From 42acb72c54e84149e4102cd22bca107f3e5a9e19 Mon Sep 17 00:00:00 2001
From: Andrew Brown <andrew.brown@intel.com>
Date: Tue, 10 Aug 2021 11:08:30 -0700
Subject: [PATCH] fuzz: retrieve the WebAssembly spec repository in `build.rs`

To avoid the large download size of the spec repository mentioned
[here](https://github.com/bytecodealliance/wasmtime/pull/3124#discussion_r684605984),
this change removes it as a submodule and instead clones it shallowly
when the directory is empty (or not present) when `build.rs` is run.
---
 .gitmodules                                   |  4 ---
 crates/fuzzing/wasm-spec-interpreter/build.rs | 33 ++++++++++++++++++-
 .../wasm-spec-interpreter/ocaml/.gitignore    |  1 +
 .../fuzzing/wasm-spec-interpreter/ocaml/spec  |  1 -
 4 files changed, 33 insertions(+), 6 deletions(-)
 delete mode 160000 crates/fuzzing/wasm-spec-interpreter/ocaml/spec

diff --git a/.gitmodules b/.gitmodules
index 725b4b4dc28c..ee264b99c477 100644
--- a/.gitmodules
+++ b/.gitmodules
@@ -13,7 +13,3 @@
 [submodule "crates/wasi-crypto/spec"]
 	path = crates/wasi-crypto/spec
 	url = https://github.com/WebAssembly/wasi-crypto.git
-[submodule "crates/fuzzing/wasm-spec-interpreter/ocaml/spec"]
-	path = crates/fuzzing/wasm-spec-interpreter/ocaml/spec
-	url = https://github.com/WebAssembly/spec
-	shallow = true
diff --git a/crates/fuzzing/wasm-spec-interpreter/build.rs b/crates/fuzzing/wasm-spec-interpreter/build.rs
index 61e8c316eb31..4d074febbc13 100644
--- a/crates/fuzzing/wasm-spec-interpreter/build.rs
+++ b/crates/fuzzing/wasm-spec-interpreter/build.rs
@@ -5,10 +5,12 @@
 /// approach to avoid missing symbols was to imitate `dune`: I observed `rm -rf
 /// _build && dune build ./ocaml/interpret.exe.o --display=verbose` and used
 /// that as a pattern, now encoded in `ocaml/Makefile` for easier debugging.
-use std::{env, process::Command};
+use std::{env, path::PathBuf, process::Command};
 
 const LIB_NAME: &'static str = "interpret";
 const OCAML_DIR: &'static str = "ocaml";
+const SPEC_DIR: &'static str = "ocaml/spec";
+const SPEC_REPOSITORY: &'static str = "https://github.com/bytecodealliance/wasm-spec-mirror";
 
 fn main() {
     if cfg!(feature = "build-libinterpret") {
@@ -28,6 +30,11 @@ fn build() {
         println!("cargo:rustc-link-search={}", other_dir.to_str().unwrap());
         println!("cargo:rustc-link-lib=static={}", LIB_NAME);
     } else {
+        // Ensure the spec repository is present.
+        if is_spec_repository_empty(SPEC_DIR) {
+            retrieve_spec_repository(SPEC_REPOSITORY, SPEC_DIR)
+        }
+
         // Build the library to link to.
         build_lib(out_dir, OCAML_DIR);
         println!("cargo:rustc-link-search={}", out_dir);
@@ -52,3 +59,27 @@ fn build_lib(out_dir: &str, ocaml_dir: &str) {
         "Failed to build the OCaml library using 'make'."
     )
 }
+
+// Check if the spec repository directory contains any files.
+fn is_spec_repository_empty(destination: &str) -> bool {
+    PathBuf::from(destination)
+        .read_dir()
+        .map(|mut i| i.next().is_none())
+        .unwrap_or(true)
+}
+
+// Clone the spec repository into `destination`. This exists due to the large
+// size of the dependencies (e.g. KaTeX) that are pulled if this were cloned
+// recursively as a submodule.
+fn retrieve_spec_repository(repository: &str, destination: &str) {
+    let status = Command::new("git")
+        .arg("clone")
+        .arg("--depth")
+        .arg("1")
+        .arg(repository)
+        .arg(destination)
+        .status()
+        .expect("Failed to execute 'git' command to clone spec repository.");
+
+    assert!(status.success(), "Failed to retrieve the spec repository.")
+}
diff --git a/crates/fuzzing/wasm-spec-interpreter/ocaml/.gitignore b/crates/fuzzing/wasm-spec-interpreter/ocaml/.gitignore
index e35d8850c968..1e82b679cf42 100644
--- a/crates/fuzzing/wasm-spec-interpreter/ocaml/.gitignore
+++ b/crates/fuzzing/wasm-spec-interpreter/ocaml/.gitignore
@@ -1 +1,2 @@
 _build
+spec
diff --git a/crates/fuzzing/wasm-spec-interpreter/ocaml/spec b/crates/fuzzing/wasm-spec-interpreter/ocaml/spec
deleted file mode 160000
index 46cf27ce076e..000000000000
--- a/crates/fuzzing/wasm-spec-interpreter/ocaml/spec
+++ /dev/null
@@ -1 +0,0 @@
-Subproject commit 46cf27ce076e1213e45b89ffdc99df4499b3cb33