Hi Kasper,
After some fight, I was able to write on an empty CSV, I needed to change
the :
CsvCreateTableBuilder.java
public Table execute() {
CsvUpdateCallback csvUpdateCallback = getUpdateCallback();
MutableTable table = getTable();
String[] columnNames = table.getColumnNames();
CsvDataContext csvDataContext =
(CsvDataContext)csvUpdateCallback.getDataContext();
if(csvDataContext.getConfiguration().getColumnNameLineNumber() !=
CsvConfiguration.NO_COLUMN_NAME_LINE) {
csvUpdateCallback.writeRow(columnNames, false);
}
CsvSchema schema = (CsvSchema) table.getSchema();
CsvTable csvTable = new CsvTable(schema, table.getName(),
table.getColumnNames());
schema.setTable(csvTable);
return csvTable;
}
Then I've changed the
CsvDataContextTest.testEmptyFile
public void testEmptyFile() throws Exception {
CsvConfiguration csvConfiguration =
new CsvConfiguration(CsvConfiguration.NO_COLUMN_NAME_LINE,
FileHelper.DEFAULT_ENCODING, CsvConfiguration.DEFAULT_SEPARATOR_CHAR,
CsvConfiguration.NOT_A_CHAR,
CsvConfiguration.DEFAULT_ESCAPE_CHAR);
final CsvDataContext dc = new CsvDataContext(new
File("src/test/resources/empty_file.csv"), csvConfiguration);
assertEquals(1, dc.getDefaultSchema().getTableCount());
dc.executeUpdate(new UpdateScript() {
@Override
public void run(UpdateCallback callback) {
callback.createTable(dc.getDefaultSchema(),
"new_table").withColumn("COL_1").withColumn("COL_2").execute();
callback.insertInto("new_table").value(0,
"1").value(1, 2).execute();
}
});
CsvDataContext dc1 = new CsvDataContext(new
File("src/test/resources/empty_file.csv"), csvConfiguration);
System.out.println(dc1.query().from("empty_file.csv").selectAll().execute().toRows().size());
Table table = dc1.getDefaultSchema().getTables()[0];
assertEquals("empty_file.csv", table.getName());
assertEquals(0, table.getColumnCount());
}
The test is breaking, but it wrote the lines in the empty csv.
One thing I notice is that when we call the dropTable, if the
DataContext was created pointing to a file, it simply delete the file,
and that's the why I'm not calling it.
Do you think it is working as expected after this change?
Thanks,
Junior
2014-07-16 18:29 GMT-03:00 Kasper Sørensen <[email protected]>:
> I think you're right about that. Would you be able to maybe give it a shot?
> We can add a unittest for this particular case to show if it works.
>
>
> 2014-07-16 23:26 GMT+02:00 Júnior <[email protected]>:
>
> > As I could see it is not possible to create without header because of
> this
> > line on CsvCreateTableBuilder:
> >
> > csvUpdateCallback.writeRow(columnNames, false);
> >
> > on execute method, it is always creating the header column, maybe just a
> if
> > before that would fix that
> >
> >
> >
> > 2014-07-16 18:22 GMT-03:00 Kasper Sørensen <
> [email protected]
> > >:
> >
> > > I'm just now simultaniously looking into the MM csv code ... I get the
> > > feeling you're maybe a bit in untested territory. At least it seems
> that
> > > CsvDataContextTest does not exhibit a lot of test cases around having
> no
> > > column headers ... Might be we need your feedback here if there are
> > issues.
> > >
> > >
> > > 2014-07-16 23:19 GMT+02:00 Kasper Sørensen <
> > [email protected]
> > > >:
> > >
> > > > Ah, you want a header-less CSV file? You then need to instantiate
> your
> > > > CsvDataContext with a CsvConfiguration that
> > > > specifies CsvConfiguration.NO_COLUMN_NAME_LINE as the
> > > > 'columnNameLineNumber' property. If you otherwise want defaults, that
> > > would
> > > > be like this:
> > > >
> > > > new CsvConfiguration(CsvConfiguration.NO_COLUMN_NAME_LINE);
> > > >
> > > >
> > > > 2014-07-16 23:14 GMT+02:00 Júnior <[email protected]>:
> > > >
> > > > Right, what I'd like to know is to create the table without the
> > header, I
> > > >> think the only way would be changing the CsvCreateTableBuilder.
> > > >> Thanks,
> > > >>
> > > >>
> > > >>
> > > >> 2014-07-16 18:12 GMT-03:00 Kasper Sørensen <
> > > >> [email protected]>:
> > > >>
> > > >> > Hmm actually I just now realized that you can completely omit the
> > Drop
> > > >> > Table part ... if you use create table on an empty file, it will
> > > simply
> > > >> > create the header.
> > > >> >
> > > >> >
> > > >> > 2014-07-16 23:10 GMT+02:00 Kasper Sørensen <
> > > >> [email protected]
> > > >> > >:
> > > >> >
> > > >> > > Yes, you do that while creating the table. Like this:
> > > >> > >
> > > >> > > CsvDataContext dc = ...;
> > > >> > > dc.executeUpdate(new UpdateScript() {
> > > >> > > @Override
> > > >> > > public void run(UpdateCallback callback) {
> > > >> > >
> > > callback.dropTable(dc.getDefaultSchema().getTable(0));
> > > >> > >
> > > >> > > callback.createTable(dc.getDefaultSchema(),
> > > >> > > "newtable").withColumn("foo").withColumn("bar").execute();
> > > >> > > }
> > > >> > > });
> > > >> > >
> > > >> > > (replace with your own column names of course)
> > > >> > >
> > > >> > >
> > > >> > > 2014-07-16 23:04 GMT+02:00 Júnior <[email protected]>:
> > > >> > >
> > > >> > > Hi Kasper,
> > > >> > >>
> > > >> > >> thanks for your reply, but what about the columns?
> > > >> > >>
> > > >> > >> Do I need to add columns to this newly created table?
> > > >> > >>
> > > >> > >> Att.
> > > >> > >>
> > > >> > >>
> > > >> > >> 2014-07-16 18:00 GMT-03:00 Kasper Sørensen <
> > > >> > >> [email protected]>:
> > > >> > >>
> > > >> > >> > Hi there,
> > > >> > >> >
> > > >> > >> > There is a little known trick to get around that issue, so I
> > > guess
> > > >> > it's
> > > >> > >> a
> > > >> > >> > very fair question.
> > > >> > >> >
> > > >> > >> > The trick is to first do a drop table, and then a new create
> > > table.
> > > >> > >> >
> > > >> > >> > When MM is initialized with an existing empty file, it will
> > > >> determine
> > > >> > >> that
> > > >> > >> > there IS a table, since the file is there, but that table has
> > no
> > > >> > >> columns,
> > > >> > >> > because the file does not define any columns. Maybe it's a
> bit
> > > >> silly
> > > >> > - I
> > > >> > >> > actually think that's worth discussing ... But the reason it
> is
> > > >> there
> > > >> > >> is I
> > > >> > >> > guess to indicate that at least the file is there, it's not a
> > > >> > >> non-existing
> > > >> > >> > file (in which case there would not be any table).
> > > >> > >> >
> > > >> > >> > Best regards,
> > > >> > >> > Kasper
> > > >> > >> >
> > > >> > >> >
> > > >> > >> > 2014-07-16 21:37 GMT+02:00 Júnior <[email protected]>:
> > > >> > >> >
> > > >> > >> > > Hi,
> > > >> > >> > >
> > > >> > >> > > I'm trying to use the metamodel to write on an empty csv,
> > > >> > >> > >
> > > >> > >> > > But I'm getting errors saying that there is no column.
> > > >> > >> > >
> > > >> > >> > > I was able to do that calling create table with a different
> > > name
> > > >> > from
> > > >> > >> the
> > > >> > >> > > file.
> > > >> > >> > > Then added the columns.
> > > >> > >> > >
> > > >> > >> > > I worked, but it created the header line with the column
> > names,
> > > >> > would
> > > >> > >> it
> > > >> > >> > be
> > > >> > >> > > possible to do that without adding the column names in the
> > csv
> > > >> file?
> > > >> > >> > >
> > > >> > >> > > Att.
> > > >> > >> > > --
> > > >> > >> > > Francisco Ribeiro
> > > >> > >> > > *SCEA|SCJP|SCWCD|IBM Certified SOA Associate*
> > > >> > >> > >
> > > >> > >> >
> > > >> > >>
> > > >> > >>
> > > >> > >>
> > > >> > >> --
> > > >> > >> Francisco Ribeiro
> > > >> > >> *SCEA|SCJP|SCWCD|IBM Certified SOA Associate*
> > > >> > >>
> > > >> > >
> > > >> > >
> > > >> >
> > > >>
> > > >>
> > > >>
> > > >> --
> > > >> Francisco Ribeiro
> > > >> *SCEA|SCJP|SCWCD|IBM Certified SOA Associate*
> > > >>
> > > >
> > > >
> > >
> >
> >
> >
> > --
> > Francisco Ribeiro
> > *SCEA|SCJP|SCWCD|IBM Certified SOA Associate*
> >
>
--
Francisco Ribeiro
*SCEA|SCJP|SCWCD|IBM Certified SOA Associate*
