Comments are being stripped

[bitslasher]

When clicking "Format SQL (selection)" any comments in the SQL is stripped. This includes /* multi-line */ and --single line comments

To Reproduce
Steps to reproduce the behavior:

1. Have a sql script with any kind of valid comment in it.
2. Click on "Format SQL (selection)"
3. See all comments have been removed.

Expected behavior
Code comments would remain untouched

[madskristensen]

Please add to this existing open bug on the SqlScriptDOM repo: microsoft/SqlScriptDOM#20

[bitslasher]

Thank you @madskristensen. I looked at that bug, it's quite fascinating. It seems their "white space" token filter is treating comments like white space, which is odd. I'm guessing folks have experimented with the obvious, which is removing those two token types for comments from the break condition.

[marcRobertHabib]

I think that it's crucial for any code to be commented. Is there some workarround for this?

[rwestMSFT]

There is no workaround for this. The SqlScriptDOM project is stripping out comments.

[ErikEJ]

@rwestMSFT But could it be changed to not do that (optionslly) ??

[rwestMSFT]

This doesn't seem like a conversation for this issue on this repository, wouldn't you say?

[ErikEJ]

@rwestMSFT Sorry, I missed that you were replying to an older question.

[daiplusplus]

There is no workaround for this. The SqlScriptDOM project is stripping out comments.

There kinda is a possible workaround, provided the SQL-formatting process doesn't remove or re-order keywords or identifiers.

1. Prior to sending the SQL text to the formatter, use regexen to capture all comments in the text, including the character indexes of the start and end of the comments, then anchors those to adjacent keywords or identifiers (assuming that the formatting process would not re-order those anchors).
2. Run the formatting process, which strips comments.
3. Look for the anchors in the formatted output and re-insert the comments accordingly.

• Some tweaking may be required such as converting single-line -- comments to /**/ comments if the formatting removed a comment-terminating newline; and consecutive comments would need to be coalesced into a single comment for lack of adjacent anchors. This process also assumes the comment regexes used exactly match T-SQL's rules, especially the handling of nested comments.

This workaround suggestion will fail in a few places, such as if a comment is surrounded by non-keyword/non-identifier T-SQL tokens like symbols/punctuation and significant parentheses - but in that case the SqlFormatter could gather-up all now-orphaned comments (and their surrounding text context) and write them all at the bottom of the output with an an apology note.

[rwestMSFT]

There kinda is a possible workaround, provided the SQL-formatting process doesn't remove or re-order keywords or identifiers.

I know, because I did it internally at Microsoft for our article maintenance tooling. It does remove and reorder keywords. Just one example: DELETE FROM is changed to DELETE. It'll be much better if it's fixed at source.

[ErikEJ]

Finally some progress!

microsoft/SqlScriptDOM#172

[marcRobertHabib]

Thanks.

I installed last sqlFormatter version from VS marketplace and comments are till removed. Have I something to do ?

[ErikEJ]

@marcRobertHabib Wait. Nothing has been merged yet.