Tab Spacing in YAML based DSL

[GSri30]

I am trying to implement a DSL (YAML based) which looks something like this

guid: 9c358115-372e-4da6-aad2-ed9dc01bec20
table:
    name: rls_date_general
    db: rls_test
    schema: falcon_default_schema 
    db_table: rls_date_general

Now the following is the grammar I wrote

Model:
    objectEDocs+=ObjectEDoc*;

ObjectEDoc:
    GUIDElement | TableElement
;

GUIDElement:
	'guid' ':' guid=GUID
;

TableElement:
	'table' ':' table=TABLE
;

TABLE:
	table_rows+=IndentedTableRow+;

IndentedTableRow:
    '\t' table_row=TableRow;

TableRow:
	(table_key=TABLE_KEY ':' table_value=TABLE_VALUE)
;

terminal TABLE_KEY:
	'name' | 'db' | 'schema' | 'db_table';

terminal TABLE_VALUE:
	ID
;

When I am trying to introduce IndentedTableRow rule, the tests are failing for the above sample text. But I want to have a strict indentation rules setup in my grammar.

Can anyone provide me any leads over this.

[cdietrich]

Did you check out https://www.eclipse.org/Xtext/documentation/307_special_languages.html#whitespace-aware-languages

[GSri30]

Yeah. But couldn't find any leads.

[cdietrich]

what do you mean by "lead"?

[GSri30]

I couldn't figure out why I'm getting the above error

[cdietrich]

did you try to use the documented begin and end tokens instead of hardcoding tab into keyword?

[GSri30]

Yeah, using begin and end, it's working properly.

Though, the problem now I'm facing is that when I use some nested key-value pairs using begin and end, its not able to identify the same key names.

Example

guid: 9c358115-372e-4da6-aad2-ed9dc01bec20
table:
    name: rls_date_general
    db: rls_test
    schema: falcon_default_schema 
    db_table: rls_date_general
    columns:
        name: some_name

I'm not able to distinguish between the name key inside columns and inside table.

[cdietrich]

Can you give sample grammar and unit tests

[GSri30]

Grammar

Model:
    objectEDocs+=ObjectEDoc*;

ObjectEDoc:
    GUIDElement | TableElement
;

GUIDElement:
	'guid' ':' guid=GUID
;

TableElement:
	'table' ':' table=TABLE
;

TABLE:
	BEGIN
		(table_rows+=TableRow)+
	END
;

TableRow:
	(table_key=TABLE_KEY ':' table_value=TableValue)
;

terminal TABLE_KEY:
	'name' | 'db' | 'columns';

TableValue:
	{ID} id=ID | COLUMNS
;

COLUMNS:
	(columns+=COLUMN)+
;

COLUMN: 
	'-'
	BEGIN
	(column_rows+=ColumnRows)+
	END
;

ColumnRows:
	((column_key=COLUMN_KEY ':' column_value=ID))
;

terminal COLUMN_KEY:
	'name' | 'properties'
;

terminal GUID:
	FRAGMENT FRAGMENT FRAGMENT FRAGMENT FRAGMENT FRAGMENT FRAGMENT FRAGMENT '-' 
	FRAGMENT FRAGMENT FRAGMENT FRAGMENT '-'
	FRAGMENT FRAGMENT FRAGMENT FRAGMENT '-'
	FRAGMENT FRAGMENT FRAGMENT FRAGMENT '-'
	FRAGMENT FRAGMENT FRAGMENT FRAGMENT FRAGMENT FRAGMENT FRAGMENT FRAGMENT FRAGMENT FRAGMENT FRAGMENT FRAGMENT;

terminal fragment FRAGMENT:
	(LETTER|DIGIT)
;

terminal fragment DIGIT:
	('0'..'9')
;

terminal fragment LETTER:
	('a'..'z')
	;

terminal BEGIN: 'synthetic:BEGIN';
terminal END: 'synthetic:END';

Testcase

			guid: 9c358115-372e-4da6-aad2-ed9dc01bec20
			table:
				name: rls_date_general
				db: rls_test
				columns:
				-
					name: some_name 

But the grammar is working properly by making the following changes in

grammar

terminal COLUMN_KEY:
	'nameOther' | 'properties'
;

Testcase

					nameOther: some_name 

[cdietrich]

maybe the problem is you try to do everything with terminals.
the lexer is context free and can lex "name" only as one kind of token.
please have a look at the concept of DataType rules