verify_rust_std: add diff subcommand to emit a diff merge.json

Author: zjp-CN

src/bin/verify_rust_std/diff.rs

@@ -0,0 +1,31 @@
+use crate::{Result, read_json};
+use clap::Parser;
+use distributed_verification::diff::{MergeHashKaniList, diff};
+
+pub fn run(args: &[String]) -> Result<()> {
+    SubCmdDiff::parse_from(args).run()
+}
+
+/// `verify_rust_std diff` subcommand that emits a diff merge.json.
+#[derive(Parser, Debug)]
+struct SubCmdDiff {
+    /// Path to a previous merge.json that is generated by `verify_rust_std merge`.
+    #[arg(long)]
+    old: String,
+
+    /// Path to a recent merge.json that is generated by `verify_rust_std merge`.
+    #[arg(long)]
+    new: String,
+}
+
+impl SubCmdDiff {
+    fn run(self) -> Result<()> {
+        let old: Vec<MergeHashKaniList> = read_json(&self.old)?;
+        let new: Vec<MergeHashKaniList> = read_json(&self.new)?;
+
+        let difference = diff(&old, &new);
+        serde_json::to_writer_pretty(std::io::stdout(), &difference)?;
+
+        Ok(())
+    }
+}

src/bin/verify_rust_std/main.rs

@@ -1,6 +1,7 @@
 //! `VERIFY_RUST_STD_LIBRARY=path/to/verify-rust-std/library` and
 //! `KANI_DIR=path/to/kani` should be set beforehand.
 
+use distributed_verification::logger;
 use eyre::{Context, Result};
 use std::process::{Command, Stdio};
 
@@ -12,6 +13,7 @@ extern crate eyre;
 mod env;
 use env::ENV;
 
+mod diff;
 mod merge;
 
 fn main() -> Result<()> {
@@ -41,8 +43,11 @@ fn main() -> Result<()> {
             run("rustc", rustc_args, &[])
         }
     } else if args.get(1).map(|arg| arg == "merge").unwrap_or(false) {
-        distributed_verification::logger::init();
+        logger::init();
         merge::run(&args[1..])
+    } else if args.get(1).map(|arg| arg == "diff").unwrap_or(false) {
+        logger::init();
+        diff::run(&args[1..])
     } else {
         run(
             "cargo",
@@ -85,3 +90,9 @@ fn build_core(args: Vec<String>) -> Result<()> {
     new_args.extend(args);
     run(&ENV.DISTRIBUTED_VERIFICATION, &new_args, &[])
 }
+
+fn read_json<T: serde::de::DeserializeOwned>(path: &str) -> Result<T> {
+    let _span = error_span!("read_json", path).entered();
+    let file = std::fs::File::open(path)?;
+    Ok(serde_json::from_reader(file)?)
+}

src/bin/verify_rust_std/merge.rs

@@ -4,7 +4,7 @@ use distributed_verification::{
     SerFunction,
     diff::{KaniListJson, MergeHashKaniList},
 };
-use std::{collections::HashMap, fs::File};
+use std::collections::HashMap;
 
 pub fn run(args: &[String]) -> Result<()> {
     SubCmdMerge::parse_from(args).run()
@@ -34,7 +34,11 @@ impl SubCmdMerge {
     fn run(self) -> Result<()> {
         debug!(?self);
 
-        let (mut hash_json, mut kani_list) = read_json(&self.hash_json, &self.kani_list)?;
+        let mut hash_json: Vec<SerFunction> = crate::read_json(&self.hash_json)?;
+
+        let mut kani_list: KaniListJson = crate::read_json(&self.kani_list)?;
+        // normalize paths like core/src/../../portable-simd/crates/core_simd/src/masks.rs
+        kani_list.normalize_file_path();
 
         if !self.strip_kani_list_prefix.is_empty() {
             kani_list.strip_path_prefix_raw(&self.strip_kani_list_prefix);
@@ -65,24 +69,3 @@ impl SubCmdMerge {
         Ok(())
     }
 }
-
-fn read_json(
-    hash_json_path: &str,
-    kani_list_path: &str,
-) -> Result<(Vec<SerFunction>, KaniListJson)> {
-    let hash_json: Vec<SerFunction> = {
-        let _span = error_span!("hash_json_path", hash_json_path).entered();
-        let file = File::open(hash_json_path)?;
-        serde_json::from_reader(file)?
-    };
-
-    let mut kani_list: KaniListJson = {
-        let _span = error_span!("kani_list_path", kani_list_path).entered();
-        let file = File::open(kani_list_path)?;
-        serde_json::from_reader(file)?
-    };
-    // normalize paths like core/src/../../portable-simd/crates/core_simd/src/masks.rs
-    kani_list.normalize_file_path();
-
-    Ok((hash_json, kani_list))
-}

src/diff.rs

@@ -256,9 +256,12 @@ pub struct MergeHashKaniList {
     pub hash: Option<Box<str>>,
 }
 
-/// Compare two `MergeHashKaniList`, and returns the ones from new that don't have hash values or
-/// whoes hash values changed.
+/// Compare two `MergeHashKaniList`, and returns the ones from new that don't match the old,
+/// usually the ones don't have hash values or whoes hash values changed.
+///
+/// If new is sorted, especially directly from the stdout of verify_rust_std merge subcommand,
+/// the result is sorted.
 pub fn diff(old: &[MergeHashKaniList], new: &[MergeHashKaniList]) -> Vec<MergeHashKaniList> {
     let set: HashSet<_> = old.iter().collect();
-    new.iter().filter(|item| !set.contains(item)).cloned().collect()
+    new.iter().filter(|item| item.hash.is_none() || !set.contains(item)).cloned().collect()
 }