@exulu/backend 3.7.2 → 3.7.4

package/dist/{chunk-T6JVFT7L.js → chunk-27K2CO47.js}

@@ -39,6 +39,10 @@ function getSetupScriptPath(packageRoot) {
 function getVenvPath(packageRoot) {
   return resolve(packageRoot, "ee/python/.venv");
 }
+function getPythonVenvPath(packageRoot) {
+  const root = packageRoot ?? getPackageRoot();
+  return isPythonEnvironmentSetup(root) ? getVenvPath(root) : void 0;
+}
 function isPythonEnvironmentSetup(packageRoot) {
   const root = packageRoot ?? getPackageRoot();
   const venvPath = getVenvPath(root);
@@ -205,6 +209,7 @@ Or manually run the setup script:
 
 export {
   getPackageRoot,
+  getPythonVenvPath,
   isPythonEnvironmentSetup,
   setupPythonEnvironment,
   getPythonSetupInstructions,

package/dist/{chunk-BNTL6LYY.js → chunk-AWMU6QXB.js}

@@ -1716,7 +1716,7 @@ var ExuluTool = class _ExuluTool {
     if (!agent) {
       throw new Error("Agent not found.");
     }
-    const { convertExuluToolsToAiSdkTools: convertExuluToolsToAiSdkTools2 } = await import("./convert-exulu-tools-to-ai-sdk-tools-UQSLJDXE.js");
+    const { convertExuluToolsToAiSdkTools: convertExuluToolsToAiSdkTools2 } = await import("./convert-exulu-tools-to-ai-sdk-tools-XNQ6Q3X6.js");
     const tools = await convertExuluToolsToAiSdkTools2(
       [this],
       [],
@@ -5723,6 +5723,33 @@ var createUppyRoutes = async (app, config) => {
   return app;
 };
 
+// ee/invoke-skills/artifact-filter.ts
+var IGNORED_SEGMENTS = /* @__PURE__ */ new Set([
+  "venv",
+  ".venv",
+  "env",
+  "node_modules",
+  "__pycache__",
+  "site-packages",
+  "dist-packages",
+  ".cache",
+  ".git",
+  ".pytest_cache",
+  ".mypy_cache",
+  ".ipynb_checkpoints"
+]);
+function isIgnoredArtifactPath(relativePath) {
+  return relativePath.split(/[\\/]+/).some((segment) => IGNORED_SEGMENTS.has(segment));
+}
+var DEFAULT_ARTIFACT_CAP = 50;
+function capArtifacts(artifacts, max = DEFAULT_ARTIFACT_CAP) {
+  if (artifacts.length <= max) return { kept: artifacts, omitted: 0 };
+  return { kept: artifacts.slice(0, max), omitted: artifacts.length - max };
+}
+function needsDownload(localSize, remoteSize) {
+  return localSize === void 0 || localSize !== remoteSize;
+}
+
 // src/exulu/system-dependencies.ts
 import { exec } from "child_process";
 import { existsSync as existsSync2 } from "fs";
@@ -5959,7 +5986,7 @@ function resolveSessionPath(inputPath, sessionDir) {
   }
   return resolved;
 }
-async function restoreArtifactsFromS3(sessionDir, sessionId, userId, config) {
+async function restoreArtifactsFromS3(sessionDir, sessionId, userId, config, opts = {}) {
   const userPrefix = `user_${userId}/sessions/${sessionId}/`;
   let objects;
   try {
@@ -5979,7 +6006,17 @@ async function restoreArtifactsFromS3(sessionDir, sessionId, userId, config) {
     const idx = obj.key.indexOf(userPrefix);
     const relativePath = idx >= 0 ? obj.key.slice(idx + userPrefix.length) : "";
     if (!relativePath) continue;
+    if (isIgnoredArtifactPath(relativePath)) continue;
     const localPath = join2(sessionDir, relativePath);
+    if (opts.onlyMissing) {
+      let localSize;
+      try {
+        localSize = (await stat(localPath)).size;
+      } catch {
+        localSize = void 0;
+      }
+      if (!needsDownload(localSize, obj.size)) continue;
+    }
     try {
       const bytes = await getS3ObjectBytes(obj.key, config);
       await mkdir(dirname(localPath), { recursive: true });
@@ -6013,6 +6050,15 @@ async function downloadKeyIntoSandbox(opts) {
   await writeFile(localPath, bytes);
   return { written: true, localPath };
 }
+async function resolvePythonVenvPath() {
+  try {
+    const { getPythonVenvPath } = await import("./python-setup-JZGHWQCG.js");
+    return getPythonVenvPath();
+  } catch (err) {
+    console.warn("[SKILLS] Could not resolve the Python venv for the session sandbox; skill scripts fall back to the system python.", err);
+    return void 0;
+  }
+}
 async function createSessionSandbox(sessionId, skills, config, userId) {
   const cached = sandboxCache.get(sessionId);
   if (cached) {
@@ -6026,6 +6072,13 @@ async function createSessionSandbox(sessionId, skills, config, userId) {
       await downloadSkill(skill, skillsDirectory2, config);
       cached.installedSkills.set(skill.id, skill.current_version);
     }
+    if (userId && config.fileUploads) {
+      try {
+        await restoreArtifactsFromS3(cached.handle.sessionDir, sessionId, userId, config, { onlyMissing: true });
+      } catch (err) {
+        console.error(`[SKILLS] Failed to re-sync S3 session files for session ${sessionId}; continuing.`, err);
+      }
+    }
     return cached.handle;
   }
   const sessionDir = join2("/tmp", "exulu-sessions", sessionId);
@@ -6043,8 +6096,8 @@ async function createSessionSandbox(sessionId, skills, config, userId) {
       `[SKILLS] S3 artifact persistence disabled for session ${sessionId} (userId=${userId ?? "missing"}, fileUploads=${config.fileUploads ? "configured" : "missing"})`
     );
   }
-  if (userId && config.fileUploads && !dirExisted) {
-    await restoreArtifactsFromS3(sessionDir, sessionId, userId, config);
+  if (userId && config.fileUploads) {
+    await restoreArtifactsFromS3(sessionDir, sessionId, userId, config, { onlyMissing: dirExisted });
   }
   const probe = await probeSandboxSupport();
   const useDirectExec = !probe.canSandbox;
@@ -6079,6 +6132,7 @@ Probe error: ${probe.reason ?? "(no detail)"}`
     await SandboxManager.initialize(baselineSandboxConfig);
   }
   const npmGlobalRoot = await getNpmGlobalRoot();
+  const pythonVenvPath = await resolvePythonVenvPath();
   const sessionSandboxConfig = {
     network: {
       allowedDomains: [],
@@ -6093,7 +6147,8 @@ Probe error: ${probe.reason ?? "(no detail)"}`
         // Allow Node to read globally-installed packages from inside
         // the sandbox. Without this, `require('docx')` fails with
         // EPERM even when NODE_PATH points the resolver here.
-        ...npmGlobalRoot ? [npmGlobalRoot] : []
+        ...npmGlobalRoot ? [npmGlobalRoot] : [],
+        ...pythonVenvPath ? [pythonVenvPath] : []
       ],
       allowWrite: [sessionDir],
       denyWrite: []
@@ -6111,7 +6166,8 @@ Probe error: ${probe.reason ?? "(no detail)"}`
   const sandboxedExecEnv = {
     ...configuredVariables,
     ...process.env,
-    ...npmGlobalRoot ? { NODE_PATH: npmGlobalRoot } : {}
+    ...npmGlobalRoot ? { NODE_PATH: npmGlobalRoot } : {},
+    ...pythonVenvPath ? { PATH: `${join2(pythonVenvPath, "bin")}:${process.env.PATH ?? ""}`, VIRTUAL_ENV: pythonVenvPath } : {}
   };
   const wrapIfNeeded = async (command) => {
     if (useDirectExec) return command;
@@ -6236,6 +6292,7 @@ Probe error: ${probe.reason ?? "(no detail)"}`
       for (const entry of entries) {
         const full = join2(dir, entry.name);
         if (full === skillsDir) continue;
+        if (isIgnoredArtifactPath(relative(sessionDir, full))) continue;
         if (entry.isDirectory()) {
           await walk(full);
         } else if (entry.isFile()) {
@@ -6331,19 +6388,24 @@ Probe error: ${probe.reason ?? "(no detail)"}`
         }
       }
       let stdout = result?.stdout ?? "";
-      const withUrls = artifacts.filter((a) => a.url);
+      const { kept, omitted } = capArtifacts(artifacts);
+      const withUrls = kept.filter((a) => a.url);
       if (withUrls.length > 0) {
         const lines = ["", "[exulu-artifacts]"];
         for (const a of withUrls) {
           lines.push(`  ${a.relativePath}: ${a.url}`);
         }
+        if (omitted > 0) {
+          lines.push(`  \u2026 ${omitted} more file(s) were created and mirrored but are not listed here.`);
+        }
         stdout = `${stdout}
 ${lines.join("\n")}`;
       }
       return {
         ...result,
         stdout,
-        artifacts
+        artifacts: kept,
+        ...omitted > 0 ? { artifactsOmitted: omitted } : {}
       };
     }
   });
@@ -6561,12 +6623,21 @@ var buildAuthToolModelOutput = (tool3) => ({ output }) => {
 
 // src/templates/tools/session-file-read-tool.ts
 import { z as z12 } from "zod";
+
+// src/exulu/session-files.ts
+function sessionFilePrefix(ownerId, sessionId, s3prefix) {
+  const general = s3prefix ? `${s3prefix.replace(/\/+$/, "")}/` : "";
+  return `${general}user_${ownerId}/sessions/${sessionId}/`;
+}
+
+// src/templates/tools/session-file-read-tool.ts
 var DEFAULT_LIMIT = 250;
 var MAX_CONTENT_CHARS = 16e3;
 var createSessionFileReadTool = ({
   sessionID,
   user,
-  exuluConfig
+  exuluConfig,
+  ownerId
 }) => {
   if (!sessionID || !exuluConfig?.fileUploads?.s3Bucket) return void 0;
   const readSessionFileExecute = async ({ filename, offset, limit }) => {
@@ -6577,8 +6648,7 @@ var createSessionFileReadTool = ({
       };
     }
     const uploads = exuluConfig.fileUploads;
-    const generalPrefix = uploads.s3prefix ? `${uploads.s3prefix.replace(/\/$/, "")}/` : "";
-    const key = `${generalPrefix}user_${user?.id ?? "api"}/sessions/${sessionID}/${safeName}`;
+    const key = `${sessionFilePrefix(ownerId ?? user?.id ?? "api", sessionID, uploads.s3prefix)}${safeName}`;
     try {
       const url = await getPresignedUrl(uploads.s3Bucket, key, exuluConfig);
       const res = await fetch(url);
@@ -6681,6 +6751,18 @@ async function renderPdfPageToPng(pdf, page, scaleTo) {
 var DEFAULT_LIMIT2 = 250;
 var MAX_CONTENT_CHARS2 = 16e3;
 var MIN_CHARS_PER_PAGE = 20;
+function looksLikeGarbledTextLayer(text) {
+  let control = 0;
+  let visible = 0;
+  for (const ch of text) {
+    const code = ch.charCodeAt(0);
+    if (code === 9 || code === 10 || code === 12 || code === 13 || code === 32) continue;
+    visible++;
+    if (code < 32 || code === 127) control++;
+  }
+  if (visible < 40) return false;
+  return control / visible > 0.01;
+}
 var OFFICE_EXTENSIONS = /* @__PURE__ */ new Set([
   ".docx",
   ".doc",
@@ -6697,7 +6779,8 @@ var pagesPattern = /^(\d+)(?:-(\d+))?$/;
 var createParseDocumentTool = ({
   sessionID,
   user,
-  exuluConfig
+  exuluConfig,
+  ownerId
 }) => {
   if (!sessionID || !exuluConfig?.fileUploads?.s3Bucket) return void 0;
   const parseDocumentExecute = async ({
@@ -6722,8 +6805,7 @@ var createParseDocumentTool = ({
       return { error: `The pages option is only supported for PDF files \u2014 "${ext}" documents are extracted whole.` };
     }
     const uploads = exuluConfig.fileUploads;
-    const generalPrefix = uploads.s3prefix ? `${uploads.s3prefix.replace(/\/$/, "")}/` : "";
-    const key = `${generalPrefix}user_${user?.id ?? "api"}/sessions/${sessionID}/${safeName}`;
+    const key = `${sessionFilePrefix(ownerId ?? user?.id ?? "api", sessionID, uploads.s3prefix)}${safeName}`;
     try {
       const url = await getPresignedUrl(uploads.s3Bucket, key, exuluConfig);
       const res = await fetch(url);
@@ -6738,6 +6820,11 @@ var createParseDocumentTool = ({
         const pageTexts = raw.replace(/\f$/, "").split("\f");
         totalPages = pageTexts.length;
         const nonWhitespace = raw.replace(/\s/g, "").length;
+        if (looksLikeGarbledTextLayer(raw)) {
+          return {
+            error: `"${safeName}" has a text layer that is unreadable (its font encoding maps glyphs to the wrong characters, so words and especially numbers come out wrong or vanish). Do not use extracted text from this file. Use view_document_page to read the pages visually, or suggest the user add the document to a knowledge base with a document processor for full OCR.`
+          };
+        }
         if (nonWhitespace < totalPages * MIN_CHARS_PER_PAGE) {
           return {
             error: `"${safeName}" has no extractable text layer (likely a scan or image-based PDF). Use view_document_page to look at pages visually, or suggest the user add the document to a knowledge base with a document processor for full OCR.`
@@ -6985,7 +7072,8 @@ var OFFICE_EXTENSIONS2 = /* @__PURE__ */ new Set([
 var createViewDocumentPageTool = ({
   sessionID,
   user,
-  exuluConfig
+  exuluConfig,
+  ownerId
 }) => {
   if (!sessionID || !exuluConfig?.fileUploads?.s3Bucket) return void 0;
   const viewDocumentPageExecute = async ({ filename, page, model }, options) => {
@@ -7018,8 +7106,7 @@ var createViewDocumentPageTool = ({
       }
     }
     const uploads = exuluConfig.fileUploads;
-    const generalPrefix = uploads.s3prefix ? `${uploads.s3prefix.replace(/\/$/, "")}/` : "";
-    const key = `${generalPrefix}user_${user?.id ?? "api"}/sessions/${sessionID}/${safeName}`;
+    const key = `${sessionFilePrefix(ownerId ?? user?.id ?? "api", sessionID, uploads.s3prefix)}${safeName}`;
     const pageNumber = page ?? 1;
     try {
       let imageBytes;
@@ -7704,7 +7791,7 @@ var hydrateVariables = async (tool3) => {
   await Promise.all(promises);
   return tool3;
 };
-var convertExuluToolsToAiSdkTools = async (currentTools, currentSkills, approvedTools, allExuluTools, configs, contexts, user, exuluConfig, sessionID, req, project, sessionItems, model, agent, memoryItems, contextWindow, disabledTools) => {
+var convertExuluToolsToAiSdkTools = async (currentTools, currentSkills, approvedTools, allExuluTools, configs, contexts, user, exuluConfig, sessionID, req, project, sessionItems, model, agent, memoryItems, contextWindow, disabledTools, sessionOwnerId) => {
   if (!currentTools) return {};
   if (!allExuluTools) {
     allExuluTools = [];
@@ -7721,7 +7808,7 @@ var convertExuluToolsToAiSdkTools = async (currentTools, currentSkills, approved
         sessionID,
         currentSkills || [],
         exuluConfig,
-        user?.id
+        sessionOwnerId ?? user?.id
       );
     } catch (err) {
       console.error(
@@ -7786,15 +7873,15 @@ var convertExuluToolsToAiSdkTools = async (currentTools, currentSkills, approved
       currentTools.push(sessionItemsRetrievalTool);
     }
   }
-  const sessionFileReadTool = createSessionFileReadTool({ sessionID, user, exuluConfig });
+  const sessionFileReadTool = createSessionFileReadTool({ sessionID, user, exuluConfig, ownerId: sessionOwnerId });
   if (sessionFileReadTool && !disabled.has(sessionFileReadTool.id)) {
     currentTools.push(sessionFileReadTool);
   }
-  const parseDocumentTool = createParseDocumentTool({ sessionID, user, exuluConfig });
+  const parseDocumentTool = createParseDocumentTool({ sessionID, user, exuluConfig, ownerId: sessionOwnerId });
   if (parseDocumentTool && !disabled.has(parseDocumentTool.id)) {
     currentTools.push(parseDocumentTool);
   }
-  const viewDocumentPageTool = createViewDocumentPageTool({ sessionID, user, exuluConfig });
+  const viewDocumentPageTool = createViewDocumentPageTool({ sessionID, user, exuluConfig, ownerId: sessionOwnerId });
   if (viewDocumentPageTool && !disabled.has(viewDocumentPageTool.id)) {
     currentTools.push(viewDocumentPageTool);
   }

package/dist/cli/start-whisper.js

@@ -2,7 +2,7 @@
 import "dotenv/config";
 import {
   getPackageRoot
-} from "../chunk-T6JVFT7L.js";
+} from "../chunk-27K2CO47.js";
 
 // src/cli/start-whisper.ts
 import "dotenv/config";

package/dist/{convert-exulu-tools-to-ai-sdk-tools-UQSLJDXE.js → convert-exulu-tools-to-ai-sdk-tools-XNQ6Q3X6.js}

@@ -2,7 +2,7 @@ import "dotenv/config";
 import {
   convertExuluToolsToAiSdkTools,
   hydrateVariables
-} from "./chunk-BNTL6LYY.js";
+} from "./chunk-AWMU6QXB.js";
 import "./chunk-4PDWNVNT.js";
 export {
   convertExuluToolsToAiSdkTools,