Indent

In autobean-refactor, indent is not considered spacing and handled separately.

The beancount indentation

The indentation in beancount v2 is handled in a somewhat surprising way. The only distinction is whether a line is indented or not, while the indentation characters / levels don’t matter.

Valid	Formatted
2000-01-01 * foo: 1 Assets:Foo 100.00 USD Assets:Bar -100.00 USD bar: 4	2000-01-01 * foo: 1 Assets:Foo 100.00 USD Assets:Bar -100.00 USD bar: 4

Basic access

In autobean-refactor, indent is a children of indentable models (e.g. Posting, Meta, BlockComment), which can be accessed as a simple string model.

txn = p.parse('''\
2000-01-01 *
    foo: 1
   Assets:Foo 100.00 USD
  Assets:Bar -100.00 USD
 bar: 4''', models.Transaction)
txn.postings[1].indent

'  '

txn.postings[1].indent = ' ' * 4
_print_model(txn)

2000-01-01 *
    foo: 1
   Assets:Foo 100.00 USD
    Assets:Bar -100.00 USD
 bar: 4

Note that the indent of each line is independent and therefore changing the indent of a posting will have no impact on its meta.

indent_by

When new children are added to a repeated field, its indent is determined as follows:

• For raw models, their own indent is used as is;

• Otherwise, if there are existing children, the indent of the last one is copied over;

• Otherwise, the indent is constructed from the parent’s indentation and its indent_by.

indent_by defaults to four spaces and can be explicitly set at construction in from_children or from_value, or after construction with the indent_by attribute.

txn = p.parse('2000-01-01 *', models.Transaction)
txn.indent_by = ' ' * 2
txn.meta['foo'] = 'foo'
txn.indent_by = ' ' * 8
txn.meta['bar'] = 'bar'
_print_model(txn)

2000-01-01 *
  foo: "foo"
  bar: "bar"

txn.meta.clear()
txn.meta['bar'] = 'bar'
_print_model(txn)

2000-01-01 *
        bar: "bar"

txn.meta.clear()
meta_item = models.MetaItem.from_value(
    key='qux',
    value=decimal.Decimal(4),
    indent=' ' * 1)
txn.raw_meta.append(meta_item)
_print_model(txn)

2000-01-01 *
 qux: 4