Merge pull request #506 from tweag/505-user-configuration-file

Read user defined configuration file from $ROOT/.topiary/languages.toml

Cargo.toml

@@ -20,6 +20,7 @@ assert_cmd = "2.0"
 cfg-if = "1.0.0"
 clap = "4.3"
 criterion = "0.4"
+directories = "5.0"
 env_logger = "0.10"
 futures = "0.3.28"
 itertools = "0.10"
@@ -28,6 +29,7 @@ pretty_assertions = "1.3"
 prettydiff = "0.6.4"
 regex = "1.8.2"
 serde = "1.0.163"
+serde-toml-merge = "0.3"
 serde_json = "1.0"
 tempfile = "3.5.0"
 test-log = "0.2.11"

README.md

@@ -171,6 +171,7 @@ Options:
   -i, --in-place                     Format the input file in place
   -v, --visualise[=<OUTPUT_FORMAT>]  Visualise the syntax tree, rather than format [possible values: json, dot]
   -s, --skip-idempotence             Do not check that formatting twice gives the same output
+      --output-configuration         Output the full configuration to stderr before continuing
   -h, --help                         Print help
   -V, --version                      Print version
 ```

topiary-cli/Cargo.toml

@@ -28,7 +28,9 @@ path = "src/main.rs"
 # Eventually we will want to dynamically load them, like Helix does.
 clap = { workspace = true, features = ["derive"] }
 env_logger = { workspace = true }
+directories = { workspace = true }
 log = { workspace = true }
+serde-toml-merge = { workspace = true }
 tempfile = { workspace = true }
 tokio = { workspace = true, features = ["rt-multi-thread", "macros"] }
 toml = { workspace = true }

topiary-cli/src/configuration.rs

@@ -0,0 +1,123 @@
+use directories::ProjectDirs;
+use std::{env::current_dir, path::PathBuf};
+use topiary::{default_configuration_toml, Configuration};
+
+use crate::error::{CLIResult, TopiaryError};
+
+pub fn parse_configuration() -> CLIResult<Configuration> {
+    user_configuration_toml()?
+        .try_into()
+        .map_err(TopiaryError::from)
+}
+
+/// User configured languages.toml file, merged with the default config.
+fn user_configuration_toml() -> CLIResult<toml::Value> {
+    let config = [find_configuration_dir(), find_workspace().join(".topiary")]
+        .into_iter()
+        .map(|path| path.join("languages.toml"))
+        .filter_map(|file| {
+            std::fs::read_to_string(file)
+                .map(|config| toml::from_str(&config))
+                .ok()
+        })
+        .collect::<Result<Vec<_>, _>>()?
+        .into_iter()
+        .fold(default_configuration_toml(), |a, b| {
+            merge_toml_values(a, b, 3)
+        });
+
+    Ok(config)
+}
+
+/// Merge two TOML documents, merging values from `right` onto `left`
+///
+/// When an array exists in both `left` and `right`, `right`'s array is
+/// used. When a table exists in both `left` and `right`, the merged table
+/// consists of all keys in `left`'s table unioned with all keys in `right`
+/// with the values of `right` being merged recursively onto values of
+/// `left`.
+///
+/// `merge_toplevel_arrays` controls whether a top-level array in the TOML
+/// document is merged instead of overridden. This is useful for TOML
+/// documents that use a top-level array of values like the `languages.toml`,
+/// where one usually wants to override or add to the array instead of
+/// replacing it altogether.
+///
+/// NOTE: This merge function is taken from Helix:
+/// https://github.com/helix-editor/helix licensed under MPL-2.0. There
+/// it is defined under: helix-loader/src/lib.rs. Taken from commit df09490
+pub fn merge_toml_values(left: toml::Value, right: toml::Value, merge_depth: usize) -> toml::Value {
+    use toml::Value;
+
+    fn get_name(v: &Value) -> Option<&str> {
+        v.get("name").and_then(Value::as_str)
+    }
+
+    match (left, right) {
+        (Value::Array(mut left_items), Value::Array(right_items)) => {
+            // The top-level arrays should be merged but nested arrays should
+            // act as overrides. For the `languages.toml` config, this means
+            // that you can specify a sub-set of languages in an overriding
+            // `languages.toml` but that nested arrays like file extensions
+            // arguments are replaced instead of merged.
+            if merge_depth > 0 {
+                left_items.reserve(right_items.len());
+                for rvalue in right_items {
+                    let lvalue = get_name(&rvalue)
+                        .and_then(|rname| {
+                            left_items.iter().position(|v| get_name(v) == Some(rname))
+                        })
+                        .map(|lpos| left_items.remove(lpos));
+                    let mvalue = match lvalue {
+                        Some(lvalue) => merge_toml_values(lvalue, rvalue, merge_depth - 1),
+                        None => rvalue,
+                    };
+                    left_items.push(mvalue);
+                }
+                Value::Array(left_items)
+            } else {
+                Value::Array(right_items)
+            }
+        }
+        (Value::Table(mut left_map), Value::Table(right_map)) => {
+            if merge_depth > 0 {
+                for (rname, rvalue) in right_map {
+                    match left_map.remove(&rname) {
+                        Some(lvalue) => {
+                            let merged_value = merge_toml_values(lvalue, rvalue, merge_depth - 1);
+                            left_map.insert(rname, merged_value);
+                        }
+                        None => {
+                            left_map.insert(rname, rvalue);
+                        }
+                    }
+                }
+                Value::Table(left_map)
+            } else {
+                Value::Table(right_map)
+            }
+        }
+        // Catch everything else we didn't handle, and use the right value
+        (_, value) => value,
+    }
+}
+
+fn find_configuration_dir() -> PathBuf {
+    ProjectDirs::from("", "", "topiary")
+        .expect("Could not access the OS's Home directory")
+        .config_dir()
+        .to_owned()
+}
+
+pub fn find_workspace() -> PathBuf {
+    let current_dir = current_dir().expect("Could not get current working directory");
+    for ancestor in current_dir.ancestors() {
+        if ancestor.join(".topiary").exists() {
+            return ancestor.to_owned();
+        }
+    }
+
+    // Default to the current dir if we could not find an ancestor with the .topiary directory
+    // If current_dir does not contain a .topiary, it will be filtered our in the `user_lang_toml` function.
+    current_dir
+}

topiary-cli/src/error.rs

@@ -116,3 +116,21 @@ where
         )
     }
 }
+
+impl From<toml::de::Error> for TopiaryError {
+    fn from(e: toml::de::Error) -> Self {
+        TopiaryError::Bin(
+            "Could not parse user configuration".to_owned(),
+            Some(CLIError::Generic(Box::new(e))),
+        )
+    }
+}
+
+impl From<serde_toml_merge::Error> for TopiaryError {
+    fn from(e: serde_toml_merge::Error) -> Self {
+        TopiaryError::Bin(
+            format!("Could not merge the default configuration and user configurations. Error occured while merging: {}", e.path),
+            None,
+        )
+    }
+}

topiary-cli/src/main.rs

@@ -1,8 +1,10 @@
+mod configuration;
 mod error;
 mod output;
 mod visualise;
 
 use std::{
+    eprintln,
     error::Error,
     fs::File,
     io::{stdin, BufReader, BufWriter, Read},
@@ -11,13 +13,14 @@ use std::{
 };
 
 use clap::{ArgGroup, Parser};
+use configuration::parse_configuration;
 
 use crate::{
     error::{CLIError, CLIResult, TopiaryError},
     output::OutputFile,
     visualise::Visualisation,
 };
-use topiary::{formatter, Configuration, Language, Operation, SupportedLanguage};
+use topiary::{formatter, Language, Operation, SupportedLanguage};
 
 #[derive(Parser, Debug)]
 #[command(author, version, about, long_about = None)]
@@ -64,6 +67,10 @@ struct Args {
     /// Do not check that formatting twice gives the same output
     #[arg(short, long, display_order = 7)]
     skip_idempotence: bool,
+
+    /// Output the full configuration to stderr before continuing
+    #[arg(long, display_order = 8)]
+    output_configuration: bool,
 }
 
 #[tokio::main]
@@ -80,7 +87,11 @@ async fn run() -> CLIResult<()> {
     env_logger::init();
     let args = Args::parse();
 
-    let configuration = Configuration::parse_default_config();
+    let configuration = parse_configuration()?;
+
+    if args.output_configuration {
+        eprintln!("{:#?}", configuration);
+    }
 
     // The as_deref() gives us an Option<&str>, which we can match against
     // string literals

topiary-playground/build.rs

@@ -51,7 +51,8 @@ fn main() {
     for path in input_files {
         let path = path.unwrap().path();
         if let Some(ext) = path.extension().map(|ext| ext.to_string_lossy()) {
-            if !Configuration::parse_default_config()
+            if !Configuration::parse_default_configuration()
+                .unwrap()
                 .known_extensions()
                 .contains(&*ext)
                 || ext == "mli"

topiary-playground/src/lib.rs

@@ -45,7 +45,7 @@ async fn format_inner(
 ) -> FormatterResult<String> {
     let mut output = Vec::new();
 
-    let configuration = Configuration::parse_default_config();
+    let configuration = Configuration::parse_default_configuration()?;
     let language = configuration.get_language(language_name)?;
     let grammars = language.grammars_wasm().await?;
 

topiary/benches/benchmark.rs

@@ -8,7 +8,7 @@ use topiary::{formatter, Operation};
 async fn format() {
     let input = fs::read_to_string("tests/samples/input/ocaml.ml").unwrap();
     let query = fs::read_to_string("../languages/ocaml.scm").unwrap();
-    let configuration = Configuration::parse_default_config();
+    let configuration = Configuration::parse_default_configuration().unwrap();
     let language = configuration.get_language("ocaml_implementation").unwrap();
     let grammars = language.grammars().await.unwrap();
 

topiary/src/configuration.rs

@@ -3,20 +3,14 @@ use std::{collections::HashSet, str::from_utf8};
 use crate::{language::Language, FormatterError, FormatterResult};
 use serde::{Deserialize, Serialize};
 
-#[derive(Deserialize, Serialize)]
+#[derive(Deserialize, Serialize, Debug)]
 pub struct Configuration {
     pub language: Vec<Language>,
 }
 
 impl Configuration {
-    // TODO: Should be able to take a filepath.
-    // TODO: Should return a FormatterResult rather than panicking.
-    #[must_use]
-    pub fn parse_default_config() -> Self {
-        let default_config = include_bytes!("../../languages.toml");
-        let default_config = toml::from_str(from_utf8(default_config).unwrap())
-            .expect("Could not parse built-in languages.toml");
-        default_config
+    pub fn new() -> Self {
+        Configuration { language: vec![] }
     }
 
     #[must_use]
@@ -40,4 +34,23 @@ impl Configuration {
             name.as_ref().to_string(),
         ));
     }
+
+    pub fn parse_default_configuration() -> FormatterResult<Self> {
+        default_configuration_toml()
+            .try_into()
+            .map_err(FormatterError::from)
+    }
+}
+
+impl Default for Configuration {
+    fn default() -> Self {
+        Self::new()
+    }
+}
+
+/// Default built-in languages.toml.
+pub fn default_configuration_toml() -> toml::Value {
+    let default_config = include_bytes!("../../languages.toml");
+    toml::from_str(from_utf8(default_config).unwrap())
+        .expect("Could not parse built-in languages.toml to valid toml")
 }

topiary/src/error.rs

@@ -207,3 +207,12 @@ impl From<tree_sitter_facade::ParserError> for FormatterError {
         Self::Internal("Error while parsing".into(), Some(Box::new(e)))
     }
 }
+
+impl From<toml::de::Error> for FormatterError {
+    fn from(e: toml::de::Error) -> Self {
+        Self::Internal(
+            "Error while parsing the internal configuration file".to_owned(),
+            Some(Box::new(e)),
+        )
+    }
+}