modelstat 0.3.0 → 0.3.1

package/dist/cli.mjs

@@ -5200,7 +5200,7 @@ var init_schemas = __esm({
       scripts: external_exports.array(external_exports.object({ token: external_exports.string().max(200), summary: external_exports.string().max(200) })).max(8).default([]),
       /** Extractor confidence in [0, 1]. */
       confidence: external_exports.number().min(0).max(1).default(0),
-      /** Provenance of the extraction, e.g. `shell.v2`. */
+      /** Provenance of the extraction, e.g. `shell.v3`. */
       extractor: external_exports.string().max(40)
     }).strict();
     ToolCallWire = external_exports.object({
@@ -5390,6 +5390,214 @@ var init_src = __esm({
   }
 });
 
+// ../../packages/parsers/src/tool-action/executable.ts
+function splitStatements(command) {
+  const out = [];
+  let cur = "";
+  let single = false;
+  let double = false;
+  for (let i = 0; i < command.length; i++) {
+    const c = command[i];
+    if (single) {
+      cur += c;
+      if (c === "'") single = false;
+      continue;
+    }
+    if (double) {
+      if (c === "\\" && i + 1 < command.length) {
+        cur += c + command[i + 1];
+        i++;
+        continue;
+      }
+      cur += c;
+      if (c === '"') double = false;
+      continue;
+    }
+    if (c === "'") {
+      single = true;
+      cur += c;
+      continue;
+    }
+    if (c === '"') {
+      double = true;
+      cur += c;
+      continue;
+    }
+    if (c === "\\" && i + 1 < command.length) {
+      cur += c + command[i + 1];
+      i++;
+      continue;
+    }
+    if (c === "#" && (cur === "" || /\s$/.test(cur))) {
+      out.push(cur);
+      cur = "";
+      while (i + 1 < command.length && command[i + 1] !== "\n") i++;
+      continue;
+    }
+    if (c === "\n" || c === ";") {
+      out.push(cur);
+      cur = "";
+      continue;
+    }
+    if (c === "&" || c === "|") {
+      out.push(cur);
+      cur = "";
+      if (command[i + 1] === c) i++;
+      continue;
+    }
+    cur += c;
+  }
+  out.push(cur);
+  return out;
+}
+function basename(token) {
+  const parts = token.split("/");
+  return parts[parts.length - 1] ?? token;
+}
+function stripOpener(token) {
+  let t = token;
+  if (t.startsWith("$(")) t = t.slice(2);
+  else if (t.startsWith("(")) t = t.slice(1);
+  if (t.startsWith("`")) t = t.slice(1);
+  return t;
+}
+function substitutionProgram(token) {
+  const rhs = token.slice(token.indexOf("=") + 1);
+  if (rhs.startsWith("$((")) return null;
+  if (rhs.startsWith("$(") || rhs.startsWith("`")) {
+    const inner = stripOpener(rhs);
+    return inner || null;
+  }
+  return null;
+}
+function looksLikeProgram(cand) {
+  if (!PROGRAM.test(cand)) return false;
+  if (/^\d+$/.test(cand)) return false;
+  if ((cand.match(/\./g)?.length ?? 0) >= 2) return false;
+  const lower = cand.toLowerCase();
+  return !DATA_EXTENSIONS.some((ext) => lower.endsWith(ext));
+}
+function scanStatement(stmt) {
+  for (const tok of stmt.split(/\s+/).filter(Boolean)) {
+    let t = tok;
+    if (BRACKETS.has(t) || FUNCTION_DEF.test(t) || WRAPPERS.has(t)) continue;
+    if (ASSIGNMENT.test(t)) {
+      const sub = substitutionProgram(t);
+      if (sub) t = sub;
+      else continue;
+    }
+    const cand = basename(stripOpener(t).replace(/[)"'`;,]+$/, "")).toLowerCase();
+    if (!cand || cand.startsWith("-")) break;
+    if (NOISE_BUILTINS.has(cand)) return { builtin: cand };
+    if (looksLikeProgram(cand)) return { program: cand };
+    break;
+  }
+  return {};
+}
+function extractExecutable(command) {
+  let fallback = null;
+  for (const raw of splitStatements(command)) {
+    const stmt = raw.trim();
+    if (!stmt || stmt.startsWith("#")) continue;
+    const { program, builtin } = scanStatement(stmt);
+    if (program) {
+      return program.length > MAX_EXECUTABLE_CHARS ? OTHER_BUCKET : program;
+    }
+    if (builtin && !fallback) fallback = builtin;
+  }
+  return fallback ?? OTHER_BUCKET;
+}
+var OTHER_BUCKET, MAX_EXECUTABLE_CHARS, WRAPPERS, BRACKETS, NOISE_BUILTINS, DATA_EXTENSIONS, ASSIGNMENT, FUNCTION_DEF, PROGRAM;
+var init_executable = __esm({
+  "../../packages/parsers/src/tool-action/executable.ts"() {
+    "use strict";
+    OTHER_BUCKET = "(other)";
+    MAX_EXECUTABLE_CHARS = 80;
+    WRAPPERS = /* @__PURE__ */ new Set([
+      "sudo",
+      "doas",
+      "env",
+      "command",
+      "exec",
+      "builtin",
+      "nohup",
+      "setsid",
+      "time",
+      "nice",
+      "ionice",
+      "chrt",
+      "stdbuf",
+      "xargs",
+      // control-flow openers that immediately precede a command
+      "then",
+      "do",
+      "else"
+    ]);
+    BRACKETS = /* @__PURE__ */ new Set(["{", "}", "(", ")"]);
+    NOISE_BUILTINS = /* @__PURE__ */ new Set([
+      "cd",
+      "pushd",
+      "popd",
+      "echo",
+      "printf",
+      "export",
+      "unset",
+      "set",
+      "readonly",
+      "typeset",
+      "declare",
+      "local",
+      "alias",
+      "source",
+      ".",
+      "eval",
+      ":",
+      "true",
+      "false",
+      "read",
+      "wait",
+      "trap",
+      "umask",
+      "shift",
+      "return",
+      "getopts",
+      "hash",
+      "let",
+      "test",
+      "[",
+      "[[",
+      // loop / conditional keywords
+      "for",
+      "while",
+      "until",
+      "if",
+      "elif",
+      "fi",
+      "case",
+      "esac",
+      "select",
+      "function",
+      "done"
+    ]);
+    DATA_EXTENSIONS = [
+      ".output",
+      ".txt",
+      ".log",
+      ".json",
+      ".jsonl",
+      ".md",
+      ".csv",
+      ".tmp",
+      ".out",
+      ".git",
+      ".lock"
+    ];
+    ASSIGNMENT = /^[A-Za-z_][A-Za-z0-9_]*=/;
+    FUNCTION_DEF = /^[A-Za-z_][A-Za-z0-9_]*\(\)?$/;
+    PROGRAM = /^[A-Za-z0-9][A-Za-z0-9._+-]*$/;
+  }
+});
+
 // ../../packages/parsers/src/tool-action/scripts.ts
 function detectScriptRefs(command) {
   const refs = [];
@@ -5462,9 +5670,9 @@ function extractToolAction(call) {
   let param_shape = null;
   let command_redacted = null;
   if (command != null) {
-    const [head = "", ...rest] = command.trim().split(/\s+/);
-    executable = basename(head) || null;
-    param_shape = clampChars(paramShape(rest.join(" ")), MAX_FIELD_CHARS) || null;
+    executable = extractExecutable(command);
+    const args = command.trim().split(/\s+/).slice(1).join(" ");
+    param_shape = clampChars(paramShape(args), MAX_FIELD_CHARS) || null;
     command_redacted = clampChars(redact(command, call.cwd ?? void 0).text, MAX_FIELD_CHARS) || null;
   }
   return {
@@ -5479,7 +5687,9 @@ function extractToolAction(call) {
     command_redacted,
     scripts: [],
     confidence: 0,
-    extractor: `${surface}.v1`
+    // Per-surface provenance. shell bumped to v3 (normalized executable, see
+    // `extractExecutable`); builtin/mcp extraction is unchanged → still v1.
+    extractor: `${surface}.${surface === "shell" ? "v3" : "v1"}`
   };
 }
 function extractLocalToolContext(call) {
@@ -5500,14 +5710,13 @@ function shellCommandOf(input) {
   }
   return null;
 }
-function basename(token) {
-  return token.split("/").pop() ?? token;
-}
 var MAX_FIELD_CHARS;
 var init_tool_action = __esm({
   "../../packages/parsers/src/tool-action/index.ts"() {
     "use strict";
     init_src();
+    init_executable();
+    init_executable();
     init_scripts();
     MAX_FIELD_CHARS = 16384;
   }
@@ -47573,7 +47782,7 @@ var init_scan = __esm({
     init_api();
     init_config2();
     init_pipeline2();
-    DAEMON_VERSION = true ? "daemon-0.3.0" : "daemon-dev";
+    DAEMON_VERSION = true ? "daemon-0.3.1" : "daemon-dev";
     BATCH_MAX_EVENTS = INGEST_BATCH_MAX_EVENTS;
     BATCH_MAX_TOOL_CALLS = 2e4;
     BATCH_BUFFER_HARD_CAP = BATCH_MAX_EVENTS * 2;
@@ -50083,7 +50292,7 @@ var init_daemon = __esm({
     init_machine_key();
     init_scan();
     init_single_flight();
-    DAEMON_VERSION2 = true ? "daemon-0.3.0" : "daemon-dev";
+    DAEMON_VERSION2 = true ? "daemon-0.3.1" : "daemon-dev";
     HEARTBEAT_INTERVAL_MS = 1e4;
     SCAN_INTERVAL_MS = 5 * 60 * 1e3;
     DISCOVERY_INTERVAL_MS = 6e4;
@@ -50685,7 +50894,7 @@ function tryOpenBrowser(url) {
     return false;
   }
 }
-var DAEMON_VERSION3 = true ? "daemon-0.3.0" : "daemon-dev";
+var DAEMON_VERSION3 = true ? "daemon-0.3.1" : "daemon-dev";
 function osFamily() {
   const p = platform5();
   if (p === "darwin") return "macos";