@vertesia/workflow 0.78.0-dev-28b447d → 0.78.0-dev.1

package/lib/cjs/activities/renditions/generateVideoRendition.js

@@ -0,0 +1,200 @@
+"use strict";
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.generateVideoRendition = generateVideoRendition;
+const activity_1 = require("@temporalio/activity");
+const child_process_1 = require("child_process");
+const fs_1 = __importDefault(require("fs"));
+const os_1 = __importDefault(require("os"));
+const path_1 = __importDefault(require("path"));
+const util_1 = require("util");
+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 renditions_js_1 = require("../../utils/renditions.js");
+const execAsync = (0, util_1.promisify)(child_process_1.exec);
+async function getVideoMetadata(videoPath) {
+    try {
+        const command = `ffprobe -v quiet -print_format json -show_format -show_streams "${videoPath}"`;
+        const { stdout } = await execAsync(command);
+        const metadata = JSON.parse(stdout);
+        const videoStream = metadata.streams.find((stream) => stream.codec_type === "video");
+        const duration = parseFloat(metadata.format.duration) || 0;
+        const width = videoStream?.width || 0;
+        const height = videoStream?.height || 0;
+        return { duration, width, height };
+    }
+    catch (error) {
+        activity_1.log.error(`Failed to get video metadata: ${error instanceof Error ? error.message : "Unknown error"}`);
+        throw new Error(`Failed to probe video metadata: ${error instanceof Error ? error.message : "Unknown error"}`);
+    }
+}
+async function generateThumbnail(videoPath, outputDir, timestamp, maxSize) {
+    //pad timestamp to 5 digits as filename
+    const outputFile = path_1.default.join(outputDir, `thumb-${timestamp.toString().padStart(5, "0")}.jpg`);
+    // FFmpeg command to extract thumbnail at specific timestamp
+    // Use proper scale filter syntax: scale=w:h:force_original_aspect_ratio=decrease
+    const scaleFilter = `scale=${maxSize}:${maxSize}:force_original_aspect_ratio=decrease`;
+    const command = [
+        "ffmpeg",
+        "-y", // Overwrite output files
+        "-ss",
+        timestamp.toString(), // Seek to timestamp
+        "-i",
+        `"${videoPath}"`, // Input file
+        "-vframes",
+        "1", // Extract only 1 frame
+        "-vf",
+        `"${scaleFilter}"`, // Scale maintaining aspect ratio
+        "-q:v",
+        "2", // High quality
+        `"${outputFile}"`,
+    ].join(" ");
+    activity_1.log.info(`Generating thumbnail at ${timestamp}s`), { command };
+    try {
+        const { stderr } = await execAsync(command);
+        // Log any warnings from ffmpeg
+        if (stderr && !stderr.includes("frame=")) {
+            activity_1.log.debug(`FFmpeg stderr for thumbnail at ${timestamp}s: ${stderr}`);
+        }
+        // Verify the file was created
+        if (fs_1.default.existsSync(outputFile)) {
+            activity_1.log.debug(`Generated thumbnail at ${timestamp}s`);
+            return outputFile;
+        }
+        else {
+            activity_1.log.warn(`Thumbnail not generated for timestamp ${timestamp}s`);
+            return undefined;
+        }
+    }
+    catch (error) {
+        activity_1.log.error(`Failed to generate thumbnail at ${timestamp}s: ${error instanceof Error ? error.message : "Unknown error"}`);
+        return undefined;
+    }
+}
+async function generateVideoRendition(payload) {
+    const { client, objectId, params: originParams, } = await (0, ActivityContext_js_1.setupActivity)(payload);
+    // Fix: Use maxHeightWidth if max_hw is not provided
+    const params = {
+        ...originParams,
+        max_hw: originParams.max_hw || originParams.maxHeightWidth || 1024, // Default to 1024 if both are missing
+        format: originParams.format || originParams.format_output || "png", // Default to png if format is missing
+    };
+    activity_1.log.info(`Generating video rendition for ${objectId}`, {
+        originParams,
+        params,
+    });
+    const inputObject = await client.objects.retrieve(objectId).catch((err) => {
+        activity_1.log.error(`Failed to retrieve document ${objectId}`, { err });
+        if (err.message.includes("not found")) {
+            throw new errors_js_1.DocumentNotFoundError(`Document ${objectId} not found`, [
+                objectId,
+            ]);
+        }
+        throw err;
+    });
+    if (!params.format) {
+        activity_1.log.error(`Format not found`);
+        throw new errors_js_1.WorkflowParamNotFoundError(`format`);
+    }
+    if (!inputObject.content?.source) {
+        activity_1.log.error(`Document ${objectId} has no source`);
+        throw new errors_js_1.DocumentNotFoundError(`Document ${objectId} has no source`, [
+            objectId,
+        ]);
+    }
+    if (!inputObject.content.type ||
+        !inputObject.content.type?.startsWith("video/")) {
+        activity_1.log.error(`Document ${objectId} is not a video: ${inputObject.content.type}`);
+        throw new errors_js_1.DocumentNotFoundError(`Document ${objectId} is not a video: ${inputObject.content.type}`, [objectId]);
+    }
+    //array of rendition files to upload
+    let renditionPages = [];
+    const videoFile = await (0, blobs_js_1.saveBlobToTempFile)(client, inputObject.content.source);
+    const tempOutputDir = fs_1.default.mkdtempSync(path_1.default.join(os_1.default.tmpdir(), "video-rendition-"));
+    try {
+        // Get video metadata using command line ffprobe
+        const metadata = await getVideoMetadata(videoFile);
+        const duration = metadata.duration;
+        // Calculate optimal number of thumbnails based on video length
+        const calculateThumbnailCount = (videoDuration) => {
+            if (videoDuration <= 60)
+                return 3; // Short videos: 3 thumbnails
+            if (videoDuration <= 300)
+                return 5; // 5min videos: 5 thumbnails
+            if (videoDuration <= 600)
+                return 8; // 10min videos: 8 thumbnails
+            if (videoDuration <= 1800)
+                return 12; // 30min videos: 12 thumbnails
+            if (videoDuration <= 3600)
+                return 16; // 1hr videos: 16 thumbnails
+            return 20; // Longer videos: max 20 thumbnails
+        };
+        const thumbnailCount = calculateThumbnailCount(duration);
+        // Generate evenly spaced timestamps, avoiding very beginning and end
+        const timestamps = [];
+        const startOffset = Math.min(duration * 0.05, 5); // Skip first 5% or 5 seconds
+        const endOffset = Math.min(duration * 0.05, 5); // Skip last 5% or 5 seconds
+        const usableDuration = duration - startOffset - endOffset;
+        for (let i = 0; i < thumbnailCount; i++) {
+            const progress = (i + 1) / (thumbnailCount + 1); // Evenly distribute
+            const timestamp = startOffset + usableDuration * progress;
+            timestamps.push(Math.max(timestamp, 1));
+        }
+        activity_1.log.info(`Generating ${thumbnailCount} thumbnails for ${duration}s video`, {
+            objectId,
+            duration,
+            thumbnailCount,
+            timestamps: timestamps.map((t) => Math.round(t)),
+            tempOutputDir,
+        });
+        // Generate thumbnails using command line ffmpeg
+        const generatedThumbnails = await Promise.all(timestamps.map(async (timestamp) => {
+            return await generateThumbnail(videoFile, tempOutputDir, timestamp, params.max_hw);
+        }));
+        if (generatedThumbnails.length === 0) {
+            activity_1.log.info(`No thumbnails were generated for video ${objectId}`, {
+                objectId,
+                thumbnailCount,
+                tempOutputDir,
+            });
+            throw new Error(`No thumbnails were generated for video ${objectId}`);
+        }
+        renditionPages.push(...generatedThumbnails.filter((thumbnail) => thumbnail !== undefined));
+        activity_1.log.info(`Successfully generated ${generatedThumbnails.length} thumbnails for ${objectId}`, {
+            objectId,
+            generatedCount: generatedThumbnails.length,
+            requestedCount: thumbnailCount,
+        });
+    }
+    catch (error) {
+        activity_1.log.error(`Error generating thumbnails for video: ${error instanceof Error ? error.message : "Unknown error"}`);
+        throw new Error(`Failed to generate thumbnails for video: ${objectId}`);
+    }
+    finally {
+        // Clean up temporary video file
+        try {
+            if (fs_1.default.existsSync(videoFile)) {
+                fs_1.default.unlinkSync(videoFile);
+            }
+        }
+        catch (cleanupError) {
+            activity_1.log.warn(`Failed to cleanup temporary video file: ${videoFile}`);
+        }
+    }
+    if (!inputObject.content?.etag) {
+        activity_1.log.warn(`Document ${objectId} has no etag, using object id as etag`);
+    }
+    const etag = inputObject.content.etag ?? inputObject.id;
+    // Update the final upload call to handle multiple thumbnails
+    const uploaded = await (0, renditions_js_1.uploadRenditionPages)(client, etag, renditionPages, params);
+    return {
+        uploads: uploaded.map((u) => u),
+        format: params.format,
+        thumbnailCount: renditionPages.length,
+        status: "success",
+    };
+}
+//# sourceMappingURL=generateVideoRendition.js.map

package/lib/cjs/activities/renditions/generateVideoRendition.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"generateVideoRendition.js","sourceRoot":"","sources":["../../../../src/activities/renditions/generateVideoRendition.ts"],"names":[],"mappings":";;;;;AAgHA,wDAoLC;AApSD,mDAA2C;AAE3C,iDAAqC;AACrC,4CAAoB;AACpB,4CAAoB;AACpB,gDAAwB;AACxB,+BAAiC;AACjC,2EAAmE;AACnE,+CAAoF;AACpF,mDAA0D;AAC1D,6DAGmC;AAEnC,MAAM,SAAS,GAAG,IAAA,gBAAS,EAAC,oBAAI,CAAC,CAAC;AAelC,KAAK,UAAU,gBAAgB,CAAC,SAAiB;IAC7C,IAAI,CAAC;QACD,MAAM,OAAO,GAAG,mEAAmE,SAAS,GAAG,CAAC;QAChG,MAAM,EAAE,MAAM,EAAE,GAAG,MAAM,SAAS,CAAC,OAAO,CAAC,CAAC;QAC5C,MAAM,QAAQ,GAAG,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,CAAC;QAEpC,MAAM,WAAW,GAAG,QAAQ,CAAC,OAAO,CAAC,IAAI,CACrC,CAAC,MAAW,EAAE,EAAE,CAAC,MAAM,CAAC,UAAU,KAAK,OAAO,CACjD,CAAC;QACF,MAAM,QAAQ,GAAG,UAAU,CAAC,QAAQ,CAAC,MAAM,CAAC,QAAQ,CAAC,IAAI,CAAC,CAAC;QAC3D,MAAM,KAAK,GAAG,WAAW,EAAE,KAAK,IAAI,CAAC,CAAC;QACtC,MAAM,MAAM,GAAG,WAAW,EAAE,MAAM,IAAI,CAAC,CAAC;QAExC,OAAO,EAAE,QAAQ,EAAE,KAAK,EAAE,MAAM,EAAE,CAAC;IACvC,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACb,cAAG,CAAC,KAAK,CACL,iCAAiC,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,eAAe,EAAE,CAC9F,CAAC;QACF,MAAM,IAAI,KAAK,CACX,mCAAmC,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,eAAe,EAAE,CAChG,CAAC;IACN,CAAC;AACL,CAAC;AAED,KAAK,UAAU,iBAAiB,CAC5B,SAAiB,EACjB,SAAiB,EACjB,SAAiB,EACjB,OAAe;IAEf,uCAAuC;IACvC,MAAM,UAAU,GAAG,cAAI,CAAC,IAAI,CACxB,SAAS,EACT,SAAS,SAAS,CAAC,QAAQ,EAAE,CAAC,QAAQ,CAAC,CAAC,EAAE,GAAG,CAAC,MAAM,CACvD,CAAC;IAEF,4DAA4D;IAC5D,iFAAiF;IACjF,MAAM,WAAW,GAAG,SAAS,OAAO,IAAI,OAAO,uCAAuC,CAAC;IAEvF,MAAM,OAAO,GAAG;QACZ,QAAQ;QACR,IAAI,EAAE,yBAAyB;QAC/B,KAAK;QACL,SAAS,CAAC,QAAQ,EAAE,EAAE,oBAAoB;QAC1C,IAAI;QACJ,IAAI,SAAS,GAAG,EAAE,aAAa;QAC/B,UAAU;QACV,GAAG,EAAE,uBAAuB;QAC5B,KAAK;QACL,IAAI,WAAW,GAAG,EAAE,iCAAiC;QACrD,MAAM;QACN,GAAG,EAAE,eAAe;QACpB,IAAI,UAAU,GAAG;KACpB,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IACZ,cAAG,CAAC,IAAI,CAAC,2BAA2B,SAAS,GAAG,CAAC,EAAE,EAAE,OAAO,EAAE,CAAC;IAC/D,IAAI,CAAC;QACD,MAAM,EAAE,MAAM,EAAE,GAAG,MAAM,SAAS,CAAC,OAAO,CAAC,CAAC;QAE5C,+BAA+B;QAC/B,IAAI,MAAM,IAAI,CAAC,MAAM,CAAC,QAAQ,CAAC,QAAQ,CAAC,EAAE,CAAC;YACvC,cAAG,CAAC,KAAK,CACL,kCAAkC,SAAS,MAAM,MAAM,EAAE,CAC5D,CAAC;QACN,CAAC;QAED,8BAA8B;QAC9B,IAAI,YAAE,CAAC,UAAU,CAAC,UAAU,CAAC,EAAE,CAAC;YAC5B,cAAG,CAAC,KAAK,CAAC,0BAA0B,SAAS,GAAG,CAAC,CAAC;YAClD,OAAO,UAAU,CAAC;QACtB,CAAC;aAAM,CAAC;YACJ,cAAG,CAAC,IAAI,CAAC,yCAAyC,SAAS,GAAG,CAAC,CAAC;YAChE,OAAO,SAAS,CAAC;QACrB,CAAC;IACL,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACb,cAAG,CAAC,KAAK,CACL,mCAAmC,SAAS,MAAM,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,eAAe,EAAE,CAC/G,CAAC;QACF,OAAO,SAAS,CAAC;IACrB,CAAC;AACL,CAAC;AAEM,KAAK,UAAU,sBAAsB,CACxC,OAAkE;IAElE,MAAM,EACF,MAAM,EACN,QAAQ,EACR,MAAM,EAAE,YAAY,GACvB,GAAG,MAAM,IAAA,kCAAa,EAA+B,OAAO,CAAC,CAAC;IAE/D,oDAAoD;IACpD,MAAM,MAAM,GAAG;QACX,GAAG,YAAY;QACf,MAAM,EACF,YAAY,CAAC,MAAM,IAAK,YAAoB,CAAC,cAAc,IAAI,IAAI,EAAE,sCAAsC;QAC/G,MAAM,EACF,YAAY,CAAC,MAAM,IAAK,YAAoB,CAAC,aAAa,IAAI,KAAK,EAAE,sCAAsC;KAClH,CAAC;IAEF,cAAG,CAAC,IAAI,CAAC,kCAAkC,QAAQ,EAAE,EAAE;QACnD,YAAY;QACZ,MAAM;KACT,CAAC,CAAC;IAEH,MAAM,WAAW,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,QAAQ,CAAC,QAAQ,CAAC,CAAC,KAAK,CAAC,CAAC,GAAG,EAAE,EAAE;QACtE,cAAG,CAAC,KAAK,CAAC,+BAA+B,QAAQ,EAAE,EAAE,EAAE,GAAG,EAAE,CAAC,CAAC;QAC9D,IAAI,GAAG,CAAC,OAAO,CAAC,QAAQ,CAAC,WAAW,CAAC,EAAE,CAAC;YACpC,MAAM,IAAI,iCAAqB,CAAC,YAAY,QAAQ,YAAY,EAAE;gBAC9D,QAAQ;aACX,CAAC,CAAC;QACP,CAAC;QACD,MAAM,GAAG,CAAC;IACd,CAAC,CAAC,CAAC;IAEH,IAAI,CAAC,MAAM,CAAC,MAAM,EAAE,CAAC;QACjB,cAAG,CAAC,KAAK,CAAC,kBAAkB,CAAC,CAAC;QAC9B,MAAM,IAAI,sCAA0B,CAAC,QAAQ,CAAC,CAAC;IACnD,CAAC;IAED,IAAI,CAAC,WAAW,CAAC,OAAO,EAAE,MAAM,EAAE,CAAC;QAC/B,cAAG,CAAC,KAAK,CAAC,YAAY,QAAQ,gBAAgB,CAAC,CAAC;QAChD,MAAM,IAAI,iCAAqB,CAAC,YAAY,QAAQ,gBAAgB,EAAE;YAClE,QAAQ;SACX,CAAC,CAAC;IACP,CAAC;IAED,IACI,CAAC,WAAW,CAAC,OAAO,CAAC,IAAI;QACzB,CAAC,WAAW,CAAC,OAAO,CAAC,IAAI,EAAE,UAAU,CAAC,QAAQ,CAAC,EACjD,CAAC;QACC,cAAG,CAAC,KAAK,CACL,YAAY,QAAQ,oBAAoB,WAAW,CAAC,OAAO,CAAC,IAAI,EAAE,CACrE,CAAC;QACF,MAAM,IAAI,iCAAqB,CAC3B,YAAY,QAAQ,oBAAoB,WAAW,CAAC,OAAO,CAAC,IAAI,EAAE,EAClE,CAAC,QAAQ,CAAC,CACb,CAAC;IACN,CAAC;IAED,oCAAoC;IACpC,IAAI,cAAc,GAAa,EAAE,CAAC;IAElC,MAAM,SAAS,GAAG,MAAM,IAAA,6BAAkB,EACtC,MAAM,EACN,WAAW,CAAC,OAAO,CAAC,MAAM,CAC7B,CAAC;IACF,MAAM,aAAa,GAAG,YAAE,CAAC,WAAW,CAChC,cAAI,CAAC,IAAI,CAAC,YAAE,CAAC,MAAM,EAAE,EAAE,kBAAkB,CAAC,CAC7C,CAAC;IAEF,IAAI,CAAC;QACD,gDAAgD;QAChD,MAAM,QAAQ,GAAG,MAAM,gBAAgB,CAAC,SAAS,CAAC,CAAC;QACnD,MAAM,QAAQ,GAAG,QAAQ,CAAC,QAAQ,CAAC;QAEnC,+DAA+D;QAC/D,MAAM,uBAAuB,GAAG,CAAC,aAAqB,EAAU,EAAE;YAC9D,IAAI,aAAa,IAAI,EAAE;gBAAE,OAAO,CAAC,CAAC,CAAC,6BAA6B;YAChE,IAAI,aAAa,IAAI,GAAG;gBAAE,OAAO,CAAC,CAAC,CAAC,4BAA4B;YAChE,IAAI,aAAa,IAAI,GAAG;gBAAE,OAAO,CAAC,CAAC,CAAC,6BAA6B;YACjE,IAAI,aAAa,IAAI,IAAI;gBAAE,OAAO,EAAE,CAAC,CAAC,8BAA8B;YACpE,IAAI,aAAa,IAAI,IAAI;gBAAE,OAAO,EAAE,CAAC,CAAC,4BAA4B;YAClE,OAAO,EAAE,CAAC,CAAC,mCAAmC;QAClD,CAAC,CAAC;QAEF,MAAM,cAAc,GAAG,uBAAuB,CAAC,QAAQ,CAAC,CAAC;QAEzD,qEAAqE;QACrE,MAAM,UAAU,GAAa,EAAE,CAAC;QAChC,MAAM,WAAW,GAAG,IAAI,CAAC,GAAG,CAAC,QAAQ,GAAG,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC,6BAA6B;QAC/E,MAAM,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,QAAQ,GAAG,IAAI,EAAE,CAAC,CAAC,CAAC,CAAC,4BAA4B;QAC5E,MAAM,cAAc,GAAG,QAAQ,GAAG,WAAW,GAAG,SAAS,CAAC;QAE1D,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,cAAc,EAAE,CAAC,EAAE,EAAE,CAAC;YACtC,MAAM,QAAQ,GAAG,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,CAAC,cAAc,GAAG,CAAC,CAAC,CAAC,CAAC,oBAAoB;YACrE,MAAM,SAAS,GAAG,WAAW,GAAG,cAAc,GAAG,QAAQ,CAAC;YAC1D,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,SAAS,EAAE,CAAC,CAAC,CAAC,CAAC;QAC5C,CAAC;QAED,cAAG,CAAC,IAAI,CACJ,cAAc,cAAc,mBAAmB,QAAQ,SAAS,EAChE;YACI,QAAQ;YACR,QAAQ;YACR,cAAc;YACd,UAAU,EAAE,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC;YAChD,aAAa;SAChB,CACJ,CAAC;QAEF,gDAAgD;QAChD,MAAM,mBAAmB,GAAG,MAAM,OAAO,CAAC,GAAG,CACzC,UAAU,CAAC,GAAG,CAAC,KAAK,EAAE,SAAS,EAAE,EAAE;YAC/B,OAAO,MAAM,iBAAiB,CAC1B,SAAS,EACT,aAAa,EACb,SAAS,EACT,MAAM,CAAC,MAAM,CAChB,CAAC;QACN,CAAC,CAAC,CACL,CAAC;QAEF,IAAI,mBAAmB,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACnC,cAAG,CAAC,IAAI,CAAC,0CAA0C,QAAQ,EAAE,EAAE;gBAC3D,QAAQ;gBACR,cAAc;gBACd,aAAa;aAChB,CAAC,CAAC;YACH,MAAM,IAAI,KAAK,CACX,0CAA0C,QAAQ,EAAE,CACvD,CAAC;QACN,CAAC;QAED,cAAc,CAAC,IAAI,CACf,GAAG,mBAAmB,CAAC,MAAM,CACzB,CAAC,SAAS,EAAE,EAAE,CAAC,SAAS,KAAK,SAAS,CACzC,CACJ,CAAC;QACF,cAAG,CAAC,IAAI,CACJ,0BAA0B,mBAAmB,CAAC,MAAM,mBAAmB,QAAQ,EAAE,EACjF;YACI,QAAQ;YACR,cAAc,EAAE,mBAAmB,CAAC,MAAM;YAC1C,cAAc,EAAE,cAAc;SACjC,CACJ,CAAC;IACN,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACb,cAAG,CAAC,KAAK,CACL,0CAA0C,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,eAAe,EAAE,CACvG,CAAC;QACF,MAAM,IAAI,KAAK,CAAC,4CAA4C,QAAQ,EAAE,CAAC,CAAC;IAC5E,CAAC;YAAS,CAAC;QACP,gCAAgC;QAChC,IAAI,CAAC;YACD,IAAI,YAAE,CAAC,UAAU,CAAC,SAAS,CAAC,EAAE,CAAC;gBAC3B,YAAE,CAAC,UAAU,CAAC,SAAS,CAAC,CAAC;YAC7B,CAAC;QACL,CAAC;QAAC,OAAO,YAAY,EAAE,CAAC;YACpB,cAAG,CAAC,IAAI,CAAC,2CAA2C,SAAS,EAAE,CAAC,CAAC;QACrE,CAAC;IACL,CAAC;IAED,IAAI,CAAC,WAAW,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC;QAC7B,cAAG,CAAC,IAAI,CAAC,YAAY,QAAQ,uCAAuC,CAAC,CAAC;IAC1E,CAAC;IACD,MAAM,IAAI,GAAG,WAAW,CAAC,OAAO,CAAC,IAAI,IAAI,WAAW,CAAC,EAAE,CAAC;IAExD,6DAA6D;IAC7D,MAAM,QAAQ,GAAG,MAAM,IAAA,oCAAoB,EACvC,MAAM,EACN,IAAI,EACJ,cAAc,EACd,MAAM,CACT,CAAC;IAEF,OAAO;QACH,OAAO,EAAE,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC;QAC/B,MAAM,EAAE,MAAM,CAAC,MAAM;QACrB,cAAc,EAAE,cAAc,CAAC,MAAM;QACrC,MAAM,EAAE,SAAS;KACpB,CAAC;AACN,CAAC"}

package/lib/cjs/activities/setDocumentStatus.js

@@ -0,0 +1,15 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.setDocumentStatus = setDocumentStatus;
+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 setDocumentStatus(payload) {
+    const { client, params, objectId } = await (0, ActivityContext_js_1.setupActivity)(payload);
+    const res = await client.objects.update(objectId, { status: params.status });
+    return res.status;
+}
+//# sourceMappingURL=setDocumentStatus.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"setDocumentStatus.js","sourceRoot":"","sources":["../../../src/activities/setDocumentStatus.ts"],"names":[],"mappings":";;AAiBA,8CAOC;AAvBD,wEAAgE;AAWhE;;;;GAIG;AACI,KAAK,UAAU,iBAAiB,CAAC,OAA6D;IACjG,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,IAAA,kCAAa,EAA0B,OAAO,CAAC,CAAC;IAE3F,MAAM,GAAG,GAAG,MAAM,MAAM,CAAC,OAAO,CAAC,MAAM,CAAC,QAAQ,EAAE,EAAE,MAAM,EAAE,MAAM,CAAC,MAAM,EAAE,CAAC,CAAC;IAE7E,OAAO,GAAG,CAAC,MAAM,CAAC;AAEtB,CAAC"}

package/lib/cjs/conversion/TextractProcessor.js

@@ -0,0 +1,417 @@
+"use strict";
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.TextractProcessor = void 0;
+const client_s3_1 = require("@aws-sdk/client-s3");
+const client_textract_1 = require("@aws-sdk/client-textract");
+const papaparse_1 = __importDefault(require("papaparse"));
+class TextractProcessor {
+    textractClient;
+    s3Client;
+    fileKey;
+    bucket;
+    log;
+    detectImages;
+    /**
+     * Whether or not to include confidence values in CSV output for tables.
+     */
+    includeConfidenceInTables;
+    constructor({ fileKey, region, bucket, credentials, log, detectImages = false, includeConfidenceInTables = false // NEW default = false
+     }) {
+        this.fileKey = fileKey;
+        this.bucket = bucket;
+        this.log = log;
+        this.detectImages = detectImages;
+        this.includeConfidenceInTables = includeConfidenceInTables;
+        this.textractClient = new client_textract_1.TextractClient({
+            region,
+            credentials
+        });
+        this.s3Client = new client_s3_1.S3Client({
+            region,
+            credentials
+        });
+    }
+    getText(result, blocksMap) {
+        let text = '';
+        if (result.Relationships) {
+            for (const relationship of result.Relationships) {
+                if (relationship.Type === 'CHILD') {
+                    for (const childId of relationship.Ids || []) {
+                        const word = blocksMap[childId];
+                        if (word.BlockType === 'WORD') {
+                            const wordText = word.Text || '';
+                            // Example logic to quote numeric text with commas
+                            if (wordText.includes(',') &&
+                                wordText.replace(',', '').match(/^\d+$/)) {
+                                text += `"${wordText}" `;
+                            }
+                            else {
+                                text += `${wordText} `;
+                            }
+                        }
+                        if (word.BlockType === 'SELECTION_ELEMENT' &&
+                            word.SelectionStatus === 'SELECTED') {
+                            text += 'X ';
+                        }
+                    }
+                }
+            }
+        }
+        return text.trim();
+    }
+    isBlockInTable(block, blocksMap) {
+        if (block.BlockType !== 'LINE') {
+            return false;
+        }
+        if (block.Relationships) {
+            for (const relationship of block.Relationships) {
+                if (relationship.Type === 'CHILD') {
+                    for (const childId of relationship.Ids || []) {
+                        const wordBlock = blocksMap[childId];
+                        if (this.isWordInTableCell(wordBlock, blocksMap)) {
+                            return true;
+                        }
+                    }
+                }
+            }
+        }
+        return false;
+    }
+    isWordInTableCell(wordBlock, blocksMap) {
+        // Check if the wordBlock is a descendant of any TABLE->CELL block
+        for (const blockId in blocksMap) {
+            const potentialTable = blocksMap[blockId];
+            if (potentialTable.BlockType === 'TABLE' && potentialTable.Relationships) {
+                for (const relationship of potentialTable.Relationships) {
+                    if (relationship.Type === 'CHILD') {
+                        for (const cellId of relationship.Ids || []) {
+                            const cell = blocksMap[cellId];
+                            if (cell.BlockType === 'CELL' && cell.Relationships) {
+                                for (const cellRel of cell.Relationships) {
+                                    if (cellRel.Type === 'CHILD' &&
+                                        cellRel.Ids?.includes(wordBlock.Id)) {
+                                        return true;
+                                    }
+                                }
+                            }
+                        }
+                    }
+                }
+            }
+        }
+        return false;
+    }
+    /**
+     * NEW: Helper type to store row and column text along with confidence.
+     */
+    getRowsColumnsMap(tableResult, blocksMap) {
+        const rows = [];
+        tableResult.Relationships?.forEach(relationship => {
+            if (relationship.Type === 'CHILD') {
+                relationship.Ids?.forEach(childId => {
+                    const cell = blocksMap[childId];
+                    if (cell.BlockType === 'CELL') {
+                        const rowIndex = cell.RowIndex || 1;
+                        const colIndex = cell.ColumnIndex || 1;
+                        // Expand the array if needed
+                        if (!rows[rowIndex - 1]) {
+                            rows[rowIndex - 1] = [];
+                        }
+                        // Prepare cell text and confidence
+                        const text = this.getText(cell, blocksMap);
+                        const confidence = cell.Confidence || 0;
+                        // If there's a gap, fill it with placeholders
+                        // so that we can safely place text at colIndex - 1
+                        for (let i = rows[rowIndex - 1].length; i < colIndex - 1; i++) {
+                            rows[rowIndex - 1].push({ text: '', confidence: 0 });
+                        }
+                        rows[rowIndex - 1][colIndex - 1] = { text, confidence };
+                    }
+                });
+            }
+        });
+        return { rows };
+    }
+    generateTableCSV(tableResult, blocksMap, _tableIndex, _pageNumber) {
+        const { rows } = this.getRowsColumnsMap(tableResult, blocksMap);
+        let totalConfidence = 0;
+        let cellCount = 0;
+        // Prepare CSV data
+        const csvData = [];
+        for (const row of rows) {
+            const rowData = [];
+            for (const cell of row) {
+                // Add to CSV
+                rowData.push(cell.text.trim());
+                // Accumulate confidence
+                totalConfidence += cell.confidence;
+                cellCount++;
+            }
+            csvData.push(rowData);
+        }
+        // Compute average confidence (or any other method you prefer)
+        const tableConfidence = cellCount > 0 ? (totalConfidence / cellCount) : 0;
+        // Convert to CSV
+        const csv = papaparse_1.default.unparse(csvData, {
+            delimiter: ',',
+            quotes: true,
+            quoteChar: '"',
+            escapeChar: '"',
+            header: false,
+            newline: '\n',
+            skipEmptyLines: false
+        });
+        return { csv, tableConfidence };
+    }
+    async upload(fileBuf) {
+        this.log.info('Uploading file to S3', { fileKey: this.fileKey });
+        const command = new client_s3_1.PutObjectCommand({
+            Bucket: this.bucket,
+            Key: this.fileKey,
+            Body: fileBuf,
+        });
+        await this.s3Client.send(command);
+    }
+    async startAnalysis(s3Key) {
+        const command = new client_textract_1.StartDocumentAnalysisCommand({
+            DocumentLocation: {
+                S3Object: {
+                    Bucket: this.bucket,
+                    Name: s3Key
+                }
+            },
+            FeatureTypes: ["TABLES"]
+        });
+        const response = await this.textractClient.send(command);
+        return response.JobId;
+    }
+    async checkJobStatus(jobId) {
+        const command = new client_textract_1.GetDocumentAnalysisCommand({ JobId: jobId });
+        const response = await this.textractClient.send(command);
+        return response.JobStatus;
+    }
+    getImagePlaceholder(block) {
+        const geometry = block.Geometry?.BoundingBox;
+        if (!geometry)
+            return '';
+        const area = (geometry.Width || 0) * (geometry.Height || 0);
+        if (area < 0.05)
+            return ''; // skip small images
+        const top = geometry.Top || 0;
+        const left = geometry.Left || 0;
+        let position = '';
+        if (top < 0.3)
+            position += 'TOP_';
+        else if (top > 0.7)
+            position += 'BOTTOM_';
+        if (left < 0.3)
+            position += 'LEFT';
+        else if (left > 0.7)
+            position += 'RIGHT';
+        else
+            position += 'CENTER';
+        return `[IMAGE_${position}]\n`;
+    }
+    getIndentationLevel(block) {
+        const left = block.Geometry?.BoundingBox?.Left || 0;
+        if (left < 0.15)
+            return 0;
+        if (left < 0.25)
+            return 1;
+        return 2;
+    }
+    isLikelyHeader(block, prevBlock) {
+        if (!prevBlock)
+            return true;
+        const gap = (block.Geometry?.BoundingBox?.Top || 0) -
+            ((prevBlock.Geometry?.BoundingBox?.Top || 0) +
+                (prevBlock.Geometry?.BoundingBox?.Height || 0));
+        return gap > 0.03;
+    }
+    formatTextBlock(block, prevBlock) {
+        const text = block.Text || '';
+        const indentLevel = this.getIndentationLevel(block);
+        const indent = '    '.repeat(indentLevel);
+        if (this.isLikelyHeader(block, prevBlock)) {
+            return `\n${indent}${text}\n`;
+        }
+        return `${indent}${text}\n`;
+    }
+    shouldMergeLines(prev, current) {
+        const prevBottom = (prev.Geometry?.BoundingBox?.Top || 0)
+            + (prev.Geometry?.BoundingBox?.Height || 0);
+        const currentTop = current.Geometry?.BoundingBox?.Top || 0;
+        const gap = currentTop - prevBottom;
+        // For example, if gap < 0.02, treat them as contiguous
+        if (gap < 0.02) {
+            return true;
+        }
+        return false;
+    }
+    async processResults(jobId) {
+        let nextToken;
+        let allBlocks = [];
+        do {
+            const command = new client_textract_1.GetDocumentAnalysisCommand({
+                JobId: jobId,
+                NextToken: nextToken
+            });
+            const response = await this.textractClient.send(command);
+            allBlocks = allBlocks.concat(response.Blocks || []);
+            nextToken = response.NextToken;
+        } while (nextToken);
+        // Create blocks map
+        const blocksMap = {};
+        for (const block of allBlocks) {
+            blocksMap[block.Id] = block;
+        }
+        // We'll store each page's content in sequence
+        const pageContents = [];
+        let currentPage = null;
+        // We'll keep track of a "current text block" that we're building
+        let currentTextContent = "";
+        let prevLineBlock = null;
+        // Sort by page and vertical position
+        allBlocks.sort((a, b) => {
+            if (a.Page !== b.Page)
+                return (a.Page || 0) - (b.Page || 0);
+            return (a.Geometry?.BoundingBox?.Top || 0) - (b.Geometry?.BoundingBox?.Top || 0);
+        });
+        for (const block of allBlocks) {
+            if (block.BlockType === 'PAGE') {
+                // If we were building a text block, push it before starting a new page
+                if (currentTextContent.trim().length > 0 && currentPage) {
+                    currentPage.blocks.push({
+                        type: 'text',
+                        content: currentTextContent
+                    });
+                }
+                if (currentPage) {
+                    pageContents.push(currentPage);
+                }
+                currentPage = {
+                    pageNumber: block.Page || 0,
+                    blocks: []
+                };
+                currentTextContent = "";
+                prevLineBlock = null;
+            }
+            else if (currentPage && block.Page === currentPage.pageNumber) {
+                // TABLE handling
+                if (block.BlockType === 'TABLE') {
+                    // If there's a pending text block, push it first
+                    if (currentTextContent.trim().length > 0) {
+                        currentPage.blocks.push({
+                            type: 'text',
+                            content: currentTextContent
+                        });
+                        currentTextContent = "";
+                    }
+                    const { csv, tableConfidence } = this.generateTableCSV(block, blocksMap, currentPage.blocks.filter(b => b.type === 'table').length + 1, currentPage.pageNumber);
+                    currentPage.blocks.push({
+                        type: 'table',
+                        content: csv,
+                        confidence: tableConfidence
+                    });
+                    prevLineBlock = null;
+                }
+                // LINE handling (merge or start new)
+                else if (block.BlockType === 'LINE' && !this.isBlockInTable(block, blocksMap)) {
+                    if (prevLineBlock && this.shouldMergeLines(prevLineBlock, block)) {
+                        // If we consider this line to be part of the same paragraph,
+                        // just append the text. We'll call formatTextBlock to get
+                        // indentation/header logic, but we won't add a leading newline.
+                        const formatted = this.formatTextBlock(block, prevLineBlock);
+                        // formatTextBlock might include a leading newline if isLikelyHeader = true
+                        // so you can strip it out if you want them truly "merged" into one paragraph:
+                        const mergedText = formatted.replace(/^\s*\n/, " ");
+                        currentTextContent += " " + mergedText.trim();
+                    }
+                    else {
+                        // If there's an existing text block, push it
+                        if (currentTextContent.trim().length > 0) {
+                            currentPage.blocks.push({
+                                type: 'text',
+                                content: currentTextContent
+                            });
+                        }
+                        // Start a new text block
+                        currentTextContent = this.formatTextBlock(block, prevLineBlock).trim();
+                    }
+                    prevLineBlock = block;
+                }
+                // IMAGES (if detectImages)
+                else if (this.detectImages) {
+                    const geometry = block.Geometry?.BoundingBox;
+                    if (geometry && geometry.Width && geometry.Height) {
+                        const imagePlaceholder = this.getImagePlaceholder(block);
+                        if (imagePlaceholder) {
+                            // If there's a pending text block, push it first
+                            if (currentTextContent.trim().length > 0) {
+                                currentPage.blocks.push({
+                                    type: 'text',
+                                    content: currentTextContent
+                                });
+                                currentTextContent = "";
+                            }
+                            currentPage.blocks.push({
+                                type: 'image',
+                                content: imagePlaceholder,
+                                left: geometry.Left,
+                                top: geometry.Top,
+                                width: geometry.Width,
+                                height: geometry.Height
+                            });
+                        }
+                    }
+                    // No line update to prevLineBlock here
+                }
+            }
+        }
+        // Handle last page
+        if (currentPage) {
+            if (currentTextContent.trim().length > 0) {
+                currentPage.blocks.push({
+                    type: 'text',
+                    content: currentTextContent
+                });
+            }
+            pageContents.push(currentPage);
+        }
+        // Build final output
+        let fullText = '';
+        let imgNumber = 1;
+        let tableNumber = 1;
+        for (const page of pageContents) {
+            fullText += `<page number="${page.pageNumber}">\n`;
+            for (const block of page.blocks) {
+                if (block.type === 'text') {
+                    fullText += `<text>\n${block.content}\n</text>\n\n`;
+                }
+                else if (block.type === 'table') {
+                    const confidenceAttr = block.confidence !== undefined && this.includeConfidenceInTables
+                        ? ` confidence="${block.confidence.toFixed(2)}"`
+                        : '';
+                    fullText += `<table number=${tableNumber++} type="csv" ${confidenceAttr}>\n`;
+                    fullText += `${block.content}\n`;
+                    fullText += `</table>\n\n`;
+                }
+                else if (block.type === 'image') {
+                    // Include geometry if you like
+                    const leftAttr = block.left ? ` left="${block.left.toFixed(4)}"` : '';
+                    const topAttr = block.top ? ` top="${block.top.toFixed(4)}"` : '';
+                    const widthAttr = block.width ? ` width="${block.width.toFixed(4)}"` : '';
+                    const heightAttr = block.height ? ` height="${block.height.toFixed(4)}"` : '';
+                    fullText += `<image id="${imgNumber++}" ${leftAttr}${topAttr}${widthAttr}${heightAttr}>\n${block.content.trim()}\n</image>\n\n`;
+                }
+            }
+            fullText += `</page>\n\n`;
+        }
+        return fullText;
+    }
+}
+exports.TextractProcessor = TextractProcessor;
+//# sourceMappingURL=TextractProcessor.js.map