Merge pull request #23 from full-stack-plugins/feat/rust-hook-prompt-guidance

按需校验语法 worker 的固定 grammar 资产

Author: loong10k

crates/codeguard-adapters/src/grammar_asset_manifest.rs

@@ -93,10 +93,39 @@ pub struct GrammarAsset {
 /// 读取仓内固定清单，并检查来源、许可证及候选资产字节。
 pub fn bundled_grammar_candidates() -> Result<GrammarAssetManifest, String> {
     let manifest = parse_grammar_asset_manifest(include_bytes!("../../../grammars/manifest.json"))?;
-    let codegraph_license = include_bytes!("../../../grammars/LICENSE.codegraph");
-    if digest(codegraph_license) != CODEGRAPH_LICENSE_SHA256 {
+    verify_codegraph_license()?;
+    verify_dependency_license(&manifest)?;
+    for asset in &manifest.assets {
+        verify_bundled_asset(asset)?;
+    }
+    Ok(manifest)
+}
+
+/// 只核对指定内置 grammar 的字节与许可，供按需加载的语法工作进程使用。
+/// 参数为固定语言 ID；返回清单身份和静态 WASM 字节，未知语言返回错误。
+pub fn bundled_grammar_candidate(language: &str) -> Result<(GrammarAsset, &'static [u8]), String> {
+    let manifest = parse_grammar_asset_manifest(include_bytes!("../../../grammars/manifest.json"))?;
+    let asset = manifest
+        .assets
+        .iter()
+        .find(|asset| asset.language == language)
+        .ok_or("不支持的 grammar 语种")?;
+    verify_codegraph_license()?;
+    if matches!(language, "objc" | "solidity") {
+        verify_dependency_license(&manifest)?;
+    }
+    let wasm = verify_bundled_asset(asset)?;
+    Ok((asset.clone(), wasm))
+}
+
+fn verify_codegraph_license() -> Result<(), String> {
+    if digest(include_bytes!("../../../grammars/LICENSE.codegraph")) != CODEGRAPH_LICENSE_SHA256 {
         return Err("CodeGraph 许可证字节与固定来源不符".into());
     }
+    Ok(())
+}
+
+fn verify_dependency_license(manifest: &GrammarAssetManifest) -> Result<(), String> {
     if digest(include_bytes!(
         "../../../grammars/LICENSE.tree-sitter-wasms"
     )) != manifest
@@ -106,90 +135,92 @@ pub fn bundled_grammar_candidates() -> Result<GrammarAssetManifest, String> {
     {
         return Err("依赖包许可证字节与固定来源不符".into());
     }
-    for asset in &manifest.assets {
-        let (wasm, license) = match asset.language.as_str() {
-            "c" => (
-                include_bytes!("../../../grammars/c/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/c/LICENSE").as_slice(),
-            ),
-            "cpp" => (
-                include_bytes!("../../../grammars/cpp/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/cpp/LICENSE").as_slice(),
-            ),
-            "csharp" => (
-                include_bytes!("../../../grammars/csharp/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/csharp/LICENSE").as_slice(),
-            ),
-            "go" => (
-                include_bytes!("../../../grammars/go/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/go/LICENSE").as_slice(),
-            ),
-            "javascript" => (
-                include_bytes!("../../../grammars/javascript/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/javascript/LICENSE").as_slice(),
-            ),
-            "lua" => (
-                include_bytes!("../../../grammars/lua/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/lua/LICENSE").as_slice(),
-            ),
-            "luau" => (
-                include_bytes!("../../../grammars/luau/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/luau/LICENSE").as_slice(),
-            ),
-            "java" => (
-                include_bytes!("../../../grammars/java/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/java/LICENSE").as_slice(),
-            ),
-            "typescript" => (
-                include_bytes!("../../../grammars/typescript/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/typescript/LICENSE").as_slice(),
-            ),
-            "python" => (
-                include_bytes!("../../../grammars/python/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/python/LICENSE").as_slice(),
-            ),
-            "tsx" => (
-                include_bytes!("../../../grammars/tsx/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/typescript/LICENSE").as_slice(),
-            ),
-            "zig" => (
-                include_bytes!("../../../grammars/zig/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/zig/LICENSE").as_slice(),
-            ),
-            "objc" => (
-                include_bytes!("../../../grammars/objc/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/objc/LICENSE").as_slice(),
-            ),
-            "solidity" => (
-                include_bytes!("../../../grammars/solidity/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/solidity/LICENSE").as_slice(),
-            ),
-            "rust" => (
-                include_bytes!("../../../grammars/rust/parser.wasm").as_slice(),
-                include_bytes!("../../../grammars/rust/LICENSE").as_slice(),
-            ),
-            _ => return Err("候选语言未知".into()),
+    Ok(())
+}
+
+fn verify_bundled_asset(asset: &GrammarAsset) -> Result<&'static [u8], String> {
+    let (wasm, license) = match asset.language.as_str() {
+        "c" => (
+            include_bytes!("../../../grammars/c/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/c/LICENSE").as_slice(),
+        ),
+        "cpp" => (
+            include_bytes!("../../../grammars/cpp/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/cpp/LICENSE").as_slice(),
+        ),
+        "csharp" => (
+            include_bytes!("../../../grammars/csharp/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/csharp/LICENSE").as_slice(),
+        ),
+        "go" => (
+            include_bytes!("../../../grammars/go/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/go/LICENSE").as_slice(),
+        ),
+        "javascript" => (
+            include_bytes!("../../../grammars/javascript/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/javascript/LICENSE").as_slice(),
+        ),
+        "lua" => (
+            include_bytes!("../../../grammars/lua/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/lua/LICENSE").as_slice(),
+        ),
+        "luau" => (
+            include_bytes!("../../../grammars/luau/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/luau/LICENSE").as_slice(),
+        ),
+        "java" => (
+            include_bytes!("../../../grammars/java/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/java/LICENSE").as_slice(),
+        ),
+        "typescript" => (
+            include_bytes!("../../../grammars/typescript/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/typescript/LICENSE").as_slice(),
+        ),
+        "python" => (
+            include_bytes!("../../../grammars/python/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/python/LICENSE").as_slice(),
+        ),
+        "tsx" => (
+            include_bytes!("../../../grammars/tsx/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/typescript/LICENSE").as_slice(),
+        ),
+        "zig" => (
+            include_bytes!("../../../grammars/zig/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/zig/LICENSE").as_slice(),
+        ),
+        "objc" => (
+            include_bytes!("../../../grammars/objc/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/objc/LICENSE").as_slice(),
+        ),
+        "solidity" => (
+            include_bytes!("../../../grammars/solidity/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/solidity/LICENSE").as_slice(),
+        ),
+        "rust" => (
+            include_bytes!("../../../grammars/rust/parser.wasm").as_slice(),
+            include_bytes!("../../../grammars/rust/LICENSE").as_slice(),
+        ),
+        _ => return Err("候选语言未知".into()),
+    };
+    verify_grammar_asset(asset, wasm, license)?;
+    if asset.language == "zig"
+        && crate::adapt_zig_wasm(include_bytes!("../../../grammars/zig/source.wasm"))? != wasm
+    {
+        return Err("Zig grammar 来源适配关系不符".into());
+    }
+    if matches!(asset.language.as_str(), "objc" | "solidity") {
+        let source = if asset.language == "objc" {
+            include_bytes!("../../../grammars/objc/source.wasm").as_slice()
+        } else {
+            include_bytes!("../../../grammars/solidity/source.wasm").as_slice()
         };
-        verify_grammar_asset(asset, wasm, license)?;
-        if asset.language == "zig"
-            && crate::adapt_zig_wasm(include_bytes!("../../../grammars/zig/source.wasm"))? != wasm
+        if asset.source_sha256.as_deref() != Some(digest(source).as_str())
+            || crate::adapt_legacy_dylink(&asset.language, source)? != wasm
         {
-            return Err("Zig grammar 来源适配关系不符".into());
-        }
-        if matches!(asset.language.as_str(), "objc" | "solidity") {
-            let source = if asset.language == "objc" {
-                include_bytes!("../../../grammars/objc/source.wasm").as_slice()
-            } else {
-                include_bytes!("../../../grammars/solidity/source.wasm").as_slice()
-            };
-            if asset.source_sha256.as_deref() != Some(digest(source).as_str())
-                || crate::adapt_legacy_dylink(&asset.language, source)? != wasm
-            {
-                return Err("依赖 grammar 来源适配关系不符".into());
-            }
+            return Err("依赖 grammar 来源适配关系不符".into());
         }
     }
-    Ok(manifest)
+    Ok(wasm)
 }
 
 /// 解析来源固定的候选清单；拒绝重复字段、漂移来源和伪造的已验收状态。

crates/codeguard-adapters/src/lib.rs

@@ -67,8 +67,8 @@ mod grammar_asset_manifest;
 mod legacy_dylink_compat;
 mod zig_wasm_compat;
 pub use grammar_asset_manifest::{
-    GrammarAsset, GrammarAssetManifest, bundled_grammar_candidates, parse_grammar_asset_manifest,
-    verify_grammar_asset,
+    GrammarAsset, GrammarAssetManifest, bundled_grammar_candidate, bundled_grammar_candidates,
+    parse_grammar_asset_manifest, verify_grammar_asset,
 };
 pub use legacy_dylink_compat::adapt_legacy_dylink;
 pub use zig_wasm_compat::adapt_zig_wasm;

crates/codeguard-adapters/tests/grammar_asset_manifest.rs

@@ -1,7 +1,18 @@
 use codeguard_adapters::{
-    adapt_legacy_dylink, adapt_zig_wasm, bundled_grammar_candidates, parse_grammar_asset_manifest,
-    verify_grammar_asset,
+    adapt_legacy_dylink, adapt_zig_wasm, bundled_grammar_candidate, bundled_grammar_candidates,
+    parse_grammar_asset_manifest, verify_grammar_asset,
 };
+
+#[test]
+fn selected_candidate_matches_full_manifest_without_promoting_unknown_language() {
+    let full = bundled_grammar_candidates().expect("all bundled assets");
+    for asset in &full.assets {
+        let (selected, wasm) = bundled_grammar_candidate(&asset.language).expect("selected asset");
+        assert_eq!(&selected, asset);
+        assert_eq!(wasm.len(), asset.bytes);
+    }
+    assert!(bundled_grammar_candidate("unknown").is_err());
+}
 use serde_json::Value;
 
 #[test]

crates/codeguard-cli/src/syntax_worker_command.rs

@@ -2,7 +2,7 @@
 
 use crate::syntax_worker_envelope::SyntaxWorkerEnvelope;
 use crate::syntax_worker_recovery::SyntaxWorkerRecovery;
-use codeguard_adapters::bundled_grammar_candidates;
+use codeguard_adapters::bundled_grammar_candidate;
 use codeguard_runtime::{WasmGrammar, scan_wasm_recoveries};
 use sha2::{Digest, Sha256};
 use std::io::Read;
@@ -54,33 +54,10 @@ pub fn run(args: &[String]) -> ExitCode {
 }
 
 fn observe(language: &str, source: &[u8]) -> Result<SyntaxWorkerEnvelope, String> {
-    let manifest = bundled_grammar_candidates()?;
-    let asset = manifest
-        .assets
-        .iter()
-        .find(|asset| asset.language == language)
-        .ok_or("不支持的 grammar 语种")?;
+    let (asset, wasm) = bundled_grammar_candidate(language)?;
     if asset.codeguard_runtime_validation != "rust_loader_smoke_passed" {
         return Err("grammar 与当前 Rust WASM 运行时不兼容".into());
     }
-    let wasm = match language {
-        "c" => include_bytes!("../../../grammars/c/parser.wasm").as_slice(),
-        "cpp" => include_bytes!("../../../grammars/cpp/parser.wasm").as_slice(),
-        "csharp" => include_bytes!("../../../grammars/csharp/parser.wasm").as_slice(),
-        "go" => include_bytes!("../../../grammars/go/parser.wasm").as_slice(),
-        "javascript" => include_bytes!("../../../grammars/javascript/parser.wasm").as_slice(),
-        "lua" => include_bytes!("../../../grammars/lua/parser.wasm").as_slice(),
-        "luau" => include_bytes!("../../../grammars/luau/parser.wasm").as_slice(),
-        "java" => include_bytes!("../../../grammars/java/parser.wasm").as_slice(),
-        "python" => include_bytes!("../../../grammars/python/parser.wasm").as_slice(),
-        "typescript" => include_bytes!("../../../grammars/typescript/parser.wasm").as_slice(),
-        "tsx" => include_bytes!("../../../grammars/tsx/parser.wasm").as_slice(),
-        "zig" => include_bytes!("../../../grammars/zig/parser.wasm").as_slice(),
-        "objc" => include_bytes!("../../../grammars/objc/parser.wasm").as_slice(),
-        "solidity" => include_bytes!("../../../grammars/solidity/parser.wasm").as_slice(),
-        "rust" => include_bytes!("../../../grammars/rust/parser.wasm").as_slice(),
-        _ => return Err("不支持的 grammar 语种".into()),
-    };
     let mut grammar = WasmGrammar::load(
         asset.loader_symbol.as_deref().unwrap_or(language),
         wasm,

openspec/changes/introduce-rust-codeguard-cli/tasks.md

@@ -1047,3 +1047,5 @@ TDD缺自动同步先RED。相关普通：build9、work sync15、next6，共30
 - [ ] 14.17 逐语言误报与性能评测。责任：评测/adapters。规格：SP06,SP12。依赖：14.3,14.4,14.11。验收：独立合法/非法 oracle、版本/方言语料、原生对照；分别报告误报、漏报、未知覆盖及冷暖启动/包体/内存；先实测再定预算。
 - [ ] 14.18 打包兼容与分阶段发布验收。责任：发布/CLI。规格：SP04,SP11,SP12。依赖：14.12-14.17。验收：每个平台验证内置资产/许可证/散列和离线 npx；保持退出码、统一入口及旧报告读取；同步中英文示例；未验收语言/宿主仍标 gap。
 - [ ] 14.19 CodeGraph 32 种 grammar 全量接入与 Zig 误报回归。责任：adapters/runtime/CLI/发布。规格：SP01,SP04,SP06,SP11,SP12。依赖：14.1-14.18。验收：逐语言固定可再分发来源、许可证、原始和发行字节、真实 ABI、Rust worker 离线加载、合法/非法版本语料、原生工具对照、原生优先统一命令、任务与对话反馈、资源预算及发行包实装；覆盖 arkts/c/cfml/cfquery/cfscript/cobol/cpp/csharp/dart/erlang/go/java/javascript/kotlin/lua/luau/nix/objc/pascal/php/python/r/ruby/rust/scala/solidity/swift/terraform/tsx/typescript/vbnet/zig。当前十五份候选（C/C++/C#/Go/Java/JavaScript/Lua/Luau/Objective-C/Python/Rust/Solidity/TypeScript/TSX/Zig），已验收发行 0；COBOL 超出 8 MiB 上限，其他 17 种仍无资产清单和执行路由。C++/C#/Lua/Luau 的固定来源、许可证、真实 ABI、离线加载及基础正反例见[局部验收](../../../tests/acceptance/cpp-csharp-lua-luau-grammar-candidates.md)。C/Go/JavaScript/Rust 的固定来源、许可证、离线加载及基础正反例见[局部验收](../../../tests/acceptance/mainstream-grammar-candidates.md)。不得以 `grammar status` 库存、WASM 可加载或未验收候选替代原生 lint 或语言验收。
+
+14.19 性能子项进展：候选 worker 已改为仅校验所选固定 grammar 及相关许可证，避免每次单文件初检遍历全部 WASM；全量库存命令保持完整校验。选择接口先 RED 后通过 15 语种身份一致性测试，单次调试构建启动观察见[局部验收](../../../tests/acceptance/syntax-worker-selected-asset.md)。这不是正式性能评测，也不改变 15/32 候选和 0 份发行验收的状态。

tests/acceptance/syntax-worker-selected-asset.md

@@ -0,0 +1,11 @@
+# 语法 worker 按需资产校验局部验收
+
+日期：2026-09-29。范围：已固定的 15 份候选 grammar，仍非语言能力或发行验收。
+
+原实现的每次 `__syntax-worker <language>` 都先调用全量 `bundled_grammar_candidates()`，对所有已打包 WASM 和许可证重新计算摘要，再选择一份资产。随着语言数增加，单文件反馈会承担无关语言的校验开销。
+
+现在 worker 通过 `bundled_grammar_candidate(language)` 仅选择固定清单中的目标语言，校验 CodeGraph 许可证、该语言的 WASM 与许可证，以及该语言适用的来源适配关系；Objective-C/Solidity 额外校验依赖包许可证。未知语言拒绝。全量 `grammar status` 仍校验全部资产。测试先因缺少选择接口而编译失败，随后证明 15 份选择结果与全量清单身份及字节长度一致，未知语言拒绝。
+
+调试构建、同一主机上的一次前后观察：Lua worker 从约 2.14 秒到 1.63 秒，C# 从约 3.25 秒到 1.62 秒。此观察包含进程启动和 Rust WASM 解析开销，不能分离摘要计算贡献，也不能当作正式冷/暖启动性能基准。
+
+未完成：32 份资产覆盖、逐语言语料与原生对照、公开原生优先路由、宿主实装及发行验收。候选仍不能代替原生 lint；结果仍为 `incomplete`。