Build script: replace string-based codegen with quote-based codegen

Cargo.lock

@@ -141,6 +141,9 @@ dependencies = [
  "path_abs",
  "plist",
  "predicates",
+ "prettyplease",
+ "proc-macro2",
+ "quote",
  "regex",
  "semver",
  "serde",
@@ -149,6 +152,7 @@ dependencies = [
  "serde_yaml",
  "serial_test",
  "shell-words",
+ "syn",
  "syntect",
  "tempfile",
  "terminal-colorsaurus",
@@ -1163,10 +1167,20 @@ dependencies = [
 ]
 
 [[package]]
-name = "proc-macro2"
+name = "prettyplease"
-version = "1.0.92"
+version = "0.2.35"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "37d3544b3f2748c54e147655edb5025752e2303145b5aefb3c3ea2c78b973bb0"
+checksum = "061c1221631e079b26479d25bbf2275bfe5917ae8419cd7e34f13bfc2aa7539a"
+dependencies = [
+ "proc-macro2",
+ "syn",
+]
+
+[[package]]
+name = "proc-macro2"
+version = "1.0.95"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "02b3e5e68a3a1a02aad3ec490a98007cbc13c37cbe84a3cd7b8e406d76e7f778"
 dependencies = [
  "unicode-ident",
 ]
@@ -1182,9 +1196,9 @@ dependencies = [
 
 [[package]]
 name = "quote"
-version = "1.0.38"
+version = "1.0.40"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "0e4dccaaaf89514f546c693ddc140f729f958c247918a13380cccc6078391acc"
+checksum = "1885c039570dc00dcb4ff087a89e185fd56bae234ddc7f056a945bf36467248d"
 dependencies = [
  "proc-macro2",
 ]
@@ -1440,9 +1454,9 @@ checksum = "7da8b5736845d9f2fcb837ea5d9e2628564b3b043a70948a3f0b778838c5fb4f"
 
 [[package]]
 name = "syn"
-version = "2.0.95"
+version = "2.0.104"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "46f71c0377baf4ef1cc3e3402ded576dccc315800fbc62dfc7fe04b009773b4a"
+checksum = "17b6f705963418cdb9927482fa304bc562ece2fdd4f616084c50b7023b435a40"
 dependencies = [
  "proc-macro2",
  "quote",

Cargo.toml

@@ -108,10 +108,14 @@ anyhow = "1.0.97"
 indexmap = { version = "2.8.0", features = ["serde"] }
 itertools = "0.14.0"
 once_cell = "1.20"
+prettyplease = "0.2.35"
+proc-macro2 = "1.0.95"
+quote = "1.0.40"
 regex = "1.10.6"
 serde = "1.0"
 serde_derive = "1.0"
 serde_with = { version = "3.12.0", default-features = false, features = ["macros"] }
+syn = { version = "2.0.104", features = ["full"] }
 toml = { version = "0.8.19", features = ["preserve_order"] }
 walkdir = "2.5"
 

build/syntax_mapping.rs

@@ -9,6 +9,8 @@ use anyhow::{anyhow, bail};
 use indexmap::IndexMap;
 use itertools::Itertools;
 use once_cell::sync::Lazy;
+use proc_macro2::TokenStream;
+use quote::{quote, ToTokens, TokenStreamExt};
 use regex::Regex;
 use serde_derive::Deserialize;
 use serde_with::DeserializeFromStr;
@@ -34,13 +36,14 @@ impl FromStr for MappingTarget {
         }
     }
 }
-impl MappingTarget {
+impl ToTokens for MappingTarget {
-    fn codegen(&self) -> String {
+    fn to_tokens(&self, tokens: &mut TokenStream) {
-        match self {
+        let t = match self {
-            Self::MapTo(syntax) => format!(r###"MappingTarget::MapTo(r#"{syntax}"#)"###),
+            Self::MapTo(syntax) => quote! { MappingTarget::MapTo(#syntax) },
-            Self::MapToUnknown => "MappingTarget::MapToUnknown".into(),
+            Self::MapToUnknown => quote! { MappingTarget::MapToUnknown },
-            Self::MapExtensionToUnknown => "MappingTarget::MapExtensionToUnknown".into(),
+            Self::MapExtensionToUnknown => quote! { MappingTarget::MapExtensionToUnknown },
-        }
+        };
+        tokens.append_all(t);
     }
 }
 
@@ -116,22 +119,17 @@ impl FromStr for Matcher {
         Ok(Self(non_empty_segments))
     }
 }
-impl Matcher {
+impl ToTokens for Matcher {
-    fn codegen(&self) -> String {
+    fn to_tokens(&self, tokens: &mut TokenStream) {
-        match self.0.len() {
+        let t = match self.0.as_slice() {
-            0 => unreachable!("0-length matcher should never be created"),
+            [] => unreachable!("0-length matcher should never be created"),
-            // if-let guard would be ideal here
+            [MatcherSegment::Text(text)] => {
-            // see: https://github.com/rust-lang/rust/issues/51114
+                quote! { Lazy::new(|| Some(build_matcher_fixed(#text)))}
-            1 if self.0[0].is_text() => {
-                let s = self.0[0].text().unwrap();
-                format!(r###"Lazy::new(|| Some(build_matcher_fixed(r#"{s}"#)))"###)
             }
             // parser logic ensures that this case can only happen when there are dynamic segments
-            _ => {
+            segs @ [_, ..] => quote! { Lazy::new(|| build_matcher_dynamic(&[ #(#segs),* ]))},
-                let segs = self.0.iter().map(MatcherSegment::codegen).join(", ");
+        };
-                format!(r###"Lazy::new(|| build_matcher_dynamic(&[{segs}]))"###)
+        tokens.append_all(t);
-            }
-        }
     }
 }
 
@@ -143,6 +141,15 @@ enum MatcherSegment {
     Text(String),
     Env(String),
 }
+impl ToTokens for MatcherSegment {
+    fn to_tokens(&self, tokens: &mut TokenStream) {
+        let t = match self {
+            Self::Text(text) => quote! { MatcherSegment::Text(#text)},
+            Self::Env(env) => quote! {MatcherSegment::Env(#env)},
+        };
+        tokens.append_all(t);
+    }
+}
 #[allow(dead_code)]
 impl MatcherSegment {
     fn is_text(&self) -> bool {
@@ -163,12 +170,6 @@ impl MatcherSegment {
             Self::Env(t) => Some(t),
         }
     }
-    fn codegen(&self) -> String {
-        match self {
-            Self::Text(s) => format!(r###"MatcherSegment::Text(r#"{s}"#)"###),
-            Self::Env(s) => format!(r###"MatcherSegment::Env(r#"{s}"#)"###),
-        }
-    }
 }
 
 /// A struct that models a single .toml file in /src/syntax_mapping/builtins/.
@@ -194,22 +195,19 @@ impl MappingDefModel {
 
 #[derive(Clone, Debug)]
 struct MappingList(Vec<(Matcher, MappingTarget)>);
-impl MappingList {
+impl ToTokens for MappingList {
-    fn codegen(&self) -> String {
+    fn to_tokens(&self, tokens: &mut TokenStream) {
-        let array_items: Vec<_> = self
+        let len = self.0.len();
+        let array_items = self
             .0
             .iter()
-            .map(|(matcher, target)| {
+            .map(|(matcher, target)| quote! { (#matcher, #target) });
-                format!("({m}, {t})", m = matcher.codegen(), t = target.codegen())
-            })
-            .collect();
-        let len = array_items.len();
 
-        format!(
+        let t = quote! {
-            "/// Generated by build script from /src/syntax_mapping/builtins/.\n\
+            /// Generated by build script from /src/syntax_mapping/builtins/.
-            pub(crate) static BUILTIN_MAPPINGS: [(Lazy<Option<GlobMatcher>>, MappingTarget); {len}] = [\n{items}\n];",
+            pub(crate) static BUILTIN_MAPPINGS: [(Lazy<Option<GlobMatcher>>, MappingTarget); #len] = [#(#array_items),*];
-            items = array_items.join(",\n")
+        };
-        )
+        tokens.append_all(t);
     }
 }
 
@@ -290,11 +288,13 @@ pub fn build_static_mappings() -> anyhow::Result<()> {
     println!("cargo:rerun-if-changed=src/syntax_mapping/builtins/");
 
     let mappings = read_all_mappings()?;
+    let rs_src = syn::parse_file(&mappings.to_token_stream().to_string())?;
+    let rs_src_pretty = prettyplease::unparse(&rs_src);
 
     let codegen_path = Path::new(&env::var_os("OUT_DIR").ok_or(anyhow!("OUT_DIR is unset"))?)
         .join("codegen_static_syntax_mappings.rs");
 
-    fs::write(codegen_path, mappings.codegen())?;
+    fs::write(codegen_path, rs_src_pretty)?;
 
     Ok(())
 }