Add crisp tools

Cargo.toml

@@ -21,6 +21,7 @@ exclude = [
     "analysis/tests",
     "examples",
     "tests",
+    "tools",
 ]
 
 [workspace.package]

tools/README.md

@@ -0,0 +1,20 @@
+# C2Rust CRISP tools
+
+- `merge_rust`
+- `split_rust`
+- `split_ffi_entry_points`
+
+## Building
+
+These tools rely on rust-analyzer's libraries, so they need a relatively recent version of Rust. Run `cargo build --release` in their respective directories to build them.
+
+## Running
+
+`split_rust` and `merge_rust` expect the root source file of a Rust project as their first argument, e.g. `lib.rs` or `main.rs`.
+
+- `merge_rust` modifies the specified codebase in-place.
+- `split_rust` emits JSON on standard output.
+
+`split_ffi_entry_points` expects a Rust project directory (the directory containing a `Cargo.toml` file) as its only argument.
+
+It modifies that Rust project in-place.

tools/merge_rust/Cargo.toml

@@ -0,0 +1,14 @@
+[package]
+name = "merge_rust"
+version = "0.1.0"
+edition = "2024"
+
+[dependencies]
+rust_util = { path = "../rust_util" }
+
+syn = { version = "2.0.108", features = ["full", "visit"] }
+proc-macro2 = { version = "1", features = ["span-locations"] }
+clap = { version = "4.5", features = ["derive"] }
+quote = "1"
+
+serde_json = "1"

tools/merge_rust/src/main.rs

@@ -0,0 +1,70 @@
+use clap::Parser;
+use rust_util::collect::FileCollector;
+use rust_util::item_span::item_spans;
+use serde_json;
+use std::collections::HashMap;
+use std::fs::{self, File};
+use std::path::PathBuf;
+
+/// Merge updated item definitions into a Rust codebase.
+#[derive(Parser)]
+struct Args {
+    /// Root Rust source file to update (`lib.rs` or `main.rs`).
+    src_root_path: PathBuf,
+    /// JSON file containing mapping from Rust item paths to desired new contents.
+    new_snippets_file: PathBuf,
+}
+
+fn main() {
+    let args = Args::parse();
+    let src_root_path = args.src_root_path;
+    let new_snippet_json_path = args.new_snippets_file;
+
+    let new_snippets_file = File::open(&new_snippet_json_path).unwrap();
+    let new_snippets: HashMap<String, String> = serde_json::from_reader(new_snippets_file).unwrap();
+
+    let mut fc = FileCollector::default();
+    fc.parse(&src_root_path, vec![], true).unwrap();
+    for &(ref file_path, ref mod_path, ref ast) in &fc.files {
+        eprintln!("visit {:?}", file_path);
+        let old_src = fs::read_to_string(file_path).unwrap();
+
+        let mut rewrites = Vec::new();
+        for (item_path, lo, hi) in item_spans(mod_path.to_owned(), ast) {
+            let old_snippet = &old_src[lo..hi];
+            let item_path_str = item_path.join("::");
+            if let Some(new_snippet) = new_snippets.get(&item_path_str) {
+                if new_snippet != old_snippet {
+                    rewrites.push((lo, hi, new_snippet));
+                }
+            }
+        }
+
+        if rewrites.len() == 0 {
+            continue;
+        }
+
+        rewrites.sort_by_key(|&(lo, _, _)| lo);
+        let mut new_src = String::with_capacity(old_src.len());
+        let mut pos = 0;
+        for &(lo, hi, new_snippet) in &rewrites {
+            assert!(
+                lo >= pos,
+                "overlapping rewrites: previous rewrite ended at {}, \
+                but current rewrite covers {} .. {}",
+                pos,
+                lo,
+                hi
+            );
+            new_src.push_str(&old_src[pos..lo]);
+            new_src.push_str(new_snippet);
+            pos = hi;
+        }
+        new_src.push_str(&old_src[pos..]);
+
+        let tmp_path = file_path.with_extension(".new");
+        fs::write(&tmp_path, &new_src).unwrap();
+        fs::rename(&tmp_path, file_path).unwrap();
+        eprintln!("applied {} rewrites to {:?}", rewrites.len(), file_path);
+    }
+}