For doc strings, the rawValue (on either the String_ or
InterpolatedStringPrts) does not include the leading indentation
(which is available as docIndentation) or the trailing newline on
the last part.
I did this to start with, but then alignment kept being broken
during refactorings, and at some point I switched to not aligning,
and now we have a big mess.
Add a php-cs-fixer rule to consistently not align phpdoc tags.
Types omitted in two places where we violate them currently:
Namespace_::$stmts can be null during parsing, and Enum_::$scalarType
can be a complex type for invalid programs.
The Lexer now only provides the tokens to the parser, while the
parser is responsible for determining which attributes are placed
on notes. This only needs to be done when the attributes are
actually needed, rather than for all tokens.
This removes the usedAttributes lexer option (and lexer options
entirely). The attributes are now enabled unconditionally. They
have less overhead now, and the need to explicitly enable them for
some use cases (e.g. formatting-preserving printing) doesn't seem
like a good tradeoff anymore.
There are some additional changes to the Lexer interface that
should be done after this, and the docs / upgrading guide haven't
been adjusted yet.
Now that destructuring is always represented using List nodes,
make sure that Array nodes can no longer contain null elements,
so well-typed code doesn't have to deal with them unnecessarily.
If an array does contain empty elements, these are now result in
an error and are represented as a ArrayItem with Error value if
error recovery is used.
The implementation is a bit tricky because at the time the Array
node is created, we cannot tell whether it will be used in a
creation or destructuring context. For this reason the error
reporting is delayed parsing has finished.
Closes#876.
The formatting in this project has become something of a mess,
because it changed over time. Add a CS fixer config and reformat
to the desired style, which is PSR-12, but with sane brace placement.
Empty productions are supposed to be assigned the start attributes
of the lookahead token. Currently, this happens by assigning above
the current stack position when the token it read.
This fails in a situation where we first reduce an empty production
higher up in the stack, and then again reduce an empty production
lower in the stack, without consuming the lookahead token in the
meantime.
Fix this by moving the assignment into the reduction phase. We
also need to do this for error productions, which are effectively
empty.
Previously zero-length nop nodes used the lookahead start attributes
and current end attributes. This choice ends up being somewhat weird,
because the end attributes will be the at the last non-whitespace,
non-comment token, which might be quite far back. More problematically,
we may not have encountered any non-discarded token if we're at the
start of the file, in which case we will have no end attributes to
assign.
Change things to use a canonical "zero-length" node representation,
where the end position (token & file) will be exactly one before the
start position.
Fixes#589.
