Generate environment variables doc from code (#8493)

## Summary

Resolves #8417

I've just begun learning procedural macros, so this PR is more of a
proof of concept. It's still a work in progress, and I welcome any
assistance or feedback.

Author: j178

.prettierignore

@@ -3,5 +3,6 @@ CHANGELOG.md
 PREVIEW-CHANGELOG.md
 docs/reference/cli.md
 docs/reference/settings.md
+docs/configuration/environment.md
 ecosystem/home-assistant-core/LICENSE.md
 docs/guides/integration/gitlab.md

Cargo.lock

@@ -2,7 +2,10 @@
 
 use anyhow::Result;
 
-use crate::{generate_cli_reference, generate_json_schema, generate_options_reference};
+use crate::{
+    generate_cli_reference, generate_env_vars_reference, generate_json_schema,
+    generate_options_reference,
+};
 
 #[derive(clap::Args)]
 pub(crate) struct Args {
@@ -27,5 +30,6 @@ pub(crate) fn main(args: &Args) -> Result<()> {
     generate_json_schema::main(&generate_json_schema::Args { mode: args.mode })?;
     generate_options_reference::main(&generate_options_reference::Args { mode: args.mode })?;
     generate_cli_reference::main(&generate_cli_reference::Args { mode: args.mode })?;
+    generate_env_vars_reference::main(&generate_env_vars_reference::Args { mode: args.mode })?;
     Ok(())
 }

crates/uv-dev/src/generate_all.rs

@@ -31,7 +31,6 @@ const SHOW_HIDDEN_COMMANDS: &[&str] = &["generate-shell-completion"];
 
 #[derive(clap::Args)]
 pub(crate) struct Args {
-    /// Write the generated output to stdout (rather than to `settings.md`).
     #[arg(long, default_value_t, value_enum)]
     pub(crate) mode: Mode,
 }

crates/uv-dev/src/generate_cli_reference.rs

@@ -0,0 +1,98 @@
+//! Generate the environment variables reference from `uv_static::EnvVars`.
+
+use anyhow::bail;
+use pretty_assertions::StrComparison;
+use std::path::PathBuf;
+
+use uv_static::EnvVars;
+
+use crate::generate_all::Mode;
+use crate::ROOT_DIR;
+
+#[derive(clap::Args)]
+pub(crate) struct Args {
+    #[arg(long, default_value_t, value_enum)]
+    pub(crate) mode: Mode,
+}
+
+pub(crate) fn main(args: &Args) -> anyhow::Result<()> {
+    let reference_string = generate();
+    let filename = "environment.md";
+    let reference_path = PathBuf::from(ROOT_DIR)
+        .join("docs")
+        .join("configuration")
+        .join(filename);
+
+    match args.mode {
+        Mode::DryRun => {
+            anstream::println!("{reference_string}");
+        }
+        Mode::Check => match fs_err::read_to_string(reference_path) {
+            Ok(current) => {
+                if current == reference_string {
+                    anstream::println!("Up-to-date: {filename}");
+                } else {
+                    let comparison = StrComparison::new(&current, &reference_string);
+                    bail!("{filename} changed, please run `cargo dev generate-env-vars-reference`:\n{comparison}");
+                }
+            }
+            Err(err) if err.kind() == std::io::ErrorKind::NotFound => {
+                bail!("{filename} not found, please run `cargo dev generate-env-vars-reference`");
+            }
+            Err(err) => {
+                bail!(
+                    "{filename} changed, please run `cargo dev generate-env-vars-reference`:\n{err}"
+                );
+            }
+        },
+        Mode::Write => match fs_err::read_to_string(&reference_path) {
+            Ok(current) => {
+                if current == reference_string {
+                    anstream::println!("Up-to-date: {filename}");
+                } else {
+                    anstream::println!("Updating: {filename}");
+                    fs_err::write(reference_path, reference_string.as_bytes())?;
+                }
+            }
+            Err(err) if err.kind() == std::io::ErrorKind::NotFound => {
+                anstream::println!("Updating: {filename}");
+                fs_err::write(reference_path, reference_string.as_bytes())?;
+            }
+            Err(err) => {
+                bail!("{filename} changed, please run `cargo dev generate-env-vars-reference`:\n{err}");
+            }
+        },
+    }
+
+    Ok(())
+}
+
+fn generate() -> String {
+    let mut output = String::new();
+
+    output.push_str("# Environment variables\n\n");
+    output.push_str("uv respects the following environment variables:\n\n");
+
+    for (var, doc) in EnvVars::metadata() {
+        // Remove empty lines and ddd two spaces to the beginning from the second line.
+        let doc = doc
+            .lines()
+            .enumerate()
+            .filter(|(_, line)| !line.trim().is_empty())
+            .map(|(i, line)| {
+                if i == 0 {
+                    line.to_string()
+                } else {
+                    format!("  {line}")
+                }
+            })
+            .collect::<Vec<_>>()
+            .join("\n");
+        output.push_str(&format!("- `{var}`: {doc}\n"));
+    }
+
+    output
+}
+
+#[cfg(test)]
+mod tests;

crates/uv-dev/src/generate_env_vars_reference.rs

@@ -0,0 +1,19 @@
+use std::env;
+
+use anyhow::Result;
+
+use uv_static::EnvVars;
+
+use crate::generate_all::Mode;
+
+use super::{main, Args};
+
+#[test]
+fn test_generate_env_vars_reference() -> Result<()> {
+    let mode = if env::var(EnvVars::UV_UPDATE_SCHEMA).as_deref() == Ok("1") {
+        Mode::Write
+    } else {
+        Mode::Check
+    };
+    main(&Args { mode })
+}

crates/uv-dev/src/generate_env_vars_reference/tests.rs

@@ -27,7 +27,6 @@ struct CombinedOptions {
 
 #[derive(clap::Args)]
 pub(crate) struct Args {
-    /// Write the generated output to stdout (rather than to `uv.schema.json`).
     #[arg(long, default_value_t, value_enum)]
     pub(crate) mode: Mode,
 }

crates/uv-dev/src/generate_json_schema.rs

@@ -34,7 +34,6 @@ struct CombinedOptions {
 
 #[derive(clap::Args)]
 pub(crate) struct Args {
-    /// Write the generated output to stdout (rather than to `settings.md`).
     #[arg(long, default_value_t, value_enum)]
     pub(crate) mode: Mode,
 }

crates/uv-dev/src/generate_options_reference.rs

@@ -20,6 +20,7 @@ use crate::clear_compile::ClearCompileArgs;
 use crate::compile::CompileArgs;
 use crate::generate_all::Args as GenerateAllArgs;
 use crate::generate_cli_reference::Args as GenerateCliReferenceArgs;
+use crate::generate_env_vars_reference::Args as GenerateEnvVarsReferenceArgs;
 use crate::generate_json_schema::Args as GenerateJsonSchemaArgs;
 use crate::generate_options_reference::Args as GenerateOptionsReferenceArgs;
 #[cfg(feature = "render")]
@@ -31,6 +32,7 @@ mod clear_compile;
 mod compile;
 mod generate_all;
 mod generate_cli_reference;
+mod generate_env_vars_reference;
 mod generate_json_schema;
 mod generate_options_reference;
 mod render_benchmarks;
@@ -54,6 +56,8 @@ enum Cli {
     GenerateOptionsReference(GenerateOptionsReferenceArgs),
     /// Generate the CLI reference for the documentation.
     GenerateCliReference(GenerateCliReferenceArgs),
+    /// Generate the environment variables reference for the documentation.
+    GenerateEnvVarsReference(GenerateEnvVarsReferenceArgs),
     #[cfg(feature = "render")]
     /// Render the benchmarks.
     RenderBenchmarks(RenderBenchmarksArgs),
@@ -70,6 +74,7 @@ async fn run() -> Result<()> {
         Cli::GenerateJSONSchema(args) => generate_json_schema::main(&args)?,
         Cli::GenerateOptionsReference(args) => generate_options_reference::main(&args)?,
         Cli::GenerateCliReference(args) => generate_cli_reference::main(&args)?,
+        Cli::GenerateEnvVarsReference(args) => generate_env_vars_reference::main(&args)?,
         #[cfg(feature = "render")]
         Cli::RenderBenchmarks(args) => render_benchmarks::render_benchmarks(&args)?,
     }

crates/uv-dev/src/main.rs

@@ -2,7 +2,7 @@ mod options_metadata;
 
 use proc_macro::TokenStream;
 use quote::quote;
-use syn::{parse_macro_input, DeriveInput};
+use syn::{parse_macro_input, Attribute, DeriveInput, ImplItem, ItemImpl, LitStr};
 
 #[proc_macro_derive(OptionsMetadata, attributes(option, doc, option_group))]
 pub fn derive_options_metadata(input: TokenStream) -> TokenStream {
@@ -49,3 +49,92 @@ fn impl_combine(ast: &DeriveInput) -> TokenStream {
     };
     gen.into()
 }
+
+fn get_doc_comment(attrs: &[Attribute]) -> String {
+    attrs
+        .iter()
+        .filter_map(|attr| {
+            if attr.path().is_ident("doc") {
+                if let syn::Meta::NameValue(meta) = &attr.meta {
+                    if let syn::Expr::Lit(expr) = &meta.value {
+                        if let syn::Lit::Str(str) = &expr.lit {
+                            return Some(str.value().trim().to_string());
+                        }
+                    }
+                }
+            }
+            None
+        })
+        .collect::<Vec<_>>()
+        .join("\n")
+}
+
+fn get_env_var_pattern_from_attr(attrs: &[Attribute]) -> Option<String> {
+    attrs
+        .iter()
+        .find(|attr| attr.path().is_ident("attr_env_var_pattern"))
+        .and_then(|attr| attr.parse_args::<LitStr>().ok())
+        .map(|lit_str| lit_str.value())
+}
+
+fn is_hidden(attrs: &[Attribute]) -> bool {
+    attrs.iter().any(|attr| attr.path().is_ident("attr_hidden"))
+}
+
+/// This attribute is used to generate environment variables metadata for [`uv_static::EnvVars`].
+#[proc_macro_attribute]
+pub fn attribute_env_vars_metadata(_attr: TokenStream, input: TokenStream) -> TokenStream {
+    let ast = parse_macro_input!(input as ItemImpl);
+
+    let constants: Vec<_> = ast
+        .items
+        .iter()
+        .filter_map(|item| match item {
+            ImplItem::Const(item) if !is_hidden(&item.attrs) => {
+                let name = item.ident.to_string();
+                let doc = get_doc_comment(&item.attrs);
+                Some((name, doc))
+            }
+            ImplItem::Fn(item) if !is_hidden(&item.attrs) => {
+                // Extract the environment variable patterns.
+                if let Some(pattern) = get_env_var_pattern_from_attr(&item.attrs) {
+                    let doc = get_doc_comment(&item.attrs);
+                    Some((pattern, doc))
+                } else {
+                    None // Skip if pattern extraction fails.
+                }
+            }
+            _ => None,
+        })
+        .collect();
+
+    let struct_name = &ast.self_ty;
+    let pairs = constants.iter().map(|(name, doc)| {
+        quote! {
+            (#name, #doc)
+        }
+    });
+
+    let expanded = quote! {
+        #ast
+
+        impl #struct_name {
+            /// Returns a list of pairs of env var and their documentation defined in this impl block.
+            pub fn metadata<'a>() -> &'a [(&'static str, &'static str)] {
+                &[#(#pairs),*]
+            }
+        }
+    };
+
+    expanded.into()
+}
+
+#[proc_macro_attribute]
+pub fn attr_hidden(_attr: TokenStream, item: TokenStream) -> TokenStream {
+    item
+}
+
+#[proc_macro_attribute]
+pub fn attr_env_var_pattern(_attr: TokenStream, item: TokenStream) -> TokenStream {
+    item
+}

crates/uv-macros/src/lib.rs

@@ -16,3 +16,4 @@ doctest = false
 workspace = true
 
 [dependencies]
+uv-macros = { workspace = true }