This is a Python 3.6+ package that wraps google-diff-match-patch's C++ implementation for performing very fast string comparisons. This package was previously known as diff_match_patch_python.
google-diff-match-patch is a Google library for computing differences between text files (http://code.google.com/p/google-diff-match-patch). There are implementations in various languages. Although there is a Python port, it's slow on very large documents, and I have a need for speed. I wanted to use the C++ implementation, but I'm a Python guy so I'd prefer to use it from Python.
Google's library depends on Qt 4, so some other folks rewrote it using the standard C++ library classes instead, making it more portable. That's at https://github.com/leutloff/diff-match-patch-cpp-stl. This package uses that library.
(This package can hopefully still be built for Python 2.7 also but a pre-built package is not available.)
pip3 install fast_diff_match_patch
Then write (this is Python 3):
from fast_diff_match_patch import diff changes = diff("Hello world.", "Goodbye moon.") for op, length in changes: if op == "-": print ("next", length, "characters are deleted") if op == "=": print ("next", length, "characters are in common") if op == "+": print ("next", length, "characters are inserted")
The two textual arguments can be either strings or bytes (Unicode or str on Python 2.x).
Some keyword arguments are also available:
timelimit (default 0) gives the maximum running time in seconds if you
want to ensure the result comes quickly. According to the Google docs,
the diff will stop working after the time is exceeded and will return a
valid diff, but it might not be the best one.
checklines is also a
Google thing and might speed up diffs that are over lined-based text
True) is the same argument in the diff_main
subroutine of the main library.
to run the corresponding cleanup subroutine after performing the diff.
False to have the returned value be an array of
tuples of operations and corresponding strings rather than operations
and the lengths of those strings.
True, the diff is returned in patch format
as a string.
On Windows, an exception will be thrown if either of the two text strings has characters
outside of the Basic Multilingual Plane because the native platform character type
is a two-byte character. The
fast_diff_match_patch.CHAR_WIDTH field, which will either
be 2 or 4, can be used to determine whether these characters are supported ahead of time.
The Global Interpreter Lock (GIL) is released while performing the diff so that this library can be used in a multi-threaded application.
fast_diff_match_patchto avoid an import naming collision with https://pypi.org/project/diff-match-patch/ and the package name has been updated to match the import name.
diff_str(Py2) methods were available. They have been merged into a single
diffmethod that checks the type of the arguments passed.)
cleanup_semantichas been renamed to
cleanup, which takes one of three options (see above)
To build from these sources, you will need:
git submodule update --init.
Then build/install the binary module using:
python setup.py build python setup.py install
To build everything (for testing):
git submodule update && rm -rf build && python3 setup.py build
To test without installing:
PYTHONPATH=build/lib.linux-x86_64-*/ python3 -m unittest
To upload a new release to PyPi:
python3 -m pip install --upgrade twine
python3 -m twine upload path-to-artifact-files/*. Username:
__token__; Password: A PyPi API token.