From a3ab844ba7f016e3650e2228a8381aa500ae2add Mon Sep 17 00:00:00 2001
From: Schrottkatze <git@schrottkatze.de>
Date: Sun, 13 Oct 2024 15:32:26 +0200
Subject: [PATCH] pawarser(init): start extracting the parser lib

---
 Cargo.lock                 |  9 +++++++++
 Cargo.toml                 |  2 +-
 crates/pawarser/Cargo.toml | 12 ++++++++++++
 crates/pawarser/src/lib.rs | 39 ++++++++++++++++++++++++++++++++++++++
 4 files changed, 61 insertions(+), 1 deletion(-)
 create mode 100644 crates/pawarser/Cargo.toml
 create mode 100644 crates/pawarser/src/lib.rs

diff --git a/Cargo.lock b/Cargo.lock
index 25bef26..4304154 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -1159,6 +1159,15 @@ version = "1.0.15"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "57c0d7b74b563b49d38dae00a0c37d4d6de9b432382b2892f0574ddcae73fd0a"
 
+[[package]]
+name = "pawarser"
+version = "0.1.0"
+dependencies = [
+ "drop_bomb",
+ "enumset",
+ "rowan",
+]
+
 [[package]]
 name = "petgraph"
 version = "0.6.5"
diff --git a/Cargo.toml b/Cargo.toml
index 6b774ed..e5c6dc7 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -6,7 +6,7 @@ members = [
   "crates/lang", 
   "crates/svg-filters",
   "crates/prowocessing", 
-  "crates/executor-poc",
+  "crates/executor-poc", "crates/pawarser",
 ]
 resolver = "2"
 
diff --git a/crates/pawarser/Cargo.toml b/crates/pawarser/Cargo.toml
new file mode 100644
index 0000000..787cb2f
--- /dev/null
+++ b/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
diff --git a/crates/pawarser/src/lib.rs b/crates/pawarser/src/lib.rs
new file mode 100644
index 0000000..f1413cb
--- /dev/null
+++ b/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,
+                }
+            }
+        }
+    }
+}