Rich file comparison with a focus on structured and tabular data
This a WIP implementation of Myers' algorithm for constructing meaningful diffs beyond text files.
pip install git+git://github.com/pulkin/rdiff.git
WIP
-
sequence comparison
from rdiff.sequence import diff print(diff(['apples', 'bananas', 'carrots', 'dill'], ['apples', 'carrots', 'dill', 'eggplant']).to_string())
Diff(0.7500): # 6 out of 8 tokens in both sequences were found aligned ··a=b: ['apples'] # ['apples'] slice of both lists is aligned ··a≠b: ['bananas'] ≠ [] # but ['bananas'] is misaligned with an empty slice [] ··a=b: ['carrots', 'dill'] ··a≠b: [] ≠ ['eggplant'] -
nested sequences
from rdiff.sequence import diff_nested print(diff_nested([0, 1, ["alice", "bob"]], [0, 1, ["alice2", "bob2"]]).to_string())
Diff(1.0000): # Diff(1.0) does not necessarily mean sequences are equal exactly ··a≈b: # it rather means that all elements can be aligned ····a=b 0 # elements 1 and 2 are equal exactly ····a=b 1 ····Diff(1.0000): # element 3 can be aligned ······a≈b: ········Diff(0.9091): # characters in "alice" and "alice2" cannot be aligned exactly ··········a=b: 'alice' # (thus, Diff(0.9091)) ··········a≠b: '' ≠ '2' # but, as objects, they are aligned in the respecting containers ········Diff(0.8571): ··········a=b: 'bob' ··········a≠b: '' ≠ '2'