feat: set up auto-format

engine/baml-schema-wasm/src/lib.rs

@@ -1,6 +1,7 @@
 #[cfg(target_arch = "wasm32")]
 pub mod runtime_wasm;
 
+use internal_baml_core::internal_baml_schema_ast::{format_schema, FormatOptions};
 use std::env;
 use wasm_bindgen::prelude::*;
 
@@ -9,3 +10,24 @@ pub fn version() -> String {
     // register_panic_hook();
     env!("CARGO_PKG_VERSION").to_string()
 }
+
+#[wasm_bindgen]
+pub fn format_document(path: String, text: String) -> Option<String> {
+    log::info!("Trying to format document (rust): {}", path);
+    match format_schema(
+        &text,
+        FormatOptions {
+            indent_width: 2,
+            fail_on_unhandled_rule: false,
+        },
+    ) {
+        Ok(formatted) => {
+            log::info!("Formatted document: {}", formatted);
+            Some(formatted)
+        }
+        Err(e) => {
+            log::error!("Failed to format document: {} {:?}", path, e);
+            None
+        }
+    }
+}

engine/baml-schema-wasm/src/runtime_wasm/mod.rs

@@ -57,7 +57,8 @@ pub fn on_wasm_init() {
         if #[cfg(debug_assertions)] {
             const LOG_LEVEL: log::Level = log::Level::Debug;
         } else {
-            const LOG_LEVEL: log::Level = log::Level::Warn;
+            // const LOG_LEVEL: log::Level = log::Level::Warn;
+            const LOG_LEVEL: log::Level = log::Level::Debug;
         }
     };
     // I dont think we need this line anymore -- seems to break logging if you add it.

engine/cli/src/commands.rs

@@ -68,7 +68,12 @@ impl RuntimeCli {
                 args.from = BamlRuntime::parse_baml_src_path(&args.from)?;
                 t.block_on(async { args.run_async().await })
             }
-            Commands::Format(args) => args.run(),
+            Commands::Format(args) => {
+                // We deliberately don't apply parse_baml_src_path here
+                // see format.rs for more details
+                // args.from = BamlRuntime::parse_baml_src_path(&args.from)?;
+                args.run()
+            }
         }
     }
 }

engine/cli/src/format.rs

@@ -1,31 +1,64 @@
 use std::{fs, path::PathBuf};
 
 use anyhow::Result;
+use baml_runtime::{baml_src_files, BamlRuntime};
 use clap::Args;
 use internal_baml_core::internal_baml_schema_ast::{format_schema, FormatOptions};
 
 #[derive(Args, Debug)]
 pub struct FormatArgs {
     #[arg(long, help = "path/to/baml_src", default_value = "./baml_src")]
     pub from: PathBuf,
+
+    #[arg(
+        help = "Specific files to format. If none provided, formats all files in the baml_src directory"
+    )]
+    pub paths: Vec<PathBuf>,
+
+    #[arg(
+        short = 'n',
+        long = "dry-run",
+        help = "Write formatter changes to stdout instead of files",
+        default_value = "false"
+    )]
+    pub dry_run: bool,
 }
 
 impl FormatArgs {
     pub fn run(&self) -> Result<()> {
-        let source = fs::read_to_string(&self.from)?;
-        let formatted = format_schema(
-            &source,
-            FormatOptions {
-                indent_width: 4,
-                fail_on_unhandled_rule: false,
-            },
-        )?;
-
-        let mut to = self.from.clone();
-        to.set_extension("formatted.baml");
-        fs::write(&to, formatted)?;
-
-        log::info!("Formatted {} to {}", self.from.display(), to.display());
+        let paths = if self.paths.is_empty() {
+            // Usually this is done in commands.rs, but fmt is a special case
+            // because it doesn't need to actually load the BAML runtime to parse
+            // BAML files.
+            let from = BamlRuntime::parse_baml_src_path(&self.from)?;
+            baml_src_files(&from)?
+        } else {
+            self.paths.clone()
+        };
+
+        for path in paths.iter() {
+            let source = fs::read_to_string(&path)?;
+            match format_schema(
+                &source,
+                FormatOptions {
+                    indent_width: 2,
+                    fail_on_unhandled_rule: false,
+                },
+            ) {
+                Ok(formatted) => {
+                    if self.dry_run {
+                        println!("{}", formatted);
+                    } else {
+                        fs::write(&path, formatted)?;
+                    }
+                }
+                Err(e) => {
+                    log::error!("Failed to format {}: {}", path.display(), e);
+                }
+            }
+        }
+
+        log::info!("Formatted {} files", paths.len());
 
         Ok(())
     }

integ-tests/baml_src/clients.baml

@@ -19,7 +19,7 @@ client<llm> GPT4 {
     model gpt-4o
     api_key env.OPENAI_API_KEY
   }
-}  
+}
 
 
 client<llm> GPT4o {
@@ -28,7 +28,7 @@ client<llm> GPT4o {
     model gpt-4o
     api_key env.OPENAI_API_KEY
   }
-} 
+}
 
 
 client<llm> GPT4Turbo {
@@ -38,7 +38,7 @@ client<llm> GPT4Turbo {
     model gpt-4-turbo
     api_key env.OPENAI_API_KEY
   }
-} 
+}
 
 client<llm> GPT35 {
   provider openai
@@ -164,7 +164,7 @@ client<llm> Resilient_SimpleSyntax {
       Lottery_SimpleSyntax
     ]
   }
-} 
+}
 
 client<llm> Lottery_SimpleSyntax {
   provider baml-round-robin

integ-tests/baml_src/fiddle-examples/extract-receipt-info.baml

@@ -6,9 +6,9 @@ class ReceiptItem {
 }
 
 class ReceiptInfo {
-    items ReceiptItem[]
-    total_cost float?
-    venue "barisa" | "ox_burger"
+  items ReceiptItem[]
+  total_cost float?
+  venue "barisa" | "ox_burger"
 }
 
 function ExtractReceiptInfo(email: string, reason: "curiosity" | "personal_finance") -> ReceiptInfo {

integ-tests/baml_src/test-files/dynamic/dynamic.baml

@@ -13,7 +13,6 @@ enum DynEnumTwo {
 class SomeClassNestedDynamic {
   hi string
   @@dynamic
-
 }
 
 class DynamicClassTwo {

integ-tests/baml_src/test-files/functions/input/named-args/single/named-class.baml

@@ -1,10 +1,7 @@
 class NamedArgsSingleClass {
   key string
   key_two bool
-  key_three int
-  // TODO: doesn't work with keys with numbers
-  // key2 bool
-  // key3 int
+  key_three int // TODO: doesn't work with keys with numbers // key2 bool // key3 int
 }
 
 function TestFnNamedArgsSingleClass(myArg: NamedArgsSingleClass) -> string {

integ-tests/baml_src/test-files/functions/input/named-args/single/named-string-optional.baml

@@ -1,6 +1,6 @@
 
 
- // string[]
+// string[]
 function FnNamedArgsSingleStringOptional(myString: string?) -> string {
   client GPT35
   prompt #"

integ-tests/baml_src/test-files/functions/input/named-args/single/testcase_audio.baml

@@ -15,4 +15,3 @@ test TestAudioInput {
 }
 
 
-

integ-tests/baml_src/test-files/functions/output/enum-list.baml

@@ -7,7 +7,7 @@ function FnEnumListOutput(input: string) -> EnumOutput[] {
 
     Answer:
   "#
-} 
+}
 
 test FnEnumListOutput {
   functions [FnEnumListOutput]

integ-tests/baml_src/test-files/testing_pipeline/output-format.baml

@@ -1,15 +1,16 @@
 class Recipe {
+  // what do comments
     ingredients map<string, Quantity>
-    recipe_type "breakfast" | "dinner"
+  recipe_type "breakfast" | "dinner"
 }
 
 class Quantity {
-    amount int | float
-    unit string?
+  amount int | float
+  unit string?
 }
 
 function AaaSamOutputFormat(recipe: string) -> Recipe {
-  client GPT35
+    client GPT35
   prompt #"
     Return this value back to me: {{recipe}}
 

integ-tests/baml_src/test-files/testing_pipeline/resume.baml

@@ -1,18 +1,18 @@
 class Resume {
-    name string
-    email string
-    phone string
-    experience Education[]
-    education string[]
-    skills string[]
+  name string
+  email string
+  phone string
+  experience Education[]
+  education string[]
+  skills string[]
 }
 
 class Education {
-    institution string
-    location string
-    degree string
-    major string[]
-    graduation_date string?
+  institution string
+  location string
+  degree string
+  major string[]
+  graduation_date string?
 }
 
 template_string AddRole(foo: string) #"