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

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 {
-    fn codegen(&self) -> String {
-        match self {
-            Self::MapTo(syntax) => format!(r###"MappingTarget::MapTo(r#"{syntax}"#)"###),
-            Self::MapToUnknown => "MappingTarget::MapToUnknown".into(),
-            Self::MapExtensionToUnknown => "MappingTarget::MapExtensionToUnknown".into(),
-        }
+impl ToTokens for MappingTarget {
+    fn to_tokens(&self, tokens: &mut TokenStream) {
+        let t = match self {
+            Self::MapTo(syntax) => quote! { MappingTarget::MapTo(#syntax) },
+            Self::MapToUnknown => quote! { MappingTarget::MapToUnknown },
+            Self::MapExtensionToUnknown => quote! { MappingTarget::MapExtensionToUnknown },
+        };
+        tokens.append_all(t);
     }
 }
 
@@ -116,22 +119,17 @@ impl FromStr for Matcher {
         Ok(Self(non_empty_segments))
     }
 }
-impl Matcher {
-    fn codegen(&self) -> String {
-        match self.0.len() {
-            0 => unreachable!("0-length matcher should never be created"),
-            // if-let guard would be ideal here
-            // see: https://github.com/rust-lang/rust/issues/51114
-            1 if self.0[0].is_text() => {
-                let s = self.0[0].text().unwrap();
-                format!(r###"Lazy::new(|| Some(build_matcher_fixed(r#"{s}"#)))"###)
+impl ToTokens for Matcher {
+    fn to_tokens(&self, tokens: &mut TokenStream) {
+        let t = match self.0.as_slice() {
+            [] => unreachable!("0-length matcher should never be created"),
+            [MatcherSegment::Text(text)] => {
+                quote! { Lazy::new(|| Some(build_matcher_fixed(#text)))}
             }
             // parser logic ensures that this case can only happen when there are dynamic segments
-            _ => {
-                let segs = self.0.iter().map(MatcherSegment::codegen).join(", ");
-                format!(r###"Lazy::new(|| build_matcher_dynamic(&[{segs}]))"###)
-            }
-        }
+            segs @ [_, ..] => quote! { 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 {
-    fn codegen(&self) -> String {
-        let array_items: Vec<_> = self
+impl ToTokens for MappingList {
+    fn to_tokens(&self, tokens: &mut TokenStream) {
+        let len = self.0.len();
+        let array_items = self
             .0
             .iter()
-            .map(|(matcher, target)| {
-                format!("({m}, {t})", m = matcher.codegen(), t = target.codegen())
-            })
-            .collect();
-        let len = array_items.len();
+            .map(|(matcher, target)| quote! { (#matcher, #target) });
 
-        format!(
-            "/// Generated by build script from /src/syntax_mapping/builtins/.\n\
-            pub(crate) static BUILTIN_MAPPINGS: [(Lazy<Option<GlobMatcher>>, MappingTarget); {len}] = [\n{items}\n];",
-            items = array_items.join(",\n")
-        )
+        let t = quote! {
+            /// Generated by build script from /src/syntax_mapping/builtins/.
+            pub(crate) static BUILTIN_MAPPINGS: [(Lazy<Option<GlobMatcher>>, MappingTarget); #len] = [#(#array_items),*];
+        };
+        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(())
 }