Feature/insert row key

[jbyte]

Description (what is it you've done, has anything changed?)

• updated mysql#PostBufferFormat so it automatically deletes all lines before the expected output (everything before a line containing +----- for all SQH filetypes except SQHInsert, or everything before a line containing , for SQHInsert filetype) -- this could probably use some more tweeking
• added function mysql#GetTableInfoFromQuery(query): returns list [database, table] names (used in mysql#EditRow and mysql#InsertRow
• added function mysql#CreateInsertFromCSV(): creates an INSERT sql statement from the csv in the current buffer
• added function mysql#InsertRow(): opens a SQHInsert buffer with the headings in csv format
• added key mapping for calling provider non-specific xxxx#InsertRow function
• added simple vader tests for mysql#CreateInsertFromCSV and mysql#CreateUpdateFromCSV

Checklist

• I have written my code in a way that it can be easily tested
• I have written tests for my code
• I have documented new features in doc/sqhell.txt
• I have added information to the README (where relevant)
• I have ran the automated tests and performed manual tests.

[jbyte]

Well I also fixed the travis issue where it didn't use the correct branch/pull request for Vader tests.
Although the mac tests still fail because the macOS sort works different than the linux one.

[joereynolds]

Hey @jbyte

All looks good, I'm going to fix the Mac Vs Linux sort bug before merging any pull requests

Once I've done that I'll comment on here

[jbyte]

I'm fine with that.

[joereynolds]

Hey @jbyte,

I've fixed the tests/code, can you pull down master and try again?

[jbyte]

The tests are fine now 👍

[joereynolds]

I think these are modifying the jump list, meaning if I do <c-o> or <c-i> I need to press them a few times, can we avoid adding to the jump list?

[joereynolds]

When I try to ZZ on an i in the SQHResult buffer with this data

id,prefix,location,date,artist,url
"22", "With", "Wales", "2018-07-03", "The Three Stooges", "http://google.com"

I get

Incorrect number of values.
Expected: 6, got: 11.

[jbyte]

It should be fixed now.

[joereynolds]

I think now that we do everything in one buffer, your calls to :bd can be removed otherwise we get some undesirable jumps. For example if I do the following

SQHDatabase -> SQHTable -> SQHResult -> e to edit result -> <c-o>

Instead of going back to the SQHResult as expected, it goes to SQHTable

This also happens on the i to insert a new record.

Looking good so far though :D

[jbyte]

Well I wouldn't remove them, but change them. Because they are used for "reloading" the buffer.
Example:
SQHResult -> e to edit -> edit data -> ZZ to save and close -> updated SQHResult
I'll just change the :bd to something like dG and put newresults

Edit:
Never mind it's fine to remove them, I'll just move the "reloading" functionality to ZZ in SQHResult

Edit2:
The :bd in mysql#EditRow and mysql#InsertRow should be changed and moved to ZZ.
The :bd in mysql#DropDatabase and mysql#DropTableFromDatabase should be changed in place.

But in the end I think this would need quite a bit of refactoring so maybe should be its own PR?

[jbyte]

Thoughts on the refactoring: #46.

[joereynolds]

Sounds good to me, we can test the query generation better this way too.

Do we need to delete all lines in the buffer each time? We use enew! inside the InsertResultsToNewBuffer function anyway so the previous buffer would be removed (though still available in the jump list I guess).

Also it sounds like the pros outweight the cons so feel free to take a stab it, I would definitely like a cleaner codebase for this