@vertesia/workflow 0.42.2

package/lib/cjs/conversion/TextractProcessor.js

@@ -0,0 +1,416 @@
+"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;
+        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="${block.confidence.toFixed(2)}"`
+                        : '';
+                    fulltext += `<table type="csv"${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

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

@@ -0,0 +1 @@
+{"version":3,"file":"TextractProcessor.js","sourceRoot":"","sources":["../../../src/conversion/TextractProcessor.ts"],"names":[],"mappings":";;;;;;AAAA,kDAAgE;AAEhE,8DAIkC;AAElC,0DAA6B;AAmC7B,MAAa,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,gCAAc,CAAC;YACrC,MAAM;YACN,WAAW;SACd,CAAC,CAAC;QACH,IAAI,CAAC,QAAQ,GAAG,IAAI,oBAAQ,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,mBAAI,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,4BAAgB,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,8CAA4B,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,4CAA0B,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,4CAA0B,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,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,oBAAoB,cAAc,KAAK,CAAC;oBACpD,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;AA7cD,8CA6cC"}

package/lib/cjs/conversion/image.js

@@ -0,0 +1,22 @@
+"use strict";
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.imageResizer = imageResizer;
+const sharp_1 = __importDefault(require("sharp"));
+/**
+ * Resizes an image to a maximum height or width
+ * @param max_hw
+ * @param format
+ * @returns
+ */
+function imageResizer(max_hw, format) {
+    return (0, sharp_1.default)().resize({
+        width: max_hw,
+        height: max_hw,
+        fit: sharp_1.default.fit.inside,
+        withoutEnlargement: true,
+    }).toFormat(format);
+}
+//# sourceMappingURL=image.js.map

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

@@ -0,0 +1 @@
+{"version":3,"file":"image.js","sourceRoot":"","sources":["../../../src/conversion/image.ts"],"names":[],"mappings":";;;;;AAUA,oCAUC;AAnBD,kDAA0B;AAG1B;;;;;GAKG;AACH,SAAgB,YAAY,CAAC,MAAc,EAAE,MAA8B;IAEvE,OAAO,IAAA,eAAK,GAAE,CAAC,MAAM,CAAC;QAClB,KAAK,EAAE,MAAM;QACb,MAAM,EAAE,MAAM;QACd,GAAG,EAAE,eAAK,CAAC,GAAG,CAAC,MAAM;QACrB,kBAAkB,EAAE,IAAI;KAE3B,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC;AAExB,CAAC"}

package/lib/cjs/conversion/mutool.js

@@ -0,0 +1,147 @@
+"use strict";
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.mutoolPdfToText = mutoolPdfToText;
+exports.pdfToImages = pdfToImages;
+exports.pdfExtractPages = pdfExtractPages;
+const activity_1 = require("@temporalio/activity");
+const child_process_1 = require("child_process");
+const fs_1 = __importDefault(require("fs"));
+const tmp_1 = __importDefault(require("tmp"));
+/**
+ * Convert a pdf file to text
+ * TODO: pass file reference instead of Buffer
+ */
+function mutoolPdfToText(buffer) {
+    const inputFile = tmp_1.default.fileSync({ postfix: '.pdf' });
+    const targetFileName = tmp_1.default.tmpNameSync({ postfix: '.txt' });
+    fs_1.default.writeSync(inputFile.fd, buffer);
+    return new Promise((resolve, reject) => {
+        activity_1.log.info("Converting pdf to text", { inputFile: inputFile.name, targetFileName });
+        const command = (0, child_process_1.spawn)("mutool", ["convert", "-o", targetFileName, inputFile.name]);
+        command.on('exit', function (code) {
+            if (code) {
+                reject(new Error(`mutool exited with code ${code}`));
+            }
+        });
+        command.on('close', function (code) {
+            if (code) {
+                reject(new Error(`mutool exited with code ${code}`));
+            }
+            else {
+                return fs_1.default.readFile(targetFileName, 'utf8', (err, data) => {
+                    if (err) {
+                        reject(err);
+                    }
+                    return resolve(data);
+                });
+            }
+            ;
+        });
+        command.on('error', (err) => {
+            reject(err);
+        });
+    });
+}
+/**
+ *
+ * Convert a pdf files to images (one image per page), as PNG format
+ *
+ * @param file
+ * @param pages
+ * @returns
+ */
+async function pdfToImages(file, pages) {
+    const workDir = tmp_1.default.dirSync();
+    activity_1.log.info(`Converting pdf to images`, { workDir: workDir.name, input_type: typeof file, pages });
+    if (file instanceof Buffer) {
+        fs_1.default.writeFileSync(`${workDir.name}/input.pdf`, file);
+    }
+    const filename = typeof file === 'string' ? file : `${workDir.name}/input.pdf`;
+    const args = [
+        "draw",
+        "-o", `${workDir.name}/page-%d.png`,
+        filename,
+    ];
+    if (pages) {
+        args.push(pages.join(','));
+    }
+    return new Promise((resolve, reject) => {
+        const command = (0, child_process_1.spawn)("mutool", args);
+        activity_1.log.info(`Executing mutool command`, { workDir: workDir.name, filename, command: command.spawnargs });
+        let errors = '';
+        command.stderr.on('data', (data) => {
+            errors += data;
+        });
+        command.on('exit', function (code) {
+            if (code) {
+                activity_1.log.error(`mutool exited with code ${code}`, { errors });
+                reject(new Error(`mutool exited with code ${code}`));
+            }
+        });
+        command.on('close', function (code) {
+            if (code) {
+                reject(new Error(`mutool finished with code ${code}`));
+            }
+            else {
+                const files = fs_1.default.readdirSync(workDir.name);
+                const images = files.filter(f => f.endsWith('.png')).map(f => `${workDir.name}/${f}`);
+                activity_1.log.info(`Converted pdf to ${images.length} images`, files);
+                return resolve(images);
+            }
+            ;
+        });
+        command.on('error', (err) => {
+            reject(err);
+        });
+    });
+}
+/**
+ * Get somes pages from a PDF to create a new one
+ */
+async function pdfExtractPages(file, pages) {
+    const workDir = tmp_1.default.dirSync();
+    activity_1.log.info(`Getting pages from pdf`, { workDir: workDir.name, input_type: typeof file, pages });
+    if (file instanceof Buffer) {
+        fs_1.default.writeFileSync(`${workDir.name}/input.pdf`, file);
+    }
+    const filename = typeof file === 'string' ? file : `${workDir.name}/input.pdf`;
+    const args = [
+        "merge",
+        "-o", `${workDir.name}/output.pdf`,
+        "-O", "garbage=compact,sanitize",
+        filename,
+        pages.join(','),
+    ];
+    return new Promise((resolve, reject) => {
+        const command = (0, child_process_1.spawn)("mutool", args);
+        activity_1.log.info(`Executing mutool command`, { workDir: workDir.name, filename, command: command.spawnargs });
+        let errors = '';
+        command.stderr.on('data', (data) => {
+            errors += data;
+        });
+        command.on('exit', function (code) {
+            if (code) {
+                activity_1.log.error(`mutool exited with code ${code}`, { errors });
+                reject(new Error(`mutool exited with code ${code}`));
+            }
+        });
+        command.on('close', function (code) {
+            if (code) {
+                reject(new Error(`mutool finished with code ${code}`));
+            }
+            else {
+                const file = `${workDir.name}/output.pdf`;
+                activity_1.log.info(`Extracted pages from pdf`, { pages, file });
+                return resolve(file);
+            }
+            ;
+        });
+        command.on('error', (err) => {
+            reject(err);
+        });
+    });
+}
+//# sourceMappingURL=mutool.js.map

package/lib/cjs/conversion/mutool.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"mutool.js","sourceRoot":"","sources":["../../../src/conversion/mutool.ts"],"names":[],"mappings":";;;;;AAYA,0CAwCC;AAUD,kCAwDC;AAOD,0CAsDC;AAnLD,mDAA2C;AAC3C,iDAAsC;AACtC,4CAAoB;AACpB,8CAAsB;AAItB;;;GAGG;AAEH,SAAgB,eAAe,CAAC,MAAc;IAE1C,MAAM,SAAS,GAAG,aAAG,CAAC,QAAQ,CAAC,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC,CAAC;IACpD,MAAM,cAAc,GAAG,aAAG,CAAC,WAAW,CAAC,EAAE,OAAO,EAAE,MAAM,EAAE,CAAC,CAAC;IAE5D,YAAE,CAAC,SAAS,CAAC,SAAS,CAAC,EAAE,EAAE,MAAM,CAAC,CAAC;IAEnC,OAAO,IAAI,OAAO,CAAC,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;QAGnC,cAAG,CAAC,IAAI,CAAC,wBAAwB,EAAE,EAAE,SAAS,EAAE,SAAS,CAAC,IAAI,EAAE,cAAc,EAAE,CAAC,CAAC;QAElF,MAAM,OAAO,GAAG,IAAA,qBAAK,EAAC,QAAQ,EAAE,CAAC,SAAS,EAAE,IAAI,EAAE,cAAc,EAAE,SAAS,CAAC,IAAI,CAAC,CAAC,CAAC;QAEnF,OAAO,CAAC,EAAE,CAAC,MAAM,EAAE,UAAU,IAAI;YAC7B,IAAI,IAAI,EAAE,CAAC;gBACP,MAAM,CAAC,IAAI,KAAK,CAAC,2BAA2B,IAAI,EAAE,CAAC,CAAC,CAAC;YACzD,CAAC;QACL,CAAC,CAAC,CAAC;QAGH,OAAO,CAAC,EAAE,CAAC,OAAO,EAAE,UAAU,IAAI;YAC9B,IAAI,IAAI,EAAE,CAAC;gBACP,MAAM,CAAC,IAAI,KAAK,CAAC,2BAA2B,IAAI,EAAE,CAAC,CAAC,CAAC;YACzD,CAAC;iBAAM,CAAC;gBACJ,OAAO,YAAE,CAAC,QAAQ,CAAC,cAAc,EAAE,MAAM,EAAE,CAAC,GAAG,EAAE,IAAI,EAAE,EAAE;oBACrD,IAAI,GAAG,EAAE,CAAC;wBACN,MAAM,CAAC,GAAG,CAAC,CAAC;oBAChB,CAAC;oBACD,OAAO,OAAO,CAAC,IAAI,CAAC,CAAC;gBACzB,CAAC,CAAC,CAAC;YACP,CAAC;YAAA,CAAC;QACN,CAAC,CAAC,CAAC;QAEH,OAAO,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE;YACxB,MAAM,CAAC,GAAG,CAAC,CAAC;QAChB,CAAC,CAAC,CAAC;IAEP,CAAC,CAAC,CAAC;AAEP,CAAC;AAED;;;;;;;GAOG;AACI,KAAK,UAAU,WAAW,CAAC,IAAqB,EAAE,KAAgB;IAErE,MAAM,OAAO,GAAG,aAAG,CAAC,OAAO,EAAE,CAAC;IAC9B,cAAG,CAAC,IAAI,CAAC,0BAA0B,EAAE,EAAE,OAAO,EAAE,OAAO,CAAC,IAAI,EAAE,UAAU,EAAE,OAAO,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC;IAEhG,IAAI,IAAI,YAAY,MAAM,EAAE,CAAC;QACzB,YAAE,CAAC,aAAa,CAAC,GAAG,OAAO,CAAC,IAAI,YAAY,EAAE,IAAI,CAAC,CAAC;IACxD,CAAC;IACD,MAAM,QAAQ,GAAG,OAAO,IAAI,KAAK,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,OAAO,CAAC,IAAI,YAAY,CAAC;IAE/E,MAAM,IAAI,GAAG;QACT,MAAM;QACN,IAAI,EAAE,GAAG,OAAO,CAAC,IAAI,cAAc;QACnC,QAAQ;KACX,CAAC;IAEF,IAAI,KAAK,EAAE,CAAC;QACR,IAAI,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC,CAAC;IAC/B,CAAC;IAED,OAAO,IAAI,OAAO,CAAC,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;QAEnC,MAAM,OAAO,GAAG,IAAA,qBAAK,EAAC,QAAQ,EAAE,IAAI,CAAC,CAAC;QACtC,cAAG,CAAC,IAAI,CAAC,0BAA0B,EAAE,EAAE,OAAO,EAAE,OAAO,CAAC,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE,OAAO,CAAC,SAAS,EAAE,CAAC,CAAC;QAEtG,IAAI,MAAM,GAAG,EAAE,CAAC;QAEhB,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,EAAE;YAC/B,MAAM,IAAI,IAAI,CAAC;QACnB,CAAC,CAAC,CAAC;QAEH,OAAO,CAAC,EAAE,CAAC,MAAM,EAAE,UAAU,IAAI;YAE7B,IAAI,IAAI,EAAE,CAAC;gBACP,cAAG,CAAC,KAAK,CAAC,2BAA2B,IAAI,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC;gBACzD,MAAM,CAAC,IAAI,KAAK,CAAC,2BAA2B,IAAI,EAAE,CAAC,CAAC,CAAC;YACzD,CAAC;QACL,CAAC,CAAC,CAAC;QAEH,OAAO,CAAC,EAAE,CAAC,OAAO,EAAE,UAAU,IAAI;YAC9B,IAAI,IAAI,EAAE,CAAC;gBACP,MAAM,CAAC,IAAI,KAAK,CAAC,6BAA6B,IAAI,EAAE,CAAC,CAAC,CAAC;YAC3D,CAAC;iBAAM,CAAC;gBACJ,MAAM,KAAK,GAAG,YAAE,CAAC,WAAW,CAAC,OAAO,CAAC,IAAI,CAAC,CAAC;gBAC3C,MAAM,MAAM,GAAG,KAAK,CAAC,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,QAAQ,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,IAAI,IAAI,CAAC,EAAE,CAAC,CAAC;gBACtF,cAAG,CAAC,IAAI,CAAC,oBAAoB,MAAM,CAAC,MAAM,SAAS,EAAE,KAAK,CAAC,CAAC;gBAC5D,OAAO,OAAO,CAAC,MAAM,CAAC,CAAC;YAC3B,CAAC;YAAA,CAAC;QACN,CAAC,CAAC,CAAC;QAEH,OAAO,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE;YACxB,MAAM,CAAC,GAAG,CAAC,CAAC;QAChB,CAAC,CAAC,CAAC;IAEP,CAAC,CAAC,CAAC;AAEP,CAAC;AAGD;;GAEG;AAEI,KAAK,UAAU,eAAe,CAAC,IAAqB,EAAE,KAAe;IAExE,MAAM,OAAO,GAAG,aAAG,CAAC,OAAO,EAAE,CAAC;IAC9B,cAAG,CAAC,IAAI,CAAC,wBAAwB,EAAE,EAAE,OAAO,EAAE,OAAO,CAAC,IAAI,EAAE,UAAU,EAAE,OAAO,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC;IAE9F,IAAI,IAAI,YAAY,MAAM,EAAE,CAAC;QACzB,YAAE,CAAC,aAAa,CAAC,GAAG,OAAO,CAAC,IAAI,YAAY,EAAE,IAAI,CAAC,CAAC;IACxD,CAAC;IACD,MAAM,QAAQ,GAAG,OAAO,IAAI,KAAK,QAAQ,CAAC,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,GAAG,OAAO,CAAC,IAAI,YAAY,CAAC;IAE/E,MAAM,IAAI,GAAG;QACT,OAAO;QACP,IAAI,EAAE,GAAG,OAAO,CAAC,IAAI,aAAa;QAClC,IAAI,EAAE,0BAA0B;QAChC,QAAQ;QACR,KAAK,CAAC,IAAI,CAAC,GAAG,CAAC;KAClB,CAAC;IAEF,OAAO,IAAI,OAAO,CAAC,CAAC,OAAO,EAAE,MAAM,EAAE,EAAE;QAEnC,MAAM,OAAO,GAAG,IAAA,qBAAK,EAAC,QAAQ,EAAE,IAAI,CAAC,CAAC;QACtC,cAAG,CAAC,IAAI,CAAC,0BAA0B,EAAE,EAAE,OAAO,EAAE,OAAO,CAAC,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE,OAAO,CAAC,SAAS,EAAE,CAAC,CAAC;QAEtG,IAAI,MAAM,GAAG,EAAE,CAAC;QAEhB,OAAO,CAAC,MAAM,CAAC,EAAE,CAAC,MAAM,EAAE,CAAC,IAAI,EAAE,EAAE;YAC/B,MAAM,IAAI,IAAI,CAAC;QACnB,CAAC,CAAC,CAAC;QAEH,OAAO,CAAC,EAAE,CAAC,MAAM,EAAE,UAAU,IAAI;YAE7B,IAAI,IAAI,EAAE,CAAC;gBACP,cAAG,CAAC,KAAK,CAAC,2BAA2B,IAAI,EAAE,EAAE,EAAE,MAAM,EAAE,CAAC,CAAC;gBACzD,MAAM,CAAC,IAAI,KAAK,CAAC,2BAA2B,IAAI,EAAE,CAAC,CAAC,CAAC;YACzD,CAAC;QACL,CAAC,CAAC,CAAC;QAEH,OAAO,CAAC,EAAE,CAAC,OAAO,EAAE,UAAU,IAAI;YAC9B,IAAI,IAAI,EAAE,CAAC;gBACP,MAAM,CAAC,IAAI,KAAK,CAAC,6BAA6B,IAAI,EAAE,CAAC,CAAC,CAAC;YAC3D,CAAC;iBAAM,CAAC;gBACJ,MAAM,IAAI,GAAG,GAAG,OAAO,CAAC,IAAI,aAAa,CAAC;gBAC1C,cAAG,CAAC,IAAI,CAAC,0BAA0B,EAAE,EAAE,KAAK,EAAE,IAAI,EAAE,CAAC,CAAC;gBACtD,OAAO,OAAO,CAAC,IAAI,CAAC,CAAC;YACzB,CAAC;YAAA,CAAC;QACN,CAAC,CAAC,CAAC;QAEH,OAAO,CAAC,EAAE,CAAC,OAAO,EAAE,CAAC,GAAG,EAAE,EAAE;YACxB,MAAM,CAAC,GAAG,CAAC,CAAC;QAChB,CAAC,CAAC,CAAC;IAEP,CAAC,CAAC,CAAC;AAGP,CAAC"}

package/lib/cjs/conversion/pandoc.js

@@ -0,0 +1,39 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.manyToMarkdown = manyToMarkdown;
+const activity_1 = require("@temporalio/activity");
+const child_process_1 = require("child_process");
+const stream_1 = require("stream");
+function manyToMarkdown(buffer, fromFormat) {
+    const fromType = undefined;
+    return new Promise((resolve, reject) => {
+        activity_1.log.info(`Converting ${fromType} to markdown`);
+        const input = new stream_1.PassThrough();
+        input.end(buffer);
+        let result = [];
+        const command = (0, child_process_1.spawn)("pandoc", ["-t", "markdown", '-f', fromFormat], {
+            stdio: 'pipe',
+        });
+        input.pipe(command.stdin);
+        command.stdout.on('data', function (data) {
+            result.push(data.toString());
+        });
+        command.on('exit', function (code) {
+            if (code) {
+                reject(new Error(`pandoc exited with code ${code}`));
+            }
+        });
+        command.on('close', function (code) {
+            if (code) {
+                reject(new Error(`pandoc exited with code ${code}`));
+            }
+            else {
+                resolve(result.join(''));
+            }
+        });
+        command.on('error', (err) => {
+            reject(err);
+        });
+    });
+}
+//# sourceMappingURL=pandoc.js.map