@proxysoul/soulforge 2.11.0 → 2.12.1

package/dist/workers/intelligence.worker.js

@@ -23304,7 +23304,7 @@ var init_tree_sitter2 = __esm(() => {
         language
       };
     }
-    static MIN_HASH_LINES = 5;
+    static MIN_HASH_LINES = 12;
     static HASHABLE_KEYWORDS = ["function", "method", "class", "impl", "struct", "trait", "module", "constructor"];
     static isHashableType(nodeType) {
       return TreeSitterBackend.HASHABLE_KEYWORDS.some((kw) => nodeType.includes(kw));
@@ -24194,6 +24194,14 @@ import { stat as statAsync } from "fs/promises";
 import { dirname as dirname2, extname as extname2, join as join6, relative, resolve as resolve4 } from "path";
 
 class RepoMap {
+  static testFileMatch(alias = "f") {
+    const a2 = alias;
+    return `(${a2}.path LIKE 'tests/%' OR ${a2}.path LIKE 'test/%' OR ${a2}.path LIKE 'spec/%' OR ${a2}.path LIKE 'src/test/%' OR ${a2}.path LIKE '%.test.%' OR ${a2}.path LIKE '%.spec.%' OR ${a2}.path LIKE '%_test.%' OR ${a2}.path LIKE '%_spec.%' OR ${a2}.path LIKE '%/test_%' OR ${a2}.path LIKE '%/__tests__/%')`;
+  }
+  static notTestFile(alias = "f") {
+    const a2 = alias;
+    return `${a2}.path NOT LIKE 'tests/%' AND ${a2}.path NOT LIKE 'test/%' AND ${a2}.path NOT LIKE 'spec/%' AND ${a2}.path NOT LIKE 'src/test/%' AND ${a2}.path NOT LIKE '%.test.%' AND ${a2}.path NOT LIKE '%.spec.%' AND ${a2}.path NOT LIKE '%_test.%' AND ${a2}.path NOT LIKE '%_spec.%' AND ${a2}.path NOT LIKE '%/test_%' AND ${a2}.path NOT LIKE '%/__tests__/%'`;
+  }
   db;
   cwd;
   scanPromise = null;
@@ -25370,11 +25378,8 @@ class RepoMap {
   }
   linkTestFiles() {
     const testFiles = this.db.query(`SELECT f.id, f.path FROM files f
-         WHERE (f.path LIKE '%.test.%' OR f.path LIKE '%.spec.%'
-           OR f.path LIKE '%_test.%' OR f.path LIKE '%_spec.%'
-           OR f.path LIKE 'tests/%' OR f.path LIKE 'test/%'
-           OR f.path LIKE '%/__tests__/%')
-         AND NOT EXISTS (SELECT 1 FROM edges WHERE source_file_id = f.id)`).all();
+         WHERE ${RepoMap.testFileMatch()}
+           AND NOT EXISTS (SELECT 1 FROM edges WHERE source_file_id = f.id)`).all();
     if (testFiles.length === 0)
       return;
     const insert = this.db.prepare("INSERT OR IGNORE INTO edges (source_file_id, target_file_id, weight, confidence) VALUES (?, ?, 1.0, 1)");
@@ -26835,7 +26840,7 @@ class RepoMap {
                WHERE s2.name = s.name AND s2.is_exported = 1
              ) = 1)
            )
-           AND (rf.path LIKE 'tests/%' OR rf.path LIKE '%.test.%' OR rf.path LIKE '%.spec.%' OR rf.path LIKE '%/__tests__/%')
+           AND ${RepoMap.testFileMatch("rf")}
          )
          AND NOT EXISTS (
            SELECT 1 FROM refs r
@@ -26848,7 +26853,7 @@ class RepoMap {
                WHERE s2.name = s.name AND s2.is_exported = 1
              ) = 1)
            )
-           AND rf.path NOT LIKE 'tests/%' AND rf.path NOT LIKE '%.test.%' AND rf.path NOT LIKE '%.spec.%' AND rf.path NOT LIKE '%/__tests__/%'
+           AND ${RepoMap.notTestFile("rf")}
          )
          ORDER BY f.path`).all().map((r4) => ({
       name: r4.name,
@@ -26984,13 +26989,14 @@ class RepoMap {
     });
     tx();
   }
-  getNearDuplicates(threshold = 0.7, limit = 20) {
+  getNearDuplicates(threshold = 0.8, limit = 20) {
     if (!this.ready)
       return [];
     const rows = this.db.query(`SELECT ts.name, COALESCE(s.kind, '') as kind, s.signature as sig, f.path, ts.line, ts.end_line, ts.minhash
          FROM token_signatures ts
          JOIN files f ON f.id = ts.file_id
          LEFT JOIN symbols s ON s.file_id = ts.file_id AND s.name = ts.name AND s.line = ts.line
+         WHERE ${RepoMap.notTestFile()}
          ORDER BY f.pagerank DESC
          LIMIT 500`).all();
     const pairs = [];
@@ -27070,8 +27076,9 @@ class RepoMap {
            FROM token_fragments tf
            JOIN files f ON f.id = tf.file_id
            WHERE tf.hash = ?
+               AND ${RepoMap.notTestFile()}
            ORDER BY f.path, tf.line
-           LIMIT 20`).all(cluster.hash);
+             LIMIT 20`).all(cluster.hash);
       const uniqueFiles = new Set(locs.map((l3) => `${l3.path}:${l3.name}`));
       if (uniqueFiles.size < 2)
         continue;
@@ -27089,13 +27096,15 @@ class RepoMap {
   getDuplicateStructures(limit = 20) {
     if (!this.ready)
       return [];
-    const clusters = this.db.query(`SELECT shape_hash, kind, node_count, COUNT(*) as cnt
-         FROM shape_hashes
-         WHERE node_count >= ?
-         GROUP BY shape_hash
+    const clusters = this.db.query(`SELECT sh.shape_hash, sh.kind, sh.node_count, COUNT(*) as cnt
+         FROM shape_hashes sh
+         JOIN files f ON f.id = sh.file_id
+         WHERE sh.node_count >= ?
+             AND ${RepoMap.notTestFile()}
+         GROUP BY sh.shape_hash
          HAVING cnt > 1
-         ORDER BY node_count * cnt DESC
-         LIMIT ?`).all(10, limit);
+           ORDER BY sh.node_count * cnt DESC
+           LIMIT ?`).all(20, limit);
     const results = [];
     for (const cluster of clusters) {
       const members = this.db.query(`SELECT sh.name, f.path, sh.line, sh.end_line
@@ -27123,22 +27132,84 @@ class RepoMap {
     const fileRow = this.db.query("SELECT id FROM files WHERE path = ?").get(relPath);
     if (!fileRow)
       return [];
-    const hashes = this.db.query("SELECT name, line, shape_hash FROM shape_hashes WHERE file_id = ?").all(fileRow.id);
+    const rawSigs = this.db.query("SELECT name, line, end_line, minhash FROM token_signatures WHERE file_id = ?").all(fileRow.id);
+    const seen = new Set;
+    const sigs = rawSigs.filter((s2) => {
+      const key2 = `${s2.name}:${String(s2.line)}`;
+      if (seen.has(key2))
+        return false;
+      seen.add(key2);
+      return true;
+    });
+    if (sigs.length === 0)
+      return [];
+    const candidates = this.db.query(`SELECT ts.name, f.path, ts.line, ts.end_line, ts.minhash, s.signature as sig, ts.file_id
+           FROM token_signatures ts
+           JOIN files f ON f.id = ts.file_id
+         LEFT JOIN symbols s ON s.file_id = ts.file_id AND s.name = ts.name AND s.line = ts.line
+         WHERE ts.file_id != ?
+           AND ${RepoMap.notTestFile()}`).all(fileRow.id);
+    const toSig = (buf) => {
+      if (buf.byteOffset % 4 === 0) {
+        return new Uint32Array(buf.buffer, buf.byteOffset, 128);
+      }
+      const copy = new Uint32Array(128);
+      new Uint8Array(copy.buffer).set(new Uint8Array(buf.buffer, buf.byteOffset, 512));
+      return copy;
+    };
     const results = [];
-    for (const h2 of hashes) {
-      const clones = this.db.query(`SELECT sh.name, f.path, sh.line
-           FROM shape_hashes sh
-           JOIN files f ON f.id = sh.file_id
-           WHERE sh.shape_hash = ? AND sh.file_id != ?
-           ORDER BY f.pagerank DESC`).all(h2.shape_hash, fileRow.id);
+    for (const s2 of sigs) {
+      const sigA = toSig(s2.minhash);
+      const clones = [];
+      let bestSim = 0;
+      for (const c of candidates) {
+        if (c.path === relPath) {
+          if (s2.line <= c.line && s2.end_line >= c.end_line)
+            continue;
+          if (c.line <= s2.line && c.end_line >= s2.end_line)
+            continue;
+        }
+        const sigB = toSig(c.minhash);
+        let matches2 = 0;
+        for (let i4 = 0;i4 < 128; i4++) {
+          if (sigA[i4] === sigB[i4])
+            matches2++;
+        }
+        const sim = matches2 / 128;
+        if (sim < 0.8 || sim >= 1)
+          continue;
+        if (sim < 0.95 && c.sig) {
+          const aSig = this.db.query("SELECT signature FROM symbols WHERE file_id = ? AND name = ? AND line = ? LIMIT 1").get(fileRow.id, s2.name, s2.line);
+          if (aSig?.signature) {
+            const tokA = new Set(aSig.signature.toLowerCase().split(/[\s,(){}:;|&=<>]+/).filter(Boolean));
+            const tokB = new Set(c.sig.toLowerCase().split(/[\s,(){}:;|&=<>]+/).filter(Boolean));
+            let shared = 0;
+            for (const t of tokA)
+              if (tokB.has(t))
+                shared++;
+            const sigSim = shared / Math.max(tokA.size, tokB.size);
+            if (sigSim < 0.3)
+              continue;
+          }
+        }
+        clones.push({
+          name: c.name,
+          path: c.path,
+          line: c.line
+        });
+        if (sim > bestSim)
+          bestSim = sim;
+      }
       if (clones.length > 0) {
         results.push({
-          name: h2.name,
-          line: h2.line,
-          clones
+          name: s2.name,
+          line: s2.line,
+          similarity: bestSim,
+          clones: clones.slice(0, 5)
         });
       }
     }
+    results.sort((a2, b3) => b3.similarity - a2.similarity);
     return results;
   }
   getTopFiles(limit = 20) {
@@ -29480,6 +29551,39 @@ var init_process_tracker = __esm(() => {
   trackedBun = new Set;
 });
 
+// src/core/intelligence/backends/lsp/pid-tracker.ts
+import { existsSync as existsSync8, mkdirSync as mkdirSync3, readFileSync as readFileSync5, renameSync, writeFileSync as writeFileSync4 } from "fs";
+import { homedir as homedir6 } from "os";
+import { join as join10 } from "path";
+function ensureDir() {
+  const dir = join10(homedir6(), ".soulforge");
+  if (!existsSync8(dir))
+    mkdirSync3(dir, {
+      recursive: true
+    });
+}
+function flush() {
+  try {
+    ensureDir();
+    const tmp = `${PID_FILE}.${String(process.pid)}.tmp`;
+    writeFileSync4(tmp, JSON.stringify([...activePids]), "utf-8");
+    renameSync(tmp, PID_FILE);
+  } catch {}
+}
+function trackLspPid(pid) {
+  activePids.add(pid);
+  flush();
+}
+function untrackLspPid(pid) {
+  activePids.delete(pid);
+  flush();
+}
+var PID_FILE, activePids;
+var init_pid_tracker = __esm(() => {
+  PID_FILE = join10(homedir6(), ".soulforge", "lsp-pids.json");
+  activePids = new Set;
+});
+
 // src/core/intelligence/backends/lsp/standalone-client.ts
 import { spawn } from "child_process";
 import { readFile as readFile3 } from "fs/promises";
@@ -29507,11 +29611,16 @@ class StandaloneLspClient {
   async start() {
     this.process = spawn(this.config.command, this.config.args, {
       stdio: ["pipe", "pipe", "pipe"],
-      cwd: this.cwd
+      cwd: this.cwd,
+      detached: true
     });
     trackProcess(this.process);
+    if (this.process.pid)
+      trackLspPid(this.process.pid);
     this.process.stdout?.on("data", (chunk) => this.onData(chunk));
     this.process.on("exit", (code, signal) => {
+      if (this.process?.pid)
+        untrackLspPid(this.process.pid);
       this.process = null;
       if (code != null && code !== 0) {
         logBackgroundError(`LSP:${this.config.command}`, `exited with code ${code}`);
@@ -30012,9 +30121,25 @@ class StandaloneLspClient {
       this.notify("exit", null);
     } catch {}
     const proc = this.process;
-    if (proc) {
-      setTimeout(() => proc.kill("SIGKILL"), 2000);
-      proc.kill("SIGTERM");
+    const pid = proc?.pid;
+    if (pid) {
+      try {
+        process.kill(-pid, "SIGTERM");
+      } catch {
+        try {
+          proc.kill("SIGTERM");
+        } catch {}
+      }
+      setTimeout(() => {
+        try {
+          process.kill(-pid, "SIGKILL");
+        } catch {
+          try {
+            proc.kill("SIGKILL");
+          } catch {}
+        }
+      }, 2000);
+      untrackLspPid(pid);
     }
     this.process = null;
     this.initialized = false;
@@ -30050,6 +30175,7 @@ var LANGUAGE_ID_MAP;
 var init_standalone_client = __esm(() => {
   init_errors();
   init_process_tracker();
+  init_pid_tracker();
   init_protocol();
   LANGUAGE_ID_MAP = {
     typescript: "typescript",
@@ -30087,9 +30213,9 @@ var exports_lsp = {};
 __export(exports_lsp, {
   LspBackend: () => LspBackend
 });
-import { existsSync as existsSync8, readdirSync as readdirSync3 } from "fs";
+import { existsSync as existsSync9, readdirSync as readdirSync3 } from "fs";
 import { readdir as readdir2, readFile as readFile4 } from "fs/promises";
-import { dirname as dirname3, join as join10, resolve as resolve5 } from "path";
+import { dirname as dirname3, join as join11, resolve as resolve5 } from "path";
 function getNvimBridge() {
   if (_nvimBridge === null) {
     try {
@@ -30116,18 +30242,18 @@ class LspBackend {
   async findAnchorFile() {
     const candidates = ["src/index.ts", "src/main.ts", "src/app.ts", "index.ts", "main.ts", "src/index.js", "src/main.js", "index.js", "main.py", "src/main.py", "main.go", "src/main.go", "src/lib.rs", "src/main.rs"];
     for (const candidate of candidates) {
-      const full = join10(this.cwd, candidate);
-      if (existsSync8(full))
+      const full = join11(this.cwd, candidate);
+      if (existsSync9(full))
         return full;
     }
-    for (const dir of [join10(this.cwd, "src"), this.cwd]) {
-      if (!existsSync8(dir))
+    for (const dir of [join11(this.cwd, "src"), this.cwd]) {
+      if (!existsSync9(dir))
         continue;
       try {
         const files = await readdir2(dir);
         const source = files.find((f3) => f3.endsWith(".ts") || f3.endsWith(".js") || f3.endsWith(".py") || f3.endsWith(".go") || f3.endsWith(".rs"));
         if (source)
-          return join10(dir, source);
+          return join11(dir, source);
       } catch {}
     }
     return null;
@@ -31278,7 +31404,7 @@ function findProjectRootForLanguage(file, language) {
       } catch {}
     }
     for (const marker of nameMarkers) {
-      if (existsSync8(join10(dir, marker)))
+      if (existsSync9(join11(dir, marker)))
         return dir;
     }
     dir = dirname3(dir);
@@ -201914,7 +202040,7 @@ var require_path_browserify = __commonJS((exports, module2) => {
       assertPath(path);
       return path.length > 0 && path.charCodeAt(0) === 47;
     },
-    join: function join11() {
+    join: function join12() {
       if (arguments.length === 0)
         return ".";
       var joined;

package/dist/workers/io.worker.js

@@ -30797,6 +30797,26 @@ function isStatFile(stat2) {
   }
 }
 
+// src/core/llm/model-utils.ts
+function inferModelGroup(modelId) {
+  const id = modelId.toLowerCase();
+  if (id.startsWith("claude"))
+    return "anthropic";
+  if (id.startsWith("gpt") || id.startsWith("o1-") || id.startsWith("o3-") || id.startsWith("o4-") || id.startsWith("chatgpt"))
+    return "openai";
+  if (id.startsWith("gemini"))
+    return "google";
+  if (id.startsWith("grok"))
+    return "xai";
+  if (id.startsWith("llama") || id.startsWith("meta-"))
+    return "meta";
+  if (id.startsWith("mistral") || id.startsWith("codestral") || id.startsWith("pixtral"))
+    return "mistral";
+  if (id.startsWith("deepseek"))
+    return "deepseek";
+  return "other";
+}
+
 // src/core/workers/io.worker.ts
 var MAX_READ_LINES = 2000;
 var MAX_LINE_LENGTH = 2000;
@@ -31067,24 +31087,6 @@ var handlers = {
     function tc(s) {
       return s.charAt(0).toUpperCase() + s.slice(1);
     }
-    function inferGroup(modelId) {
-      const id = modelId.toLowerCase();
-      if (id.startsWith("claude"))
-        return "anthropic";
-      if (id.startsWith("gpt") || id.startsWith("o1-") || id.startsWith("o3-") || id.startsWith("o4-") || id.startsWith("chatgpt"))
-        return "openai";
-      if (id.startsWith("gemini"))
-        return "google";
-      if (id.startsWith("grok"))
-        return "xai";
-      if (id.startsWith("llama") || id.startsWith("meta-"))
-        return "meta";
-      if (id.startsWith("mistral") || id.startsWith("codestral") || id.startsWith("pixtral"))
-        return "mistral";
-      if (id.startsWith("deepseek"))
-        return "deepseek";
-      return "other";
-    }
     try {
       const res = await fetch(url2, {
         headers
@@ -31117,11 +31119,11 @@ var handlers = {
           const si = m.id.indexOf("/");
           group = si >= 0 ? m.id.slice(0, si).toLowerCase() : "other";
         } else if (isLG) {
-          group = m.family?.toLowerCase() || inferGroup(m.id);
+          group = m.family?.toLowerCase() || inferModelGroup(m.id);
         } else {
           if (m.type && m.type !== "language")
             continue;
-          group = m.owned_by ?? inferGroup(m.id);
+          group = m.owned_by ?? inferModelGroup(m.id);
         }
         const arr = gMap[group] || [];
         gMap[group] = arr;

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@proxysoul/soulforge",
-  "version": "2.11.0",
+  "version": "2.12.1",
   "description": "Graph-powered code intelligence — multi-agent coding with codebase-aware AI",
   "repository": {
     "type": "git",
@@ -82,6 +82,7 @@
     "ai": "6.0.154",
     "ghostty-opentui": "1.4.10",
     "isbinaryfile": "6.0.0",
+    "jsonrepair": "^3.13.3",
     "linkedom": "0.18.12",
     "linkify-it": "5.0.0",
     "marked": "17.0.5",