Compares two XML documents by diffing their text, ignoring structure, and wraps changed text in <del>/<ins> tags.
Compares the text inside two XML documents and marks up the differences with <del> and <ins> tags.
This is the result of about 7 years of trying to get this right and coded simply. I’ve used code like this in one form or another to compare bill text on GovTrack.us <https://www.govtrack.us>.
The comparison is completely blind to the structure of the two XML documents. It does a word-by-word comparison on the text content only, and then it goes back into the original documents and wraps changed text in new <del> and <ins> wrapper elements.
The documents are then concatenated to form a new document and the new document is printed on standard output. Or use this as a library and call compare yourself with two lxml.etree.Element nodes (the roots of your documents).
The script is written in Python 3.
Comparing these two documents:
<html> Here is <b>some bold</b> text. </html>
<html> Here is <i>some italic</i> content that shows how <tt>xml_diff</tt> works. </html>
<documents> <html> Here is <b>some <del>bold</del></b><del> text</del>. </html> <html> Here is <i>some <ins>italic</ins></i><ins> content that shows how </ins><tt><ins>xml_diff</ins></tt><ins> works</ins>. </html> </documents>
On Ubuntu, get dependencies with:
apt-get install python3-lxml libxml2-dev libxslt1-dev
For really fast comparisons, get Google’s Diff Match Patch library <https://code.google.com/p/google-diff-match-patch/>, as re-written and sped-up by @leutloff <https://github.com/leutloff/diff-match-patch-cpp-stl> and then turned into a Python extension module by me <https://github.com/JoshData/diff_match_patch-python>:
pip3 install diff_match_patch_python
Or if you can’t install that for any reason, use the pure-Python library:
pip3 install diff-match-patch
This is also at <https://code.google.com/p/google-diff-match-patch/source/browse/trunk/python3/diff_match_patch.py>. xml_diff will use whichever is installed.
Finally, install this module:
pip3 install xml_diff
Then call the module from the command line:
python3 -m xml_diff --tags del,ins doc1.xml doc2.xml > changes.xml
Or use the module from Python:
import lxml.etree from xml_diff import compare dom1 = lxml.etree.parse("doc1.xml").getroot() dom2 = lxml.etree.parse("doc2.xml").getroot() comparison = compare(dom1, dom2)
The two DOMs are modified in-place.
The compare function takes other optional keyword arguments:
merge is a boolean (default false) that indicates whether the comparison function should perform a merge. If true, dom1 will contain not just <del> nodes but also <ins> nodes and, similarly, dom2 will contain not just <ins> nodes but also <del> nodes. Although the two DOMs will now contain the same semantic information about changes, and the same text content, each preserves their original structure — since the comparison is only over text and not structure. The new ins/del nodes contain content from the other document (including whole subtrees), and so there’s no guarantee that the final documents will conform to any particular structural schema after this operation.
word_separator_regex (default r"\s+|[^\s\w]") is a regular expression for how to separate words. The default splits on one or more spaces in a row and single instances of non-word characters.
differ is a function that takes two arguments (text1, text2) and returns an iterator over difference operations given as tuples of the form (operation, text_length), where operation is one of "=" (no change in text), "+" (text inserted into text2), or "-" (text deleted from text1). (See xml_diff/__init__.py’s default_differ function for how the default differ works.)
tags is a two-tuple of tag names to use for deleted and inserted content. The default is ('del', 'ins').
make_tag_func is a function that takes one argument, which is either "ins" or "del", and returns a new lxml.etree.Element to be inserted into the DOM to wrap changed content. If given, the tags argument is ignored.