@vertesia/workflow 0.82.1 → 0.82.3

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

@@ -0,0 +1 @@
+{"version":3,"file":"generateVideoRendition.js","sourceRoot":"","sources":["../../../../src/activities/renditions/generateVideoRendition.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAE3C,OAAO,EAAE,IAAI,EAAE,MAAM,eAAe,CAAC;AACrC,OAAO,EAAE,MAAM,IAAI,CAAC;AACpB,OAAO,EAAE,MAAM,IAAI,CAAC;AACpB,OAAO,IAAI,MAAM,MAAM,CAAC;AACxB,OAAO,EAAE,SAAS,EAAE,MAAM,MAAM,CAAC;AACjC,OAAO,EAAE,aAAa,EAAE,MAAM,oCAAoC,CAAC;AACnE,OAAO,EAAE,qBAAqB,EAAE,0BAA0B,EAAE,MAAM,iBAAiB,CAAC;AACpF,OAAO,EAAE,kBAAkB,EAAE,MAAM,sBAAsB,CAAC;AAC1D,OAAO,EAEH,oBAAoB,GACvB,MAAM,2BAA2B,CAAC;AAEnC,MAAM,SAAS,GAAG,SAAS,CAAC,IAAI,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,GAAG,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,IAAI,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,GAAG,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,GAAG,CAAC,KAAK,CACL,kCAAkC,SAAS,MAAM,MAAM,EAAE,CAC5D,CAAC;QACN,CAAC;QAED,8BAA8B;QAC9B,IAAI,EAAE,CAAC,UAAU,CAAC,UAAU,CAAC,EAAE,CAAC;YAC5B,GAAG,CAAC,KAAK,CAAC,0BAA0B,SAAS,GAAG,CAAC,CAAC;YAClD,OAAO,UAAU,CAAC;QACtB,CAAC;aAAM,CAAC;YACJ,GAAG,CAAC,IAAI,CAAC,yCAAyC,SAAS,GAAG,CAAC,CAAC;YAChE,OAAO,SAAS,CAAC;QACrB,CAAC;IACL,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACb,GAAG,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;AAED,MAAM,CAAC,KAAK,UAAU,sBAAsB,CACxC,OAAkE;IAElE,MAAM,EACF,MAAM,EACN,QAAQ,EACR,MAAM,EAAE,YAAY,GACvB,GAAG,MAAM,aAAa,CAA+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,GAAG,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,GAAG,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,qBAAqB,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,GAAG,CAAC,KAAK,CAAC,kBAAkB,CAAC,CAAC;QAC9B,MAAM,IAAI,0BAA0B,CAAC,QAAQ,CAAC,CAAC;IACnD,CAAC;IAED,IAAI,CAAC,WAAW,CAAC,OAAO,EAAE,MAAM,EAAE,CAAC;QAC/B,GAAG,CAAC,KAAK,CAAC,YAAY,QAAQ,gBAAgB,CAAC,CAAC;QAChD,MAAM,IAAI,qBAAqB,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,GAAG,CAAC,KAAK,CACL,YAAY,QAAQ,oBAAoB,WAAW,CAAC,OAAO,CAAC,IAAI,EAAE,CACrE,CAAC;QACF,MAAM,IAAI,qBAAqB,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,kBAAkB,CACtC,MAAM,EACN,WAAW,CAAC,OAAO,CAAC,MAAM,CAC7B,CAAC;IACF,MAAM,aAAa,GAAG,EAAE,CAAC,WAAW,CAChC,IAAI,CAAC,IAAI,CAAC,EAAE,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,GAAG,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,GAAG,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,GAAG,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,GAAG,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,EAAE,CAAC,UAAU,CAAC,SAAS,CAAC,EAAE,CAAC;gBAC3B,EAAE,CAAC,UAAU,CAAC,SAAS,CAAC,CAAC;YAC7B,CAAC;QACL,CAAC;QAAC,OAAO,YAAY,EAAE,CAAC;YACpB,GAAG,CAAC,IAAI,CAAC,2CAA2C,SAAS,EAAE,CAAC,CAAC;QACrE,CAAC;IACL,CAAC;IAED,IAAI,CAAC,WAAW,CAAC,OAAO,EAAE,IAAI,EAAE,CAAC;QAC7B,GAAG,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,oBAAoB,CACvC,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/esm/activities/setDocumentStatus.js

@@ -0,0 +1,12 @@
+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 setDocumentStatus(payload) {
+    const { client, params, objectId } = await setupActivity(payload);
+    const res = await client.objects.update(objectId, { status: params.status });
+    return res.status;
+}
+//# sourceMappingURL=setDocumentStatus.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"setDocumentStatus.js","sourceRoot":"","sources":["../../../src/activities/setDocumentStatus.ts"],"names":[],"mappings":"AACA,OAAO,EAAE,aAAa,EAAE,MAAM,iCAAiC,CAAC;AAWhE;;;;GAIG;AACH,MAAM,CAAC,KAAK,UAAU,iBAAiB,CAAC,OAA6D;IACjG,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,QAAQ,EAAE,GAAG,MAAM,aAAa,CAA0B,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/esm/conversion/TextractProcessor.js

@@ -0,0 +1,410 @@
+import { PutObjectCommand, S3Client } from "@aws-sdk/client-s3";
+import { GetDocumentAnalysisCommand, StartDocumentAnalysisCommand, TextractClient } from "@aws-sdk/client-textract";
+import Papa from 'papaparse';
+export 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 TextractClient({
+            region,
+            credentials
+        });
+        this.s3Client = new 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 = Papa.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 PutObjectCommand({
+            Bucket: this.bucket,
+            Key: this.fileKey,
+            Body: fileBuf,
+        });
+        await this.s3Client.send(command);
+    }
+    async startAnalysis(s3Key) {
+        const command = new 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 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 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;
+    }
+}
+//# sourceMappingURL=TextractProcessor.js.map

package/lib/esm/conversion/TextractProcessor.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"TextractProcessor.js","sourceRoot":"","sources":["../../../src/conversion/TextractProcessor.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,gBAAgB,EAAE,QAAQ,EAAE,MAAM,oBAAoB,CAAC;AAEhE,OAAO,EACH,0BAA0B,EAC1B,4BAA4B,EAC5B,cAAc,EACjB,MAAM,0BAA0B,CAAC;AAElC,OAAO,IAAI,MAAM,WAAW,CAAC;AAmC7B,MAAM,OAAO,iBAAiB;IAClB,cAAc,CAAiB;IAC/B,QAAQ,CAAW;IACnB,OAAO,CAAS;IAChB,MAAM,CAAS;IACf,GAAG,CAAM;IACT,YAAY,CAAU;IAC9B;;OAEG;IACK,yBAAyB,CAAU;IAE3C,YAAY,EACR,OAAO,EACP,MAAM,EACN,MAAM,EACN,WAAW,EACX,GAAG,EACH,YAAY,GAAG,KAAK,EACpB,yBAAyB,GAAG,KAAK,CAAE,sBAAsB;MAClC;QACvB,IAAI,CAAC,OAAO,GAAG,OAAO,CAAC;QACvB,IAAI,CAAC,MAAM,GAAG,MAAM,CAAC;QACrB,IAAI,CAAC,GAAG,GAAG,GAAG,CAAC;QACf,IAAI,CAAC,YAAY,GAAG,YAAY,CAAC;QACjC,IAAI,CAAC,yBAAyB,GAAG,yBAAyB,CAAC;QAE3D,IAAI,CAAC,cAAc,GAAG,IAAI,cAAc,CAAC;YACrC,MAAM;YACN,WAAW;SACd,CAAC,CAAC;QACH,IAAI,CAAC,QAAQ,GAAG,IAAI,QAAQ,CAAC;YACzB,MAAM;YACN,WAAW;SACd,CAAC,CAAC;IACP,CAAC;IAEO,OAAO,CAAC,MAAa,EAAE,SAAoB;QAC/C,IAAI,IAAI,GAAG,EAAE,CAAC;QACd,IAAI,MAAM,CAAC,aAAa,EAAE,CAAC;YACvB,KAAK,MAAM,YAAY,IAAI,MAAM,CAAC,aAAa,EAAE,CAAC;gBAC9C,IAAI,YAAY,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;oBAChC,KAAK,MAAM,OAAO,IAAI,YAAY,CAAC,GAAG,IAAI,EAAE,EAAE,CAAC;wBAC3C,MAAM,IAAI,GAAG,SAAS,CAAC,OAAO,CAAC,CAAC;wBAChC,IAAI,IAAI,CAAC,SAAS,KAAK,MAAM,EAAE,CAAC;4BAC5B,MAAM,QAAQ,GAAG,IAAI,CAAC,IAAI,IAAI,EAAE,CAAC;4BACjC,kDAAkD;4BAClD,IAAI,QAAQ,CAAC,QAAQ,CAAC,GAAG,CAAC;gCACtB,QAAQ,CAAC,OAAO,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,EAAE,CAAC;gCAC3C,IAAI,IAAI,IAAI,QAAQ,IAAI,CAAC;4BAC7B,CAAC;iCAAM,CAAC;gCACJ,IAAI,IAAI,GAAG,QAAQ,GAAG,CAAC;4BAC3B,CAAC;wBACL,CAAC;wBACD,IACI,IAAI,CAAC,SAAS,KAAK,mBAAmB;4BACtC,IAAI,CAAC,eAAe,KAAK,UAAU,EACrC,CAAC;4BACC,IAAI,IAAI,IAAI,CAAC;wBACjB,CAAC;oBACL,CAAC;gBACL,CAAC;YACL,CAAC;QACL,CAAC;QACD,OAAO,IAAI,CAAC,IAAI,EAAE,CAAC;IACvB,CAAC;IAEO,cAAc,CAAC,KAAY,EAAE,SAAoB;QACrD,IAAI,KAAK,CAAC,SAAS,KAAK,MAAM,EAAE,CAAC;YAC7B,OAAO,KAAK,CAAC;QACjB,CAAC;QACD,IAAI,KAAK,CAAC,aAAa,EAAE,CAAC;YACtB,KAAK,MAAM,YAAY,IAAI,KAAK,CAAC,aAAa,EAAE,CAAC;gBAC7C,IAAI,YAAY,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;oBAChC,KAAK,MAAM,OAAO,IAAI,YAAY,CAAC,GAAG,IAAI,EAAE,EAAE,CAAC;wBAC3C,MAAM,SAAS,GAAG,SAAS,CAAC,OAAO,CAAC,CAAC;wBACrC,IAAI,IAAI,CAAC,iBAAiB,CAAC,SAAS,EAAE,SAAS,CAAC,EAAE,CAAC;4BAC/C,OAAO,IAAI,CAAC;wBAChB,CAAC;oBACL,CAAC;gBACL,CAAC;YACL,CAAC;QACL,CAAC;QACD,OAAO,KAAK,CAAC;IACjB,CAAC;IAEO,iBAAiB,CAAC,SAAgB,EAAE,SAAoB;QAC5D,kEAAkE;QAClE,KAAK,MAAM,OAAO,IAAI,SAAS,EAAE,CAAC;YAC9B,MAAM,cAAc,GAAG,SAAS,CAAC,OAAO,CAAC,CAAC;YAC1C,IAAI,cAAc,CAAC,SAAS,KAAK,OAAO,IAAI,cAAc,CAAC,aAAa,EAAE,CAAC;gBACvE,KAAK,MAAM,YAAY,IAAI,cAAc,CAAC,aAAa,EAAE,CAAC;oBACtD,IAAI,YAAY,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;wBAChC,KAAK,MAAM,MAAM,IAAI,YAAY,CAAC,GAAG,IAAI,EAAE,EAAE,CAAC;4BAC1C,MAAM,IAAI,GAAG,SAAS,CAAC,MAAM,CAAC,CAAC;4BAC/B,IAAI,IAAI,CAAC,SAAS,KAAK,MAAM,IAAI,IAAI,CAAC,aAAa,EAAE,CAAC;gCAClD,KAAK,MAAM,OAAO,IAAI,IAAI,CAAC,aAAa,EAAE,CAAC;oCACvC,IACI,OAAO,CAAC,IAAI,KAAK,OAAO;wCACxB,OAAO,CAAC,GAAG,EAAE,QAAQ,CAAC,SAAS,CAAC,EAAG,CAAC,EACtC,CAAC;wCACC,OAAO,IAAI,CAAC;oCAChB,CAAC;gCACL,CAAC;4BACL,CAAC;wBACL,CAAC;oBACL,CAAC;gBACL,CAAC;YACL,CAAC;QACL,CAAC;QACD,OAAO,KAAK,CAAC;IACjB,CAAC;IAED;;OAEG;IACK,iBAAiB,CACrB,WAAkB,EAClB,SAAoB;QAIpB,MAAM,IAAI,GAAuD,EAAE,CAAC;QAEpE,WAAW,CAAC,aAAa,EAAE,OAAO,CAAC,YAAY,CAAC,EAAE;YAC9C,IAAI,YAAY,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;gBAChC,YAAY,CAAC,GAAG,EAAE,OAAO,CAAC,OAAO,CAAC,EAAE;oBAChC,MAAM,IAAI,GAAG,SAAS,CAAC,OAAO,CAAC,CAAC;oBAChC,IAAI,IAAI,CAAC,SAAS,KAAK,MAAM,EAAE,CAAC;wBAC5B,MAAM,QAAQ,GAAG,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC;wBACpC,MAAM,QAAQ,GAAG,IAAI,CAAC,WAAW,IAAI,CAAC,CAAC;wBAEvC,6BAA6B;wBAC7B,IAAI,CAAC,IAAI,CAAC,QAAQ,GAAG,CAAC,CAAC,EAAE,CAAC;4BACtB,IAAI,CAAC,QAAQ,GAAG,CAAC,CAAC,GAAG,EAAE,CAAC;wBAC5B,CAAC;wBAED,mCAAmC;wBACnC,MAAM,IAAI,GAAG,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,SAAS,CAAC,CAAC;wBAC3C,MAAM,UAAU,GAAG,IAAI,CAAC,UAAU,IAAI,CAAC,CAAC;wBAExC,8CAA8C;wBAC9C,mDAAmD;wBACnD,KAAK,IAAI,CAAC,GAAG,IAAI,CAAC,QAAQ,GAAG,CAAC,CAAC,CAAC,MAAM,EAAE,CAAC,GAAG,QAAQ,GAAG,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC;4BAC5D,IAAI,CAAC,QAAQ,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,EAAE,IAAI,EAAE,EAAE,EAAE,UAAU,EAAE,CAAC,EAAE,CAAC,CAAC;wBACzD,CAAC;wBACD,IAAI,CAAC,QAAQ,GAAG,CAAC,CAAC,CAAC,QAAQ,GAAG,CAAC,CAAC,GAAG,EAAE,IAAI,EAAE,UAAU,EAAE,CAAC;oBAC5D,CAAC;gBACL,CAAC,CAAC,CAAC;YACP,CAAC;QACL,CAAC,CAAC,CAAC;QAEH,OAAO,EAAE,IAAI,EAAE,CAAC;IACpB,CAAC;IAEO,gBAAgB,CACpB,WAAkB,EAClB,SAAoB,EACpB,WAAmB,EACnB,WAAmB;QAEnB,MAAM,EAAE,IAAI,EAAE,GAAG,IAAI,CAAC,iBAAiB,CAAC,WAAW,EAAE,SAAS,CAAC,CAAC;QAEhE,IAAI,eAAe,GAAG,CAAC,CAAC;QACxB,IAAI,SAAS,GAAG,CAAC,CAAC;QAElB,mBAAmB;QACnB,MAAM,OAAO,GAAe,EAAE,CAAC;QAC/B,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;YACrB,MAAM,OAAO,GAAa,EAAE,CAAC;YAC7B,KAAK,MAAM,IAAI,IAAI,GAAG,EAAE,CAAC;gBACrB,aAAa;gBACb,OAAO,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,CAAC;gBAC/B,wBAAwB;gBACxB,eAAe,IAAI,IAAI,CAAC,UAAU,CAAC;gBACnC,SAAS,EAAE,CAAC;YAChB,CAAC;YACD,OAAO,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QAC1B,CAAC;QAED,8DAA8D;QAC9D,MAAM,eAAe,GAAG,SAAS,GAAG,CAAC,CAAC,CAAC,CAAC,CAAC,eAAe,GAAG,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;QAE1E,iBAAiB;QACjB,MAAM,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,OAAO,EAAE;YAC9B,SAAS,EAAE,GAAG;YACd,MAAM,EAAE,IAAI;YACZ,SAAS,EAAE,GAAG;YACd,UAAU,EAAE,GAAG;YACf,MAAM,EAAE,KAAK;YACb,OAAO,EAAE,IAAI;YACb,cAAc,EAAE,KAAK;SACxB,CAAC,CAAC;QAEH,OAAO,EAAE,GAAG,EAAE,eAAe,EAAE,CAAC;IACpC,CAAC;IAED,KAAK,CAAC,MAAM,CAAC,OAAe;QACxB,IAAI,CAAC,GAAG,CAAC,IAAI,CAAC,sBAAsB,EAAE,EAAE,OAAO,EAAE,IAAI,CAAC,OAAO,EAAE,CAAC,CAAC;QACjE,MAAM,OAAO,GAAG,IAAI,gBAAgB,CAAC;YACjC,MAAM,EAAE,IAAI,CAAC,MAAM;YACnB,GAAG,EAAE,IAAI,CAAC,OAAO;YACjB,IAAI,EAAE,OAAO;SAChB,CAAC,CAAC;QACH,MAAM,IAAI,CAAC,QAAQ,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IACtC,CAAC;IAED,KAAK,CAAC,aAAa,CAAC,KAAa;QAC7B,MAAM,OAAO,GAAG,IAAI,4BAA4B,CAAC;YAC7C,gBAAgB,EAAE;gBACd,QAAQ,EAAE;oBACN,MAAM,EAAE,IAAI,CAAC,MAAM;oBACnB,IAAI,EAAE,KAAK;iBACd;aACJ;YACD,YAAY,EAAE,CAAC,QAAQ,CAAC;SAC3B,CAAC,CAAC;QACH,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QACzD,OAAO,QAAQ,CAAC,KAAM,CAAC;IAC3B,CAAC;IAED,KAAK,CAAC,cAAc,CAAC,KAAa;QAC9B,MAAM,OAAO,GAAG,IAAI,0BAA0B,CAAC,EAAE,KAAK,EAAE,KAAK,EAAE,CAAC,CAAC;QACjE,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;QACzD,OAAO,QAAQ,CAAC,SAAU,CAAC;IAC/B,CAAC;IAEO,mBAAmB,CAAC,KAAY;QACpC,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,EAAE,WAAW,CAAC;QAC7C,IAAI,CAAC,QAAQ;YAAE,OAAO,EAAE,CAAC;QACzB,MAAM,IAAI,GAAG,CAAC,QAAQ,CAAC,KAAK,IAAI,CAAC,CAAC,GAAG,CAAC,QAAQ,CAAC,MAAM,IAAI,CAAC,CAAC,CAAC;QAC5D,IAAI,IAAI,GAAG,IAAI;YAAE,OAAO,EAAE,CAAC,CAAC,oBAAoB;QAEhD,MAAM,GAAG,GAAG,QAAQ,CAAC,GAAG,IAAI,CAAC,CAAC;QAC9B,MAAM,IAAI,GAAG,QAAQ,CAAC,IAAI,IAAI,CAAC,CAAC;QAEhC,IAAI,QAAQ,GAAG,EAAE,CAAC;QAClB,IAAI,GAAG,GAAG,GAAG;YAAE,QAAQ,IAAI,MAAM,CAAC;aAC7B,IAAI,GAAG,GAAG,GAAG;YAAE,QAAQ,IAAI,SAAS,CAAC;QAE1C,IAAI,IAAI,GAAG,GAAG;YAAE,QAAQ,IAAI,MAAM,CAAC;aAC9B,IAAI,IAAI,GAAG,GAAG;YAAE,QAAQ,IAAI,OAAO,CAAC;;YACpC,QAAQ,IAAI,QAAQ,CAAC;QAE1B,OAAO,UAAU,QAAQ,KAAK,CAAC;IACnC,CAAC;IAEO,mBAAmB,CAAC,KAAY;QACpC,MAAM,IAAI,GAAG,KAAK,CAAC,QAAQ,EAAE,WAAW,EAAE,IAAI,IAAI,CAAC,CAAC;QACpD,IAAI,IAAI,GAAG,IAAI;YAAE,OAAO,CAAC,CAAC;QAC1B,IAAI,IAAI,GAAG,IAAI;YAAE,OAAO,CAAC,CAAC;QAC1B,OAAO,CAAC,CAAC;IACb,CAAC;IAEO,cAAc,CAAC,KAAY,EAAE,SAAuB;QACxD,IAAI,CAAC,SAAS;YAAE,OAAO,IAAI,CAAC;QAC5B,MAAM,GAAG,GAAG,CAAC,KAAK,CAAC,QAAQ,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,CAAC;YAC/C,CAAC,CAAC,SAAS,CAAC,QAAQ,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,CAAC;gBACxC,CAAC,SAAS,CAAC,QAAQ,EAAE,WAAW,EAAE,MAAM,IAAI,CAAC,CAAC,CAAC,CAAC;QACxD,OAAO,GAAG,GAAG,IAAI,CAAC;IACtB,CAAC;IAEO,eAAe,CAAC,KAAY,EAAE,SAAuB;QACzD,MAAM,IAAI,GAAG,KAAK,CAAC,IAAI,IAAI,EAAE,CAAC;QAC9B,MAAM,WAAW,GAAG,IAAI,CAAC,mBAAmB,CAAC,KAAK,CAAC,CAAC;QACpD,MAAM,MAAM,GAAG,MAAM,CAAC,MAAM,CAAC,WAAW,CAAC,CAAC;QAE1C,IAAI,IAAI,CAAC,cAAc,CAAC,KAAK,EAAE,SAAS,CAAC,EAAE,CAAC;YACxC,OAAO,KAAK,MAAM,GAAG,IAAI,IAAI,CAAC;QAClC,CAAC;QACD,OAAO,GAAG,MAAM,GAAG,IAAI,IAAI,CAAC;IAChC,CAAC;IAEO,gBAAgB,CAAC,IAAW,EAAE,OAAc;QAChD,MAAM,UAAU,GAAG,CAAC,IAAI,CAAC,QAAQ,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,CAAC;cACnD,CAAC,IAAI,CAAC,QAAQ,EAAE,WAAW,EAAE,MAAM,IAAI,CAAC,CAAC,CAAC;QAChD,MAAM,UAAU,GAAG,OAAO,CAAC,QAAQ,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,CAAC;QAC3D,MAAM,GAAG,GAAG,UAAU,GAAG,UAAU,CAAC;QAEpC,uDAAuD;QACvD,IAAI,GAAG,GAAG,IAAI,EAAE,CAAC;YACb,OAAO,IAAI,CAAC;QAChB,CAAC;QACD,OAAO,KAAK,CAAC;IACjB,CAAC;IAED,KAAK,CAAC,cAAc,CAAC,KAAa;QAC9B,IAAI,SAA6B,CAAC;QAClC,IAAI,SAAS,GAAY,EAAE,CAAC;QAE5B,GAAG,CAAC;YACA,MAAM,OAAO,GAAG,IAAI,0BAA0B,CAAC;gBAC3C,KAAK,EAAE,KAAK;gBACZ,SAAS,EAAE,SAAS;aACvB,CAAC,CAAC;YACH,MAAM,QAAQ,GAAG,MAAM,IAAI,CAAC,cAAc,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;YACzD,SAAS,GAAG,SAAS,CAAC,MAAM,CAAC,QAAQ,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC;YACpD,SAAS,GAAG,QAAQ,CAAC,SAAS,CAAC;QACnC,CAAC,QAAQ,SAAS,EAAE;QAEpB,oBAAoB;QACpB,MAAM,SAAS,GAAc,EAAE,CAAC;QAChC,KAAK,MAAM,KAAK,IAAI,SAAS,EAAE,CAAC;YAC5B,SAAS,CAAC,KAAK,CAAC,EAAG,CAAC,GAAG,KAAK,CAAC;QACjC,CAAC;QAED,8CAA8C;QAC9C,MAAM,YAAY,GAAkB,EAAE,CAAC;QACvC,IAAI,WAAW,GAAuB,IAAI,CAAC;QAE3C,iEAAiE;QACjE,IAAI,kBAAkB,GAAG,EAAE,CAAC;QAC5B,IAAI,aAAa,GAAiB,IAAI,CAAC;QAEvC,qCAAqC;QACrC,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE;YACpB,IAAI,CAAC,CAAC,IAAI,KAAK,CAAC,CAAC,IAAI;gBAAE,OAAO,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,IAAI,CAAC,CAAC,CAAC;YAC5D,OAAO,CAAC,CAAC,CAAC,QAAQ,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,QAAQ,EAAE,WAAW,EAAE,GAAG,IAAI,CAAC,CAAC,CAAC;QACrF,CAAC,CAAC,CAAC;QAEH,KAAK,MAAM,KAAK,IAAI,SAAS,EAAE,CAAC;YAC5B,IAAI,KAAK,CAAC,SAAS,KAAK,MAAM,EAAE,CAAC;gBAC7B,uEAAuE;gBACvE,IAAI,kBAAkB,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,IAAI,WAAW,EAAE,CAAC;oBACtD,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC;wBACpB,IAAI,EAAE,MAAM;wBACZ,OAAO,EAAE,kBAAkB;qBAC9B,CAAC,CAAC;gBACP,CAAC;gBACD,IAAI,WAAW,EAAE,CAAC;oBACd,YAAY,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;gBACnC,CAAC;gBACD,WAAW,GAAG;oBACV,UAAU,EAAE,KAAK,CAAC,IAAI,IAAI,CAAC;oBAC3B,MAAM,EAAE,EAAE;iBACb,CAAC;gBACF,kBAAkB,GAAG,EAAE,CAAC;gBACxB,aAAa,GAAG,IAAI,CAAC;YACzB,CAAC;iBACI,IAAI,WAAW,IAAI,KAAK,CAAC,IAAI,KAAK,WAAW,CAAC,UAAU,EAAE,CAAC;gBAC5D,iBAAiB;gBACjB,IAAI,KAAK,CAAC,SAAS,KAAK,OAAO,EAAE,CAAC;oBAC9B,iDAAiD;oBACjD,IAAI,kBAAkB,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;wBACvC,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC;4BACpB,IAAI,EAAE,MAAM;4BACZ,OAAO,EAAE,kBAAkB;yBAC9B,CAAC,CAAC;wBACH,kBAAkB,GAAG,EAAE,CAAC;oBAC5B,CAAC;oBACD,MAAM,EAAE,GAAG,EAAE,eAAe,EAAE,GAAG,IAAI,CAAC,gBAAgB,CAClD,KAAK,EACL,SAAS,EACT,WAAW,CAAC,MAAM,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,KAAK,OAAO,CAAC,CAAC,MAAM,GAAG,CAAC,EAC7D,WAAW,CAAC,UAAU,CACzB,CAAC;oBACF,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC;wBACpB,IAAI,EAAE,OAAO;wBACb,OAAO,EAAE,GAAG;wBACZ,UAAU,EAAE,eAAe;qBAC9B,CAAC,CAAC;oBACH,aAAa,GAAG,IAAI,CAAC;gBACzB,CAAC;gBACD,qCAAqC;qBAChC,IAAI,KAAK,CAAC,SAAS,KAAK,MAAM,IAAI,CAAC,IAAI,CAAC,cAAc,CAAC,KAAK,EAAE,SAAS,CAAC,EAAE,CAAC;oBAC5E,IAAI,aAAa,IAAI,IAAI,CAAC,gBAAgB,CAAC,aAAa,EAAE,KAAK,CAAC,EAAE,CAAC;wBAC/D,6DAA6D;wBAC7D,0DAA0D;wBAC1D,gEAAgE;wBAChE,MAAM,SAAS,GAAG,IAAI,CAAC,eAAe,CAAC,KAAK,EAAE,aAAa,CAAC,CAAC;wBAE7D,2EAA2E;wBAC3E,8EAA8E;wBAC9E,MAAM,UAAU,GAAG,SAAS,CAAC,OAAO,CAAC,QAAQ,EAAE,GAAG,CAAC,CAAC;wBAEpD,kBAAkB,IAAI,GAAG,GAAG,UAAU,CAAC,IAAI,EAAE,CAAC;oBAClD,CAAC;yBAAM,CAAC;wBACJ,6CAA6C;wBAC7C,IAAI,kBAAkB,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;4BACvC,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC;gCACpB,IAAI,EAAE,MAAM;gCACZ,OAAO,EAAE,kBAAkB;6BAC9B,CAAC,CAAC;wBACP,CAAC;wBACD,yBAAyB;wBACzB,kBAAkB,GAAG,IAAI,CAAC,eAAe,CAAC,KAAK,EAAE,aAAa,CAAC,CAAC,IAAI,EAAE,CAAC;oBAC3E,CAAC;oBACD,aAAa,GAAG,KAAK,CAAC;gBAC1B,CAAC;gBACD,2BAA2B;qBACtB,IAAI,IAAI,CAAC,YAAY,EAAE,CAAC;oBACzB,MAAM,QAAQ,GAAG,KAAK,CAAC,QAAQ,EAAE,WAAW,CAAC;oBAC7C,IAAI,QAAQ,IAAI,QAAQ,CAAC,KAAK,IAAI,QAAQ,CAAC,MAAM,EAAE,CAAC;wBAChD,MAAM,gBAAgB,GAAG,IAAI,CAAC,mBAAmB,CAAC,KAAK,CAAC,CAAC;wBACzD,IAAI,gBAAgB,EAAE,CAAC;4BACnB,iDAAiD;4BACjD,IAAI,kBAAkB,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gCACvC,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC;oCACpB,IAAI,EAAE,MAAM;oCACZ,OAAO,EAAE,kBAAkB;iCAC9B,CAAC,CAAC;gCACH,kBAAkB,GAAG,EAAE,CAAC;4BAC5B,CAAC;4BAED,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC;gCACpB,IAAI,EAAE,OAAO;gCACb,OAAO,EAAE,gBAAgB;gCACzB,IAAI,EAAE,QAAQ,CAAC,IAAI;gCACnB,GAAG,EAAE,QAAQ,CAAC,GAAG;gCACjB,KAAK,EAAE,QAAQ,CAAC,KAAK;gCACrB,MAAM,EAAE,QAAQ,CAAC,MAAM;6BAC1B,CAAC,CAAC;wBACP,CAAC;oBACL,CAAC;oBACD,uCAAuC;gBAC3C,CAAC;YACL,CAAC;QACL,CAAC;QAED,mBAAmB;QACnB,IAAI,WAAW,EAAE,CAAC;YACd,IAAI,kBAAkB,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;gBACvC,WAAW,CAAC,MAAM,CAAC,IAAI,CAAC;oBACpB,IAAI,EAAE,MAAM;oBACZ,OAAO,EAAE,kBAAkB;iBAC9B,CAAC,CAAC;YACP,CAAC;YACD,YAAY,CAAC,IAAI,CAAC,WAAW,CAAC,CAAC;QACnC,CAAC;QAED,qBAAqB;QACrB,IAAI,QAAQ,GAAG,EAAE,CAAC;QAClB,IAAI,SAAS,GAAG,CAAC,CAAC;QAClB,IAAI,WAAW,GAAG,CAAC,CAAC;QACpB,KAAK,MAAM,IAAI,IAAI,YAAY,EAAE,CAAC;YAC9B,QAAQ,IAAI,iBAAiB,IAAI,CAAC,UAAU,MAAM,CAAC;YACnD,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,MAAM,EAAE,CAAC;gBAC9B,IAAI,KAAK,CAAC,IAAI,KAAK,MAAM,EAAE,CAAC;oBACxB,QAAQ,IAAI,WAAW,KAAK,CAAC,OAAO,eAAe,CAAC;gBACxD,CAAC;qBAAM,IAAI,KAAK,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;oBAChC,MAAM,cAAc,GAAG,KAAK,CAAC,UAAU,KAAK,SAAS,IAAI,IAAI,CAAC,yBAAyB;wBACnF,CAAC,CAAC,gBAAgB,KAAK,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG;wBAChD,CAAC,CAAC,EAAE,CAAC;oBACT,QAAQ,IAAI,iBAAiB,WAAW,EAAE,eAAe,cAAc,KAAK,CAAC;oBAC7E,QAAQ,IAAI,GAAG,KAAK,CAAC,OAAO,IAAI,CAAC;oBACjC,QAAQ,IAAI,cAAc,CAAC;gBAC/B,CAAC;qBAAM,IAAI,KAAK,CAAC,IAAI,KAAK,OAAO,EAAE,CAAC;oBAChC,+BAA+B;oBAC/B,MAAM,QAAQ,GAAG,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,UAAU,KAAK,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;oBACtE,MAAM,OAAO,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,SAAS,KAAK,CAAC,GAAG,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;oBAClE,MAAM,SAAS,GAAG,KAAK,CAAC,KAAK,CAAC,CAAC,CAAC,WAAW,KAAK,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;oBAC1E,MAAM,UAAU,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,YAAY,KAAK,CAAC,MAAM,CAAC,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC;oBAE9E,QAAQ,IAAI,cAAc,SAAS,EAAE,KAAK,QAAQ,GAAG,OAAO,GAAG,SAAS,GAAG,UAAU,MAAM,KAAK,CAAC,OAAO,CAAC,IAAI,EAAE,gBAAgB,CAAC;gBACpI,CAAC;YACL,CAAC;YACD,QAAQ,IAAI,aAAa,CAAC;QAC9B,CAAC;QAED,OAAO,QAAQ,CAAC;IACpB,CAAC;CAEJ"}

package/lib/esm/conversion/image.js

@@ -0,0 +1,143 @@
+import { log } from "@temporalio/activity";
+import { execFile as execFileCallback } from "child_process";
+import fs from "fs";
+import { file } from "tmp-promise";
+import { promisify } from "util";
+const execFile = promisify(execFileCallback);
+/**
+ * Resizes an image to a maximum height or width using ImageMagick
+ * with progressive loading when supported and colorspace correction
+ * @param inputPath Input file path
+ * @param max_hw Maximum height or width
+ * @param format Output format
+ * @param progressive Enable progressive loading for supported formats (defaults to true)
+ * @param colorspaceCorrection Enable colorspace correction (defaults to true), not recommended for Q8 image magick.
+ * @param colorspace Colorspace to use for processing ('RGB', 'LAB', 'LUV', 'sigmoidal') (defaults to 'RGB')
+ * @returns Path to the resized image
+ */
+export async function imageResizer(inputPath, max_hw, format, progressive = true, colorspaceCorrection = true, colorspace = 'RGB') {
+    log.info(`[image-resizer] Resizing image: ${inputPath} to max_hw: ${max_hw}, format: ${format}, progressive: ${progressive}, colorspaceCorrection: ${colorspaceCorrection ? colorspace : 'disabled'}`);
+    const allowedFormats = ["jpg", "jpeg", "png", "webp"];
+    if (!format || format.trim() === "") {
+        throw new Error(`Invalid format: ${format}.Supported : ${allowedFormats.join(", ")}`);
+    }
+    //check that max_hw is valid
+    if (!Number.isInteger(max_hw) || max_hw <= 0) {
+        throw new Error(`Invalid max_hw value: ${max_hw}`);
+    }
+    //check that inputPath exists
+    if (!fs.existsSync(inputPath)) {
+        throw new Error(`Input file does not exist: ${inputPath}`);
+    }
+    // Create a temporary file
+    const { path: outputPath, cleanup } = await file({ postfix: `.${format}` });
+    try {
+        // Check if input file exists
+        if (!fs.existsSync(inputPath)) {
+            throw new Error(`Input file does not exist: ${inputPath}`);
+        }
+        // Validate max_hw
+        if (!Number.isInteger(max_hw) || max_hw <= 0) {
+            throw new Error(`Invalid max_hw value: ${max_hw}`);
+        }
+        // Progressive loading options
+        let conversionOption = "";
+        // Only add progressive option for formats that support it
+        if (progressive) {
+            // JPEG and some other formats support progressive loading
+            const lowerFormat = format.toLowerCase();
+            if (lowerFormat === "jpg" || lowerFormat === "jpeg") {
+                conversionOption = "-interlace JPEG";
+                log.info(`Enabling interlaced ${lowerFormat.toUpperCase()} format`);
+            }
+            else if (lowerFormat === "png") {
+                conversionOption = "-interlace PNG";
+                log.info(`Enabling interlaced ${lowerFormat.toUpperCase()} format`);
+            }
+            else if (lowerFormat === "gif") {
+                conversionOption = "-interlace GIF";
+                log.info(`Enabling interlaced ${lowerFormat.toUpperCase()} format`);
+            }
+        }
+        log.info(`Resizing image using ImageMagick: ${inputPath} -> ${outputPath}`);
+        const command = `convert`;
+        let args = [inputPath];
+        // Add JPEG shrink-on-load optimization
+        args.push("-define", `jpeg:size=${max_hw * 3}x${max_hw * 3}`);
+        // Remove metadata
+        args.push("-strip");
+        // https://usage.imagemagick.org/filter/nicolas/#downsample
+        // Add colorspace correction if enabled
+        if (colorspaceCorrection) {
+            switch (colorspace) {
+                case 'RGB':
+                    // Linear light, recommended default
+                    // Convert from sRGB to linear RGB for processing
+                    args.push("-colorspace", "RGB");
+                    log.info("Using linear RGB colorspace for resize processing");
+                    break;
+                case 'LAB':
+                    // Perceptual linear light
+                    // Use LAB colorspace which separates intensity from color
+                    // Better for avoiding color clipping and distortion
+                    args.push("-colorspace", "LAB");
+                    log.info("Using LAB colorspace for resize processing");
+                    break;
+                case 'LUV':
+                    // Perceptual linear light
+                    // Alternative to LAB with perceptually uniform color deltas
+                    args.push("-colorspace", "LUV");
+                    log.info("Using LUV colorspace for resize processing");
+                    break;
+                case 'sigmoidal':
+                    // Sigmoidal colorspace modification to reduce ringing artifacts
+                    args.push("-colorspace", "RGB");
+                    args.push("+sigmoidal-contrast", "6.5,50%");
+                    log.info("Using sigmoidal contrast modification for resize processing");
+                    break;
+            }
+        }
+        // Resize operation
+        args.push("-resize", `${max_hw}x${max_hw}>`);
+        // Restore colorspace after processing
+        if (colorspaceCorrection) {
+            switch (colorspace) {
+                case 'RGB':
+                case 'LAB':
+                case 'LUV':
+                    // Convert back to sRGB for output
+                    args.push("-colorspace", "sRGB");
+                    break;
+                case 'sigmoidal':
+                    // Restore from sigmoidal modification and convert to sRGB
+                    args.push("-sigmoidal-contrast", "6.5,50%");
+                    args.push("-colorspace", "sRGB");
+                    break;
+            }
+        }
+        // Add progressive/interlace options
+        if (conversionOption) {
+            args.push(...conversionOption.split(" "));
+        }
+        // Output path
+        args.push(outputPath);
+        log.info(`ImageMagick command: ${command} ${args.join(" ")}`);
+        const { stderr } = await execFile(command, args);
+        if (stderr) {
+            log.warn(`ImageMagick warning: ${stderr}`);
+        }
+        // Verify output exists and has content
+        if (!fs.existsSync(outputPath) || fs.statSync(outputPath).size === 0) {
+            throw new Error(`ImageMagick conversion failed: output file not created or empty`);
+        }
+        return outputPath;
+    }
+    catch (error) {
+        // Clean up the temporary file
+        await cleanup();
+        const errorMessage = error instanceof Error ? error.message : String(error);
+        log.error(`Image conversion failed: ${errorMessage}`);
+        throw new Error(`Image conversion failed: ${errorMessage}`);
+    }
+}
+//# sourceMappingURL=image.js.map

package/lib/esm/conversion/image.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"image.js","sourceRoot":"","sources":["../../../src/conversion/image.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,GAAG,EAAE,MAAM,sBAAsB,CAAC;AAC3C,OAAO,EAAE,QAAQ,IAAI,gBAAgB,EAAE,MAAM,eAAe,CAAC;AAC7D,OAAO,EAAE,MAAM,IAAI,CAAC;AACpB,OAAO,EAAE,IAAI,EAAE,MAAM,aAAa,CAAC;AACnC,OAAO,EAAE,SAAS,EAAE,MAAM,MAAM,CAAC;AACjC,MAAM,QAAQ,GAAG,SAAS,CAAC,gBAAgB,CAAC,CAAC;AAE7C;;;;;;;;;;GAUG;AACH,MAAM,CAAC,KAAK,UAAU,YAAY,CAC9B,SAAiB,EACjB,MAAc,EACd,MAAc,EACd,cAAuB,IAAI,EAC3B,uBAAgC,IAAI,EACpC,aAAkD,KAAK;IAEvD,GAAG,CAAC,IAAI,CAAC,mCAAmC,SAAS,eAAe,MAAM,aAAa,MAAM,kBAAkB,WAAW,2BAA2B,oBAAoB,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,UAAU,EAAE,CAAC,CAAC;IAEvM,MAAM,cAAc,GAAG,CAAC,KAAK,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,CAAC,CAAC;IAEtD,IAAI,CAAC,MAAM,IAAI,MAAM,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE,CAAC;QAClC,MAAM,IAAI,KAAK,CAAC,mBAAmB,MAAM,gBAAgB,cAAc,CAAC,IAAI,CAAC,IAAI,CAAC,EAAE,CAAC,CAAC;IAC1F,CAAC;IAED,4BAA4B;IAC5B,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,MAAM,CAAC,IAAI,MAAM,IAAI,CAAC,EAAE,CAAC;QAC3C,MAAM,IAAI,KAAK,CAAC,yBAAyB,MAAM,EAAE,CAAC,CAAC;IACvD,CAAC;IAED,6BAA6B;IAC7B,IAAI,CAAC,EAAE,CAAC,UAAU,CAAC,SAAS,CAAC,EAAE,CAAC;QAC5B,MAAM,IAAI,KAAK,CAAC,8BAA8B,SAAS,EAAE,CAAC,CAAC;IAC/D,CAAC;IAED,0BAA0B;IAC1B,MAAM,EAAE,IAAI,EAAE,UAAU,EAAE,OAAO,EAAE,GAAG,MAAM,IAAI,CAAC,EAAE,OAAO,EAAE,IAAI,MAAM,EAAE,EAAE,CAAC,CAAC;IAC5E,IAAI,CAAC;QACD,6BAA6B;QAC7B,IAAI,CAAC,EAAE,CAAC,UAAU,CAAC,SAAS,CAAC,EAAE,CAAC;YAC5B,MAAM,IAAI,KAAK,CAAC,8BAA8B,SAAS,EAAE,CAAC,CAAC;QAC/D,CAAC;QACD,kBAAkB;QAClB,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,MAAM,CAAC,IAAI,MAAM,IAAI,CAAC,EAAE,CAAC;YAC3C,MAAM,IAAI,KAAK,CAAC,yBAAyB,MAAM,EAAE,CAAC,CAAC;QACvD,CAAC;QAED,8BAA8B;QAC9B,IAAI,gBAAgB,GAAG,EAAE,CAAC;QAE1B,0DAA0D;QAC1D,IAAI,WAAW,EAAE,CAAC;YACd,0DAA0D;YAC1D,MAAM,WAAW,GAAG,MAAM,CAAC,WAAW,EAAE,CAAC;YACzC,IAAI,WAAW,KAAK,KAAK,IAAI,WAAW,KAAK,MAAM,EAAE,CAAC;gBAClD,gBAAgB,GAAG,iBAAiB,CAAC;gBACrC,GAAG,CAAC,IAAI,CAAC,uBAAuB,WAAW,CAAC,WAAW,EAAE,SAAS,CAAC,CAAC;YACxE,CAAC;iBAAM,IAAI,WAAW,KAAK,KAAK,EAAE,CAAC;gBAC/B,gBAAgB,GAAG,gBAAgB,CAAC;gBACpC,GAAG,CAAC,IAAI,CAAC,uBAAuB,WAAW,CAAC,WAAW,EAAE,SAAS,CAAC,CAAC;YACxE,CAAC;iBAAM,IAAI,WAAW,KAAK,KAAK,EAAE,CAAC;gBAC/B,gBAAgB,GAAG,gBAAgB,CAAC;gBACpC,GAAG,CAAC,IAAI,CAAC,uBAAuB,WAAW,CAAC,WAAW,EAAE,SAAS,CAAC,CAAC;YACxE,CAAC;QACL,CAAC;QAED,GAAG,CAAC,IAAI,CAAC,qCAAqC,SAAS,OAAO,UAAU,EAAE,CAAC,CAAC;QAE5E,MAAM,OAAO,GAAG,SAAS,CAAA;QACzB,IAAI,IAAI,GAAG,CAAC,SAAS,CAAC,CAAC;QAEvB,uCAAuC;QACvC,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,aAAa,MAAM,GAAG,CAAC,IAAI,MAAM,GAAG,CAAC,EAAE,CAAC,CAAC;QAE9D,kBAAkB;QAClB,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QAEpB,2DAA2D;QAC3D,uCAAuC;QACvC,IAAI,oBAAoB,EAAE,CAAC;YACvB,QAAQ,UAAU,EAAE,CAAC;gBACjB,KAAK,KAAK;oBACN,oCAAoC;oBACpC,iDAAiD;oBACjD,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,KAAK,CAAC,CAAC;oBAChC,GAAG,CAAC,IAAI,CAAC,mDAAmD,CAAC,CAAC;oBAC9D,MAAM;gBACV,KAAK,KAAK;oBACN,0BAA0B;oBAC1B,0DAA0D;oBAC1D,oDAAoD;oBACpD,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,KAAK,CAAC,CAAC;oBAChC,GAAG,CAAC,IAAI,CAAC,4CAA4C,CAAC,CAAC;oBACvD,MAAM;gBACV,KAAK,KAAK;oBACN,0BAA0B;oBAC1B,4DAA4D;oBAC5D,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,KAAK,CAAC,CAAC;oBAChC,GAAG,CAAC,IAAI,CAAC,4CAA4C,CAAC,CAAC;oBACvD,MAAM;gBACV,KAAK,WAAW;oBACZ,gEAAgE;oBAChE,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,KAAK,CAAC,CAAC;oBAChC,IAAI,CAAC,IAAI,CAAC,qBAAqB,EAAE,SAAS,CAAC,CAAC;oBAC5C,GAAG,CAAC,IAAI,CAAC,6DAA6D,CAAC,CAAC;oBACxE,MAAM;YACd,CAAC;QACL,CAAC;QAED,mBAAmB;QACnB,IAAI,CAAC,IAAI,CAAC,SAAS,EAAE,GAAG,MAAM,IAAI,MAAM,GAAG,CAAC,CAAC;QAE7C,sCAAsC;QACtC,IAAI,oBAAoB,EAAE,CAAC;YACvB,QAAQ,UAAU,EAAE,CAAC;gBACjB,KAAK,KAAK,CAAC;gBACX,KAAK,KAAK,CAAC;gBACX,KAAK,KAAK;oBACN,kCAAkC;oBAClC,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,MAAM,CAAC,CAAC;oBACjC,MAAM;gBACV,KAAK,WAAW;oBACZ,0DAA0D;oBAC1D,IAAI,CAAC,IAAI,CAAC,qBAAqB,EAAE,SAAS,CAAC,CAAC;oBAC5C,IAAI,CAAC,IAAI,CAAC,aAAa,EAAE,MAAM,CAAC,CAAC;oBACjC,MAAM;YACd,CAAC;QACL,CAAC;QAED,oCAAoC;QACpC,IAAI,gBAAgB,EAAE,CAAC;YACnB,IAAI,CAAC,IAAI,CAAC,GAAG,gBAAgB,CAAC,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC;QAC9C,CAAC;QAED,cAAc;QACd,IAAI,CAAC,IAAI,CAAC,UAAU,CAAC,CAAC;QAEtB,GAAG,CAAC,IAAI,CAAC,wBAAwB,OAAO,IAAI,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,CAAC;QAE9D,MAAM,EAAE,MAAM,EAAE,GAAG,MAAM,QAAQ,CAAC,OAAO,EAAE,IAAI,CAAC,CAAC;QAEjD,IAAI,MAAM,EAAE,CAAC;YACT,GAAG,CAAC,IAAI,CAAC,wBAAwB,MAAM,EAAE,CAAC,CAAC;QAC/C,CAAC;QAED,uCAAuC;QACvC,IAAI,CAAC,EAAE,CAAC,UAAU,CAAC,UAAU,CAAC,IAAI,EAAE,CAAC,QAAQ,CAAC,UAAU,CAAC,CAAC,IAAI,KAAK,CAAC,EAAE,CAAC;YACnE,MAAM,IAAI,KAAK,CAAC,iEAAiE,CAAC,CAAC;QACvF,CAAC;QAED,OAAO,UAAU,CAAC;IACtB,CAAC;IAAC,OAAO,KAAK,EAAE,CAAC;QACb,8BAA8B;QAC9B,MAAM,OAAO,EAAE,CAAC;QAChB,MAAM,YAAY,GAAG,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC;QAC5E,GAAG,CAAC,KAAK,CAAC,4BAA4B,YAAY,EAAE,CAAC,CAAC;QACtD,MAAM,IAAI,KAAK,CAAC,4BAA4B,YAAY,EAAE,CAAC,CAAC;IAChE,CAAC;AACL,CAAC"}

package/lib/esm/conversion/markitdown.js

@@ -0,0 +1,36 @@
+import { log } from "@temporalio/activity";
+import { spawn } from "child_process";
+import fs from "fs";
+import tmp from "tmp";
+export function markdownWithMarkitdown(buffer, ext) {
+    const inputFile = tmp.fileSync({ postfix: ext });
+    const targetFileName = tmp.tmpNameSync({ postfix: ".md" });
+    fs.writeSync(inputFile.fd, buffer);
+    return new Promise((resolve, reject) => {
+        const tool = "markitdown";
+        log.info(`Converting document to markdown with ${tool}`, { inputFile: inputFile.name, targetFileName });
+        const command = spawn(tool, [inputFile.name, "-o", targetFileName]);
+        command.on("exit", function (code) {
+            if (code) {
+                reject(new Error(`${tool} exited with code ${code}`));
+            }
+        });
+        command.on("close", function (code) {
+            if (code) {
+                reject(new Error(`${tool} exited with code ${code}`));
+            }
+            else {
+                return fs.readFile(targetFileName, "utf8", (err, data) => {
+                    if (err) {
+                        reject(err);
+                    }
+                    return resolve(data);
+                });
+            }
+        });
+        command.on("error", (err) => {
+            reject(err);
+        });
+    });
+}
+//# sourceMappingURL=markitdown.js.map