progmune-runtime 3.7.0 → 3.7.1

package/CHANGELOG.md

@@ -1,5 +1,19 @@
 # Changelog
 
+## [3.7.1] — 2026-08-23
+
+### 修复：词段匹配门控（仅项目函数适用）
+
+- `ssg-bridge` 的词段匹配（Strategy 2）增加 `projectFunctions` 门控：只对项目函数做词段匹配——它是为改名协议原语设计的（协议原语必然是项目内函数，如 S5 的 `create_active_session`），外部库调用（如 Node 的 `readFileSync`）经词段撞上 `read_file` 是纯噪声
+- 外部 API 的语义桥接不受影响：alias 配置（Strategy 0b）与 domain 关键词（Strategy 3）照常工作；未提供集合时保持旧行为（向后兼容）
+- 共享集合构造 `collectProjectFunctionNames`（`src/call-sequence.ts`，全名/裸名/小写变体三形态收录），生产引擎与协议盲测扫描器同款传入
+
+### 修复：合并形态 ir.json 恢复 IR-first（3.5.0 起静默回退的回归）
+
+- `extractCallSequencesFromIR` 与项目 IR 注解合并块兼容 `{ typeMap, functions }` 合并对象（execute/MCP 写盘形态）——此前 `Array.isArray` 守卫使所有 TS 项目自 3.5.0 起静默走正则回退，P4.5/P4.6 的 IR-first 语义在合并形态下未生效
+- 配合词段门控后实测：自身 1966 函数 451 入口序列，SSG 违规 346→**2**（均真实命中，`writeTrajectoryFile`→`write_file`），Trust 总分 60→83（APPROVED）
+- 协议盲测 v1.2 复测零漂移：66 可测金标 64 检出（Recall 97% / Precision 100% / 0 FP），S5 改名检测不受门控影响
+
 ## [3.7.0] — 2026-08-23
 
 ### 新增：P4.6 跨函数传播（入口展开 + 片段抑制）

package/dist/call-sequence.js

@@ -13,12 +13,39 @@
  * 不做数据流/指针/分支分析；跨文件依赖 IR 的 calls[] 图。
  */
 Object.defineProperty(exports, "__esModule", { value: true });
+exports.isProjectFn = isProjectFn;
+exports.collectProjectFunctionNames = collectProjectFunctionNames;
 exports.buildCallSequences = buildCallSequences;
 const MAX_DEPTH = 4;
 /** 项目函数判定：有真实文件且非外部导入条目（external 条目无函数体可内联）。 */
 function isProjectFn(f) {
     return !f.external && !!f.file && f.file !== "(external)";
 }
+/**
+ * 构建项目函数名集合——词段匹配门控用（ssg-bridge 的 projectFunctions 参数）。
+ * 每个项目函数收录三种形态：全名（FlowService.svc_x）、裸名（svc_x）与
+ * 小写变体（调用名大小写差异，如 createActiveSession vs createactivesession）。
+ */
+function collectProjectFunctionNames(ir) {
+    const names = new Set();
+    for (const f of ir) {
+        if (!isProjectFn(f))
+            continue;
+        const name = String(f.name || "");
+        if (!name)
+            continue;
+        const lower = name.toLowerCase();
+        names.add(name);
+        names.add(lower);
+        const dotIdx = name.lastIndexOf(".");
+        if (dotIdx >= 0) {
+            const bare = name.slice(dotIdx + 1);
+            names.add(bare);
+            names.add(bare.toLowerCase());
+        }
+    }
+    return names;
+}
 /**
  * 从 IR 构建验证序列：入口函数展开 + 非入口抑制。
  * @param ir - FunctionInfo 列表（TS 或 Python 提取器输出）

package/dist/trust/engine.js

@@ -727,6 +727,19 @@ async function collectProtocolViolations(ctx, callGraph) {
         const callSequences = extractCallSequencesFromProject(ctx.projectPath, ctx.language, protocolRulesData ? new Set(protocolRulesData.rules.keys()) : undefined);
         const flaggedCount = { value: 0 };
         const cleanCount = { value: 0 };
+        // ── P4.6.1: 词段匹配门控的项目函数集合（best-effort，与注解合并共用 ir.json） ──
+        // 词段匹配只对项目函数适用（改名协议原语）；外部库调用走 alias/关键词桥接。
+        let projectFunctions;
+        try {
+            const fs = require("fs");
+            const irPath = path.join(ctx.projectPath, "ir.json");
+            if (fs.existsSync(irPath)) {
+                const ir = JSON.parse(fs.readFileSync(irPath, "utf-8"));
+                const functions = Array.isArray(ir) ? ir : (ir.functions || []);
+                projectFunctions = (0, call_sequence_1.collectProjectFunctionNames)(functions);
+            }
+        }
+        catch { /* best-effort */ }
         // ── P4.5: 合并项目 IR 注解协议（IR 优先，缺 namespace 继承内置 JSON） ──
         // 内置 protocols.json 的规则是通用弱约束（如 generate_jwt pre=[]），
         // 项目文件里的 @protocol 注解才是项目真实协议（如 pre=[PASSWORD_VERIFIED]）。
@@ -737,17 +750,18 @@ async function collectProtocolViolations(ctx, callGraph) {
                 const irPath = path.join(ctx.projectPath, "ir.json");
                 if (fs.existsSync(irPath)) {
                     const ir = JSON.parse(fs.readFileSync(irPath, "utf-8"));
-                    if (Array.isArray(ir)) {
-                        for (const f of ir) {
-                            if (!f.protocol)
-                                continue;
-                            const protocol = { ...f.protocol };
-                            const existing = protocolRulesData.rules.get(String(f.name));
-                            if (existing?.namespace && !protocol.namespace) {
-                                protocol.namespace = existing.namespace;
-                            }
-                            protocolRulesData.rules.set(String(f.name), protocol);
+                    // ir.json 两种形态：extractIR/extractIRPython 的裸数组、extractProjectIR
+                    // 的 { typeMap, functions } 合并对象（execute/MCP 写盘）——统一取函数列表。
+                    const functions = Array.isArray(ir) ? ir : (ir.functions || []);
+                    for (const f of functions) {
+                        if (!f.protocol)
+                            continue;
+                        const protocol = { ...f.protocol };
+                        const existing = protocolRulesData.rules.get(String(f.name));
+                        if (existing?.namespace && !protocol.namespace) {
+                            protocol.namespace = existing.namespace;
                         }
+                        protocolRulesData.rules.set(String(f.name), protocol);
                     }
                 }
             }
@@ -844,7 +858,7 @@ async function collectProtocolViolations(ctx, callGraph) {
                 // to bridge real API names to abstract protocol function names.
                 if (protocolRulesData) {
                     try {
-                        const ssgResult = (0, ssg_bridge_1.validateSequenceWithSSG)(semantic.steps, protocolRulesData.rules, protocolRulesData.namespaceInitialStates, seq.file, protocolRulesData.aliasIndex, protocolRulesData.wildcardAliases);
+                        const ssgResult = (0, ssg_bridge_1.validateSequenceWithSSG)(semantic.steps, protocolRulesData.rules, protocolRulesData.namespaceInitialStates, seq.file, protocolRulesData.aliasIndex, protocolRulesData.wildcardAliases, projectFunctions);
                         ssgResults.push(ssgResult);
                         ssgTotalCalls += ssgResult.stats.totalCalls;
                         ssgMatchedCalls += ssgResult.stats.matchedCalls;
@@ -927,9 +941,12 @@ function extractCallSequencesFromIR(projectPath, keepNames) {
         if (!fs.existsSync(irPath))
             return [];
         const ir = JSON.parse(fs.readFileSync(irPath, "utf-8"));
-        if (!Array.isArray(ir))
+        // ir.json 两种形态兼容：裸数组（extractIR / extractIRPython）与
+        // { typeMap, functions } 合并对象（extractProjectIR，execute/MCP 写盘）。
+        const functions = Array.isArray(ir) ? ir : (ir.functions || []);
+        if (!Array.isArray(functions) || functions.length === 0)
             return [];
-        return (0, call_sequence_1.buildCallSequences)(ir, keepNames);
+        return (0, call_sequence_1.buildCallSequences)(functions, keepNames);
     }
     catch {
         return [];

package/dist/trust/ssg-bridge.js

@@ -227,7 +227,7 @@ function normalizeName(name) {
  *   2. Substring match (call contains rule name or vice versa)
  *   3. Domain-guided keyword match
  */
-function inferRuleName(apiName, domain, description, ruleNames, namespace, aliases, wildcardAliases) {
+function inferRuleName(apiName, domain, description, ruleNames, namespace, aliases, wildcardAliases, projectFunctions) {
     const normalized = normalizeName(apiName);
     const lowerApi = apiName.toLowerCase();
     const lowerDesc = description.toLowerCase();
@@ -262,7 +262,20 @@ function inferRuleName(apiName, domain, description, ruleNames, namespace, alias
     // word segment in the normalized call name. This prevents "status" from
     // matching "poll_status" via raw substring, since "status" must appear
     // as a complete _-delimited segment.
+    //
+    // 门控（P4.6.1）：词段匹配只对「项目函数」适用——它是为改名协议原语设计的
+    // （协议原语必然是项目内函数，如 S5 的 create_active_session）。外部库调用
+    // （如 Node 的 readFileSync）经词段撞上 read_file 是纯噪声：外部 API 的语义
+    // 桥接走 alias 配置（Strategy 0b）或 domain 关键词（Strategy 3），不走词段。
+    // 未提供 projectFunctions 时保持旧行为（向后兼容测试与无 IR 的调用方）。
+    const isProjectFn = !projectFunctions
+        || projectFunctions.has(apiName)
+        || projectFunctions.has(lowerApi)
+        || projectFunctions.has(normalized)
+        || (dotIdx >= 0 && projectFunctions.has(lowerApi.slice(dotIdx + 1)));
     for (const ruleName of ruleNames) {
+        if (projectFunctions && !isProjectFn)
+            continue;
         const ruleWords = ruleName.split("_");
         const callWords = normalized.split("_");
         // Rule must have at least 2 words for this strategy (single-word rules
@@ -303,7 +316,9 @@ function inferRuleName(apiName, domain, description, ruleNames, namespace, alias
  * @param file — source file path (for violation reporting)
  * @returns SSG validation result with violations and trace
  */
-function validateSequenceWithSSG(steps, rules, namespaceInitialStates, file, aliasIndex, wildcardAliases) {
+function validateSequenceWithSSG(steps, rules, namespaceInitialStates, file, aliasIndex, wildcardAliases, 
+/** 项目函数名集合（含裸名/全名/小写变体由调用方构造）——提供后词段匹配只对项目函数适用 */
+projectFunctions) {
     const allRuleNames = Array.from(rules.keys());
     const aliases = aliasIndex || new Map();
     const wildcards = wildcardAliases || new Map();
@@ -361,7 +376,7 @@ function validateSequenceWithSSG(steps, rules, namespaceInitialStates, file, ali
         // Even if the semantic domain maps to "stateless", an alias may still
         // map this call to a specific protocol rule with its own namespace.
         // Priority: wildcard alias → exact alias → name match → word match → keyword
-        const ruleName = inferRuleName(step.api, step.domain, step.description, allRuleNames, namespace, aliases, wildcards);
+        const ruleName = inferRuleName(step.api, step.domain, step.description, allRuleNames, namespace, aliases, wildcards, projectFunctions);
         // Use matched rule's namespace, falling back to domain-inferred namespace
         const effectiveNamespace = ruleName
             ? (rules.get(ruleName)?.namespace || namespace)

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "progmune-runtime",
-  "version": "3.7.0",
+  "version": "3.7.1",
   "description": "Progmune — AI Trust Decision Engine. Verify AI-generated code before it reaches production. Outputs APPROVED / NEEDS_REVIEW / BLOCKED with evidence.",
   "files": [
     "dist/",