@vertesia/workflow 0.78.0-dev-28b447d → 0.79.0

package/lib/esm/activities/generateEmbeddings.js

@@ -0,0 +1,355 @@
+import { log } from "@temporalio/activity";
+import { ImageRenditionFormat, SupportedEmbeddingTypes, } from "@vertesia/common";
+import { setupActivity } from "../dsl/setup/ActivityContext.js";
+import { DocumentNotFoundError } from "../errors.js";
+import { fetchBlobAsBase64, md5 } from "../utils/blobs.js";
+import { getContentParts } from "../utils/chunks.js";
+import { countTokens } from "../utils/tokens.js";
+export async function generateEmbeddings(payload) {
+    const { params, client, objectId, fetchProject } = await setupActivity(payload);
+    const { force, type } = params;
+    const projectData = await fetchProject();
+    const config = projectData?.configuration.embeddings[type];
+    if (!projectData) {
+        throw new DocumentNotFoundError("Project not found", [payload.project_id]);
+    }
+    if (!config) {
+        throw new DocumentNotFoundError("Embeddings configuration not found", [
+            objectId,
+        ]);
+    }
+    if (!projectData) {
+        throw new DocumentNotFoundError("Project not found", [payload.project_id]);
+    }
+    if (!projectData?.configuration.embeddings[type]?.enabled) {
+        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 ${type}`,
+        };
+    }
+    log.info(`${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 DocumentNotFoundError("Document not found", [objectId]);
+    }
+    if (!document.content) {
+        throw new DocumentNotFoundError("Document content not found", [objectId]);
+    }
+    let res;
+    switch (type) {
+        case SupportedEmbeddingTypes.text:
+            res = await generateTextEmbeddings({
+                client,
+                config,
+                document,
+                type,
+            });
+            break;
+        case SupportedEmbeddingTypes.properties:
+            res = await generateTextEmbeddings({
+                client,
+                config,
+                document,
+                type,
+            });
+            break;
+        case 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 }, parts) {
+    if (!document) {
+        return { status: "error", message: "document is null or undefined" };
+    }
+    if (type !== SupportedEmbeddingTypes.text &&
+        type !== SupportedEmbeddingTypes.properties) {
+        return {
+            id: document.id,
+            status: "failed",
+            message: `unsupported embedding type: ${type}`,
+        };
+    }
+    if (type === SupportedEmbeddingTypes.text && !document.text) {
+        return { id: document.id, status: "failed", message: "no text found" };
+    }
+    if (type === SupportedEmbeddingTypes.properties && !document?.properties) {
+        return {
+            id: document.id,
+            status: "failed",
+            message: "no properties found",
+        };
+    }
+    const { environment, model } = config;
+    const partDefinitions = parts ?? [];
+    // Count tokens if not already done
+    if (!document.tokens?.count && type === SupportedEmbeddingTypes.text) {
+        log.debug("Updating token count for document: " + document.id);
+        const tokensData = countTokens(document.text);
+        await client.objects.update(document.id, {
+            tokens: {
+                ...tokensData,
+                etag: document.text_etag ?? md5(document.text),
+            },
+        });
+        document.tokens = {
+            ...tokensData,
+            etag: document.text_etag ?? md5(document.text),
+        };
+    }
+    const maxTokens = config.max_tokens ?? 8000;
+    //generate embeddings for the main doc if document isn't too large
+    //if too large, we'll just generate embeddings for the parts
+    //then we can generate embeddings for the main document by averaging the tensors
+    log.info(`Generating ${type} embeddings for document ${document.id}`);
+    if (type === SupportedEmbeddingTypes.text &&
+        document.tokens?.count &&
+        document.tokens?.count > maxTokens) {
+        log.info("Document too large, generating embeddings for parts");
+        if (!document.text) {
+            return {
+                id: document.id,
+                status: "failed",
+                message: "no text found",
+            };
+        }
+        if (!partDefinitions || partDefinitions.length === 0) {
+            log.info("No parts found for document, skipping embeddings generation");
+            return {
+                id: document.id,
+                status: "failed",
+                message: "no parts found",
+            };
+        }
+        log.info("Generating embeddings for parts", {
+            parts: partDefinitions,
+            max_tokens: maxTokens,
+        });
+        const docParts = getContentParts(document.text, partDefinitions);
+        log.info(`Retrieved ${docParts.length} parts`);
+        const start = new Date().getTime();
+        const generatePartEmbeddings = async (partContent, i) => {
+            const localStart = new Date().getTime();
+            try {
+                log.info(`Generating embeddings for part ${i}`, {
+                    text_len: partContent.length,
+                });
+                if (!partContent) {
+                    return {
+                        id: i,
+                        number: i,
+                        result: null,
+                        status: "skipped",
+                        message: "no text found",
+                    };
+                }
+                const e = await generateEmbeddingsFromStudio(partContent, environment, client, model).catch((e) => {
+                    log.error("Error generating embeddings for part " + i, {
+                        text_length: partContent.length,
+                        error: e,
+                    });
+                    return null;
+                });
+                if (!e || !e.values) {
+                    return {
+                        id: i,
+                        number: i,
+                        result: null,
+                        message: "no embeddings generated",
+                    };
+                }
+                if (e.values.length === 0) {
+                    return {
+                        id: i,
+                        number: i,
+                        result: null,
+                        message: "no embeddings generated",
+                    };
+                }
+                log.info(`Generated embeddings for part ${i}`, {
+                    len: e.values.length,
+                    duration: new Date().getTime() - localStart,
+                });
+                return { number: i, result: e };
+            }
+            catch (err) {
+                log.info(`Error generating ${type} embeddings for part ${i} of ${document.id}`, { error: err });
+                return {
+                    number: i,
+                    result: null,
+                    message: "error generating embeddings",
+                    error: err.message,
+                };
+            }
+        };
+        const partEmbeddings = await Promise.all(docParts.map((part, i) => generatePartEmbeddings(part, i)));
+        const validPartEmbeddings = partEmbeddings
+            .filter((e) => e.result !== null)
+            .map((e) => e.result);
+        const averagedEmbedding = computeAttentionEmbedding(validPartEmbeddings.map((e) => e.values));
+        log.info(`Averaged embeddings for document ${document.id} in ${(new Date().getTime() - start) / 1000} seconds`, {
+            len: averagedEmbedding.length,
+            count: validPartEmbeddings.length,
+            max_tokens: maxTokens,
+        });
+        await client.objects.setEmbedding(document.id, type, {
+            values: averagedEmbedding,
+            model: validPartEmbeddings[0].model,
+            etag: document.text_etag,
+        });
+        log.info(`Object ${document.id} embedding set`, {
+            type,
+            len: averagedEmbedding.length,
+        });
+    }
+    else {
+        log.info(`Generating ${type} embeddings for document`);
+        const res = await generateEmbeddingsFromStudio(JSON.stringify(document[type]), environment, client);
+        if (!res || !res.values) {
+            return {
+                id: document.id,
+                status: "failed",
+                message: "no embeddings generated",
+            };
+        }
+        log.info(`${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, }) {
+    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: 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 DocumentNotFoundError("Rendition retrieval failed", [document.id]);
+    }
+    const renditions = resRnd.renditions;
+    if (!renditions?.length) {
+        throw new DocumentNotFoundError("No source found in rendition", [
+            document.id,
+        ]);
+    }
+    const rendition = renditions[0];
+    const image = await fetchBlobAsBase64(client, rendition);
+    const res = await client.environments
+        .embeddings(environment, {
+        image,
+        model,
+    })
+        .then((res) => res)
+        .catch((e) => {
+        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, 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) {
+    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) => {
+        log.error("Error generating embeddings for text", { error: e });
+        throw e;
+    });
+}
+//Simplified attention mechanism
+// This is a naive implementation and should be replaced with a more sophisticated
+// using tensorflow in a specific package
+function computeAttentionEmbedding(chunkEmbeddings) {
+    if (chunkEmbeddings.length === 0)
+        return [];
+    const start = new Date().getTime();
+    // Generate random attention weights
+    const attentionWeights = chunkEmbeddings.map(() => Math.random());
+    // Apply softmax to get attention scores
+    const expWeights = attentionWeights.map((w) => Math.exp(w));
+    const sumExpWeights = expWeights.reduce((sum, val) => sum + val, 0);
+    const attentionScores = expWeights.map((w) => w / sumExpWeights);
+    // Get embedding dimension
+    const embeddingDim = chunkEmbeddings[0].length;
+    // Initialize document embedding
+    const documentEmbedding = new Array(embeddingDim).fill(0);
+    // Weighted sum of embeddings
+    for (let i = 0; i < chunkEmbeddings.length; i++) {
+        for (let j = 0; j < embeddingDim; j++) {
+            documentEmbedding[j] += chunkEmbeddings[i][j] * attentionScores[i];
+        }
+    }
+    const duration = new Date().getTime() - start;
+    console.log(`Computed document embedding in ${duration}ms for ${chunkEmbeddings.length} chunks`);
+    return documentEmbedding;
+}
+//# sourceMappingURL=generateEmbeddings.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"generateEmbeddings.js","sourceRoot":"","sources":["../../../src/activities/generateEmbeddings.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAE3C,OAAO,EAIH,oBAAoB,EAEpB,uBAAuB,GAC1B,MAAM,kBAAkB,CAAC;AAC1B,OAAO,EAAE,aAAa,EAAE,MAAM,iCAAiC,CAAC;AAChE,OAAO,EAAE,qBAAqB,EAAE,MAAM,cAAc,CAAC;AACrD,OAAO,EAAE,iBAAiB,EAAE,GAAG,EAAE,MAAM,mBAAmB,CAAC;AAC3D,OAAO,EAAW,eAAe,EAAE,MAAM,oBAAoB,CAAC;AAC9D,OAAO,EAAE,WAAW,EAAE,MAAM,oBAAoB,CAAC;AAoCjD,MAAM,CAAC,KAAK,UAAU,kBAAkB,CACpC,OAA8D;IAE9D,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,YAAY,EAAE,GAC5C,MAAM,aAAa,CAA2B,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,qBAAqB,CAAC,mBAAmB,EAAE,CAAC,OAAO,CAAC,UAAU,CAAC,CAAC,CAAC;IAC/E,CAAC;IACD,IAAI,CAAC,MAAM,EAAE,CAAC;QACV,MAAM,IAAI,qBAAqB,CAAC,oCAAoC,EAAE;YAClE,QAAQ;SACX,CAAC,CAAC;IACP,CAAC;IAED,IAAI,CAAC,WAAW,EAAE,CAAC;QACf,MAAM,IAAI,qBAAqB,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,GAAG,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,GAAG,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,qBAAqB,CAAC,oBAAoB,EAAE,CAAC,QAAQ,CAAC,CAAC,CAAC;IACtE,CAAC;IAED,IAAI,CAAC,QAAQ,CAAC,OAAO,EAAE,CAAC;QACpB,MAAM,IAAI,qBAAqB,CAAC,4BAA4B,EAAE,CAAC,QAAQ,CAAC,CAAC,CAAC;IAC9E,CAAC;IAED,IAAI,GAAG,CAAC;IAER,QAAQ,IAAI,EAAE,CAAC;QACX,KAAK,uBAAuB,CAAC,IAAI;YAC7B,GAAG,GAAG,MAAM,sBAAsB,CAAC;gBAC/B,MAAM;gBACN,MAAM;gBACN,QAAQ;gBACR,IAAI;aACP,CAAC,CAAC;YACH,MAAM;QACV,KAAK,uBAAuB,CAAC,UAAU;YACnC,GAAG,GAAG,MAAM,sBAAsB,CAAC;gBAC/B,MAAM;gBACN,MAAM;gBACN,QAAQ;gBACR,IAAI;aACP,CAAC,CAAC;YACH,MAAM;QACV,KAAK,uBAAuB,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,EACnE,KAAiB;IAGjB,IAAI,CAAC,QAAQ,EAAE,CAAC;QACZ,OAAO,EAAE,MAAM,EAAE,OAAO,EAAE,OAAO,EAAE,+BAA+B,EAAE,CAAC;IACzE,CAAC;IAED,IACI,IAAI,KAAK,uBAAuB,CAAC,IAAI;QACrC,IAAI,KAAK,uBAAuB,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,uBAAuB,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,uBAAuB,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,KAAK,EAAE,GAAG,MAAM,CAAC;IAEtC,MAAM,eAAe,GAAG,KAAK,IAAI,EAAE,CAAC;IAEpC,mCAAmC;IACnC,IAAI,CAAC,QAAQ,CAAC,MAAM,EAAE,KAAK,IAAI,IAAI,KAAK,uBAAuB,CAAC,IAAI,EAAE,CAAC;QACnE,GAAG,CAAC,KAAK,CAAC,qCAAqC,GAAG,QAAQ,CAAC,EAAE,CAAC,CAAC;QAC/D,MAAM,UAAU,GAAG,WAAW,CAAC,QAAQ,CAAC,IAAK,CAAC,CAAC;QAC/C,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,CAAC,EAAE,EAAE;YACrC,MAAM,EAAE;gBACJ,GAAG,UAAU;gBACb,IAAI,EAAE,QAAQ,CAAC,SAAS,IAAI,GAAG,CAAC,QAAQ,CAAC,IAAK,CAAC;aAClD;SACJ,CAAC,CAAC;QACH,QAAQ,CAAC,MAAM,GAAG;YACd,GAAG,UAAU;YACb,IAAI,EAAE,QAAQ,CAAC,SAAS,IAAI,GAAG,CAAC,QAAQ,CAAC,IAAK,CAAC;SAClD,CAAC;IACN,CAAC;IAED,MAAM,SAAS,GAAG,MAAM,CAAC,UAAU,IAAI,IAAI,CAAC;IAE5C,kEAAkE;IAClE,4DAA4D;IAC5D,gFAAgF;IAChF,GAAG,CAAC,IAAI,CAAC,cAAc,IAAI,4BAA4B,QAAQ,CAAC,EAAE,EAAE,CAAC,CAAC;IACtE,IACI,IAAI,KAAK,uBAAuB,CAAC,IAAI;QACrC,QAAQ,CAAC,MAAM,EAAE,KAAK;QACtB,QAAQ,CAAC,MAAM,EAAE,KAAK,GAAG,SAAS,EACpC,CAAC;QACC,GAAG,CAAC,IAAI,CAAC,qDAAqD,CAAC,CAAC;QAEhE,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,CAAC;YACjB,OAAO;gBACH,EAAE,EAAE,QAAQ,CAAC,EAAE;gBACf,MAAM,EAAE,QAAQ;gBAChB,OAAO,EAAE,eAAe;aAC3B,CAAC;QACN,CAAC;QAED,IAAI,CAAC,eAAe,IAAI,eAAe,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACnD,GAAG,CAAC,IAAI,CACJ,6DAA6D,CAChE,CAAC;YACF,OAAO;gBACH,EAAE,EAAE,QAAQ,CAAC,EAAE;gBACf,MAAM,EAAE,QAAQ;gBAChB,OAAO,EAAE,gBAAgB;aAC5B,CAAC;QACN,CAAC;QAED,GAAG,CAAC,IAAI,CAAC,iCAAiC,EAAE;YACxC,KAAK,EAAE,eAAe;YACtB,UAAU,EAAE,SAAS;SACxB,CAAC,CAAC;QACH,MAAM,QAAQ,GAAG,eAAe,CAAC,QAAQ,CAAC,IAAI,EAAE,eAAe,CAAC,CAAC;QAEjE,GAAG,CAAC,IAAI,CAAC,aAAa,QAAQ,CAAC,MAAM,QAAQ,CAAC,CAAC;QAC/C,MAAM,KAAK,GAAG,IAAI,IAAI,EAAE,CAAC,OAAO,EAAE,CAAC;QACnC,MAAM,sBAAsB,GAAG,KAAK,EAChC,WAAmB,EACnB,CAAS,EACX,EAAE;YACA,MAAM,UAAU,GAAG,IAAI,IAAI,EAAE,CAAC,OAAO,EAAE,CAAC;YACxC,IAAI,CAAC;gBACD,GAAG,CAAC,IAAI,CAAC,kCAAkC,CAAC,EAAE,EAAE;oBAC5C,QAAQ,EAAE,WAAW,CAAC,MAAM;iBAC/B,CAAC,CAAC;gBACH,IAAI,CAAC,WAAW,EAAE,CAAC;oBACf,OAAO;wBACH,EAAE,EAAE,CAAC;wBACL,MAAM,EAAE,CAAC;wBACT,MAAM,EAAE,IAAI;wBACZ,MAAM,EAAE,SAAS;wBACjB,OAAO,EAAE,eAAe;qBAC3B,CAAC;gBACN,CAAC;gBAED,MAAM,CAAC,GAAG,MAAM,4BAA4B,CACxC,WAAW,EACX,WAAW,EACX,MAAM,EACN,KAAK,CACR,CAAC,KAAK,CAAC,CAAC,CAAC,EAAE,EAAE;oBACV,GAAG,CAAC,KAAK,CAAC,uCAAuC,GAAG,CAAC,EAAE;wBACnD,WAAW,EAAE,WAAW,CAAC,MAAM;wBAC/B,KAAK,EAAE,CAAC;qBACX,CAAC,CAAC;oBACH,OAAO,IAAI,CAAC;gBAChB,CAAC,CAAC,CAAC;gBAEH,IAAI,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC;oBAClB,OAAO;wBACH,EAAE,EAAE,CAAC;wBACL,MAAM,EAAE,CAAC;wBACT,MAAM,EAAE,IAAI;wBACZ,OAAO,EAAE,yBAAyB;qBACrC,CAAC;gBACN,CAAC;gBAED,IAAI,CAAC,CAAC,MAAM,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;oBACxB,OAAO;wBACH,EAAE,EAAE,CAAC;wBACL,MAAM,EAAE,CAAC;wBACT,MAAM,EAAE,IAAI;wBACZ,OAAO,EAAE,yBAAyB;qBACrC,CAAC;gBACN,CAAC;gBACD,GAAG,CAAC,IAAI,CAAC,iCAAiC,CAAC,EAAE,EAAE;oBAC3C,GAAG,EAAE,CAAC,CAAC,MAAM,CAAC,MAAM;oBACpB,QAAQ,EAAE,IAAI,IAAI,EAAE,CAAC,OAAO,EAAE,GAAG,UAAU;iBAC9C,CAAC,CAAC;gBAEH,OAAO,EAAE,MAAM,EAAE,CAAC,EAAE,MAAM,EAAE,CAAC,EAAE,CAAC;YACpC,CAAC;YAAC,OAAO,GAAQ,EAAE,CAAC;gBAChB,GAAG,CAAC,IAAI,CACJ,oBAAoB,IAAI,wBAAwB,CAAC,OAAO,QAAQ,CAAC,EAAE,EAAE,EACrE,EAAE,KAAK,EAAE,GAAG,EAAE,CACjB,CAAC;gBACF,OAAO;oBACH,MAAM,EAAE,CAAC;oBACT,MAAM,EAAE,IAAI;oBACZ,OAAO,EAAE,6BAA6B;oBACtC,KAAK,EAAE,GAAG,CAAC,OAAO;iBACrB,CAAC;YACN,CAAC;QACL,CAAC,CAAC;QAEF,MAAM,cAAc,GAAG,MAAM,OAAO,CAAC,GAAG,CACpC,QAAQ,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC,EAAE,EAAE,CAAC,sBAAsB,CAAC,IAAI,EAAE,CAAC,CAAC,CAAC,CAC7D,CAAC;QACF,MAAM,mBAAmB,GAAG,cAAc;aACrC,MAAM,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,KAAK,IAAI,CAAC;aAChC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC;QAC1B,MAAM,iBAAiB,GAAG,yBAAyB,CAC/C,mBAAmB,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,CAC3C,CAAC;QACF,GAAG,CAAC,IAAI,CACJ,oCAAoC,QAAQ,CAAC,EAAE,OAAO,CAAC,IAAI,IAAI,EAAE,CAAC,OAAO,EAAE,GAAG,KAAK,CAAC,GAAG,IAAI,UAAU,EACrG;YACI,GAAG,EAAE,iBAAiB,CAAC,MAAM;YAC7B,KAAK,EAAE,mBAAmB,CAAC,MAAM;YACjC,UAAU,EAAE,SAAS;SACxB,CACJ,CAAC;QACF,MAAM,MAAM,CAAC,OAAO,CAAC,YAAY,CAAC,QAAQ,CAAC,EAAE,EAAE,IAAI,EAAE;YACjD,MAAM,EAAE,iBAAiB;YACzB,KAAK,EAAE,mBAAmB,CAAC,CAAC,CAAC,CAAC,KAAK;YACnC,IAAI,EAAE,QAAQ,CAAC,SAAS;SAC3B,CAAC,CAAC;QACH,GAAG,CAAC,IAAI,CAAC,UAAU,QAAQ,CAAC,EAAE,gBAAgB,EAAE;YAC5C,IAAI;YACJ,GAAG,EAAE,iBAAiB,CAAC,MAAM;SAChC,CAAC,CAAC;IACP,CAAC;SAAM,CAAC;QACJ,GAAG,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,GAAG,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,GAAG,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,oBAAoB,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,qBAAqB,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,qBAAqB,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,iBAAiB,CAAC,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,GAAG,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,uBAAuB,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,GAAG,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,GAAG,CAAC,KAAK,CAAC,sCAAsC,EAAE,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,CAAC;QAChE,MAAM,CAAC,CAAC;IACZ,CAAC,CAAC,CAAC;AACX,CAAC;AAED,gCAAgC;AAChC,kFAAkF;AAClF,yCAAyC;AACzC,SAAS,yBAAyB,CAAC,eAA2B;IAC1D,IAAI,eAAe,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IAE5C,MAAM,KAAK,GAAG,IAAI,IAAI,EAAE,CAAC,OAAO,EAAE,CAAC;IAEnC,oCAAoC;IACpC,MAAM,gBAAgB,GAAG,eAAe,CAAC,GAAG,CAAC,GAAG,EAAE,CAAC,IAAI,CAAC,MAAM,EAAE,CAAC,CAAC;IAElE,wCAAwC;IACxC,MAAM,UAAU,GAAG,gBAAgB,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC;IAC5D,MAAM,aAAa,GAAG,UAAU,CAAC,MAAM,CAAC,CAAC,GAAG,EAAE,GAAG,EAAE,EAAE,CAAC,GAAG,GAAG,GAAG,EAAE,CAAC,CAAC,CAAC;IACpE,MAAM,eAAe,GAAG,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,aAAa,CAAC,CAAC;IAEjE,0BAA0B;IAC1B,MAAM,YAAY,GAAG,eAAe,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC;IAE/C,gCAAgC;IAChC,MAAM,iBAAiB,GAAG,IAAI,KAAK,CAAC,YAAY,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;IAE1D,6BAA6B;IAC7B,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,eAAe,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;QAC9C,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,YAAY,EAAE,CAAC,EAAE,EAAE,CAAC;YACpC,iBAAiB,CAAC,CAAC,CAAC,IAAI,eAAe,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,eAAe,CAAC,CAAC,CAAC,CAAC;QACvE,CAAC;IACL,CAAC;IAED,MAAM,QAAQ,GAAG,IAAI,IAAI,EAAE,CAAC,OAAO,EAAE,GAAG,KAAK,CAAC;IAC9C,OAAO,CAAC,GAAG,CACP,kCAAkC,QAAQ,UAAU,eAAe,CAAC,MAAM,SAAS,CACtF,CAAC;IAEF,OAAO,iBAAiB,CAAC;AAC7B,CAAC"}

package/lib/esm/activities/generateOrAssignContentType.js

@@ -0,0 +1,123 @@
+import { log } from "@temporalio/activity";
+import { ImageRenditionFormat, } from "@vertesia/common";
+import { setupActivity, } from "../dsl/setup/ActivityContext.js";
+import { truncByMaxTokens } from "../utils/tokens.js";
+import { executeInteractionFromActivity, } from "./executeInteraction.js";
+import { parseCompletionResultsToJson } from "@llumiverse/common";
+const INT_SELECT_DOCUMENT_TYPE = "sys:SelectDocumentType";
+const INT_GENERATE_METADATA_MODEL = "sys:GenerateMetadataModel";
+export async function generateOrAssignContentType(payload) {
+    const context = await setupActivity(payload);
+    const { params, client, objectId } = context;
+    const interactionName = params.interactionNames?.selectDocumentType ?? INT_SELECT_DOCUMENT_TYPE;
+    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) {
+        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"))) {
+        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
+        ? 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: 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();
+    log.info("Execute SelectDocumentType interaction on content with \nexisting types - passing full types: " +
+        existing_types.filter((t) => !t.tags?.includes("system")));
+    const res = await executeInteractionFromActivity(client, interactionName, params, {
+        existing_types,
+        content,
+        image: fileRef,
+    });
+    const jsonResult = parseCompletionResultsToJson(res.result);
+    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) {
+        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) {
+        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 executeInteractionFromActivity(client, interactionName, params, {
+        existing_types,
+        content: content,
+        human_context: project?.configuration?.human_context ?? undefined,
+        image: fileRef ? fileRef : undefined,
+    });
+    const jsonResult = parseCompletionResultsToJson(genTypeRes.result);
+    if (!jsonResult.document_type) {
+        log.error("No name generated for type", genTypeRes);
+        throw new Error("No name generated for type");
+    }
+    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/esm/activities/generateOrAssignContentType.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"generateOrAssignContentType.js","sourceRoot":"","sources":["../../../src/activities/generateOrAssignContentType.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAC3C,OAAO,EAKL,oBAAoB,GACrB,MAAM,kBAAkB,CAAC;AAC1B,OAAO,EAEL,aAAa,GACd,MAAM,iCAAiC,CAAC;AACzC,OAAO,EAAgB,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AACpE,OAAO,EAEL,8BAA8B,GAC/B,MAAM,yBAAyB,CAAC;AACjC,OAAO,EAAE,4BAA4B,EAAE,MAAM,oBAAoB,CAAC;AAElE,MAAM,wBAAwB,GAAG,wBAAwB,CAAC;AAC1D,MAAM,2BAA2B,GAAG,2BAA2B,CAAC;AAyBhE,MAAM,CAAC,KAAK,UAAU,2BAA2B,CAC/C,OAAuE;IAEvE,MAAM,OAAO,GACX,MAAM,aAAa,CAAoC,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,GAAG,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,GAAG,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,GAAG,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,gBAAgB,CAAC,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,oBAAoB,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,GAAG,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,8BAA8B,CAC9C,MAAM,EACN,eAAe,EACf,MAAM,EACN;QACE,cAAc;QACd,OAAO;QACP,KAAK,EAAE,OAAO;KACf,CACF,CAAC;IAEF,MAAM,UAAU,GAAG,4BAA4B,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC;IAE5D,GAAG,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,GAAG,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,GAAG,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,8BAA8B,CACrD,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,4BAA4B,CAAC,UAAU,CAAC,MAAM,CAAC,CAAC;IAEnE,IAAI,CAAC,UAAU,CAAC,aAAa,EAAE,CAAC;QAC9B,GAAG,CAAC,KAAK,CAAC,4BAA4B,EAAE,UAAU,CAAC,CAAC;QACpD,MAAM,IAAI,KAAK,CAAC,4BAA4B,CAAC,CAAC;IAChD,CAAC;IAED,GAAG,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/esm/activities/getObjectFromStore.js

@@ -0,0 +1,17 @@
+import { projectResult } from "../dsl/projections.js";
+import { setupActivity } from "../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
+ */
+export async function getObjectFromStore(payload) {
+    const { client, params, objectId } = await setupActivity(payload);
+    const obj = await client.objects.retrieve(objectId, params.select);
+    const projection = projectResult(payload, params, obj, obj);
+    return {
+        ...projection,
+        id: obj.id,
+    };
+}
+//# sourceMappingURL=getObjectFromStore.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"getObjectFromStore.js","sourceRoot":"","sources":["../../../src/activities/getObjectFromStore.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,aAAa,EAAE,MAAM,uBAAuB,CAAC;AACtD,OAAO,EAAE,aAAa,EAAE,MAAM,iCAAiC,CAAC;AAWhE;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,kBAAkB,CAAC,OAAqD;IAC1F,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,aAAa,CAAkB,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,aAAa,CAAC,OAAO,EAAE,MAAM,EAAE,GAAG,EAAE,GAAG,CAAC,CAAC;IAE5D,OAAO;QACH,GAAG,UAAU;QACb,EAAE,EAAE,GAAG,CAAC,EAAE;KACb,CAAA;AAEL,CAAC"}

package/lib/esm/activities/handleError.js

@@ -0,0 +1,19 @@
+import { ContentObjectStatus } from "@vertesia/common";
+import { setupActivity } from "../dsl/setup/ActivityContext.js";
+import { log } from "@temporalio/activity";
+export async function handleDslError(payload) {
+    const { client, params, objectId } = await setupActivity(payload);
+    const isIntake = payload.workflow_name === "StandardDocumentIntake" || payload.workflow_name === "StandardImageIntake";
+    if (!isIntake) {
+        log.warn(`Workflow execution failed, but no error handler registered for this workflow: ${payload.workflow_name}`, { error: params.errorMessage });
+        return;
+    }
+    try {
+        await client.objects.update(objectId, { status: ContentObjectStatus.failed });
+    }
+    catch (e) {
+        log.error("Failed to handle error", { error: e });
+    }
+    return;
+}
+//# sourceMappingURL=handleError.js.map

package/lib/esm/activities/handleError.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"handleError.js","sourceRoot":"","sources":["../../../src/activities/handleError.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,mBAAmB,EAA+B,MAAM,kBAAkB,CAAC;AACpF,OAAO,EAAE,aAAa,EAAE,MAAM,iCAAiC,CAAC;AAChE,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAA;AAM1C,MAAM,CAAC,KAAK,UAAU,cAAc,CAAC,OAA0D;IAC3F,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,aAAa,CAAuB,OAAO,CAAC,CAAC;IACxF,MAAM,QAAQ,GAAG,OAAO,CAAC,aAAa,KAAK,wBAAwB,IAAI,OAAO,CAAC,aAAa,KAAK,qBAAqB,CAAC;IACvH,IAAI,CAAC,QAAQ,EAAE,CAAC;QACZ,GAAG,CAAC,IAAI,CAAC,iFAAiF,OAAO,CAAC,aAAa,EAAE,EAC7G,EAAE,KAAK,EAAE,MAAM,CAAC,YAAY,EAAE,CACjC,CAAC;QACF,OAAO;IACX,CAAC;IAED,IAAI,CAAC;QACD,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,EAAE,EAAE,MAAM,EAAE,mBAAmB,CAAC,MAAM,EAAE,CAAC,CAAC;IAClF,CAAC;IAAC,OAAO,CAAC,EAAE,CAAC;QACT,GAAG,CAAC,KAAK,CAAC,wBAAwB,EAAE,EAAE,KAAK,EAAE,CAAC,EAAE,CAAC,CAAC;IACtD,CAAC;IACD,OAAO;AACX,CAAC"}

package/lib/esm/activities/index-dsl.js

@@ -0,0 +1,21 @@
+// Export here DSL activities
+export { createDocumentTypeFromInteractionRun } from "./advanced/createDocumentTypeFromInteractionRun.js";
+export { createOrUpdateDocumentFromInteractionRun } from "./advanced/createOrUpdateDocumentFromInteractionRun.js";
+export { updateDocumentFromInteractionRun } from "./advanced/updateDocumentFromInteractionRun.js";
+export { chunkDocument } from "./chunkDocument.js";
+export { createPdfDocumentFromSource } from "./createDocumentFromOther.js";
+export { executeInteraction } from "./executeInteraction.js";
+export { extractDocumentText } from "./extractDocumentText.js";
+export { generateDocumentProperties } from "./generateDocumentProperties.js";
+export { generateEmbeddings } from "./generateEmbeddings.js";
+export { generateImageRendition } from "./renditions/generateImageRendition.js";
+export { generateVideoRendition } from "./renditions/generateVideoRendition.js";
+export { generateOrAssignContentType } from "./generateOrAssignContentType.js";
+export { getObjectFromStore } from "./getObjectFromStore.js";
+export { handleDslError } from "./handleError.js";
+export { convertPdfToStructuredText } from "./media/processPdfWithTextract.js";
+export { transcribeMedia } from "./media/transcribeMediaWithGladia.js";
+export { notifyWebhook } from "./notifyWebhook.js";
+export { setDocumentStatus } from "./setDocumentStatus.js";
+export { checkRateLimit } from "./rateLimiter.js";
+//# sourceMappingURL=index-dsl.js.map

package/lib/esm/activities/index-dsl.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index-dsl.js","sourceRoot":"","sources":["../../../src/activities/index-dsl.ts"],"names":[],"mappings":"AAAA,6BAA6B;AAC7B,OAAO,EAAE,oCAAoC,EAAE,MAAM,oDAAoD,CAAC;AAC1G,OAAO,EAAE,wCAAwC,EAAE,MAAM,wDAAwD,CAAC;AAClH,OAAO,EAAE,gCAAgC,EAAE,MAAM,gDAAgD,CAAC;AAClG,OAAO,EAAE,aAAa,EAAE,MAAM,oBAAoB,CAAC;AACnD,OAAO,EAAE,2BAA2B,EAAE,MAAM,8BAA8B,CAAC;AAC3E,OAAO,EAAE,kBAAkB,EAAE,MAAM,yBAAyB,CAAC;AAC7D,OAAO,EAAE,mBAAmB,EAAE,MAAM,0BAA0B,CAAC;AAC/D,OAAO,EAAE,0BAA0B,EAAE,MAAM,iCAAiC,CAAC;AAC7E,OAAO,EAAE,kBAAkB,EAAE,MAAM,yBAAyB,CAAC;AAC7D,OAAO,EAAE,sBAAsB,EAAE,MAAM,wCAAwC,CAAC;AAChF,OAAO,EAAE,sBAAsB,EAAE,MAAM,wCAAwC,CAAC;AAChF,OAAO,EAAE,2BAA2B,EAAE,MAAM,kCAAkC,CAAC;AAC/E,OAAO,EAAE,kBAAkB,EAAE,MAAM,yBAAyB,CAAC;AAC7D,OAAO,EAAE,cAAc,EAAE,MAAM,kBAAkB,CAAC;AAClD,OAAO,EAAE,0BAA0B,EAAE,MAAM,mCAAmC,CAAC;AAC/E,OAAO,EAAE,eAAe,EAAE,MAAM,sCAAsC,CAAC;AACvE,OAAO,EAAE,aAAa,EAAE,MAAM,oBAAoB,CAAC;AACnD,OAAO,EAAE,iBAAiB,EAAE,MAAM,wBAAwB,CAAC;AAC3D,OAAO,EAAE,cAAc,EAAE,MAAM,kBAAkB,CAAC"}

package/lib/esm/activities/index.js

@@ -0,0 +1,5 @@
+/**
+ * Here we export all activities to be registered with the temporal worker
+ */
+export * from "./index-dsl.js";
+//# sourceMappingURL=index.js.map

package/lib/esm/activities/index.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"index.js","sourceRoot":"","sources":["../../../src/activities/index.ts"],"names":[],"mappings":"AAAA;;GAEG;AACH,cAAc,gBAAgB,CAAC"}