Firstly, this is a great library and it fits natively fit in our dev requirment, and it does make my life easier. But it seems that the algorithms being used does not handle the complex html structure (in my case table) very well.
During internal testing I came across quite a few instances where the diff shown on the table does not make sense. Problems include but not limited to
the deleted cell is shown on the wrong row
some existing css has been deleted
3.the header of the table become shorter in lengh
4: the table layout breaks
Not sure if there is any parameter or tuning options that I can leverage to better diff the table. ( I saw quite a few similar issues remain opened in this repo)
There only seems to be limited control exposed to be clients like OrphanMatchThreshold, however while the changing of it makes different on the text, it does not seem to change the output of table at all.
Here is one of the simple example.
The original table image:
The html code:
One of the highlighted diff cell is shown on the wrong row(the row next to the one actualy being deleted)
Is there any way to deal with situations like that? The layout tends to break when it comes to rows/column changes (addition&deletion or comination).
Some suggestions:
If we cannot always do a line-by-line diff correctly, can we do the diff on the whole table level?
can we just skip the table?
I am actually going to implement the skip logic my self by using some html parser to ignore the table..
mfaizan24
commented
2 years ago
Firstly, this is a great library and it fits natively fit in our dev requirment, and it does make my life easier. But it seems that the algorithms being used does not handle the complex html structure (in my case table) very well. During internal testing I came across quite a few instances where the diff shown on the table does not make sense. Problems include but not limited to
Not sure if there is any parameter or tuning options that I can leverage to better diff the table. ( I saw quite a few similar issues remain opened in this repo) There only seems to be limited control exposed to be clients like OrphanMatchThreshold, however while the changing of it makes different on the text, it does not seem to change the output of table at all. Here is one of the simple example. The original table image:
The html code:
After deleting one row:
The html:
The diff image
html:
One of the highlighted diff cell is shown on the wrong row(the row next to the one actualy being deleted)
Is there any way to deal with situations like that? The layout tends to break when it comes to rows/column changes (addition&deletion or comination). Some suggestions: