pawarser(init): start extracting the parser lib

crates/pawarser/Cargo.toml

@@ -0,0 +1,12 @@
+[package]
+name = "pawarser"
+version = "0.1.0"
+edition = "2021"
+
+[dependencies]
+rowan = "0.15.15"
+drop_bomb = "0.1.5"
+enumset = "1.1.3"
+
+[lints]
+workspace = true

crates/pawarser/src/lib.rs

@@ -0,0 +1,39 @@
+#![feature(iter_collect_into)]
+pub mod parser {
+    pub mod input {
+        use enumset::{EnumSet, EnumSetType};
+
+        struct Input<'src, 'toks, SyntaxKind: EnumSetType + Into<rowan::SyntaxKind>> {
+            raw: &'toks Vec<(SyntaxKind, &'src str)>,
+            // enumset of meaningless tokens
+            semantically_meaningless: EnumSet<SyntaxKind>,
+            // indices of non-meaningless tokens
+            meaningful_toks: Vec<usize>,
+        }
+
+        impl<'src, 'toks, SyntaxKind: EnumSetType + Into<rowan::SyntaxKind>>
+            Input<'src, 'toks, SyntaxKind>
+        {
+            pub fn new(
+                raw_toks: &'toks Vec<(SyntaxKind, &'src str)>,
+                meaningless: Option<EnumSet<SyntaxKind>>,
+            ) -> Self {
+                let mut meaningful_toks = Vec::new();
+
+                if let Some(meaningless) = meaningless {
+                    let meaningful_toks = raw_toks
+                        .iter()
+                        .enumerate()
+                        .filter_map(|(i, tok)| (!meaningless.contains(tok.0)).then_some(i))
+                        .collect_into(&mut meaningful_toks);
+                }
+
+                Self {
+                    raw: raw_toks,
+                    semantically_meaningless: meaningless.unwrap_or_default(),
+                    meaningful_toks,
+                }
+            }
+        }
+    }
+}