1 files changed, 166 insertions, 0 deletions

diff --git a/crates/ra_syntax/src/parsing/text_tree_sink.rs b/crates/ra_syntax/src/parsing/text_tree_sink.rs
new file mode 100644
index 000000000..961a91d41
--- /dev/null
+++ b/crates/ra_syntax/src/parsing/text_tree_sink.rs
@@ -0,0 +1,166 @@
+use std::mem;
+
+use ra_parser::{TreeSink, ParseError};
+
+use crate::{
+    SmolStr, SyntaxError, TextUnit, TextRange, SyntaxTreeBuilder,
+    SyntaxKind::{self, *},
+    parsing::Token,
+    syntax_node::GreenNode,
+};
+
+/// Bridges the parser with our specific syntax tree representation.
+///
+/// `TextTreeSink` also handles attachment of trivia (whitespace) to nodes.
+pub(crate) struct TextTreeSink<'a> {
+    text: &'a str,
+    tokens: &'a [Token],
+    text_pos: TextUnit,
+    token_pos: usize,
+    state: State,
+    inner: SyntaxTreeBuilder,
+}
+
+enum State {
+    PendingStart,
+    Normal,
+    PendingFinish,
+}
+
+impl<'a> TreeSink for TextTreeSink<'a> {
+    fn leaf(&mut self, kind: SyntaxKind, n_tokens: u8) {
+        match mem::replace(&mut self.state, State::Normal) {
+            State::PendingStart => unreachable!(),
+            State::PendingFinish => self.inner.finish_branch(),
+            State::Normal => (),
+        }
+        self.eat_trivias();
+        let n_tokens = n_tokens as usize;
+        let len = self.tokens[self.token_pos..self.token_pos + n_tokens]
+            .iter()
+            .map(|it| it.len)
+            .sum::<TextUnit>();
+        self.do_leaf(kind, len, n_tokens);
+    }
+
+    fn start_branch(&mut self, kind: SyntaxKind) {
+        match mem::replace(&mut self.state, State::Normal) {
+            State::PendingStart => {
+                self.inner.start_branch(kind);
+                // No need to attach trivias to previous node: there is no
+                // previous node.
+                return;
+            }
+            State::PendingFinish => self.inner.finish_branch(),
+            State::Normal => (),
+        }
+
+        let n_trivias =
+            self.tokens[self.token_pos..].iter().take_while(|it| it.kind.is_trivia()).count();
+        let leading_trivias = &self.tokens[self.token_pos..self.token_pos + n_trivias];
+        let mut trivia_end =
+            self.text_pos + leading_trivias.iter().map(|it| it.len).sum::<TextUnit>();
+
+        let n_attached_trivias = {
+            let leading_trivias = leading_trivias.iter().rev().map(|it| {
+                let next_end = trivia_end - it.len;
+                let range = TextRange::from_to(next_end, trivia_end);
+                trivia_end = next_end;
+                (it.kind, &self.text[range])
+            });
+            n_attached_trivias(kind, leading_trivias)
+        };
+        self.eat_n_trivias(n_trivias - n_attached_trivias);
+        self.inner.start_branch(kind);
+        self.eat_n_trivias(n_attached_trivias);
+    }
+
+    fn finish_branch(&mut self) {
+        match mem::replace(&mut self.state, State::PendingFinish) {
+            State::PendingStart => unreachable!(),
+            State::PendingFinish => self.inner.finish_branch(),
+            State::Normal => (),
+        }
+    }
+
+    fn error(&mut self, error: ParseError) {
+        self.inner.error(error, self.text_pos)
+    }
+}
+
+impl<'a> TextTreeSink<'a> {
+    pub(super) fn new(text: &'a str, tokens: &'a [Token]) -> TextTreeSink<'a> {
+        TextTreeSink {
+            text,
+            tokens,
+            text_pos: 0.into(),
+            token_pos: 0,
+            state: State::PendingStart,
+            inner: SyntaxTreeBuilder::default(),
+        }
+    }
+
+    pub(super) fn finish(mut self) -> (GreenNode, Vec<SyntaxError>) {
+        match mem::replace(&mut self.state, State::Normal) {
+            State::PendingFinish => {
+                self.eat_trivias();
+                self.inner.finish_branch()
+            }
+            State::PendingStart | State::Normal => unreachable!(),
+        }
+
+        self.inner.finish_raw()
+    }
+
+    fn eat_trivias(&mut self) {
+        while let Some(&token) = self.tokens.get(self.token_pos) {
+            if !token.kind.is_trivia() {
+                break;
+            }
+            self.do_leaf(token.kind, token.len, 1);
+        }
+    }
+
+    fn eat_n_trivias(&mut self, n: usize) {
+        for _ in 0..n {
+            let token = self.tokens[self.token_pos];
+            assert!(token.kind.is_trivia());
+            self.do_leaf(token.kind, token.len, 1);
+        }
+    }
+
+    fn do_leaf(&mut self, kind: SyntaxKind, len: TextUnit, n_tokens: usize) {
+        let range = TextRange::offset_len(self.text_pos, len);
+        let text: SmolStr = self.text[range].into();
+        self.text_pos += len;
+        self.token_pos += n_tokens;
+        self.inner.leaf(kind, text);
+    }
+}
+
+fn n_attached_trivias<'a>(
+    kind: SyntaxKind,
+    trivias: impl Iterator<Item = (SyntaxKind, &'a str)>,
+) -> usize {
+    match kind {
+        CONST_DEF | TYPE_DEF | STRUCT_DEF | ENUM_DEF | ENUM_VARIANT | FN_DEF | TRAIT_DEF
+        | MODULE | NAMED_FIELD_DEF => {
+            let mut res = 0;
+            for (i, (kind, text)) in trivias.enumerate() {
+                match kind {
+                    WHITESPACE => {
+                        if text.contains("\n\n") {
+                            break;
+                        }
+                    }
+                    COMMENT => {
+                        res = i + 1;
+                    }
+                    _ => (),
+                }
+            }
+            res
+        }
+        _ => 0,
+    }
+}