chore: provide a tool for generating config docs

src/common/proc_macro/Cargo.toml

@@ -22,6 +22,7 @@ quote = "1"
 proc-macro2 = { version = "1", default-features = false }
 syn = "1"
 bae = "0.1.7"
+itertools = "0.12"
 
 [lints]
 workspace = true

src/common/proc_macro/src/config_doc.rs

@@ -0,0 +1,143 @@
+use itertools::Itertools;
+use quote::quote;
+use syn::{Attribute, Data, DataStruct, DeriveInput, Field, Fields};
+
+pub fn generate_config_doc_fn(input: DeriveInput) -> proc_macro2::TokenStream {
+    let mut doc = StructFieldDocs::new();
+
+    let struct_name = input.ident;
+    match input.data {
+        Data::Struct(ref data) => doc.extract_field_docs(data),
+        _ => panic!("This macro only supports structs"),
+    };
+
+    let vec_fields = doc.token_vec_fields();
+    let call_nested_fields = doc.token_call_nested_fields();
+    quote! {
+        impl #struct_name {
+            pub fn config_docs(name: String, docs: &mut std::collections::BTreeMap<String, Vec<(String, String)>>) {
+                docs.insert(name.clone(), #vec_fields);
+                #call_nested_fields;
+            }
+        }
+    }
+}
+
+fn extract_comment(attrs: &Vec<Attribute>) -> String {
+    attrs
+        .iter()
+        .filter_map(|attr| {
+            if let Ok(meta) = attr.parse_meta() {
+                if meta.path().is_ident("doc") {
+                    if let syn::Meta::NameValue(syn::MetaNameValue {
+                        lit: syn::Lit::Str(comment),
+                        ..
+                    }) = meta
+                    {
+                        return Some(comment.value());
+                    }
+                }
+            }
+            None
+        })
+        .join(" ")
+}
+
+fn is_nested_config_field(field: &Field) -> bool {
+    field.attrs.iter().any(|attr| {
+        if let Some(attr_name) = attr.path.get_ident() {
+            attr_name == "config_doc" && attr.tokens.to_string() == "(nested)"
+        } else {
+            false
+        }
+    })
+}
+
+fn is_omitted_config_field(field: &Field) -> bool {
+    field.attrs.iter().any(|attr| {
+        if let Some(attr_name) = attr.path.get_ident() {
+            attr_name == "config_doc" && attr.tokens.to_string() == "(omitted)"
+        } else {
+            false
+        }
+    })
+}
+
+fn field_name(f: &Field) -> String {
+    f.ident
+        .as_ref()
+        .expect("field name should not be empty")
+        .to_string()
+}
+
+struct StructFieldDocs {
+    // Fields that require recursively retrieving their field docs.
+    nested_fields: Vec<(String, syn::Type)>,
+
+    fields: Vec<(String, String)>,
+}
+
+impl StructFieldDocs {
+    fn new() -> Self {
+        Self {
+            nested_fields: vec![],
+            fields: vec![],
+        }
+    }
+
+    fn extract_field_docs(&mut self, data: &DataStruct) {
+        match &data.fields {
+            Fields::Named(fields) => {
+                self.fields = fields
+                    .named
+                    .iter()
+                    .filter_map(|field| {
+                        if is_omitted_config_field(field) {
+                            return None;
+                        }
+                        if is_nested_config_field(field) {
+                            self.nested_fields
+                                .push((field_name(field), field.ty.clone()));
+                            return None;
+                        }
+                        let field_name = field.ident.as_ref()?.to_string();
+                        let rustdoc = extract_comment(&field.attrs);
+                        Some((field_name, rustdoc))
+                    })
+                    .collect_vec();
+            }
+            _ => unreachable!("field should be named"),
+        }
+    }
+
+    fn token_vec_fields(&self) -> proc_macro2::TokenStream {
+        let token_fields: Vec<proc_macro2::TokenStream> = self
+            .fields
+            .iter()
+            .map(|(name, doc)| {
+                quote! { (#name.to_string(), #doc.to_string()) }
+            })
+            .collect();
+
+        quote! {
+            vec![#(#token_fields),*]
+        }
+    }
+
+    fn token_call_nested_fields(&self) -> proc_macro2::TokenStream {
+        let tokens: Vec<proc_macro2::TokenStream> = self
+            .nested_fields
+            .iter()
+            .map(|(ident, ty)| {
+                quote! {
+                    if name.is_empty() {
+                        #ty::config_docs(#ident.to_string(), docs);
+                    } else {
+                        #ty::config_docs(format!("{}.{}", name, #ident), docs);
+                    }
+                }
+            })
+            .collect();
+        quote! { #(#tokens)* }
+    }
+}

src/common/proc_macro/src/lib.rs

@@ -23,6 +23,7 @@ use quote::quote;
 use syn::parse_macro_input;
 
 mod config;
+mod config_doc;
 mod estimate_size;
 mod session_config;
 
@@ -263,3 +264,44 @@ pub fn session_config(input: TokenStream) -> TokenStream {
     let input = parse_macro_input!(input);
     session_config::derive_config(input).into()
 }
+
+/// This proc macro recursively extracts rustdoc comments from the fields in a struct and generates a method
+/// that produces docs for each field.
+/// Unlike rustdoc, this tool focuses solely on extracting rustdoc for struct fields, without methods.
+///
+/// Example:
+///
+/// ```ignore
+/// #[derive(ConfigDoc)]
+/// pub struct Foo {
+///   /// Description for `a`.
+///   a i32,
+///
+///   #[config_doc(nested)]
+///   b Bar,
+///
+///   #[config_doc(omitted)]
+///   dummy (),
+/// }
+/// ```
+///
+/// The `#[config_doc(nested)]` attribute indicates that the field is a nested config that will be documented in a separate section.
+/// Fields marked with `#[config_doc(omitted)]` will simply be omitted from the doc.
+///
+/// Here is the method generated by this macro:
+///
+/// ```ignore
+/// impl Foo {
+///     pub fn config_docs(name: String, docs: &mut std::collections::BTreeMap<String, Vec<(String, String)>>)
+/// }
+/// ```
+///
+/// In `test_example_up_to_date`, we further process the output of this method to generate a markdown in src/config/docs.md.
+#[proc_macro_derive(ConfigDoc, attributes(config_doc))]
+pub fn config_doc(input: TokenStream) -> TokenStream {
+    let input = parse_macro_input!(input);
+
+    let gen = config_doc::generate_config_doc_fn(input);
+
+    gen.into()
+}