file2md 1.0.3

package/dist/parsers/docx-parser.js

@@ -0,0 +1,237 @@
+import JSZip from 'jszip';
+import { parseStringPromise } from 'xml2js';
+import { LayoutParser } from '../utils/layout-parser.js';
+import { ParseError, InvalidFileError } from '../types/errors.js';
+/**
+ * Parse DOCX buffer and convert to markdown with layout preservation
+ */
+export async function parseDocx(buffer, imageExtractor, chartExtractor, options = {}) {
+    try {
+        const zip = await JSZip.loadAsync(buffer);
+        const documentXml = zip.file('word/document.xml');
+        if (!documentXml) {
+            throw new InvalidFileError('Invalid DOCX file: missing document.xml');
+        }
+        // Extract images first
+        const extractedImages = options.extractImages !== false
+            ? await imageExtractor.extractImagesFromZip(zip, 'word/')
+            : [];
+        // Extract charts if enabled
+        const extractedCharts = options.extractCharts !== false
+            ? await chartExtractor.extractChartsFromZip(zip, 'word/')
+            : [];
+        // Initialize layout parser
+        const layoutParser = new LayoutParser();
+        const xmlContent = await documentXml.async('string');
+        const result = await parseStringPromise(xmlContent);
+        const body = result['w:document'][0]['w:body'][0];
+        let markdown = '';
+        // Process paragraphs
+        for (const element of body['w:p'] || []) {
+            const paragraph = await parseParagraph(element, imageExtractor, extractedImages);
+            if (paragraph.trim()) {
+                markdown += paragraph + '\n\n';
+            }
+        }
+        // Process tables
+        for (const table of body['w:tbl'] || []) {
+            const tableMarkdown = await parseAdvancedTable(table, layoutParser, imageExtractor, extractedImages);
+            if (tableMarkdown.trim()) {
+                markdown += tableMarkdown + '\n\n';
+            }
+        }
+        return {
+            markdown: markdown.trim(),
+            images: extractedImages,
+            charts: extractedCharts.map(chart => chart.data),
+            metadata: {
+                paragraphCount: (body['w:p'] || []).length,
+                tableCount: (body['w:tbl'] || []).length
+            }
+        };
+    }
+    catch (error) {
+        if (error instanceof InvalidFileError) {
+            throw error;
+        }
+        const message = error instanceof Error ? error.message : 'Unknown error';
+        throw new ParseError('DOCX', message, error);
+    }
+}
+async function parseAdvancedTable(table, layoutParser, imageExtractor, extractedImages) {
+    const tableData = table;
+    const rows = tableData['w:tr'] || [];
+    if (rows.length === 0)
+        return '';
+    const tableStruct = { rows: [] };
+    for (const row of rows) {
+        const cells = row['w:tc'] || [];
+        const rowData = { cells: [] };
+        for (const cell of cells) {
+            const cellData = {
+                text: '',
+                bold: false,
+                italic: false,
+                alignment: 'left',
+                backgroundColor: undefined,
+                colSpan: 1,
+                rowSpan: 1,
+                merged: false
+            };
+            // Extract cell properties
+            const tcPr = cell['w:tcPr'];
+            if (tcPr?.[0]) {
+                // Check for merged cells
+                if (tcPr[0]['w:gridSpan']) {
+                    cellData.colSpan = parseInt(tcPr[0]['w:gridSpan'][0].$.val) || 1;
+                }
+                if (tcPr[0]['w:vMerge']) {
+                    cellData.merged = true;
+                }
+                // Check for background color
+                if (tcPr[0]['w:shd']?.[0]?.$.fill) {
+                    cellData.backgroundColor = tcPr[0]['w:shd'][0].$.fill;
+                }
+            }
+            // Extract cell content
+            if (cell['w:p']) {
+                const cellTexts = [];
+                for (const paragraph of cell['w:p']) {
+                    const paragraphData = await parseAdvancedParagraph(paragraph, imageExtractor, extractedImages);
+                    if (paragraphData.text.trim()) {
+                        cellTexts.push(paragraphData.text);
+                        // Inherit formatting from paragraph
+                        if (paragraphData.bold)
+                            cellData.bold = true;
+                        if (paragraphData.italic)
+                            cellData.italic = true;
+                        if (paragraphData.alignment !== 'left')
+                            cellData.alignment = paragraphData.alignment;
+                    }
+                }
+                cellData.text = cellTexts.join(' ');
+            }
+            rowData.cells.push(cellData);
+        }
+        tableStruct.rows.push(rowData);
+    }
+    return layoutParser.parseAdvancedTable(tableStruct, {
+        preserveAlignment: true,
+        showBorders: true,
+        preserveColors: true
+    });
+}
+async function parseAdvancedParagraph(paragraph, imageExtractor, extractedImages) {
+    const para = paragraph;
+    let text = '';
+    let bold = false;
+    let italic = false;
+    let alignment = 'left';
+    let fontSize = 'normal';
+    let isList = false;
+    let listLevel = 0;
+    // Check paragraph properties
+    const pPr = para['w:pPr'];
+    if (pPr?.[0]) {
+        // Check alignment
+        if (pPr[0]['w:jc']?.[0]?.$.val) {
+            const alignValue = pPr[0]['w:jc'][0].$.val;
+            alignment = (['left', 'center', 'right', 'justify'].includes(alignValue)
+                ? alignValue
+                : 'left');
+        }
+        // Check if it's a list
+        if (pPr[0]['w:numPr']) {
+            isList = true;
+            if (pPr[0]['w:numPr'][0]['w:ilvl']) {
+                listLevel = parseInt(pPr[0]['w:numPr'][0]['w:ilvl'][0].$.val) || 0;
+            }
+        }
+    }
+    if (para['w:r']) {
+        for (const run of para['w:r']) {
+            // Check for images/drawings
+            if (run['w:drawing'] || run['w:pict']) {
+                const imageRef = await extractImageFromRun(run, imageExtractor, extractedImages);
+                if (imageRef) {
+                    text += imageRef + '\n';
+                }
+            }
+            // Extract text with formatting
+            if (run['w:t']) {
+                let runText = '';
+                for (const textElement of run['w:t']) {
+                    if (typeof textElement === 'string') {
+                        runText += textElement;
+                    }
+                    else if (textElement && typeof textElement === 'object' && '_' in textElement) {
+                        runText += textElement._;
+                    }
+                }
+                // Apply formatting
+                const rPr = run['w:rPr']?.[0];
+                if (rPr) {
+                    if (rPr['w:b']) {
+                        runText = `**${runText}**`;
+                        bold = true;
+                    }
+                    if (rPr['w:i']) {
+                        runText = `*${runText}*`;
+                        italic = true;
+                    }
+                    if (rPr['w:sz']?.[0]?.$.val) {
+                        fontSize = parseInt(rPr['w:sz'][0].$.val) / 2; // Convert half-points to points
+                    }
+                }
+                text += runText;
+            }
+        }
+    }
+    // Apply list formatting
+    if (isList && text.trim()) {
+        const indent = '  '.repeat(listLevel);
+        text = `${indent}- ${text.trim()}`;
+    }
+    // Apply heading formatting
+    if (pPr?.[0]?.['w:pStyle']?.[0]?.$.val) {
+        const styleVal = pPr[0]['w:pStyle'][0].$.val;
+        if (styleVal && (styleVal.includes('Heading') || styleVal.includes('heading'))) {
+            const match = styleVal.match(/(\d+)/);
+            if (match) {
+                const headingLevel = parseInt(match[1]);
+                const hashes = '#'.repeat(Math.min(headingLevel, 6));
+                text = `${hashes} ${text.trim()}`;
+            }
+        }
+    }
+    // Apply font size formatting
+    if (fontSize !== 'normal' && text.trim()) {
+        const layoutParser = new LayoutParser();
+        text = layoutParser.formatWithSize(text, fontSize);
+    }
+    return {
+        text,
+        bold,
+        italic,
+        alignment,
+        fontSize,
+        isList,
+        listLevel
+    };
+}
+async function parseParagraph(paragraph, imageExtractor, extractedImages) {
+    const advancedData = await parseAdvancedParagraph(paragraph, imageExtractor, extractedImages);
+    return advancedData.text;
+}
+async function extractImageFromRun(run, imageExtractor, extractedImages) {
+    // This is a simplified image extraction - in reality, we'd need to parse the drawing XML
+    // and match it with the extracted images
+    if (extractedImages.length > 0) {
+        const img = extractedImages.find(img => img.savedPath);
+        if (img) {
+            return imageExtractor.getImageMarkdown('Document Image', img.savedPath);
+        }
+    }
+    return null;
+}
+//# sourceMappingURL=docx-parser.js.map

package/dist/parsers/docx-parser.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"docx-parser.js","sourceRoot":"","sources":["../../src/parsers/docx-parser.ts"],"names":[],"mappings":"AAAA,OAAO,KAAK,MAAM,OAAO,CAAC;AAC1B,OAAO,EAAE,kBAAkB,EAAE,MAAM,QAAQ,CAAC;AAK5C,OAAO,EAAE,YAAY,EAAE,MAAM,2BAA2B,CAAC;AACzD,OAAO,EAAE,UAAU,EAAE,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AAiDlE;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,SAAS,CAC7B,MAAc,EACd,cAA8B,EAC9B,cAA8B,EAC9B,UAA4B,EAAE;IAE9B,IAAI,CAAC;QACH,MAAM,GAAG,GAAG,MAAM,KAAK,CAAC,SAAS,CAAC,MAAM,CAAC,CAAC;QAC1C,MAAM,WAAW,GAAG,GAAG,CAAC,IAAI,CAAC,mBAAmB,CAAC,CAAC;QAElD,IAAI,CAAC,WAAW,EAAE,CAAC;YACjB,MAAM,IAAI,gBAAgB,CAAC,yCAAyC,CAAC,CAAC;QACxE,CAAC;QAED,uBAAuB;QACvB,MAAM,eAAe,GAAG,OAAO,CAAC,aAAa,KAAK,KAAK;YACrD,CAAC,CAAC,MAAM,cAAc,CAAC,oBAAoB,CAAC,GAAG,EAAE,OAAO,CAAC;YACzD,CAAC,CAAC,EAAE,CAAC;QAEP,4BAA4B;QAC5B,MAAM,eAAe,GAAG,OAAO,CAAC,aAAa,KAAK,KAAK;YACrD,CAAC,CAAC,MAAM,cAAc,CAAC,oBAAoB,CAAC,GAAG,EAAE,OAAO,CAAC;YACzD,CAAC,CAAC,EAAE,CAAC;QAEP,2BAA2B;QAC3B,MAAM,YAAY,GAAG,IAAI,YAAY,EAAE,CAAC;QAExC,MAAM,UAAU,GAAG,MAAM,WAAW,CAAC,KAAK,CAAC,QAAQ,CAAC,CAAC;QACrD,MAAM,MAAM,GAAG,MAAM,kBAAkB,CAAC,UAAU,CAAiB,CAAC;QAEpE,MAAM,IAAI,GAAG,MAAM,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC;QAClD,IAAI,QAAQ,GAAG,EAAE,CAAC;QAElB,qBAAqB;QACrB,KAAK,MAAM,OAAO,IAAI,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,EAAE,CAAC;YACxC,MAAM,SAAS,GAAG,MAAM,cAAc,CAAC,OAAO,EAAE,cAAc,EAAE,eAAe,CAAC,CAAC;YACjF,IAAI,SAAS,CAAC,IAAI,EAAE,EAAE,CAAC;gBACrB,QAAQ,IAAI,SAAS,GAAG,MAAM,CAAC;YACjC,CAAC;QACH,CAAC;QAED,iBAAiB;QACjB,KAAK,MAAM,KAAK,IAAI,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,EAAE,CAAC;YACxC,MAAM,aAAa,GAAG,MAAM,kBAAkB,CAAC,KAAK,EAAE,YAAY,EAAE,cAAc,EAAE,eAAe,CAAC,CAAC;YACrG,IAAI,aAAa,CAAC,IAAI,EAAE,EAAE,CAAC;gBACzB,QAAQ,IAAI,aAAa,GAAG,MAAM,CAAC;YACrC,CAAC;QACH,CAAC;QAED,OAAO;YACL,QAAQ,EAAE,QAAQ,CAAC,IAAI,EAAE;YACzB,MAAM,EAAE,eAAe;YACvB,MAAM,EAAE,eAAe,CAAC,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC;YAChD,QAAQ,EAAE;gBACR,cAAc,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM;gBAC1C,UAAU,EAAE,CAAC,IAAI,CAAC,OAAO,CAAC,IAAI,EAAE,CAAC,CAAC,MAAM;aACzC;SACF,CAAC;IACJ,CAAC;IAAC,OAAO,KAAc,EAAE,CAAC;QACxB,IAAI,KAAK,YAAY,gBAAgB,EAAE,CAAC;YACtC,MAAM,KAAK,CAAC;QACd,CAAC;QAED,MAAM,OAAO,GAAG,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,eAAe,CAAC;QACzE,MAAM,IAAI,UAAU,CAAC,MAAM,EAAE,OAAO,EAAE,KAAc,CAAC,CAAC;IACxD,CAAC;AACH,CAAC;AAED,KAAK,UAAU,kBAAkB,CAC/B,KAAc,EACd,YAA0B,EAC1B,cAA8B,EAC9B,eAAqC;IAErC,MAAM,SAAS,GAAG,KAAY,CAAC;IAC/B,MAAM,IAAI,GAAG,SAAS,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;IACrC,IAAI,IAAI,CAAC,MAAM,KAAK,CAAC;QAAE,OAAO,EAAE,CAAC;IAEjC,MAAM,WAAW,GAAc,EAAE,IAAI,EAAE,EAAE,EAAE,CAAC;IAE5C,KAAK,MAAM,GAAG,IAAI,IAAI,EAAE,CAAC;QACvB,MAAM,KAAK,GAAG,GAAG,CAAC,MAAM,CAAC,IAAI,EAAE,CAAC;QAChC,MAAM,OAAO,GAAY,EAAE,KAAK,EAAE,EAAE,EAAE,CAAC;QAEvC,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;YACzB,MAAM,QAAQ,GAAa;gBACzB,IAAI,EAAE,EAAE;gBACR,IAAI,EAAE,KAAK;gBACX,MAAM,EAAE,KAAK;gBACb,SAAS,EAAE,MAAuB;gBAClC,eAAe,EAAE,SAAS;gBAC1B,OAAO,EAAE,CAAC;gBACV,OAAO,EAAE,CAAC;gBACV,MAAM,EAAE,KAAK;aACd,CAAC;YAEF,0BAA0B;YAC1B,MAAM,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC,CAAC;YAC5B,IAAI,IAAI,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;gBACd,yBAAyB;gBACzB,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,YAAY,CAAC,EAAE,CAAC;oBAC1B,QAAQ,CAAC,OAAO,GAAG,QAAQ,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,YAAY,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;gBACnE,CAAC;gBACD,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,EAAE,CAAC;oBACxB,QAAQ,CAAC,MAAM,GAAG,IAAI,CAAC;gBACzB,CAAC;gBAED,6BAA6B;gBAC7B,IAAI,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,IAAI,EAAE,CAAC;oBAClC,QAAQ,CAAC,eAAe,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC,OAAO,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC;gBACxD,CAAC;YACH,CAAC;YAED,uBAAuB;YACvB,IAAI,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;gBAChB,MAAM,SAAS,GAAa,EAAE,CAAC;gBAC/B,KAAK,MAAM,SAAS,IAAI,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;oBACpC,MAAM,aAAa,GAAG,MAAM,sBAAsB,CAAC,SAAS,EAAE,cAAc,EAAE,eAAe,CAAC,CAAC;oBAC/F,IAAI,aAAa,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC;wBAC9B,SAAS,CAAC,IAAI,CAAC,aAAa,CAAC,IAAI,CAAC,CAAC;wBAEnC,oCAAoC;wBACpC,IAAI,aAAa,CAAC,IAAI;4BAAE,QAAQ,CAAC,IAAI,GAAG,IAAI,CAAC;wBAC7C,IAAI,aAAa,CAAC,MAAM;4BAAE,QAAQ,CAAC,MAAM,GAAG,IAAI,CAAC;wBACjD,IAAI,aAAa,CAAC,SAAS,KAAK,MAAM;4BAAE,QAAQ,CAAC,SAAS,GAAG,aAAa,CAAC,SAAS,CAAC;oBACvF,CAAC;gBACH,CAAC;gBACD,QAAQ,CAAC,IAAI,GAAG,SAAS,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YACtC,CAAC;YAED,OAAO,CAAC,KAAK,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QAC/B,CAAC;QAED,WAAW,CAAC,IAAI,CAAC,IAAI,CAAC,OAAO,CAAC,CAAC;IACjC,CAAC;IAED,OAAO,YAAY,CAAC,kBAAkB,CAAC,WAAW,EAAE;QAClD,iBAAiB,EAAE,IAAI;QACvB,WAAW,EAAE,IAAI;QACjB,cAAc,EAAE,IAAI;KACrB,CAAC,CAAC;AACL,CAAC;AAED,KAAK,UAAU,sBAAsB,CACnC,SAAkB,EAClB,cAA8B,EAC9B,eAAqC;IAErC,MAAM,IAAI,GAAG,SAAgB,CAAC;IAC9B,IAAI,IAAI,GAAG,EAAE,CAAC;IACd,IAAI,IAAI,GAAG,KAAK,CAAC;IACjB,IAAI,MAAM,GAAG,KAAK,CAAC;IACnB,IAAI,SAAS,GAAkB,MAAM,CAAC;IACtC,IAAI,QAAQ,GAAoB,QAAQ,CAAC;IACzC,IAAI,MAAM,GAAG,KAAK,CAAC;IACnB,IAAI,SAAS,GAAG,CAAC,CAAC;IAElB,6BAA6B;IAC7B,MAAM,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,CAAC;IAC1B,IAAI,GAAG,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC;QACb,kBAAkB;QAClB,IAAI,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,GAAG,EAAE,CAAC;YAC/B,MAAM,UAAU,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC;YAC3C,SAAS,GAAG,CAAC,CAAC,MAAM,EAAE,QAAQ,EAAE,OAAO,EAAE,SAAS,CAAC,CAAC,QAAQ,CAAC,UAAU,CAAC;gBACtE,CAAC,CAAC,UAAU;gBACZ,CAAC,CAAC,MAAM,CAAkB,CAAC;QAC/B,CAAC;QAED,uBAAuB;QACvB,IAAI,GAAG,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,EAAE,CAAC;YACtB,MAAM,GAAG,IAAI,CAAC;YACd,IAAI,GAAG,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,EAAE,CAAC;gBACnC,SAAS,GAAG,QAAQ,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC;YACrE,CAAC;QACH,CAAC;IACH,CAAC;IAED,IAAI,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;QAChB,KAAK,MAAM,GAAG,IAAI,IAAI,CAAC,KAAK,CAAC,EAAE,CAAC;YAC9B,4BAA4B;YAC5B,IAAI,GAAG,CAAC,WAAW,CAAC,IAAI,GAAG,CAAC,QAAQ,CAAC,EAAE,CAAC;gBACtC,MAAM,QAAQ,GAAG,MAAM,mBAAmB,CAAC,GAAG,EAAE,cAAc,EAAE,eAAe,CAAC,CAAC;gBACjF,IAAI,QAAQ,EAAE,CAAC;oBACb,IAAI,IAAI,QAAQ,GAAG,IAAI,CAAC;gBAC1B,CAAC;YACH,CAAC;YAED,+BAA+B;YAC/B,IAAI,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;gBACf,IAAI,OAAO,GAAG,EAAE,CAAC;gBACjB,KAAK,MAAM,WAAW,IAAI,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;oBACrC,IAAI,OAAO,WAAW,KAAK,QAAQ,EAAE,CAAC;wBACpC,OAAO,IAAI,WAAW,CAAC;oBACzB,CAAC;yBAAM,IAAI,WAAW,IAAI,OAAO,WAAW,KAAK,QAAQ,IAAI,GAAG,IAAI,WAAW,EAAE,CAAC;wBAChF,OAAO,IAAK,WAAmB,CAAC,CAAC,CAAC;oBACpC,CAAC;gBACH,CAAC;gBAED,mBAAmB;gBACnB,MAAM,GAAG,GAAG,GAAG,CAAC,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC;gBAC9B,IAAI,GAAG,EAAE,CAAC;oBACR,IAAI,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;wBACf,OAAO,GAAG,KAAK,OAAO,IAAI,CAAC;wBAC3B,IAAI,GAAG,IAAI,CAAC;oBACd,CAAC;oBACD,IAAI,GAAG,CAAC,KAAK,CAAC,EAAE,CAAC;wBACf,OAAO,GAAG,IAAI,OAAO,GAAG,CAAC;wBACzB,MAAM,GAAG,IAAI,CAAC;oBAChB,CAAC;oBACD,IAAI,GAAG,CAAC,MAAM,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,GAAG,EAAE,CAAC;wBAC5B,QAAQ,GAAG,QAAQ,CAAC,GAAG,CAAC,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC,CAAC,gCAAgC;oBACjF,CAAC;gBACH,CAAC;gBAED,IAAI,IAAI,OAAO,CAAC;YAClB,CAAC;QACH,CAAC;IACH,CAAC;IAED,wBAAwB;IACxB,IAAI,MAAM,IAAI,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC;QAC1B,MAAM,MAAM,GAAG,IAAI,CAAC,MAAM,CAAC,SAAS,CAAC,CAAC;QACtC,IAAI,GAAG,GAAG,MAAM,KAAK,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC;IACrC,CAAC;IAED,2BAA2B;IAC3B,IAAI,GAAG,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,UAAU,CAAC,EAAE,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,GAAG,EAAE,CAAC;QACvC,MAAM,QAAQ,GAAG,GAAG,CAAC,CAAC,CAAC,CAAC,UAAU,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC;QAC7C,IAAI,QAAQ,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,SAAS,CAAC,IAAI,QAAQ,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC,EAAE,CAAC;YAC/E,MAAM,KAAK,GAAG,QAAQ,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC;YACtC,IAAI,KAAK,EAAE,CAAC;gBACV,MAAM,YAAY,GAAG,QAAQ,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC;gBACxC,MAAM,MAAM,GAAG,GAAG,CAAC,MAAM,CAAC,IAAI,CAAC,GAAG,CAAC,YAAY,EAAE,CAAC,CAAC,CAAC,CAAC;gBACrD,IAAI,GAAG,GAAG,MAAM,IAAI,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC;YACpC,CAAC;QACH,CAAC;IACH,CAAC;IAED,6BAA6B;IAC7B,IAAI,QAAQ,KAAK,QAAQ,IAAI,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC;QACzC,MAAM,YAAY,GAAG,IAAI,YAAY,EAAE,CAAC;QACxC,IAAI,GAAG,YAAY,CAAC,cAAc,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAC;IACrD,CAAC;IAED,OAAO;QACL,IAAI;QACJ,IAAI;QACJ,MAAM;QACN,SAAS;QACT,QAAQ;QACR,MAAM;QACN,SAAS;KACV,CAAC;AACJ,CAAC;AAED,KAAK,UAAU,cAAc,CAC3B,SAAkB,EAClB,cAA8B,EAC9B,eAAqC;IAErC,MAAM,YAAY,GAAG,MAAM,sBAAsB,CAAC,SAAS,EAAE,cAAc,EAAE,eAAe,CAAC,CAAC;IAC9F,OAAO,YAAY,CAAC,IAAI,CAAC;AAC3B,CAAC;AAED,KAAK,UAAU,mBAAmB,CAChC,GAAY,EACZ,cAA8B,EAC9B,eAAqC;IAErC,yFAAyF;IACzF,yCAAyC;IACzC,IAAI,eAAe,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QAC/B,MAAM,GAAG,GAAG,eAAe,CAAC,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,SAAS,CAAC,CAAC;QACvD,IAAI,GAAG,EAAE,CAAC;YACR,OAAO,cAAc,CAAC,gBAAgB,CAAC,gBAAgB,EAAE,GAAG,CAAC,SAAS,CAAC,CAAC;QAC1E,CAAC;IACH,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC"}

package/dist/parsers/pdf-parser.d.ts

@@ -0,0 +1,8 @@
+import type { Buffer } from 'node:buffer';
+import { type PDFParseOptions, type PDFParseResult } from '../utils/pdf-extractor.js';
+import type { ImageExtractor } from '../utils/image-extractor.js';
+/**
+ * Parse PDF buffer and convert to markdown with enhanced layout preservation
+ */
+export declare function parsePdf(buffer: Buffer, imageExtractor: ImageExtractor, options?: PDFParseOptions): Promise<PDFParseResult>;
+//# sourceMappingURL=pdf-parser.d.ts.map

package/dist/parsers/pdf-parser.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdf-parser.d.ts","sourceRoot":"","sources":["../../src/parsers/pdf-parser.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,aAAa,CAAC;AAE1C,OAAO,EAAgB,KAAK,eAAe,EAAE,KAAK,cAAc,EAAE,MAAM,2BAA2B,CAAC;AAEpG,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,6BAA6B,CAAC;AAGlE;;GAEG;AACH,wBAAsB,QAAQ,CAC5B,MAAM,EAAE,MAAM,EACd,cAAc,EAAE,cAAc,EAC9B,OAAO,GAAE,eAAoB,GAC5B,OAAO,CAAC,cAAc,CAAC,CAuFzB"}

package/dist/parsers/pdf-parser.js

@@ -0,0 +1,98 @@
+import pdfParse from 'pdf-parse';
+import { PDFExtractor } from '../utils/pdf-extractor.js';
+import { ParseError, InvalidFileError } from '../types/errors.js';
+/**
+ * Parse PDF buffer and convert to markdown with enhanced layout preservation
+ */
+export async function parsePdf(buffer, imageExtractor, options = {}) {
+    try {
+        const data = await pdfParse(buffer);
+        const pdfExtractor = new PDFExtractor(imageExtractor);
+        let markdown = '';
+        const images = [];
+        let pageCount = data.numpages || 1;
+        // Apply maxPages limit if specified
+        if (options.maxPages && options.maxPages > 0) {
+            pageCount = Math.min(pageCount, options.maxPages);
+        }
+        // Try to extract text with enhanced layout
+        if (data.text && data.text.trim()) {
+            console.log('📄 Extracting text with layout enhancement...');
+            try {
+                const enhancedText = await pdfExtractor.enhanceTextWithLayout(data.text, data);
+                markdown += enhancedText;
+            }
+            catch (layoutError) {
+                console.warn('Layout enhancement failed, falling back to basic text extraction');
+                // Fall back to basic text extraction
+                markdown = extractBasicText(data.text);
+            }
+        }
+        // If text is minimal or extraction failed, convert pages to images
+        if (!data.text || data.text.trim().length < 100) {
+            console.log('📸 Converting PDF pages to images for better preservation...');
+            try {
+                const pageImages = await pdfExtractor.extractImagesFromPDF(buffer);
+                if (pageImages.length > 0) {
+                    // Convert page images to ImageData format
+                    for (const page of pageImages) {
+                        images.push({
+                            originalPath: `page_${page.pageNumber}`,
+                            savedPath: page.imagePath,
+                            basePath: '',
+                            format: 'png',
+                            dimensions: page.dimensions
+                        });
+                    }
+                    if (markdown.trim()) {
+                        markdown += '\n\n---\n\n## Visual Content\n\n';
+                    }
+                    markdown += await pdfExtractor.createPageBreaks(pageImages);
+                }
+            }
+            catch (imageError) {
+                console.warn('Failed to extract PDF as images:', imageError instanceof Error ? imageError.message : 'Unknown error');
+            }
+        }
+        // Fallback to basic text if everything else fails
+        if (!markdown.trim()) {
+            if (data.text && data.text.trim()) {
+                markdown = extractBasicText(data.text);
+            }
+            else {
+                throw new InvalidFileError('PDF file appears to be empty or contains no extractable text');
+            }
+        }
+        return {
+            markdown,
+            images,
+            pageCount,
+            metadata: {
+                version: data.version || 'unknown',
+                info: data.info || {},
+                textLength: data.text?.length || 0
+            }
+        };
+    }
+    catch (error) {
+        if (error instanceof InvalidFileError) {
+            throw error;
+        }
+        const message = error instanceof Error ? error.message : 'Unknown error';
+        if (message.includes('Invalid PDF') || message.includes('PDF')) {
+            throw new InvalidFileError('Invalid or corrupted PDF file', error);
+        }
+        throw new ParseError('PDF', message, error);
+    }
+}
+/**
+ * Extract basic text from PDF with minimal formatting
+ */
+function extractBasicText(text) {
+    const lines = text.split('\n');
+    const cleanedLines = lines
+        .map(line => line.trim())
+        .filter(line => line.length > 0);
+    return cleanedLines.join('\n');
+}
+//# sourceMappingURL=pdf-parser.js.map

package/dist/parsers/pdf-parser.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdf-parser.js","sourceRoot":"","sources":["../../src/parsers/pdf-parser.ts"],"names":[],"mappings":"AAAA,OAAO,QAAQ,MAAM,WAAW,CAAC;AAGjC,OAAO,EAAE,YAAY,EAA6C,MAAM,2BAA2B,CAAC;AACpG,OAAO,EAAE,UAAU,EAAE,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AAIlE;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,QAAQ,CAC5B,MAAc,EACd,cAA8B,EAC9B,UAA2B,EAAE;IAE7B,IAAI,CAAC;QACH,MAAM,IAAI,GAAG,MAAM,QAAQ,CAAC,MAAM,CAAC,CAAC;QACpC,MAAM,YAAY,GAAG,IAAI,YAAY,CAAC,cAAc,CAAC,CAAC;QAEtD,IAAI,QAAQ,GAAG,EAAE,CAAC;QAClB,MAAM,MAAM,GAAgB,EAAE,CAAC;QAC/B,IAAI,SAAS,GAAG,IAAI,CAAC,QAAQ,IAAI,CAAC,CAAC;QAEnC,oCAAoC;QACpC,IAAI,OAAO,CAAC,QAAQ,IAAI,OAAO,CAAC,QAAQ,GAAG,CAAC,EAAE,CAAC;YAC7C,SAAS,GAAG,IAAI,CAAC,GAAG,CAAC,SAAS,EAAE,OAAO,CAAC,QAAQ,CAAC,CAAC;QACpD,CAAC;QAED,2CAA2C;QAC3C,IAAI,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC;YAClC,OAAO,CAAC,GAAG,CAAC,+CAA+C,CAAC,CAAC;YAC7D,IAAI,CAAC;gBACH,MAAM,YAAY,GAAG,MAAM,YAAY,CAAC,qBAAqB,CAAC,IAAI,CAAC,IAAI,EAAE,IAAI,CAAC,CAAC;gBAC/E,QAAQ,IAAI,YAAY,CAAC;YAC3B,CAAC;YAAC,OAAO,WAAoB,EAAE,CAAC;gBAC9B,OAAO,CAAC,IAAI,CAAC,kEAAkE,CAAC,CAAC;gBACjF,qCAAqC;gBACrC,QAAQ,GAAG,gBAAgB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACzC,CAAC;QACH,CAAC;QAED,mEAAmE;QACnE,IAAI,CAAC,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC,MAAM,GAAG,GAAG,EAAE,CAAC;YAChD,OAAO,CAAC,GAAG,CAAC,8DAA8D,CAAC,CAAC;YAC5E,IAAI,CAAC;gBACH,MAAM,UAAU,GAAG,MAAM,YAAY,CAAC,oBAAoB,CAAC,MAAM,CAAC,CAAC;gBACnE,IAAI,UAAU,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;oBAC1B,0CAA0C;oBAC1C,KAAK,MAAM,IAAI,IAAI,UAAU,EAAE,CAAC;wBAC9B,MAAM,CAAC,IAAI,CAAC;4BACV,YAAY,EAAE,QAAQ,IAAI,CAAC,UAAU,EAAE;4BACvC,SAAS,EAAE,IAAI,CAAC,SAAS;4BACzB,QAAQ,EAAE,EAAE;4BACZ,MAAM,EAAE,KAAK;4BACb,UAAU,EAAE,IAAI,CAAC,UAAU;yBAC5B,CAAC,CAAC;oBACL,CAAC;oBAED,IAAI,QAAQ,CAAC,IAAI,EAAE,EAAE,CAAC;wBACpB,QAAQ,IAAI,kCAAkC,CAAC;oBACjD,CAAC;oBACD,QAAQ,IAAI,MAAM,YAAY,CAAC,gBAAgB,CAAC,UAAU,CAAC,CAAC;gBAC9D,CAAC;YACH,CAAC;YAAC,OAAO,UAAmB,EAAE,CAAC;gBAC7B,OAAO,CAAC,IAAI,CAAC,kCAAkC,EAAE,UAAU,YAAY,KAAK,CAAC,CAAC,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC,CAAC,eAAe,CAAC,CAAC;YACvH,CAAC;QACH,CAAC;QAED,kDAAkD;QAClD,IAAI,CAAC,QAAQ,CAAC,IAAI,EAAE,EAAE,CAAC;YACrB,IAAI,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,IAAI,CAAC,IAAI,EAAE,EAAE,CAAC;gBAClC,QAAQ,GAAG,gBAAgB,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;YACzC,CAAC;iBAAM,CAAC;gBACN,MAAM,IAAI,gBAAgB,CAAC,8DAA8D,CAAC,CAAC;YAC7F,CAAC;QACH,CAAC;QAED,OAAO;YACL,QAAQ;YACR,MAAM;YACN,SAAS;YACT,QAAQ,EAAE;gBACR,OAAO,EAAE,IAAI,CAAC,OAAO,IAAI,SAAS;gBAClC,IAAI,EAAE,IAAI,CAAC,IAAI,IAAI,EAAE;gBACrB,UAAU,EAAE,IAAI,CAAC,IAAI,EAAE,MAAM,IAAI,CAAC;aACnC;SACF,CAAC;IAEJ,CAAC;IAAC,OAAO,KAAc,EAAE,CAAC;QACxB,IAAI,KAAK,YAAY,gBAAgB,EAAE,CAAC;YACtC,MAAM,KAAK,CAAC;QACd,CAAC;QAED,MAAM,OAAO,GAAG,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,eAAe,CAAC;QAEzE,IAAI,OAAO,CAAC,QAAQ,CAAC,aAAa,CAAC,IAAI,OAAO,CAAC,QAAQ,CAAC,KAAK,CAAC,EAAE,CAAC;YAC/D,MAAM,IAAI,gBAAgB,CAAC,+BAA+B,EAAE,KAAc,CAAC,CAAC;QAC9E,CAAC;QAED,MAAM,IAAI,UAAU,CAAC,KAAK,EAAE,OAAO,EAAE,KAAc,CAAC,CAAC;IACvD,CAAC;AACH,CAAC;AAED;;GAEG;AACH,SAAS,gBAAgB,CAAC,IAAY;IACpC,MAAM,KAAK,GAAG,IAAI,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAC/B,MAAM,YAAY,GAAG,KAAK;SACvB,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,IAAI,EAAE,CAAC;SACxB,MAAM,CAAC,IAAI,CAAC,EAAE,CAAC,IAAI,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC;IAEnC,OAAO,YAAY,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC;AACjC,CAAC"}

package/dist/parsers/pptx-parser.d.ts

@@ -0,0 +1,21 @@
+import type { Buffer } from 'node:buffer';
+import type { ImageExtractor } from '../utils/image-extractor.js';
+import type { ChartExtractor } from '../utils/chart-extractor.js';
+import type { ImageData, ChartData } from '../types/interfaces.js';
+export interface PptxParseOptions {
+    readonly preserveLayout?: boolean;
+    readonly extractImages?: boolean;
+    readonly extractCharts?: boolean;
+}
+export interface PptxParseResult {
+    readonly markdown: string;
+    readonly images: readonly ImageData[];
+    readonly charts: readonly ChartData[];
+    readonly slideCount: number;
+    readonly metadata: Record<string, unknown>;
+}
+/**
+ * Parse PPTX buffer and convert to markdown with layout preservation
+ */
+export declare function parsePptx(buffer: Buffer, imageExtractor: ImageExtractor, chartExtractor: ChartExtractor, options?: PptxParseOptions): Promise<PptxParseResult>;
+//# sourceMappingURL=pptx-parser.d.ts.map

package/dist/parsers/pptx-parser.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"pptx-parser.d.ts","sourceRoot":"","sources":["../../src/parsers/pptx-parser.ts"],"names":[],"mappings":"AAEA,OAAO,KAAK,EAAE,MAAM,EAAE,MAAM,aAAa,CAAC;AAE1C,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,6BAA6B,CAAC;AAClE,OAAO,KAAK,EAAE,cAAc,EAAE,MAAM,6BAA6B,CAAC;AAGlE,OAAO,KAAK,EACV,SAAS,EACT,SAAS,EAIV,MAAM,wBAAwB,CAAC;AAEhC,MAAM,WAAW,gBAAgB;IAC/B,QAAQ,CAAC,cAAc,CAAC,EAAE,OAAO,CAAC;IAClC,QAAQ,CAAC,aAAa,CAAC,EAAE,OAAO,CAAC;IACjC,QAAQ,CAAC,aAAa,CAAC,EAAE,OAAO,CAAC;CAClC;AAED,MAAM,WAAW,eAAe;IAC9B,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC;IAC1B,QAAQ,CAAC,MAAM,EAAE,SAAS,SAAS,EAAE,CAAC;IACtC,QAAQ,CAAC,MAAM,EAAE,SAAS,SAAS,EAAE,CAAC;IACtC,QAAQ,CAAC,UAAU,EAAE,MAAM,CAAC;IAC5B,QAAQ,CAAC,QAAQ,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAC;CAC5C;AAkBD;;GAEG;AACH,wBAAsB,SAAS,CAC7B,MAAM,EAAE,MAAM,EACd,cAAc,EAAE,cAAc,EAC9B,cAAc,EAAE,cAAc,EAC9B,OAAO,GAAE,gBAAqB,GAC7B,OAAO,CAAC,eAAe,CAAC,CAwE1B"}

package/dist/parsers/pptx-parser.js

@@ -0,0 +1,264 @@
+import JSZip from 'jszip';
+import { parseStringPromise } from 'xml2js';
+import { LayoutParser } from '../utils/layout-parser.js';
+import { ParseError } from '../types/errors.js';
+/**
+ * Parse PPTX buffer and convert to markdown with layout preservation
+ */
+export async function parsePptx(buffer, imageExtractor, chartExtractor, options = {}) {
+    try {
+        const zip = await JSZip.loadAsync(buffer);
+        // Extract images first
+        const extractedImages = options.extractImages !== false
+            ? await imageExtractor.extractImagesFromZip(zip, 'ppt/')
+            : [];
+        // Extract charts if enabled
+        const extractedCharts = options.extractCharts !== false
+            ? await chartExtractor.extractChartsFromZip(zip, 'ppt/')
+            : [];
+        // Initialize layout parser
+        const layoutParser = new LayoutParser();
+        const slideFiles = [];
+        zip.forEach((relativePath, file) => {
+            if (relativePath.startsWith('ppt/slides/slide') && relativePath.endsWith('.xml')) {
+                slideFiles.push({
+                    path: relativePath,
+                    file: file
+                });
+            }
+        });
+        slideFiles.sort((a, b) => {
+            const aNum = parseInt(a.path.match(/slide(\d+)\.xml/)?.[1] || '0');
+            const bNum = parseInt(b.path.match(/slide(\d+)\.xml/)?.[1] || '0');
+            return aNum - bNum;
+        });
+        let markdown = '';
+        for (let i = 0; i < slideFiles.length; i++) {
+            const slideFile = slideFiles[i];
+            const slideNumber = i + 1;
+            markdown += `## Slide ${slideNumber}\n\n`;
+            const xmlContent = await slideFile.file.async('string');
+            const slideContent = await extractAdvancedSlideContent(xmlContent, imageExtractor, extractedImages, slideNumber, layoutParser);
+            if (slideContent.trim()) {
+                markdown += slideContent + '\n\n';
+            }
+            else {
+                markdown += '*No content*\n\n';
+            }
+        }
+        return {
+            markdown: markdown.trim(),
+            images: extractedImages,
+            charts: extractedCharts.map(chart => chart.data),
+            slideCount: slideFiles.length,
+            metadata: {
+                totalSlides: slideFiles.length,
+                hasImages: extractedImages.length > 0,
+                hasCharts: extractedCharts.length > 0
+            }
+        };
+    }
+    catch (error) {
+        const message = error instanceof Error ? error.message : 'Unknown error';
+        throw new ParseError('PPTX', message, error);
+    }
+}
+async function extractAdvancedSlideContent(xmlContent, imageExtractor, extractedImages, slideNumber, layoutParser) {
+    try {
+        const result = await parseStringPromise(xmlContent);
+        const elements = [];
+        let imageCount = 0;
+        // Extract all shapes and their positions
+        function extractShapes(obj, parentPos = { x: 0, y: 0 }) {
+            if (typeof obj === 'object' && obj !== null) {
+                if (Array.isArray(obj)) {
+                    for (const item of obj) {
+                        extractShapes(item, parentPos);
+                    }
+                }
+                else {
+                    // Check for shape positioning
+                    let position = { ...parentPos };
+                    if (obj['a:off']?.[0]?.$) {
+                        position.x = parseInt(obj['a:off'][0].$.x) || 0;
+                        position.y = parseInt(obj['a:off'][0].$.y) || 0;
+                    }
+                    // Check for text content in shapes
+                    if (obj['a:t']) {
+                        const textElement = {
+                            type: 'text',
+                            content: '',
+                            position: position
+                        };
+                        if (Array.isArray(obj['a:t'])) {
+                            for (const textItem of obj['a:t']) {
+                                let text = '';
+                                if (typeof textItem === 'string') {
+                                    text = textItem;
+                                }
+                                else if (textItem && typeof textItem === 'object' && '_' in textItem) {
+                                    text = textItem._;
+                                }
+                                if (text && text.trim()) {
+                                    textElement.content += text.trim() + ' ';
+                                }
+                            }
+                        }
+                        if (textElement.content.trim()) {
+                            elements.push(textElement);
+                        }
+                    }
+                    // Check for tables
+                    if (obj['a:tbl']) {
+                        const tableElement = {
+                            type: 'table',
+                            content: obj['a:tbl'],
+                            position: position
+                        };
+                        elements.push(tableElement);
+                    }
+                    // Check for images
+                    if (obj['a:blip'] || obj['p:pic'] || obj['a:pic']) {
+                        const slideImages = extractedImages.filter(img => img.originalPath.includes(`slide${slideNumber}`) ||
+                            img.originalPath.includes('media/'));
+                        if (slideImages.length > imageCount) {
+                            const img = slideImages[imageCount];
+                            if (img?.savedPath) {
+                                const imageElement = {
+                                    type: 'image',
+                                    content: img.savedPath,
+                                    position: position
+                                };
+                                elements.push(imageElement);
+                                imageCount++;
+                            }
+                        }
+                    }
+                    // Recursively process nested objects
+                    for (const key in obj) {
+                        if (key !== 'a:t') {
+                            extractShapes(obj[key], position);
+                        }
+                    }
+                }
+            }
+        }
+        extractShapes(result);
+        // Sort elements by position (top to bottom, left to right)
+        const sortedElements = layoutParser.calculateRelativePosition(elements);
+        let markdown = '';
+        let currentRow = null;
+        const rowThreshold = 50; // EMUs tolerance for same row
+        for (const element of sortedElements) {
+            const elementY = element.position?.y || 0;
+            // Check if this element is in the same row as the previous
+            if (currentRow && Math.abs(elementY - currentRow.y) < rowThreshold) {
+                // Same row - add as column
+                currentRow.elements.push(element);
+            }
+            else {
+                // New row
+                if (currentRow && currentRow.elements.length > 0) {
+                    // Process previous row
+                    markdown += processSlideRow(currentRow, layoutParser, imageExtractor);
+                }
+                currentRow = {
+                    y: elementY,
+                    elements: [element]
+                };
+            }
+        }
+        // Process the last row
+        if (currentRow && currentRow.elements.length > 0) {
+            markdown += processSlideRow(currentRow, layoutParser, imageExtractor);
+        }
+        // If no organized content, fall back to simple extraction
+        if (!markdown.trim() && extractedImages.length > 0) {
+            const slideImages = extractedImages.filter(img => img.originalPath.includes(`slide${slideNumber}`) ||
+                (slideNumber === 1 && img.originalPath.includes('media/')));
+            for (const img of slideImages) {
+                if (img.savedPath) {
+                    markdown += imageExtractor.getImageMarkdown(`Slide ${slideNumber} Image`, img.savedPath) + '\n\n';
+                }
+            }
+        }
+        return markdown.trim();
+    }
+    catch (error) {
+        const message = error instanceof Error ? error.message : 'Unknown error';
+        throw new ParseError('PPTX', `Failed to extract advanced content from slide: ${message}`, error);
+    }
+}
+function processSlideRow(row, layoutParser, imageExtractor) {
+    if (row.elements.length === 1) {
+        // Single element in row
+        const element = row.elements[0];
+        return formatSlideElement(element, layoutParser, imageExtractor) + '\n\n';
+    }
+    else {
+        // Multiple elements - create columns
+        const columns = row.elements.map(element => ({
+            content: formatSlideElement(element, layoutParser, imageExtractor)
+        }));
+        return layoutParser.createColumns(columns) + '\n';
+    }
+}
+function formatSlideElement(element, layoutParser, imageExtractor) {
+    switch (element.type) {
+        case 'text': {
+            const content = element.content;
+            // Determine if it's a title based on position and length
+            if ((element.position?.y || 0) < 1000000 && content.length < 100) {
+                return `### ${content.trim()}`;
+            }
+            return content.trim();
+        }
+        case 'table':
+            // Parse PowerPoint table (simplified)
+            return parseSlideTable(element.content, layoutParser);
+        case 'image': {
+            const imagePath = element.content;
+            return imageExtractor.getImageMarkdown('Slide Image', imagePath);
+        }
+        default:
+            return typeof element.content === 'string' ? element.content : '';
+    }
+}
+function parseSlideTable(tableData, layoutParser) {
+    // Simplified table parsing for PowerPoint
+    const table = tableData;
+    if (!table?.[0]?.['a:tr']) {
+        return '';
+    }
+    const rows = table[0]['a:tr'];
+    const tableStruct = { rows: [] };
+    for (const row of rows) {
+        const cells = row['a:tc'] || [];
+        const rowData = { cells: [] };
+        for (const cell of cells) {
+            let cellText = '';
+            // Extract text from cell
+            if (cell['a:txBody']?.[0]?.['a:p']) {
+                const paragraphs = cell['a:txBody'][0]['a:p'];
+                for (const para of paragraphs) {
+                    if (para['a:r']?.[0]?.['a:t']?.[0]) {
+                        cellText += para['a:r'][0]['a:t'][0] + ' ';
+                    }
+                }
+            }
+            rowData.cells.push({
+                text: cellText.trim(),
+                bold: false,
+                italic: false,
+                alignment: 'left',
+                backgroundColor: undefined,
+                colSpan: 1,
+                rowSpan: 1,
+                merged: false
+            });
+        }
+        tableStruct.rows.push(rowData);
+    }
+    return layoutParser.parseAdvancedTable(tableStruct);
+}
+//# sourceMappingURL=pptx-parser.js.map