ComparePlus - improve alignment of similar lines

Mark Olson

@Stefan-Pendl
Have you submitted an issue in the repo?

Stefan Pendl

@Mark-Olson
no, I am not a github member.
Just wanted to check if it would be considered.

TBugReporter

@Stefan-Pendl
Without an official issue at github, the chances of it being considered are zero. (With an official issue at gihub, the chances are only slightly higher.)

Mark Olson

Yeah, after looking at your proposal again, I strongly suspect that this particular feature proposal would not be accepted. AFAIK, ComparePlus does not use its own algorithms to calculate alignments and diffs; I think it just uses external dependencies for that. I could be wrong, but if I’m right, changes like you propose would require the core devs to hack some dependencies, which tends to be rather hairy and error-prone.

guy038

Hello, @stefan-pendl, @mark-olson, @tbugreporter and All,

I’ve tried to understand what @stefan-pendl meant but, even with the help of the Compare Plus plugin, I could not understand which kind of alignment was desired :-((

@stefan-pendl, could you try to shorten your File_1, to the approximative number of lines of File_2, in order to easily identify where alignments must occur, when running the Compare Plus plugin ?

Best Regards,

guy038

Stefan Pendl

@guy038
The problem is, that if I shorten the file, the issue is gone.
It only happens when the files are really different.

Mark Olson

@Stefan-Pendl said in ComparePlus - improve alignment of similar lines:

It only happens when the files are really different.

Well then, that’s the problem. Diff/alignment algorithms try their best, but if two files are really different, you shouldn’t be too surprised if it seems like the alignment is bad. Even Git’s diff algorithm, which is probably the best around, produces diffs/alignment that seem suboptimal when the diff is big.

wonkawilly

@Stefan-Pendl said in ComparePlus - improve alignment of similar lines:

I am not completely sure I understand the right way you mean about Improve alignment but maybe these suggestions can be also useful:
replace spaces with tabs and install and activate the Elastic Tab Stop Plugin: with it is possible to get alignment like the following screenshot:

If you prefer you can use another plugin: CsvQuery to get alignment as the following screenshot:

Right click on CsvQuery plugin window and show the line numbers like the following screenshot is also possible:

with CsvQuery You can also query the database using SQL syntax to filter the rows using criteria you need.

Stefan Pendl

@wonkawilly
Thanks for that, but I did not mean the alignment of columns, but the alignment of the diff result => lines starting with the same word should be aligned, not lines with a similar word.

mpheath

@Stefan-Pendl IMO, not the correct plugin for the job. Try PythonScript plugin.

Open the 2 documents in separate views (view1 (main) and view2 (extra)).
Run the script named DBMergeReport.py and it will open a new tab on the focused view with the merged report.

import re

def db_merge_report():
    def get_dic(text):
        dic = {}
        for m in re.finditer(r'^(\w+ \w+) ([\d.]+)\r{0,1}$', text, re.M):
            k = m.group(1)
            dic[k] = m.group(2)
        return dic

    text1 = editor1.getText()
    text2 = editor2.getText()
    dic_a = get_dic(text1)
    dic_b = get_dic(text2)
    names = set(dic_a) | set(dic_b)
    names = sorted(names)
    length = max(len(item) for item in names)
    text1 = ''

    for name in names:
        a, b = '', ''
        if name in dic_a:
            a = dic_a[name]
        if name in dic_b:
            b = dic_b[name]
        text1 += ('{:<' + str(length + 1) + '} {:<20} {}\n').format(name, a, b)

    if text1:
        notepad.new()
        editor.addText(text1)

if __name__ == '__main__':
    db_merge_report()

New document tab with just the 1st 5 lines as an example.

ACCT_TABLE_INDEX ACCT_TABLE                                 62.148458872879807   89.86486486486487
ACCT_TABLE_INDEX2 ACCT_TABLE                                39.810607669115463   70.634920634920633
ACCT_TABLE_INDEX3 ACCT_TABLE                                59.226190476190474   100.0
ACCT_TABLE_INDEX4 ACCT_TABLE                                10.735930735930737   
PIPABSOCCDATA PABSOCCDATA                                   88.099315068493155

The spacing is quite wide for the 1st column as the largest value is 58 characters. This might make comparison of the data being in 1 document alittle easier as the float numbers are next to each other.

pnedev

Hello @Stefan-Pendl ,

I will check your case and if some improvements could be made but it might take time.
Could you please in the meantime tell me which lines should be aligned according to your expectations (except lines PIPDATASET_0 - I already got that)?
Thanks.

BR

pnedev

@pnedev said in ComparePlus - improve alignment of similar lines:

Could you please in the meantime tell me which lines should be aligned according to your expectations

Hello again @Stefan-Pendl ,

I analyzed your example data and I understand what you would expect as a result of the comparison. Please disregard my request for further clarifications.

BR

Stefan Pendl

@pnedev
Thanks for checking this issue.
If you need more data just ask.

pnedev

Hello @Stefan-Pendl ,

I have just released ComparePlus v1.2.0 - it has a bug-fix for and overall improved changed lines detection algo - your example files compare as expected now.
It doesn’t mean that now the plugin will always consider the most intuitive lines as changed in all cases (that is simply not possible because of the different scenarios and user expectations) but it does a much better job now.

I am sorry about the long time it took for releasing this update - there were also other pending issues that required attention and I didn’t had enough spare time to work on ComparePlus.