A block sequence entry cannot follow a flow collection that closed earlier
on the same line ('- { y: z }- invalid', '- [a] - b'): the entry would
begin a sequence inline after a complete node. The scanner accepted it,
producing extra sequence items.
The scanner now records when a flow collection closes at the top level on
the current line and rejects a following block entry. The flag resets at
each line break, so a flow collection on its own line is unaffected, and a
nested block entry ('- -') has no preceding flow close, so it stays valid.
Measured against the official YAML test-suite, parse-level conformance
rises from 276/333 (82.9%) to 277/333 (83.2%) with no regression.
Block style permits only one mapping pair per line, so a scalar followed
by ':' after a block-level value on the same line (a: b: c, a: 'b': c,
x: { y: z }in: valid, or trailing content after a quoted value) is not a
new key but a forbidden inline nested mapping. The scanner synthesized an
implicit key for any scalar followed by ':', regardless of position, so
these parsed as nested mappings instead of failing.
The scanner now records when a line already holds a block-level mapping
pair and suppresses a further implicit key on that line, in plain and
quoted scalars alike. The flag is set only for an implicit-key ':' (one
preceded by its key on the same line): a ':' that opens its line is an
explicit- or null-key value indicator, whose value may be a compact
mapping (? k then : a: 1), so that, flow mappings, next-line nesting, and
colons inside scalars stay valid.
Measured against the official YAML test-suite, parse-level conformance
rises from 256/333 (76.9%) to 260/333 (78.1%) with no regression.
The YAML grammar requires every directive to head an explicit document,
so a directive must be followed by a '---' marker. The scanner accepted a
directive with no document at all: '%YAML 1.2' alone, or '%YAML 1.2' then
a '...' document-end marker, parsed without error.
The scanner now records the line of a directive awaiting its document,
clears it at a document start, and rejects a still-pending directive at a
document-end marker and at stream end. Both scan drivers finalize through
the same stream-end path, so the check lives in one helper called from
each. The rule applies to any directive, including an unrecognized one
(which is otherwise still ignored); the directive test that relied on an
unrecognized directive parsing without a '---' is corrected to include
one, since that leniency is not permitted by the grammar.
Measured against the official YAML test-suite, parse-level conformance
rises from 253/333 (76.0%) to 255/333 (76.6%) with no regression.
A '#' that begins a comment must be preceded by whitespace or open the
line; glued to earlier same-line content (e.g. "value"# c, [ a, b ]#c,
>#c, a flow entry then ,#c) it is neither a comment nor, in those
positions, plain-scalar content, so the document is malformed. Such input
was previously accepted as a comment.
A '#' that opens a line, follows a space, or sits inside a plain scalar
(a#b, where it stays literal content and never reaches this check) is
unaffected. The scanner tracks whether a real token has been scanned on
the current physical line, reset on each line break, rather than comparing
token line numbers: the streaming tokenizer drains its token buffer
between scans, and a block scalar's blank line emits an empty scalar
attributed to the next line, either of which would misread as content.
Measured against the official YAML test-suite, parse-level conformance
rises from 243/333 (73.0%) to 247/333 (74.2%): four wrongly-accepted
invalid documents are now rejected, with no valid document regressing.
The scalar-scanning routines move to Scanner.Scalars.swift to keep the
file within the length limit.
The scanner dropped every blank line within and after a block scalar, so
a literal block lost internal blank lines, a folded block turned a blank
line into a space rather than a newline, and the keep indicator could not
preserve trailing blank lines. Capture blank lines as empty content in
both the batch scan loop and the streaming TokenCursor loop, which had
drifted as duplicate copies. Rework blockScalarValue to apply literal
versus folded line semantics and clip/strip/keep chomping over the
captured lines; content without blank lines is unchanged. Re-pin the
multiline-scalar test whose expectation encoded the old dropped-blank
behavior, and add a chomping-and-folding test matrix.
Add structured validation diagnostics, JSON and Yams comparison gates, and lazy token/event parsing so production behavior is measured without retaining full intermediate parser arrays.
