@vertesia/workflow 0.82.0 → 0.82.2

package/lib/cjs/activities/extractDocumentText.js

@@ -0,0 +1,156 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.extractDocumentText = extractDocumentText;
+const activity_1 = require("@temporalio/activity");
+const markitdown_js_1 = require("../conversion/markitdown.js");
+const mutool_js_1 = require("../conversion/mutool.js");
+const pandoc_js_1 = require("../conversion/pandoc.js");
+const ActivityContext_js_1 = require("../dsl/setup/ActivityContext.js");
+const errors_js_1 = require("../errors.js");
+const result_types_js_1 = require("../result-types.js");
+const blobs_js_1 = require("../utils/blobs.js");
+const tokens_js_1 = require("../utils/tokens.js");
+//@ts-ignore
+const JSON = {
+    name: "extractDocumentText",
+};
+async function extractDocumentText(payload) {
+    const { client, objectId } = await (0, ActivityContext_js_1.setupActivity)(payload);
+    const r = await client.objects.find({
+        query: { _id: objectId },
+        limit: 1,
+        select: "+text",
+    });
+    const doc = r[0];
+    if (!doc) {
+        activity_1.log.error(`Document ${objectId} not found`);
+        throw new errors_js_1.DocumentNotFoundError(`Document ${objectId} not found`, payload.objectIds);
+    }
+    activity_1.log.info(`Extracting text for object ${doc.id}`);
+    if (!doc.content?.type || !doc.content?.source) {
+        if (doc.text) {
+            return createResponse(doc, doc.text, result_types_js_1.TextExtractionStatus.skipped, "Text present and no source or type");
+        }
+        else {
+            return createResponse(doc, "", result_types_js_1.TextExtractionStatus.error, "No source or type found");
+        }
+    }
+    //skip if text already extracted and proper etag
+    if (doc.text && doc.text.length > 0 && doc.text_etag === doc.content.etag) {
+        return createResponse(doc, doc.text, result_types_js_1.TextExtractionStatus.skipped, "Text already extracted");
+    }
+    let fileBuffer;
+    try {
+        fileBuffer = await (0, blobs_js_1.fetchBlobAsBuffer)(client, doc.content.source);
+    }
+    catch (e) {
+        activity_1.log.error(`Error reading file: ${e}`);
+        return createResponse(doc, "", result_types_js_1.TextExtractionStatus.error, e.message);
+    }
+    let txt;
+    switch (doc.content.type) {
+        case "application/pdf":
+            txt = await (0, mutool_js_1.mutoolPdfToText)(fileBuffer);
+            break;
+        case "text/plain":
+            txt = fileBuffer.toString("utf8");
+            break;
+        //docx
+        case "application/vnd.openxmlformats-officedocument.wordprocessingml.document":
+            txt = await (0, markitdown_js_1.markdownWithMarkitdown)(fileBuffer, "docx");
+            break;
+        //pptx
+        case "application/vnd.openxmlformats-officedocument.presentationml.presentation":
+            txt = await (0, markitdown_js_1.markdownWithMarkitdown)(fileBuffer, "pptx");
+            break;
+        //html
+        case "text/html":
+            txt = await (0, pandoc_js_1.markdownWithPandoc)(fileBuffer, "html");
+            break;
+        //opendocument
+        case "application/vnd.oasis.opendocument.text":
+            txt = await (0, pandoc_js_1.markdownWithPandoc)(fileBuffer, "odt");
+            break;
+        //rtf
+        case "application/rtf":
+            txt = await (0, pandoc_js_1.markdownWithPandoc)(fileBuffer, "rtf");
+            break;
+        //markdown
+        case "text/markdown":
+            txt = fileBuffer.toString("utf8");
+            break;
+        //csv
+        case "text/csv":
+            txt = fileBuffer.toString("utf8");
+            break;
+        //typescript
+        case "application/typescript":
+            txt = fileBuffer.toString("utf8");
+            break;
+        //javascript
+        case "application/javascript":
+            txt = fileBuffer.toString("utf8");
+            break;
+        //json
+        case "application/json":
+            txt = fileBuffer.toString("utf8");
+            break;
+        default:
+            if (sniffIfText(fileBuffer)) {
+                txt = fileBuffer.toString("utf8"); //TODO: add charset detection
+                break;
+            }
+            return createResponse(doc, doc.text ?? "", result_types_js_1.TextExtractionStatus.skipped, `Unsupported mime type: ${doc.content.type}`);
+    }
+    const tokensData = (0, tokens_js_1.countTokens)(txt);
+    const etag = doc.content.etag ?? (0, blobs_js_1.md5)(txt);
+    const updateData = {
+        text: txt,
+        text_etag: etag,
+        tokens: {
+            ...tokensData,
+            etag: etag,
+        },
+    };
+    await client.objects.update(doc.id, updateData);
+    return createResponse(doc, txt, result_types_js_1.TextExtractionStatus.success);
+}
+function createResponse(doc, text, status, message) {
+    return {
+        status,
+        message,
+        tokens: doc.tokens,
+        len: text.length,
+        objectId: doc.id,
+        hasText: !!text,
+    };
+}
+function sniffIfText(buf) {
+    // If file is too large, don't even try
+    if (buf.length > 500 * 1024) {
+        return false;
+    }
+    // Count binary/control characters
+    let binaryCount = 0;
+    const sampleSize = Math.min(buf.length, 1000); // Check first 1000 bytes
+    for (let i = 0; i < sampleSize; i++) {
+        // Count control characters (except common whitespace)
+        const byte = buf[i];
+        if ((byte < 32 && ![9, 10, 13].includes(byte)) || byte === 0) {
+            binaryCount++;
+        }
+    }
+    // If more than 10% binary/control chars, probably not text
+    if (binaryCount / sampleSize > 0.1) {
+        return false;
+    }
+    // Additional check for valid UTF-8 encoding
+    try {
+        const s = buf.toString("utf8");
+        return s.length > 0 && !s.includes("\uFFFD"); // Replacement character
+    }
+    catch (e) {
+        return false;
+    }
+}
+//# sourceMappingURL=extractDocumentText.js.map

package/lib/cjs/activities/extractDocumentText.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"extractDocumentText.js","sourceRoot":"","sources":["../../../src/activities/extractDocumentText.ts"],"names":[],"mappings":";;AA4BA,kDAgIC;AA5JD,mDAA2C;AAO3C,+DAAqE;AACrE,uDAA0D;AAC1D,uDAA6D;AAC7D,wEAAgE;AAChE,4CAAqD;AACrD,wDAAgF;AAChF,gDAA2D;AAC3D,kDAAiD;AAEjD,YAAY;AACZ,MAAM,IAAI,GAAoB;IAC1B,IAAI,EAAE,qBAAqB;CAC9B,CAAC;AASK,KAAK,UAAU,mBAAmB,CACrC,OAA+D;IAE/D,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,IAAA,kCAAa,EAAC,OAAO,CAAC,CAAC;IAE1D,MAAM,CAAC,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,IAAI,CAAC;QAChC,KAAK,EAAE,EAAE,GAAG,EAAE,QAAQ,EAAE;QACxB,KAAK,EAAE,CAAC;QACR,MAAM,EAAE,OAAO;KAClB,CAAC,CAAC;IACH,MAAM,GAAG,GAAG,CAAC,CAAC,CAAC,CAAkB,CAAC;IAClC,IAAI,CAAC,GAAG,EAAE,CAAC;QACP,cAAG,CAAC,KAAK,CAAC,YAAY,QAAQ,YAAY,CAAC,CAAC;QAC5C,MAAM,IAAI,iCAAqB,CAAC,YAAY,QAAQ,YAAY,EAAE,OAAO,CAAC,SAAS,CAAC,CAAC;IACzF,CAAC;IAED,cAAG,CAAC,IAAI,CAAC,8BAA8B,GAAG,CAAC,EAAE,EAAE,CAAC,CAAC;IAEjD,IAAI,CAAC,GAAG,CAAC,OAAO,EAAE,IAAI,IAAI,CAAC,GAAG,CAAC,OAAO,EAAE,MAAM,EAAE,CAAC;QAC7C,IAAI,GAAG,CAAC,IAAI,EAAE,CAAC;YACX,OAAO,cAAc,CAAC,GAAG,EAAE,GAAG,CAAC,IAAI,EAAE,sCAAoB,CAAC,OAAO,EAAE,oCAAoC,CAAC,CAAC;QAC7G,CAAC;aAAM,CAAC;YACJ,OAAO,cAAc,CAAC,GAAG,EAAE,EAAE,EAAE,sCAAoB,CAAC,KAAK,EAAE,yBAAyB,CAAC,CAAC;QAC1F,CAAC;IACL,CAAC;IAED,gDAAgD;IAChD,IAAI,GAAG,CAAC,IAAI,IAAI,GAAG,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,IAAI,GAAG,CAAC,SAAS,KAAK,GAAG,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC;QACxE,OAAO,cAAc,CAAC,GAAG,EAAE,GAAG,CAAC,IAAI,EAAE,sCAAoB,CAAC,OAAO,EAAE,wBAAwB,CAAC,CAAC;IACjG,CAAC;IAED,IAAI,UAAkB,CAAC;IACvB,IAAI,CAAC;QACD,UAAU,GAAG,MAAM,IAAA,4BAAiB,EAAC,MAAM,EAAE,GAAG,CAAC,OAAO,CAAC,MAAM,CAAC,CAAC;IACrE,CAAC;IAAC,OAAO,CAAM,EAAE,CAAC;QACd,cAAG,CAAC,KAAK,CAAC,uBAAuB,CAAC,EAAE,CAAC,CAAC;QACtC,OAAO,cAAc,CAAC,GAAG,EAAE,EAAE,EAAE,sCAAoB,CAAC,KAAK,EAAE,CAAC,CAAC,OAAO,CAAC,CAAC;IAC1E,CAAC;IAED,IAAI,GAAW,CAAC;IAEhB,QAAQ,GAAG,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC;QACvB,KAAK,iBAAiB;YAClB,GAAG,GAAG,MAAM,IAAA,2BAAe,EAAC,UAAU,CAAC,CAAC;YACxC,MAAM;QAEV,KAAK,YAAY;YACb,GAAG,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YAClC,MAAM;QAEV,MAAM;QACN,KAAK,yEAAyE;YAC1E,GAAG,GAAG,MAAM,IAAA,sCAAsB,EAAC,UAAU,EAAE,MAAM,CAAC,CAAC;YACvD,MAAM;QAEV,MAAM;QACN,KAAK,2EAA2E;YAC5E,GAAG,GAAG,MAAM,IAAA,sCAAsB,EAAC,UAAU,EAAE,MAAM,CAAC,CAAC;YACvD,MAAM;QAEV,MAAM;QACN,KAAK,WAAW;YACZ,GAAG,GAAG,MAAM,IAAA,8BAAkB,EAAC,UAAU,EAAE,MAAM,CAAC,CAAC;YACnD,MAAM;QAEV,cAAc;QACd,KAAK,yCAAyC;YAC1C,GAAG,GAAG,MAAM,IAAA,8BAAkB,EAAC,UAAU,EAAE,KAAK,CAAC,CAAC;YAClD,MAAM;QAEV,KAAK;QACL,KAAK,iBAAiB;YAClB,GAAG,GAAG,MAAM,IAAA,8BAAkB,EAAC,UAAU,EAAE,KAAK,CAAC,CAAC;YAClD,MAAM;QAEV,UAAU;QACV,KAAK,eAAe;YAChB,GAAG,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YAClC,MAAM;QAEV,KAAK;QACL,KAAK,UAAU;YACX,GAAG,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YAClC,MAAM;QAEV,YAAY;QACZ,KAAK,wBAAwB;YACzB,GAAG,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YAClC,MAAM;QAEV,YAAY;QACZ,KAAK,wBAAwB;YACzB,GAAG,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YAClC,MAAM;QAEV,MAAM;QACN,KAAK,kBAAkB;YACnB,GAAG,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;YAClC,MAAM;QAEV;YACI,IAAI,WAAW,CAAC,UAAU,CAAC,EAAE,CAAC;gBAC1B,GAAG,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,6BAA6B;gBAChE,MAAM;YACV,CAAC;YACD,OAAO,cAAc,CACjB,GAAG,EACH,GAAG,CAAC,IAAI,IAAI,EAAE,EACd,sCAAoB,CAAC,OAAO,EAC5B,0BAA0B,GAAG,CAAC,OAAO,CAAC,IAAI,EAAE,CAC/C,CAAC;IACV,CAAC;IAED,MAAM,UAAU,GAAG,IAAA,uBAAW,EAAC,GAAG,CAAC,CAAC;IACpC,MAAM,IAAI,GAAG,GAAG,CAAC,OAAO,CAAC,IAAI,IAAI,IAAA,cAAG,EAAC,GAAG,CAAC,CAAC;IAE1C,MAAM,UAAU,GAA+B;QAC3C,IAAI,EAAE,GAAG;QACT,SAAS,EAAE,IAAI;QACf,MAAM,EAAE;YACJ,GAAG,UAAU;YACb,IAAI,EAAE,IAAI;SACb;KACJ,CAAC;IAEF,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,GAAG,CAAC,EAAE,EAAE,UAAU,CAAC,CAAC;IAEhD,OAAO,cAAc,CAAC,GAAG,EAAE,GAAG,EAAE,sCAAoB,CAAC,OAAO,CAAC,CAAC;AAClE,CAAC;AAED,SAAS,cAAc,CACnB,GAAkB,EAClB,IAAY,EACZ,MAA4B,EAC5B,OAAgB;IAEhB,OAAO;QACH,MAAM;QACN,OAAO;QACP,MAAM,EAAE,GAAG,CAAC,MAAM;QAClB,GAAG,EAAE,IAAI,CAAC,MAAM;QAChB,QAAQ,EAAE,GAAG,CAAC,EAAE;QAChB,OAAO,EAAE,CAAC,CAAC,IAAI;KAClB,CAAC;AACN,CAAC;AAED,SAAS,WAAW,CAAC,GAAW;IAC5B,uCAAuC;IACvC,IAAI,GAAG,CAAC,MAAM,GAAG,GAAG,GAAG,IAAI,EAAE,CAAC;QAC1B,OAAO,KAAK,CAAC;IACjB,CAAC;IAED,kCAAkC;IAClC,IAAI,WAAW,GAAG,CAAC,CAAC;IACpB,MAAM,UAAU,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,CAAC,MAAM,EAAE,IAAI,CAAC,CAAC,CAAC,yBAAyB;IAExE,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,UAAU,EAAE,CAAC,EAAE,EAAE,CAAC;QAClC,sDAAsD;QACtD,MAAM,IAAI,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC;QACpB,IAAI,CAAC,IAAI,GAAG,EAAE,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,IAAI,IAAI,KAAK,CAAC,EAAE,CAAC;YAC3D,WAAW,EAAE,CAAC;QAClB,CAAC;IACL,CAAC;IAED,2DAA2D;IAC3D,IAAI,WAAW,GAAG,UAAU,GAAG,GAAG,EAAE,CAAC;QACjC,OAAO,KAAK,CAAC;IACjB,CAAC;IAED,4CAA4C;IAC5C,IAAI,CAAC;QACD,MAAM,CAAC,GAAG,GAAG,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;QAC/B,OAAO,CAAC,CAAC,MAAM,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAC,wBAAwB;IAC1E,CAAC;IAAC,OAAO,CAAC,EAAE,CAAC;QACT,OAAO,KAAK,CAAC;IACjB,CAAC;AACL,CAAC"}

package/lib/cjs/activities/generateDocumentProperties.js

@@ -0,0 +1,83 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.generateDocumentProperties = generateDocumentProperties;
+const activity_1 = require("@temporalio/activity");
+const ActivityContext_js_1 = require("../dsl/setup/ActivityContext.js");
+const tokens_js_1 = require("../utils/tokens.js");
+const executeInteraction_js_1 = require("./executeInteraction.js");
+const INT_EXTRACT_INFORMATION = "sys:ExtractInformation";
+async function generateDocumentProperties(payload) {
+    const context = await (0, ActivityContext_js_1.setupActivity)(payload);
+    const { params, client, objectId } = context;
+    const interactionName = params.interactionName ?? INT_EXTRACT_INFORMATION;
+    const project = await context.fetchProject();
+    const doc = await client.objects.retrieve(objectId, "+text");
+    const type = doc.type ? await client.types.retrieve(doc.type.id) : undefined;
+    if (!doc?.text && !params.use_vision && !doc?.content?.type?.startsWith("image/")) {
+        activity_1.log.warn(`Object ${objectId} not found or text is empty`);
+        return { status: "failed", error: "no-text" };
+    }
+    if (!type || !type.object_schema) {
+        activity_1.log.info(`Object ${objectId} has no schema`);
+        return { document: objectId, status: "skipped", message: "no schema defined on type" };
+    }
+    const getImageRef = () => {
+        if (doc.content?.type?.startsWith("image/")) {
+            return "store:" + doc.id;
+        }
+        if (params.use_vision && doc.content?.type?.startsWith("application/pdf")) {
+            return "store:" + doc.id;
+        }
+        activity_1.log.info(`Object ${objectId} is not an image or pdf`);
+        return undefined;
+    };
+    const content = doc.text
+        ? (0, tokens_js_1.truncByMaxTokens)(doc.text, params.truncate || 30000)
+        : undefined;
+    const promptData = {
+        content: content,
+        image: getImageRef() ?? undefined,
+        human_context: project?.configuration?.human_context ?? undefined,
+    };
+    activity_1.log.info(` Extracting information from object ${objectId} with type ${type.name}`, payload.debug_mode ? { params } : undefined);
+    const infoRes = await (0, executeInteraction_js_1.executeInteractionFromActivity)(client, interactionName, {
+        ...params,
+        include_previous_error: true,
+        result_schema: type.object_schema,
+        validate_result: type.strict_mode,
+    }, promptData, payload.debug_mode ?? false);
+    const getText = () => {
+        if (doc.text) {
+            return undefined;
+        }
+        let text = "";
+        const jsonResult = infoRes.result.object();
+        if (jsonResult.title) {
+            text += jsonResult.title + "\n";
+        }
+        if (jsonResult.description) {
+            text += jsonResult.description;
+        }
+        if (text) {
+            return text;
+        }
+        else {
+            return undefined;
+        }
+    };
+    activity_1.log.info(`Extracted information from object ${objectId} with type ${type.name}`, { runId: infoRes.id });
+    await client.objects.update(doc.id, {
+        properties: {
+            ...infoRes.result.object(),
+            etag: doc.text_etag,
+        },
+        text: getText(),
+        generation_run_info: {
+            id: infoRes.id,
+            date: new Date().toISOString(),
+            model: infoRes.modelId,
+        },
+    });
+    return { status: "completed" };
+}
+//# sourceMappingURL=generateDocumentProperties.js.map

package/lib/cjs/activities/generateDocumentProperties.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"generateDocumentProperties.js","sourceRoot":"","sources":["../../../src/activities/generateDocumentProperties.ts"],"names":[],"mappings":";;AAsBA,gEAiGC;AAvHD,mDAA2C;AAE3C,wEAAgE;AAChE,kDAAoE;AACpE,mEAAqG;AAErG,MAAM,uBAAuB,GAAG,wBAAwB,CAAC;AAgBlD,KAAK,UAAU,0BAA0B,CAC5C,OAAsE;IAEtE,MAAM,OAAO,GAAG,MAAM,IAAA,kCAAa,EAAmC,OAAO,CAAC,CAAC;IAC/E,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,OAAO,CAAC;IAC7C,MAAM,eAAe,GAAG,MAAM,CAAC,eAAe,IAAI,uBAAuB,CAAC;IAE1E,MAAM,OAAO,GAAG,MAAM,OAAO,CAAC,YAAY,EAAE,CAAC;IAE7C,MAAM,GAAG,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;IAC7D,MAAM,IAAI,GAAG,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,MAAM,MAAM,CAAC,KAAK,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC;IAE7E,IAAI,CAAC,GAAG,EAAE,IAAI,IAAI,CAAC,MAAM,CAAC,UAAU,IAAI,CAAC,GAAG,EAAE,OAAO,EAAE,IAAI,EAAE,UAAU,CAAC,QAAQ,CAAC,EAAE,CAAC;QAChF,cAAG,CAAC,IAAI,CAAC,UAAU,QAAQ,6BAA6B,CAAC,CAAC;QAC1D,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,KAAK,EAAE,SAAS,EAAE,CAAC;IAClD,CAAC;IAED,IAAI,CAAC,IAAI,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,CAAC;QAC/B,cAAG,CAAC,IAAI,CAAC,UAAU,QAAQ,gBAAgB,CAAC,CAAC;QAC7C,OAAO,EAAE,QAAQ,EAAE,QAAQ,EAAE,MAAM,EAAE,SAAS,EAAE,OAAO,EAAE,2BAA2B,EAAE,CAAC;IAC3F,CAAC;IAED,MAAM,WAAW,GAAG,GAAG,EAAE;QACrB,IAAI,GAAG,CAAC,OAAO,EAAE,IAAI,EAAE,UAAU,CAAC,QAAQ,CAAC,EAAE,CAAC;YAC1C,OAAO,QAAQ,GAAG,GAAG,CAAC,EAAE,CAAC;QAC7B,CAAC;QAED,IAAI,MAAM,CAAC,UAAU,IAAI,GAAG,CAAC,OAAO,EAAE,IAAI,EAAE,UAAU,CAAC,iBAAiB,CAAC,EAAE,CAAC;YACxE,OAAO,QAAQ,GAAG,GAAG,CAAC,EAAE,CAAC;QAC7B,CAAC;QAED,cAAG,CAAC,IAAI,CAAC,UAAU,QAAQ,yBAAyB,CAAC,CAAC;QACtD,OAAO,SAAS,CAAC;IACrB,CAAC,CAAC;IAEF,MAAM,OAAO,GAAG,GAAG,CAAC,IAAI;QACpB,CAAC,CAAC,IAAA,4BAAgB,EAAC,GAAG,CAAC,IAAI,EAAE,MAAM,CAAC,QAAQ,IAAI,KAAK,CAAC;QACtD,CAAC,CAAC,SAAS,CAAC;IAEhB,MAAM,UAAU,GAAG;QACf,OAAO,EAAE,OAAO;QAChB,KAAK,EAAE,WAAW,EAAE,IAAI,SAAS;QACjC,aAAa,EAAE,OAAO,EAAE,aAAa,EAAE,aAAa,IAAI,SAAS;KACpE,CAAC;IAEF,cAAG,CAAC,IAAI,CACJ,uCAAuC,QAAQ,cAAc,IAAI,CAAC,IAAI,EAAE,EACxE,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC,EAAE,MAAM,EAAE,CAAC,CAAC,CAAC,SAAS,CAC9C,CAAC;IAEF,MAAM,OAAO,GAAG,MAAM,IAAA,sDAA8B,EAChD,MAAM,EACN,eAAe,EACf;QACI,GAAG,MAAM;QACT,sBAAsB,EAAE,IAAI;QAC5B,aAAa,EAAE,IAAI,CAAC,aAAa;QACjC,eAAe,EAAE,IAAI,CAAC,WAAW;KACpC,EACD,UAAU,EACV,OAAO,CAAC,UAAU,IAAI,KAAK,CAC9B,CAAC;IAEF,MAAM,OAAO,GAAG,GAAG,EAAE;QACjB,IAAI,GAAG,CAAC,IAAI,EAAE,CAAC;YACX,OAAO,SAAS,CAAC;QACrB,CAAC;QACD,IAAI,IAAI,GAAG,EAAE,CAAC;QACd,MAAM,UAAU,GAAG,OAAO,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC;QAC3C,IAAI,UAAU,CAAC,KAAK,EAAE,CAAC;YACnB,IAAI,IAAI,UAAU,CAAC,KAAK,GAAG,IAAI,CAAC;QACpC,CAAC;QACD,IAAI,UAAU,CAAC,WAAW,EAAE,CAAC;YACzB,IAAI,IAAI,UAAU,CAAC,WAAW,CAAC;QACnC,CAAC;QACD,IAAI,IAAI,EAAE,CAAC;YACP,OAAO,IAAI,CAAC;QAChB,CAAC;aAAM,CAAC;YACJ,OAAO,SAAS,CAAC;QACrB,CAAC;IACL,CAAC,CAAC;IAEF,cAAG,CAAC,IAAI,CAAC,qCAAqC,QAAQ,cAAc,IAAI,CAAC,IAAI,EAAE,EAAE,EAAE,KAAK,EAAE,OAAO,CAAC,EAAE,EAAE,CAAC,CAAC;IACxG,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,GAAG,CAAC,EAAE,EAAE;QAChC,UAAU,EAAE;YACR,GAAG,OAAO,CAAC,MAAM,CAAC,MAAM,EAAE;YAC1B,IAAI,EAAE,GAAG,CAAC,SAAS;SACtB;QACD,IAAI,EAAE,OAAO,EAAE;QACf,mBAAmB,EAAE;YACjB,EAAE,EAAE,OAAO,CAAC,EAAE;YACd,IAAI,EAAE,IAAI,IAAI,EAAE,CAAC,WAAW,EAAE;YAC9B,KAAK,EAAE,OAAO,CAAC,OAAO;SACzB;KACJ,CAAC,CAAC;IAEH,OAAO,EAAE,MAAM,EAAE,WAAW,EAAE,CAAC;AACnC,CAAC"}

package/lib/cjs/activities/generateEmbeddings.js

@@ -0,0 +1,228 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.generateEmbeddings = generateEmbeddings;
+const activity_1 = require("@temporalio/activity");
+const common_1 = require("@vertesia/common");
+const ActivityContext_js_1 = require("../dsl/setup/ActivityContext.js");
+const errors_js_1 = require("../errors.js");
+const blobs_js_1 = require("../utils/blobs.js");
+const tokens_js_1 = require("../utils/tokens.js");
+async function generateEmbeddings(payload) {
+    const { params, client, objectId, fetchProject } = await (0, ActivityContext_js_1.setupActivity)(payload);
+    const { force, type } = params;
+    const projectData = await fetchProject();
+    const config = projectData?.configuration.embeddings[type];
+    if (!projectData) {
+        throw new errors_js_1.DocumentNotFoundError("Project not found", [payload.project_id]);
+    }
+    if (!config) {
+        throw new errors_js_1.DocumentNotFoundError("Embeddings configuration not found", [
+            objectId,
+        ]);
+    }
+    if (!projectData) {
+        throw new errors_js_1.DocumentNotFoundError("Project not found", [payload.project_id]);
+    }
+    if (!projectData?.configuration.embeddings[type]?.enabled) {
+        activity_1.log.info(`Embeddings generation disabled for type ${type} on project: ${projectData.name} (${projectData.namespace})`, { config });
+        return {
+            id: objectId,
+            status: "skipped",
+            message: `Embeddings generation disabled for type ${type}`,
+        };
+    }
+    activity_1.log.info(`${type} embedding generation starting for object ${objectId}`, {
+        force,
+        config,
+    });
+    if (!config.environment) {
+        throw new Error("No environment found in project configuration. Set environment in project configuration to generate embeddings.");
+    }
+    const document = await client.objects.retrieve(objectId, "+text +parts +embeddings +tokens +properties");
+    if (!document) {
+        throw new errors_js_1.DocumentNotFoundError("Document not found", [objectId]);
+    }
+    if (!document.content) {
+        throw new errors_js_1.DocumentNotFoundError("Document content not found", [objectId]);
+    }
+    let res;
+    switch (type) {
+        case common_1.SupportedEmbeddingTypes.text:
+            res = await generateTextEmbeddings({
+                client,
+                config,
+                document,
+                type,
+            });
+            break;
+        case common_1.SupportedEmbeddingTypes.properties:
+            res = await generateTextEmbeddings({
+                client,
+                config,
+                document,
+                type,
+            });
+            break;
+        case common_1.SupportedEmbeddingTypes.image:
+            res = await generateImageEmbeddings({
+                client,
+                config,
+                document,
+                type,
+            });
+            break;
+        default:
+            res = {
+                id: objectId,
+                status: "failed",
+                message: `unsupported embedding type: ${type}`,
+            };
+    }
+    return res;
+}
+async function generateTextEmbeddings({ document, client, type, config }) {
+    if (!document) {
+        return { status: "error", message: "document is null or undefined" };
+    }
+    if (type !== common_1.SupportedEmbeddingTypes.text &&
+        type !== common_1.SupportedEmbeddingTypes.properties) {
+        return {
+            id: document.id,
+            status: "failed",
+            message: `unsupported embedding type: ${type}`,
+        };
+    }
+    if (type === common_1.SupportedEmbeddingTypes.text && !document.text) {
+        return { id: document.id, status: "failed", message: "no text found" };
+    }
+    if (type === common_1.SupportedEmbeddingTypes.properties && !document?.properties) {
+        return {
+            id: document.id,
+            status: "failed",
+            message: "no properties found",
+        };
+    }
+    const { environment } = config;
+    // Count tokens if needed, do not rely on existing token count
+    let tokenCount = undefined;
+    if (type === common_1.SupportedEmbeddingTypes.text && document.text) {
+        tokenCount = (0, tokens_js_1.countTokens)(document.text).count;
+    }
+    const maxTokens = config.max_tokens ?? 8000;
+    //generate embeddings for the main doc if document isn't too large
+    activity_1.log.info(`Generating ${type} embeddings for document ${document.id}`);
+    if (type === common_1.SupportedEmbeddingTypes.text &&
+        tokenCount !== undefined &&
+        tokenCount > maxTokens) {
+        //TODO: Review strategy for large documents
+        activity_1.log.warn(`Document too large for ${type} embeddings generation, skipping (${tokenCount} tokens)`);
+        return {
+            id: document.id,
+            status: "skipped",
+            message: `${type} embeddings generation, skipped for large document (${tokenCount} tokens)`,
+        };
+    }
+    else {
+        activity_1.log.info(`Generating ${type} embeddings for document`);
+        const res = await generateEmbeddingsFromStudio(JSON.stringify(document[type]), environment, client);
+        if (!res || !res.values) {
+            return {
+                id: document.id,
+                status: "failed",
+                message: "no embeddings generated",
+            };
+        }
+        activity_1.log.info(`${type} embeddings generated for document ${document.id}`, {
+            len: res.values.length,
+        });
+        await client.objects.setEmbedding(document.id, type, {
+            values: res.values,
+            model: res.model,
+            etag: document.text_etag,
+        });
+        return {
+            id: document.id,
+            type,
+            status: "completed",
+            len: res.values.length,
+        };
+    }
+}
+async function generateImageEmbeddings({ document, client, type, config, }) {
+    activity_1.log.info("Generating image embeddings for document " + document.id, {
+        content: document.content,
+    });
+    if (!document.content?.type?.startsWith("image/") &&
+        !document.content?.type?.includes("pdf")) {
+        return {
+            id: document.id,
+            type,
+            status: "failed",
+            message: "content is not an image",
+        };
+    }
+    const { environment, model } = config;
+    const resRnd = await client.store.objects.getRendition(document.id, {
+        format: common_1.ImageRenditionFormat.jpeg,
+        generate_if_missing: true,
+        sign_url: true,
+    });
+    if (resRnd.status === "generating") {
+        throw new Error("Rendition is generating, will retry later");
+    }
+    else if (resRnd.status === "failed" ||
+        !resRnd.renditions ||
+        !resRnd.renditions.length) {
+        throw new errors_js_1.DocumentNotFoundError("Rendition retrieval failed", [document.id]);
+    }
+    const renditions = resRnd.renditions;
+    if (!renditions?.length) {
+        throw new errors_js_1.DocumentNotFoundError("No source found in rendition", [
+            document.id,
+        ]);
+    }
+    const rendition = renditions[0];
+    const image = await (0, blobs_js_1.fetchBlobAsBase64)(client, rendition);
+    const res = await client.environments
+        .embeddings(environment, {
+        image,
+        model,
+    })
+        .then((res) => res)
+        .catch((e) => {
+        activity_1.log.error("Error generating embeddings for image", { error: e });
+        throw e;
+    });
+    if (!res || !res.values) {
+        return {
+            id: document.id,
+            status: "failed",
+            message: "no embeddings generated",
+        };
+    }
+    await client.objects.setEmbedding(document.id, common_1.SupportedEmbeddingTypes.image, {
+        values: res.values,
+        model: res.model,
+        etag: document.text_etag,
+    });
+    return {
+        id: document.id,
+        type,
+        status: "completed",
+        len: res.values.length,
+    };
+}
+async function generateEmbeddingsFromStudio(text, env, client, model) {
+    activity_1.log.info(`Generating embeddings for text of ${text.length} chars with environment ${env}`);
+    return client.environments
+        .embeddings(env, {
+        text,
+        model,
+    })
+        .then((res) => res)
+        .catch((e) => {
+        activity_1.log.error("Error generating embeddings for text", { error: e });
+        throw e;
+    });
+}
+//# sourceMappingURL=generateEmbeddings.js.map

package/lib/cjs/activities/generateEmbeddings.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"generateEmbeddings.js","sourceRoot":"","sources":["../../../src/activities/generateEmbeddings.ts"],"names":[],"mappings":";;AAmDA,gDA8FC;AAhJD,mDAA2C;AAE3C,6CAO0B;AAC1B,wEAAgE;AAChE,4CAAqD;AACrD,gDAAsD;AAEtD,kDAAiD;AAoC1C,KAAK,UAAU,kBAAkB,CACpC,OAA8D;IAE9D,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,YAAY,EAAE,GAC5C,MAAM,IAAA,kCAAa,EAA2B,OAAO,CAAC,CAAC;IAC3D,MAAM,EAAE,KAAK,EAAE,IAAI,EAAE,GAAG,MAAM,CAAC;IAE/B,MAAM,WAAW,GAAG,MAAM,YAAY,EAAE,CAAC;IACzC,MAAM,MAAM,GAAG,WAAW,EAAE,aAAa,CAAC,UAAU,CAAC,IAAI,CAAC,CAAC;IAC3D,IAAI,CAAC,WAAW,EAAE,CAAC;QACf,MAAM,IAAI,iCAAqB,CAAC,mBAAmB,EAAE,CAAC,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC;IAC/E,CAAC;IACD,IAAI,CAAC,MAAM,EAAE,CAAC;QACV,MAAM,IAAI,iCAAqB,CAAC,oCAAoC,EAAE;YAClE,QAAQ;SACX,CAAC,CAAC;IACP,CAAC;IAED,IAAI,CAAC,WAAW,EAAE,CAAC;QACf,MAAM,IAAI,iCAAqB,CAAC,mBAAmB,EAAE,CAAC,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC;IAC/E,CAAC;IAED,IAAI,CAAC,WAAW,EAAE,aAAa,CAAC,UAAU,CAAC,IAAI,CAAC,EAAE,OAAO,EAAE,CAAC;QACxD,cAAG,CAAC,IAAI,CACJ,2CAA2C,IAAI,gBAAgB,WAAW,CAAC,IAAI,KAAK,WAAW,CAAC,SAAS,GAAG,EAC5G,EAAE,MAAM,EAAE,CACb,CAAC;QACF,OAAO;YACH,EAAE,EAAE,QAAQ;YACZ,MAAM,EAAE,SAAS;YACjB,OAAO,EAAE,2CAA2C,IAAI,EAAE;SAC7D,CAAC;IACN,CAAC;IAED,cAAG,CAAC,IAAI,CAAC,GAAG,IAAI,6CAA6C,QAAQ,EAAE,EAAE;QACrE,KAAK;QACL,MAAM;KACT,CAAC,CAAC;IAEH,IAAI,CAAC,MAAM,CAAC,WAAW,EAAE,CAAC;QACtB,MAAM,IAAI,KAAK,CACX,iHAAiH,CACpH,CAAC;IACN,CAAC;IAED,MAAM,QAAQ,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,QAAQ,CAC1C,QAAQ,EACR,8CAA8C,CACjD,CAAC;IAEF,IAAI,CAAC,QAAQ,EAAE,CAAC;QACZ,MAAM,IAAI,iCAAqB,CAAC,oBAAoB,EAAE,CAAC,QAAQ,CAAC,CAAC,CAAC;IACtE,CAAC;IAED,IAAI,CAAC,QAAQ,CAAC,OAAO,EAAE,CAAC;QACpB,MAAM,IAAI,iCAAqB,CAAC,4BAA4B,EAAE,CAAC,QAAQ,CAAC,CAAC,CAAC;IAC9E,CAAC;IAED,IAAI,GAAG,CAAC;IAER,QAAQ,IAAI,EAAE,CAAC;QACX,KAAK,gCAAuB,CAAC,IAAI;YAC7B,GAAG,GAAG,MAAM,sBAAsB,CAAC;gBAC/B,MAAM;gBACN,MAAM;gBACN,QAAQ;gBACR,IAAI;aACP,CAAC,CAAC;YACH,MAAM;QACV,KAAK,gCAAuB,CAAC,UAAU;YACnC,GAAG,GAAG,MAAM,sBAAsB,CAAC;gBAC/B,MAAM;gBACN,MAAM;gBACN,QAAQ;gBACR,IAAI;aACP,CAAC,CAAC;YACH,MAAM;QACV,KAAK,gCAAuB,CAAC,KAAK;YAC9B,GAAG,GAAG,MAAM,uBAAuB,CAAC;gBAChC,MAAM;gBACN,MAAM;gBACN,QAAQ;gBACR,IAAI;aACP,CAAC,CAAC;YACH,MAAM;QACV;YACI,GAAG,GAAG;gBACF,EAAE,EAAE,QAAQ;gBACZ,MAAM,EAAE,QAAQ;gBAChB,OAAO,EAAE,+BAA+B,IAAI,EAAE;aACjD,CAAC;IACV,CAAC;IAED,OAAO,GAAG,CAAC;AACf,CAAC;AAWD,KAAK,UAAU,sBAAsB,CACjC,EAAE,QAAQ,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,EAAmC;IAGnE,IAAI,CAAC,QAAQ,EAAE,CAAC;QACZ,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,OAAO,EAAE,+BAA+B,EAAE,CAAC;IACzE,CAAC;IAED,IACI,IAAI,KAAK,gCAAuB,CAAC,IAAI;QACrC,IAAI,KAAK,gCAAuB,CAAC,UAAU,EAC7C,CAAC;QACC,OAAO;YACH,EAAE,EAAE,QAAQ,CAAC,EAAE;YACf,MAAM,EAAE,QAAQ;YAChB,OAAO,EAAE,+BAA+B,IAAI,EAAE;SACjD,CAAC;IACN,CAAC;IAED,IAAI,IAAI,KAAK,gCAAuB,CAAC,IAAI,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC;QAC1D,OAAO,EAAE,EAAE,EAAE,QAAQ,CAAC,EAAE,EAAE,MAAM,EAAE,QAAQ,EAAE,OAAO,EAAE,eAAe,EAAE,CAAC;IAC3E,CAAC;IACD,IAAI,IAAI,KAAK,gCAAuB,CAAC,UAAU,IAAI,CAAC,QAAQ,EAAE,UAAU,EAAE,CAAC;QACvE,OAAO;YACH,EAAE,EAAE,QAAQ,CAAC,EAAE;YACf,MAAM,EAAE,QAAQ;YAChB,OAAO,EAAE,qBAAqB;SACjC,CAAC;IACN,CAAC;IAED,MAAM,EAAE,WAAW,EAAE,GAAG,MAAM,CAAC;IAE/B,8DAA8D;IAC9D,IAAI,UAAU,GAAwB,SAAS,CAAC;IAChD,IAAI,IAAI,KAAK,gCAAuB,CAAC,IAAI,IAAI,QAAQ,CAAC,IAAI,EAAE,CAAC;QACzD,UAAU,GAAG,IAAA,uBAAW,EAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,KAAK,CAAC;IAClD,CAAC;IAED,MAAM,SAAS,GAAG,MAAM,CAAC,UAAU,IAAI,IAAI,CAAC;IAE5C,kEAAkE;IAClE,cAAG,CAAC,IAAI,CAAC,cAAc,IAAI,4BAA4B,QAAQ,CAAC,EAAE,EAAE,CAAC,CAAC;IACtE,IACI,IAAI,KAAK,gCAAuB,CAAC,IAAI;QACrC,UAAU,KAAK,SAAS;QACxB,UAAU,GAAG,SAAS,EACxB,CAAC;QACC,2CAA2C;QAC3C,cAAG,CAAC,IAAI,CACJ,0BAA0B,IAAI,qCAAqC,UAAU,UAAU,CAC1F,CAAC;QACF,OAAO;YACH,EAAE,EAAE,QAAQ,CAAC,EAAE;YACf,MAAM,EAAE,SAAS;YACjB,OAAO,EAAE,GAAG,IAAI,uDAAuD,UAAU,UAAU;SAC9F,CAAA;IACL,CAAC;SAAM,CAAC;QACJ,cAAG,CAAC,IAAI,CAAC,cAAc,IAAI,0BAA0B,CAAC,CAAC;QAEvD,MAAM,GAAG,GAAG,MAAM,4BAA4B,CAC1C,IAAI,CAAC,SAAS,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC,EAC9B,WAAW,EACX,MAAM,CACT,CAAC;QACF,IAAI,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,EAAE,CAAC;YACtB,OAAO;gBACH,EAAE,EAAE,QAAQ,CAAC,EAAE;gBACf,MAAM,EAAE,QAAQ;gBAChB,OAAO,EAAE,yBAAyB;aACrC,CAAC;QACN,CAAC;QAED,cAAG,CAAC,IAAI,CAAC,GAAG,IAAI,sCAAsC,QAAQ,CAAC,EAAE,EAAE,EAAE;YACjE,GAAG,EAAE,GAAG,CAAC,MAAM,CAAC,MAAM;SACzB,CAAC,CAAC;QACH,MAAM,MAAM,CAAC,OAAO,CAAC,YAAY,CAAC,QAAQ,CAAC,EAAE,EAAE,IAAI,EAAE;YACjD,MAAM,EAAE,GAAG,CAAC,MAAM;YAClB,KAAK,EAAE,GAAG,CAAC,KAAK;YAChB,IAAI,EAAE,QAAQ,CAAC,SAAS;SAC3B,CAAC,CAAC;QAEH,OAAO;YACH,EAAE,EAAE,QAAQ,CAAC,EAAE;YACf,IAAI;YACJ,MAAM,EAAE,WAAW;YACnB,GAAG,EAAE,GAAG,CAAC,MAAM,CAAC,MAAM;SACzB,CAAC;IACN,CAAC;AACL,CAAC;AAED,KAAK,UAAU,uBAAuB,CAAC,EACnC,QAAQ,EACR,MAAM,EACN,IAAI,EACJ,MAAM,GACwB;IAC9B,cAAG,CAAC,IAAI,CAAC,2CAA2C,GAAG,QAAQ,CAAC,EAAE,EAAE;QAChE,OAAO,EAAE,QAAQ,CAAC,OAAO;KAC5B,CAAC,CAAC;IACH,IACI,CAAC,QAAQ,CAAC,OAAO,EAAE,IAAI,EAAE,UAAU,CAAC,QAAQ,CAAC;QAC7C,CAAC,QAAQ,CAAC,OAAO,EAAE,IAAI,EAAE,QAAQ,CAAC,KAAK,CAAC,EAC1C,CAAC;QACC,OAAO;YACH,EAAE,EAAE,QAAQ,CAAC,EAAE;YACf,IAAI;YACJ,MAAM,EAAE,QAAQ;YAChB,OAAO,EAAE,yBAAyB;SACrC,CAAC;IACN,CAAC;IACD,MAAM,EAAE,WAAW,EAAE,KAAK,EAAE,GAAG,MAAM,CAAC;IAEtC,MAAM,MAAM,GAAG,MAAM,MAAM,CAAC,KAAK,CAAC,OAAO,CAAC,YAAY,CAAC,QAAQ,CAAC,EAAE,EAAE;QAChE,MAAM,EAAE,6BAAoB,CAAC,IAAI;QACjC,mBAAmB,EAAE,IAAI;QACzB,QAAQ,EAAE,IAAI;KACjB,CAAC,CAAC;IAEH,IAAI,MAAM,CAAC,MAAM,KAAK,YAAY,EAAE,CAAC;QACjC,MAAM,IAAI,KAAK,CAAC,2CAA2C,CAAC,CAAC;IACjE,CAAC;SAAM,IACH,MAAM,CAAC,MAAM,KAAK,QAAQ;QAC1B,CAAC,MAAM,CAAC,UAAU;QAClB,CAAC,MAAM,CAAC,UAAU,CAAC,MAAM,EAC3B,CAAC;QACC,MAAM,IAAI,iCAAqB,CAAC,4BAA4B,EAAE,CAAC,QAAQ,CAAC,EAAE,CAAC,CAAC,CAAC;IACjF,CAAC;IAED,MAAM,UAAU,GAAG,MAAM,CAAC,UAAU,CAAC;IACrC,IAAI,CAAC,UAAU,EAAE,MAAM,EAAE,CAAC;QACtB,MAAM,IAAI,iCAAqB,CAAC,8BAA8B,EAAE;YAC5D,QAAQ,CAAC,EAAE;SACd,CAAC,CAAC;IACP,CAAC;IAED,MAAM,SAAS,GAAG,UAAU,CAAC,CAAC,CAAC,CAAC;IAChC,MAAM,KAAK,GAAG,MAAM,IAAA,4BAAiB,EAAC,MAAM,EAAE,SAAS,CAAC,CAAC;IAEzD,MAAM,GAAG,GAAG,MAAM,MAAM,CAAC,YAAY;SAChC,UAAU,CAAC,WAAW,EAAE;QACrB,KAAK;QACL,KAAK;KACR,CAAC;SACD,IAAI,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC;SAClB,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE;QACT,cAAG,CAAC,KAAK,CAAC,uCAAuC,EAAE,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,CAAC;QACjE,MAAM,CAAC,CAAC;IACZ,CAAC,CAAC,CAAC;IAEP,IAAI,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,MAAM,EAAE,CAAC;QACtB,OAAO;YACH,EAAE,EAAE,QAAQ,CAAC,EAAE;YACf,MAAM,EAAE,QAAQ;YAChB,OAAO,EAAE,yBAAyB;SACrC,CAAC;IACN,CAAC;IAED,MAAM,MAAM,CAAC,OAAO,CAAC,YAAY,CAC7B,QAAQ,CAAC,EAAE,EACX,gCAAuB,CAAC,KAAK,EAC7B;QACI,MAAM,EAAE,GAAG,CAAC,MAAM;QAClB,KAAK,EAAE,GAAG,CAAC,KAAK;QAChB,IAAI,EAAE,QAAQ,CAAC,SAAS;KAC3B,CACJ,CAAC;IAEF,OAAO;QACH,EAAE,EAAE,QAAQ,CAAC,EAAE;QACf,IAAI;QACJ,MAAM,EAAE,WAAW;QACnB,GAAG,EAAE,GAAG,CAAC,MAAM,CAAC,MAAM;KACzB,CAAC;AACN,CAAC;AAED,KAAK,UAAU,4BAA4B,CACvC,IAAY,EACZ,GAAW,EACX,MAAsB,EACtB,KAAc;IAEd,cAAG,CAAC,IAAI,CACJ,qCAAqC,IAAI,CAAC,MAAM,2BAA2B,GAAG,EAAE,CACnF,CAAC;IAEF,OAAO,MAAM,CAAC,YAAY;SACrB,UAAU,CAAC,GAAG,EAAE;QACb,IAAI;QACJ,KAAK;KACR,CAAC;SACD,IAAI,CAAC,CAAC,GAAG,EAAE,EAAE,CAAC,GAAG,CAAC;SAClB,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE;QACT,cAAG,CAAC,KAAK,CAAC,sCAAsC,EAAE,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,CAAC;QAChE,MAAM,CAAC,CAAC;IACZ,CAAC,CAAC,CAAC;AACX,CAAC"}

package/lib/cjs/activities/generateOrAssignContentType.js

@@ -0,0 +1,125 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.generateOrAssignContentType = generateOrAssignContentType;
+const activity_1 = require("@temporalio/activity");
+const common_1 = require("@vertesia/common");
+const ActivityContext_js_1 = require("../dsl/setup/ActivityContext.js");
+const tokens_js_1 = require("../utils/tokens.js");
+const executeInteraction_js_1 = require("./executeInteraction.js");
+const INT_SELECT_DOCUMENT_TYPE = "sys:SelectDocumentType";
+const INT_GENERATE_METADATA_MODEL = "sys:GenerateMetadataModel";
+async function generateOrAssignContentType(payload) {
+    const context = await (0, ActivityContext_js_1.setupActivity)(payload);
+    const { params, client, objectId } = context;
+    const interactionName = params.interactionNames?.selectDocumentType ?? INT_SELECT_DOCUMENT_TYPE;
+    activity_1.log.info("SelectDocumentType for object: " + objectId, { payload });
+    const object = await client.objects.retrieve(objectId, "+text");
+    //Expects object.type to be null on first ingestion of content
+    //User initiated Content Type change via the Composable UI,
+    //sets object.type to null when they let Composable choose for them.
+    //sets object.type to chosen type (thus non-null) when user picks a type.
+    if (object.type) {
+        activity_1.log.warn(`Object ${objectId} has already a type. Skipping type creation.`);
+        return {
+            status: "skipped",
+            message: "Object already has a type: " + object.type.name,
+        };
+    }
+    if (!object ||
+        (!object.text &&
+            !object.content?.type?.startsWith("image/") &&
+            !object.content?.type?.startsWith("application/pdf"))) {
+        activity_1.log.info(`Object ${objectId} not found or text is empty and not an image`, {
+            object,
+        });
+        return { status: "failed", error: "no-text" };
+    }
+    const types = await client.types.list(undefined, {
+        schema: true,
+    });
+    //make a list of all existing types, and add hints if any
+    const existing_types = types.filter((t) => !["DocumentPart", "Rendition"].includes(t.name));
+    const content = object.text
+        ? (0, tokens_js_1.truncByMaxTokens)(object.text, params.truncate || 30000)
+        : undefined;
+    const getImage = async () => {
+        if (object.content?.type?.includes("pdf") &&
+            object.text?.length &&
+            object.text?.length < 100) {
+            return "store:" + objectId;
+        }
+        if (!object.content?.type?.startsWith("image/")) {
+            return undefined;
+        }
+        const res = await client.objects.getRendition(objectId, {
+            format: common_1.ImageRenditionFormat.jpeg,
+            generate_if_missing: true,
+        });
+        if (!res.renditions?.length && res.status === "generating") {
+            //throw to try again
+            throw new Error(`Rendition for object ${objectId} is in progress`);
+        }
+        else if (res.renditions) {
+            return "store:" + objectId;
+        }
+    };
+    const fileRef = await getImage();
+    activity_1.log.info("Execute SelectDocumentType interaction on content with \nexisting types - passing full types: " +
+        existing_types.filter((t) => !t.tags?.includes("system")));
+    const res = await (0, executeInteraction_js_1.executeInteractionFromActivity)(client, interactionName, params, {
+        existing_types,
+        content,
+        image: fileRef,
+    });
+    const jsonResult = res.result.object();
+    activity_1.log.info("Selected Content Type Result: " + JSON.stringify(jsonResult));
+    //if type is not identified or not present in the database, generate a new type
+    let selectedType = undefined;
+    selectedType = types.find((t) => t.name === jsonResult.document_type);
+    if (!selectedType) {
+        activity_1.log.warn("Document type not identified: starting type generation");
+        const newType = await generateNewType(context, existing_types, content, fileRef);
+        selectedType = { id: newType.id, name: newType.name };
+    }
+    if (!selectedType) {
+        activity_1.log.error("Type not found: ", res.result);
+        throw new Error("Type not found: " + jsonResult.document_type);
+    }
+    //update object with selected type
+    await client.objects.update(objectId, {
+        type: selectedType.id,
+    });
+    return {
+        id: selectedType.id,
+        name: selectedType.name,
+        isNew: !types.find((t) => t.name === selectedType.name),
+    };
+}
+async function generateNewType(context, existing_types, content, fileRef) {
+    const { client, params } = context;
+    const project = await context.fetchProject();
+    const interactionName = params.interactionNames?.generateMetadataModel ??
+        INT_GENERATE_METADATA_MODEL;
+    const genTypeRes = await (0, executeInteraction_js_1.executeInteractionFromActivity)(client, interactionName, params, {
+        existing_types,
+        content: content,
+        human_context: project?.configuration?.human_context ?? undefined,
+        image: fileRef ? fileRef : undefined,
+    });
+    const jsonResult = genTypeRes.result.object();
+    if (!jsonResult.document_type) {
+        activity_1.log.error("No name generated for type", genTypeRes);
+        throw new Error("No name generated for type");
+    }
+    activity_1.log.info("Generated schema for type", jsonResult.metadata_schema);
+    const typeData = {
+        name: jsonResult.document_type,
+        description: jsonResult.document_type_description,
+        object_schema: jsonResult.metadata_schema,
+        is_chunkable: jsonResult.is_chunkable,
+        table_layout: jsonResult.table_layout,
+    };
+    const type = await client.types.create(typeData);
+    return type;
+}
+//# sourceMappingURL=generateOrAssignContentType.js.map

package/lib/cjs/activities/generateOrAssignContentType.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"generateOrAssignContentType.js","sourceRoot":"","sources":["../../../src/activities/generateOrAssignContentType.ts"],"names":[],"mappings":";;AA4CA,kEA+HC;AA3KD,mDAA2C;AAC3C,6CAM0B;AAC1B,wEAGyC;AACzC,kDAAoE;AACpE,mEAGiC;AAEjC,MAAM,wBAAwB,GAAG,wBAAwB,CAAC;AAC1D,MAAM,2BAA2B,GAAG,2BAA2B,CAAC;AAyBzD,KAAK,UAAU,2BAA2B,CAC/C,OAAuE;IAEvE,MAAM,OAAO,GACX,MAAM,IAAA,kCAAa,EAAoC,OAAO,CAAC,CAAC;IAClE,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,OAAO,CAAC;IAE7C,MAAM,eAAe,GACnB,MAAM,CAAC,gBAAgB,EAAE,kBAAkB,IAAI,wBAAwB,CAAC;IAE1E,cAAG,CAAC,IAAI,CAAC,iCAAiC,GAAG,QAAQ,EAAE,EAAE,OAAO,EAAE,CAAC,CAAC;IAEpE,MAAM,MAAM,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,QAAQ,EAAE,OAAO,CAAC,CAAC;IAEhE,8DAA8D;IAC9D,2DAA2D;IAC3D,oEAAoE;IACpE,yEAAyE;IACzE,IAAI,MAAM,CAAC,IAAI,EAAE,CAAC;QAChB,cAAG,CAAC,IAAI,CAAC,UAAU,QAAQ,8CAA8C,CAAC,CAAC;QAC3E,OAAO;YACL,MAAM,EAAE,SAAS;YACjB,OAAO,EAAE,6BAA6B,GAAG,MAAM,CAAC,IAAI,CAAC,IAAI;SAC1D,CAAC;IACJ,CAAC;IAED,IACE,CAAC,MAAM;QACP,CAAC,CAAC,MAAM,CAAC,IAAI;YACX,CAAC,MAAM,CAAC,OAAO,EAAE,IAAI,EAAE,UAAU,CAAC,QAAQ,CAAC;YAC3C,CAAC,MAAM,CAAC,OAAO,EAAE,IAAI,EAAE,UAAU,CAAC,iBAAiB,CAAC,CAAC,EACvD,CAAC;QACD,cAAG,CAAC,IAAI,CAAC,UAAU,QAAQ,8CAA8C,EAAE;YACzE,MAAM;SACP,CAAC,CAAC;QACH,OAAO,EAAE,MAAM,EAAE,QAAQ,EAAE,KAAK,EAAE,SAAS,EAAE,CAAC;IAChD,CAAC;IAED,MAAM,KAAK,GAAG,MAAM,MAAM,CAAC,KAAK,CAAC,IAAI,CAAC,SAAS,EAAE;QAC/C,MAAM,EAAE,IAAI;KACb,CAAC,CAAC;IAEH,yDAAyD;IACzD,MAAM,cAAc,GAAG,KAAK,CAAC,MAAM,CACjC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,cAAc,EAAE,WAAW,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,CACvD,CAAC;IACF,MAAM,OAAO,GAAG,MAAM,CAAC,IAAI;QACzB,CAAC,CAAC,IAAA,4BAAgB,EAAC,MAAM,CAAC,IAAI,EAAE,MAAM,CAAC,QAAQ,IAAI,KAAK,CAAC;QACzD,CAAC,CAAC,SAAS,CAAC;IAEd,MAAM,QAAQ,GAAG,KAAK,IAAI,EAAE;QAC1B,IACE,MAAM,CAAC,OAAO,EAAE,IAAI,EAAE,QAAQ,CAAC,KAAK,CAAC;YACrC,MAAM,CAAC,IAAI,EAAE,MAAM;YACnB,MAAM,CAAC,IAAI,EAAE,MAAM,GAAG,GAAG,EACzB,CAAC;YACD,OAAO,QAAQ,GAAG,QAAQ,CAAC;QAC7B,CAAC;QACD,IAAI,CAAC,MAAM,CAAC,OAAO,EAAE,IAAI,EAAE,UAAU,CAAC,QAAQ,CAAC,EAAE,CAAC;YAChD,OAAO,SAAS,CAAC;QACnB,CAAC;QACD,MAAM,GAAG,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,YAAY,CAAC,QAAQ,EAAE;YACtD,MAAM,EAAE,6BAAoB,CAAC,IAAI;YACjC,mBAAmB,EAAE,IAAI;SAC1B,CAAC,CAAC;QACH,IAAI,CAAC,GAAG,CAAC,UAAU,EAAE,MAAM,IAAI,GAAG,CAAC,MAAM,KAAK,YAAY,EAAE,CAAC;YAC3D,oBAAoB;YACpB,MAAM,IAAI,KAAK,CAAC,wBAAwB,QAAQ,iBAAiB,CAAC,CAAC;QACrE,CAAC;aAAM,IAAI,GAAG,CAAC,UAAU,EAAE,CAAC;YAC1B,OAAO,QAAQ,GAAG,QAAQ,CAAC;QAC7B,CAAC;IACH,CAAC,CAAC;IAEF,MAAM,OAAO,GAAG,MAAM,QAAQ,EAAE,CAAC;IAEjC,cAAG,CAAC,IAAI,CACN,gGAAgG;QAChG,cAAc,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,IAAI,EAAE,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAC1D,CAAC;IAEF,MAAM,GAAG,GAAG,MAAM,IAAA,sDAA8B,EAC9C,MAAM,EACN,eAAe,EACf,MAAM,EACN;QACE,cAAc;QACd,OAAO;QACP,KAAK,EAAE,OAAO;KACf,CACF,CAAC;IAEF,MAAM,UAAU,GAAG,GAAG,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC;IAEvC,cAAG,CAAC,IAAI,CAAC,gCAAgC,GAAG,IAAI,CAAC,SAAS,CAAC,UAAU,CAAC,CAAC,CAAC;IAGxE,+EAA+E;IAC/E,IAAI,YAAY,GAA6C,SAAS,CAAC;IAEvE,YAAY,GAAG,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,UAAU,CAAC,aAAa,CAAC,CAAC;IAEtE,IAAI,CAAC,YAAY,EAAE,CAAC;QAClB,cAAG,CAAC,IAAI,CAAC,wDAAwD,CAAC,CAAC;QACnE,MAAM,OAAO,GAAG,MAAM,eAAe,CACnC,OAAO,EACP,cAAc,EACd,OAAO,EACP,OAAO,CACR,CAAC;QACF,YAAY,GAAG,EAAE,EAAE,EAAE,OAAO,CAAC,EAAE,EAAE,IAAI,EAAE,OAAO,CAAC,IAAI,EAAE,CAAC;IACxD,CAAC;IAED,IAAI,CAAC,YAAY,EAAE,CAAC;QAClB,cAAG,CAAC,KAAK,CAAC,kBAAkB,EAAE,GAAG,CAAC,MAAM,CAAC,CAAC;QAC1C,MAAM,IAAI,KAAK,CAAC,kBAAkB,GAAG,UAAU,CAAC,aAAa,CAAC,CAAC;IACjE,CAAC;IAED,kCAAkC;IAClC,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,EAAE;QACpC,IAAI,EAAE,YAAY,CAAC,EAAE;KACtB,CAAC,CAAC;IAEH,OAAO;QACL,EAAE,EAAE,YAAY,CAAC,EAAE;QACnB,IAAI,EAAE,YAAY,CAAC,IAAI;QACvB,KAAK,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,YAAY,CAAC,IAAI,CAAC;KACxD,CAAC;AACJ,CAAC;AAED,KAAK,UAAU,eAAe,CAC5B,OAA2D,EAC3D,cAAuC,EACvC,OAAgB,EAChB,OAAgB;IAEhB,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,GAAG,OAAO,CAAC;IAEnC,MAAM,OAAO,GAAG,MAAM,OAAO,CAAC,YAAY,EAAE,CAAC;IAC7C,MAAM,eAAe,GACnB,MAAM,CAAC,gBAAgB,EAAE,qBAAqB;QAC9C,2BAA2B,CAAC;IAE9B,MAAM,UAAU,GAAG,MAAM,IAAA,sDAA8B,EACrD,MAAM,EACN,eAAe,EACf,MAAM,EACN;QACE,cAAc;QACd,OAAO,EAAE,OAAO;QAChB,aAAa,EAAE,OAAO,EAAE,aAAa,EAAE,aAAa,IAAI,SAAS;QACjE,KAAK,EAAE,OAAO,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,SAAS;KACrC,CACF,CAAC;IAEF,MAAM,UAAU,GAAG,UAAU,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC;IAE9C,IAAI,CAAC,UAAU,CAAC,aAAa,EAAE,CAAC;QAC9B,cAAG,CAAC,KAAK,CAAC,4BAA4B,EAAE,UAAU,CAAC,CAAC;QACpD,MAAM,IAAI,KAAK,CAAC,4BAA4B,CAAC,CAAC;IAChD,CAAC;IAED,cAAG,CAAC,IAAI,CAAC,2BAA2B,EAAE,UAAU,CAAC,eAAe,CAAC,CAAC;IAClE,MAAM,QAAQ,GAAmC;QAC/C,IAAI,EAAE,UAAU,CAAC,aAAa;QAC9B,WAAW,EAAE,UAAU,CAAC,yBAAyB;QACjD,aAAa,EAAE,UAAU,CAAC,eAAe;QACzC,YAAY,EAAE,UAAU,CAAC,YAAY;QACrC,YAAY,EAAE,UAAU,CAAC,YAAY;KACtC,CAAC;IAEF,MAAM,IAAI,GAAG,MAAM,MAAM,CAAC,KAAK,CAAC,MAAM,CAAC,QAAQ,CAAC,CAAC;IAEjD,OAAO,IAAI,CAAC;AACd,CAAC"}

package/lib/cjs/activities/getObjectFromStore.js

@@ -0,0 +1,20 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.getObjectFromStore = getObjectFromStore;
+const projections_js_1 = require("../dsl/projections.js");
+const ActivityContext_js_1 = require("../dsl/setup/ActivityContext.js");
+/**
+ * We are using a union type for the status parameter since typescript enums breaks the workflow code generation
+ * @param objectId
+ * @param status
+ */
+async function getObjectFromStore(payload) {
+    const { client, params, objectId } = await (0, ActivityContext_js_1.setupActivity)(payload);
+    const obj = await client.objects.retrieve(objectId, params.select);
+    const projection = (0, projections_js_1.projectResult)(payload, params, obj, obj);
+    return {
+        ...projection,
+        id: obj.id,
+    };
+}
+//# sourceMappingURL=getObjectFromStore.js.map

package/lib/cjs/activities/getObjectFromStore.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"getObjectFromStore.js","sourceRoot":"","sources":["../../../src/activities/getObjectFromStore.ts"],"names":[],"mappings":";;AAkBA,gDAYC;AA7BD,0DAAsD;AACtD,wEAAgE;AAWhE;;;;GAIG;AACI,KAAK,UAAU,kBAAkB,CAAC,OAAqD;IAC1F,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,IAAA,kCAAa,EAAkB,OAAO,CAAC,CAAC;IAEnF,MAAM,GAAG,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC,MAAM,CAAC,CAAC;IAEnE,MAAM,UAAU,GAAG,IAAA,8BAAa,EAAC,OAAO,EAAE,MAAM,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;IAE5D,OAAO;QACH,GAAG,UAAU;QACb,EAAE,EAAE,GAAG,CAAC,EAAE;KACb,CAAA;AAEL,CAAC"}