CSVREAD looses column (1.4.196)

[George Ivanov]

Hello!

I am trying to create a table using the following CSV file (test.csv):

===

,,

0,1,2

===

CREATE TABLE A AS SELECT * FROM CSVREAD('test.csv');

SELECT * FROM A

However, in the result I see that the third column (COLUMN2) gets lost:

Could you please check?

Thank you,

George

[Evgenij Ryazanov]

Hello.

Yes, it looks like a bug. You can use some non-empty name of the last column as a workaround.

A,B,C

0, 1, 2

[Noel Grandin]

possibly we just need an error message when the number of columns in the first line does not match the number of columns
in the data rows

[Evgenij Ryazanov]

Both lines have 3 columns, it's just a bug in Csv.readHeader() / Csv.readValue().

[George Ivanov]

I cannot use empty name, the empty value comes from Confluence page table where the header cell is empty

[George Ivanov]

Is the same bug present in the latest H2 release?

[George Ivanov]

Sorry, I meant, I cannot replace the empty name, since it is empty in the Confluence table that is used as the data source in this case, and I am not necessarily the owner of this table.

On Thursday, August 8, 2019 at 7:01:48 PM UTC+3, George Ivanov wrote:

[George Ivanov]

Hi Evgenij,

could you confirm which version of H2 will have this bug fixed, please?

Thank you!

George

[Evgenij Ryazanov]

Sorry, it isn't fixed yet.

Of course, you may submit a pull request with a fix if you wish. The relevant code is in org.h2.tools.Csv.

[George Ivanov]

OK, the problem is that when readValue() does:

        while (true) {

            int ch = readChar();

it gets first ch == fieldSeparatorRead ...

I think, the right way of fixing would be somehow escaping "empty" values with "" before that point. 

So, that when you read buffer, it would read 

,

as

"",""

However, I still have no clue how to achieve this within H2 code.

[George Ivanov]

hi Evgenij,

could you help with fixing this?

[Evgenij Ryazanov]

Hello.

CSVREAD needs many other changes and I don't have time for them right now.