This reverts commit 5cd07648fd.
The separators construct is useful as an optimization. It turns out that
constructing a node for every chunk of whitespace in a document causes a
significant performance regression.
Conflicts:
src/compiler/build_tables/build_lex_table.cc
src/compiler/grammar.cc
src/runtime/parser.c
Now, grammars can handle whitespace by making it another ubiquitous
token, like comments.
For now, this has the side effect of whitespace being included in the
tree that precedes it. This was already an issue for other ubiquitous
tokens though, so it needs to be fixed anyway.
The lexer doesn't know the expected symbols, so it doesn't have enough
information to construct error nodes. Now, when it encounters an invalid
character, it returns NULL and the parser builds a correct error node.
Now, the root node of a document is always a document node.
It will often have only one child node which corresponds to the grammar's
start symbol, but not always. Currently, it may have more than one child
if there are ubiquitous tokens such as comments at the beginning of the
document. In the future, it will also be possible be possible to have multiple
for the document to have multiple children if the document is partially parsed.
There's no need for a `string` function since one already
exists for Nodes.
Now the root node is always stored on the document. This
means callers of `ts_document_root_node` don't need to release
its return value.
Generated parsers no longer export a parser constructor function.
They now export an opaque Language object which can be set on
Documents directly. This way, the logic for constructing parsers
lives entirely in the runtime. The Languages are just structs which
have no load-time dependency on the runtime
My original thought was to decouple the runtime from
the LR parser generator by making TSParser a generic
interface that LR parsers implement.
I think this was more trouble than it was worth.
