to a text file?
• Another utility that do a diff while ignoring certain character sequences, and produce structured output?

Extra credit if such a utility exists as a module or plugin for Python. Please keep in mind this must be done entirely from a command line / python script - no GUIs.

Answer 1

To some extent, the plain old diff command can do just that, i.e. compare directory contents and ignoring changes that match a certain regex pattern (Using the -I option).

From man bash:

-I regexp
      Ignore changes that just insert or delete lines that match  regexp.

Quick demo:

[me@home]$ diff images/ images2
Only in images2: x
Only in images/: y
diff images/z images2/z
1c1
< zzz
---
> zzzyy2

[me@home]$ # a less verbose version
[me@home]$ diff -q images/ images2
Only in images2: x
Only in images/: y
Files images/z and images2/z differ

[me@home]$ # ignore diffs on lines that contain "zzz"
[me@home]$ diff -q -I ".*zzz.*" images/ images2/
Only in images2/: x
Only in images/: y