refactor: remove unnecessary asm/ tree copy from standards and agglayer build scripts

crates/miden-agglayer/asm/agglayer/bridge/canonical_zeros.masm

@@ -1,8 +1,8 @@
-# This file is generated by build.rs, do not modify
+# This file contains deterministic values. Do not modify manually.
 
-# This file contains the canonical zeros for the Keccak hash function. 
+# This file contains the canonical zeros for the Keccak hash function.
 # Zero of height `n` (ZERO_N) is the root of the binary tree of height `n` with leaves equal zero.
-# 
+#
 # Since the Keccak hash is represented by eight u32 values, each constant consists of two Words.
 
 const ZERO_0_L = [0, 0, 0, 0]
@@ -102,7 +102,8 @@ const ZERO_31_L = [2340505732, 1648733876, 2660540036, 3759582231]
 const ZERO_31_R = [2389186238, 4049365781, 1653344606, 2840985724]
 
 use ::agglayer::common::utils::mem_store_double_word
-
+
+
 #! Inputs:  [zeros_ptr]
 #! Outputs: []
 pub proc load_zeros_to_memory

crates/miden-agglayer/build.rs

@@ -21,19 +21,14 @@ use miden_standards::account::mint_policies::OwnerControlled;
 // CONSTANTS
 // ================================================================================================
 
-/// Defines whether the build script should generate files in `/src`.
-/// The docs.rs build pipeline has a read-only filesystem, so we have to avoid writing to `src`,
-/// otherwise the docs will fail to build there. Note that writing to `OUT_DIR` is fine.
-const BUILD_GENERATED_FILES_IN_SRC: bool = option_env!("BUILD_GENERATED_FILES_IN_SRC").is_some();
-
 const ASSETS_DIR: &str = "assets";
 const ASM_DIR: &str = "asm";
 const ASM_NOTE_SCRIPTS_DIR: &str = "note_scripts";
 const ASM_AGGLAYER_DIR: &str = "agglayer";
 const ASM_AGGLAYER_BRIDGE_DIR: &str = "agglayer/bridge";
 const ASM_COMPONENTS_DIR: &str = "components";
 
-const AGGLAYER_ERRORS_FILE: &str = "src/errors/agglayer.rs";
+const AGGLAYER_ERRORS_RS_FILE: &str = "agglayer_errors.rs";
 const AGGLAYER_ERRORS_ARRAY_NAME: &str = "AGGLAYER_ERRORS";
 const AGGLAYER_GLOBAL_CONSTANTS_FILE_NAME: &str = "agglayer_constants.rs";
 
@@ -47,21 +42,17 @@ const AGGLAYER_GLOBAL_CONSTANTS_FILE_NAME: &str = "agglayer_constants.rs";
 fn main() -> Result<()> {
     // re-build when the MASM code changes
     println!("cargo::rerun-if-changed={ASM_DIR}/");
-    println!("cargo::rerun-if-env-changed=BUILD_GENERATED_FILES_IN_SRC");
+    println!("cargo::rerun-if-env-changed=REGENERATE_CANONICAL_ZEROS");
 
-    // Copies the MASM code to the build directory
     let crate_dir = env::var("CARGO_MANIFEST_DIR").unwrap();
     let build_dir = env::var("OUT_DIR").unwrap();
-    let src = Path::new(&crate_dir).join(ASM_DIR);
-
-    // generate canonical zeros in `asm/agglayer/bridge/canonical_zeros.masm`
-    generate_canonical_zeros(&src.join(ASM_AGGLAYER_BRIDGE_DIR))?;
 
-    let dst = Path::new(&build_dir).to_path_buf();
-    shared::copy_directory(src, &dst, ASM_DIR)?;
+    // validate (or regenerate) canonical zeros in `asm/agglayer/bridge/canonical_zeros.masm`
+    let crate_path = Path::new(&crate_dir);
+    ensure_canonical_zeros(&crate_path.join(ASM_DIR).join(ASM_AGGLAYER_BRIDGE_DIR))?;
 
-    // set source directory to {OUT_DIR}/asm
-    let source_dir = dst.join(ASM_DIR);
+    // Read MASM sources directly from the crate's asm/ directory.
+    let source_dir = crate_path.join(ASM_DIR);
 
     // set target directory to {OUT_DIR}/assets
     let target_dir = Path::new(&build_dir).join(ASSETS_DIR);
@@ -91,7 +82,7 @@ fn main() -> Result<()> {
     let constants_out_path = Path::new(&build_dir).join(AGGLAYER_GLOBAL_CONSTANTS_FILE_NAME);
     generate_agglayer_constants(constants_out_path, component_libraries)?;
 
-    generate_error_constants(&source_dir)?;
+    generate_error_constants(&source_dir, &build_dir)?;
 
     Ok(())
 }
@@ -320,23 +311,15 @@ fn generate_agglayer_constants(
 ///
 /// The function ensures that a constant is not defined twice, except if their error message is the
 /// same. This can happen across multiple files.
-///
-/// Because the error files will be written to ./src/errors, this should be a no-op if ./src is
-/// read-only. To enable writing to ./src, set the `BUILD_GENERATED_FILES_IN_SRC` environment
-/// variable.
-fn generate_error_constants(asm_source_dir: &Path) -> Result<()> {
-    if !BUILD_GENERATED_FILES_IN_SRC {
-        return Ok(());
-    }
-
+fn generate_error_constants(asm_source_dir: &Path, build_dir: &str) -> Result<()> {
     // Miden agglayer errors
     // ------------------------------------------
 
     let errors = shared::extract_all_masm_errors(asm_source_dir)
         .context("failed to extract all masm errors")?;
     shared::generate_error_file(
         shared::ErrorModule {
-            file_name: AGGLAYER_ERRORS_FILE,
+            file_path: Path::new(build_dir).join(AGGLAYER_ERRORS_RS_FILE),
             array_name: AGGLAYER_ERRORS_ARRAY_NAME,
             is_crate_local: false,
         },
@@ -346,14 +329,13 @@ fn generate_error_constants(asm_source_dir: &Path) -> Result<()> {
     Ok(())
 }
 
-// CANONICAL ZEROS FILE GENERATION
+// CANONICAL ZEROS VALIDATION
 // ================================================================================================
 
-fn generate_canonical_zeros(target_dir: &Path) -> Result<()> {
-    if !BUILD_GENERATED_FILES_IN_SRC {
-        return Ok(());
-    }
-
+/// Validates that the committed `canonical_zeros.masm` matches the expected content computed from
+/// Keccak256 canonical zeros. If the `REGENERATE_CANONICAL_ZEROS` environment variable is set,
+/// the file is regenerated instead.
+fn ensure_canonical_zeros(target_dir: &Path) -> Result<()> {
     const TREE_HEIGHT: u8 = 32;
 
     let mut zeros_by_height = Vec::with_capacity(TREE_HEIGHT as usize);
@@ -373,10 +355,10 @@ fn generate_canonical_zeros(target_dir: &Path) -> Result<()> {
     // convert the keccak digest into the sequence of u32 values and create two word constants from
     // them to represent the hash
     let mut zero_constants = String::from(
-        "# This file is generated by build.rs, do not modify\n
-# This file contains the canonical zeros for the Keccak hash function. 
+        "# This file contains deterministic values. Do not modify manually.\n
+# This file contains the canonical zeros for the Keccak hash function.
 # Zero of height `n` (ZERO_N) is the root of the binary tree of height `n` with leaves equal zero.
-# 
+#
 # Since the Keccak hash is represented by eight u32 values, each constant consists of two Words.\n",
     );
 
@@ -398,7 +380,8 @@ fn generate_canonical_zeros(target_dir: &Path) -> Result<()> {
     zero_constants.push_str(
         "
 use ::agglayer::common::utils::mem_store_double_word
-
+
+
 #! Inputs:  [zeros_ptr]
 #! Outputs: []
 pub proc load_zeros_to_memory\n",
@@ -410,9 +393,23 @@ pub proc load_zeros_to_memory\n",
 
     zero_constants.push_str("\tdrop\nend\n");
 
-    // write the resulting masm content into the file only if it changed to avoid
-    // invalidating the cargo fingerprint for the `asm/` directory
-    shared::write_if_changed(target_dir.join("canonical_zeros.masm"), zero_constants)?;
+    let file_path = target_dir.join("canonical_zeros.masm");
+
+    if option_env!("REGENERATE_CANONICAL_ZEROS").is_some() {
+        // Regeneration mode: write the file
+        shared::write_if_changed(&file_path, &zero_constants)?;
+    } else {
+        // Validation mode: ensure the committed file matches
+        let committed = fs::read_to_string(&file_path)
+            .into_diagnostic()
+            .wrap_err("canonical_zeros.masm not found - it should be committed in the repo")?;
+        if committed != zero_constants {
+            return Err(Report::msg(
+                "canonical_zeros.masm is out of date. \
+                 Run with REGENERATE_CANONICAL_ZEROS=1 to regenerate and commit the result.",
+            ));
+        }
+    }
 
     Ok(())
 }
@@ -431,54 +428,6 @@ mod shared {
     use regex::Regex;
     use walkdir::WalkDir;
 
-    /// Recursively copies `src` into `dst`.
-    ///
-    /// This function will overwrite the existing files if re-executed.
-    pub fn copy_directory<T: AsRef<Path>, R: AsRef<Path>>(
-        src: T,
-        dst: R,
-        asm_dir: &str,
-    ) -> Result<()> {
-        let mut prefix = src.as_ref().canonicalize().unwrap();
-        // keep all the files inside the `asm` folder
-        prefix.pop();
-
-        let target_dir = dst.as_ref().join(asm_dir);
-        if target_dir.exists() {
-            // Clear existing asm files that were copied earlier which may no longer exist.
-            fs::remove_dir_all(&target_dir)
-                .into_diagnostic()
-                .wrap_err("failed to remove ASM directory")?;
-        }
-
-        // Recreate the directory structure.
-        fs::create_dir_all(&target_dir)
-            .into_diagnostic()
-            .wrap_err("failed to create ASM directory")?;
-
-        let dst = dst.as_ref();
-        let mut todo = vec![src.as_ref().to_path_buf()];
-
-        while let Some(goal) = todo.pop() {
-            for entry in fs::read_dir(goal).unwrap() {
-                let path = entry.unwrap().path();
-                if path.is_dir() {
-                    let src_dir = path.canonicalize().unwrap();
-                    let dst_dir = dst.join(src_dir.strip_prefix(&prefix).unwrap());
-                    if !dst_dir.exists() {
-                        fs::create_dir_all(&dst_dir).unwrap();
-                    }
-                    todo.push(src_dir);
-                } else {
-                    let dst_file = dst.join(path.strip_prefix(&prefix).unwrap());
-                    fs::copy(&path, dst_file).unwrap();
-                }
-            }
-        }
-
-        Ok(())
-    }
-
     /// Returns a vector with paths to all MASM files in the specified directory.
     ///
     /// All non-MASM files are skipped.
@@ -657,7 +606,7 @@ mod shared {
             .into_diagnostic()?;
         }
 
-        write_if_changed(module.file_name, output)?;
+        fs::write(module.file_path, output).into_diagnostic()?;
 
         Ok(())
     }
@@ -690,9 +639,9 @@ mod shared {
         pub message: String,
     }
 
-    #[derive(Debug, Clone, PartialEq, Eq, PartialOrd, Ord)]
+    #[derive(Debug, Clone)]
     pub struct ErrorModule {
-        pub file_name: &'static str,
+        pub file_path: PathBuf,
         pub array_name: &'static str,
         pub is_crate_local: bool,
     }

crates/miden-agglayer/src/errors/mod.rs

@@ -1,3 +1,3 @@
 // Include generated error constants
 #[cfg(any(feature = "testing", test))]
-include!("agglayer.rs");
+include!(concat!(env!("OUT_DIR"), "/agglayer_errors.rs"));