officeparser 6.1.0 → 7.0.0

package/dist/parsers/OpenOfficeParser.js

@@ -23,6 +23,8 @@
  */
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.parseOpenOffice = void 0;
+const types_js_1 = require("../types.js");
+const astUtils_js_1 = require("../utils/astUtils.js");
 const chartUtils_js_1 = require("../utils/chartUtils.js");
 const errorUtils_js_1 = require("../utils/errorUtils.js");
 const imageUtils_js_1 = require("../utils/imageUtils.js");
@@ -63,6 +65,7 @@ const parseOpenOffice = async (buffer, config) => {
     }
     const mainContentFile = files.find(f => f.path === 'content.xml') || files.find(f => f.path.match(contentFileRegex));
     const stylesFile = files.find(f => f.path === 'styles.xml');
+    const stylesDom = stylesFile ? (0, xmlUtils_js_1.parseXmlString)(stylesFile.content.toString()) : undefined;
     const content = [];
     const notes = [];
     // Style Map: styleName -> TextFormatting
@@ -70,9 +73,13 @@ const parseOpenOffice = async (buffer, config) => {
     const styleMap = {};
     const paragraphStyleMap = {};
     const listCounters = {}; // Track item index per listId/level
+    let currentListId = null;
+    let lastListType = null;
+    let lastListStyle = null;
+    let listIdCounter = 0;
+    let lastWasList = false;
     // Helper to parse styles
-    const parseStyles = (xmlString) => {
-        const xml = (0, xmlUtils_js_1.parseXmlString)(xmlString);
+    const parseStyles = (xml) => {
         const styles = (0, xmlUtils_js_1.getElementsByTagName)(xml, "style:style");
         for (const style of styles) {
             const name = style.getAttribute("style:name");
@@ -151,8 +158,8 @@ const parseOpenOffice = async (buffer, config) => {
             }
         }
     };
-    if (stylesFile) {
-        parseStyles(stylesFile.content.toString());
+    if (stylesDom) {
+        parseStyles(stylesDom);
     }
     /**
      * Helper to parse a paragraph node (text:p or text:h) and extract its content.
@@ -172,9 +179,10 @@ const parseOpenOffice = async (buffer, config) => {
      */
     const parseInlineContent = (node, styleMap, config, notes, paragraphStyleMap, parentFormatting = {}, linkMetadata, sourceXml = '') => {
         const children = [];
+        const anchorIds = [];
         let fullText = '';
         if (!node.childNodes)
-            return { text: '', children: [] };
+            return { text: '', children: [], anchorIds: [] };
         for (let i = 0; i < node.childNodes.length; i++) {
             const child = node.childNodes[i];
             if (child.nodeType === 3) { // Text node
@@ -192,7 +200,12 @@ const parseOpenOffice = async (buffer, config) => {
             else if ((0, xmlUtils_js_1.isElement)(child)) {
                 const element = child;
                 const tagName = element.tagName;
-                if (tagName === 'text:s') {
+                if (tagName === 'text:bookmark' || tagName === 'text:bookmark-start') {
+                    const name = element.getAttribute('text:name');
+                    if (name)
+                        anchorIds.push(name);
+                }
+                else if (tagName === 'text:s') {
                     // Space
                     const count = parseInt(element.getAttribute('text:c') || '1');
                     const spaces = ' '.repeat(count);
@@ -221,7 +234,7 @@ const parseOpenOffice = async (buffer, config) => {
                         type: 'text',
                         text: '\n',
                         formatting: parentFormatting,
-                        metadata: linkMetadata ? { ...linkMetadata } : undefined
+                        metadata: { ...(linkMetadata || {}), isLineBreak: true }
                     });
                 }
                 else if (tagName === 'text:span') {
@@ -231,15 +244,34 @@ const parseOpenOffice = async (buffer, config) => {
                     const spanContent = parseInlineContent(element, styleMap, config, notes, paragraphStyleMap, formatting, linkMetadata, sourceXml);
                     fullText += spanContent.text;
                     children.push(...spanContent.children);
+                    anchorIds.push(...spanContent.anchorIds);
                 }
                 else if (tagName === 'text:a') {
                     // Hyperlink
-                    const href = element.getAttribute('xlink:href') || '';
-                    const linkType = href.startsWith('#') ? 'internal' : 'external';
-                    const newLinkMetadata = { link: href, linkType: linkType };
+                    let href = element.getAttribute('xlink:href') || '';
+                    const isInternal = href.startsWith('#');
+                    const linkType = isInternal ? 'internal' : 'external';
+                    if (isInternal) {
+                        // ODT internal links can be encoded and might have suffixes like |outline
+                        try {
+                            href = decodeURIComponent(href).split('|')[0];
+                        }
+                        catch (e) {
+                            href = href.split('|')[0];
+                        }
+                        // Normalize internal link: if it contains #, keep only from # onwards
+                        if (href.includes('#')) {
+                            href = '#' + href.split('#').pop();
+                        }
+                    }
+                    let newLinkMetadata;
+                    if (!isInternal || !config.ignoreInternalLinks) {
+                        newLinkMetadata = { link: href, linkType: linkType };
+                    }
                     const linkContent = parseInlineContent(element, styleMap, config, notes, paragraphStyleMap, parentFormatting, newLinkMetadata, sourceXml);
                     fullText += linkContent.text;
                     children.push(...linkContent.children);
+                    anchorIds.push(...linkContent.anchorIds);
                 }
                 else if (tagName === 'text:note' && !config.ignoreNotes) {
                     // Footnote or endnote
@@ -258,7 +290,10 @@ const parseOpenOffice = async (buffer, config) => {
                                 type: 'paragraph',
                                 text: npContent.text,
                                 children: npContent.children,
-                                metadata: npContent.alignment ? { alignment: npContent.alignment } : undefined
+                                metadata: {
+                                    ...(npContent.alignment ? { alignment: npContent.alignment } : {}),
+                                    ...(npContent.anchorIds?.length ? { anchorIds: npContent.anchorIds } : {})
+                                }
                             };
                             noteChildren.push(npNode);
                         }
@@ -318,7 +353,7 @@ const parseOpenOffice = async (buffer, config) => {
                 }
             }
         }
-        return { text: fullText, children };
+        return { text: fullText, children, anchorIds };
     };
     /**
      * Helper to parse a paragraph node (text:p or text:h) and extract its content.
@@ -389,7 +424,32 @@ const parseOpenOffice = async (buffer, config) => {
                 }
             }
         }
-        return { text: content.text, children: content.children, alignment, style: paraStyle || undefined };
+        return { text: content.text, children: content.children, alignment, style: paraStyle || undefined, anchorIds: content.anchorIds };
+    };
+    /**
+     * Splits paragraph content into multiple segments based on line breaks.
+     * Used to handle soft line breaks within list items.
+     *
+     * @param pContent - The content of a single paragraph
+     * @returns Array of content segments
+     */
+    const splitParagraphByBreaks = (pContent) => {
+        const segments = [];
+        let currentText = "";
+        let currentChildren = [];
+        for (const child of pContent.children) {
+            if (child.type === "text" && child.metadata?.isLineBreak) {
+                segments.push({ text: currentText, children: currentChildren });
+                currentText = "";
+                currentChildren = [];
+            }
+            else {
+                currentText += child.text || "";
+                currentChildren.push(child);
+            }
+        }
+        segments.push({ text: currentText, children: currentChildren });
+        return segments;
     };
     /**
      * Helper to parse a table node and extract its structure.
@@ -603,8 +663,9 @@ const parseOpenOffice = async (buffer, config) => {
                 (0, xmlUtils_js_1.getDirectChildren)(officeBody, "office:spreadsheet")[0];
             if (bodyContent) {
                 const bodyChildren = (0, xmlUtils_js_1.getDirectChildren)(bodyContent, "*");
+                const isSpreadsheet = bodyContent.tagName === "office:spreadsheet";
                 for (const child of bodyChildren) {
-                    traverse(child, content, false, xmlString);
+                    traverse(child, content, false, xmlString, isSpreadsheet);
                 }
             }
         }
@@ -616,19 +677,28 @@ const parseOpenOffice = async (buffer, config) => {
          * @param targetArray - The array to push extracted content nodes to
          * @param forceHeading - If true, treats all paragraphs as headings (used for slide titles)
          * @param sourceXml - The source XML string for raw content extraction
+         * @param asSheet - If true, treats tables as sheets (for ODS)
          */
-        function traverse(node, targetArray, forceHeading = false, sourceXml) {
+        function traverse(node, targetArray, forceHeading = false, sourceXml, asSheet = false) {
             if (node.tagName === "text:p") {
                 const pContent = parseParagraphContent(node, paragraphStyleMap, styleMap, config, sourceXml);
                 const type = (forceHeading || (node.getAttribute("text:style-name") || '').toLowerCase().includes('title')) ? 'heading' : 'paragraph';
+                const metadata = {
+                    ...(pContent.alignment ? { alignment: pContent.alignment } : {}),
+                    ...(pContent.style ? { style: pContent.style } : {}),
+                    ...(pContent.anchorIds?.length ? { anchorIds: pContent.anchorIds } : {})
+                };
+                const nodeId = node.getAttribute("xml:id") || node.getAttribute("text:id");
+                if (nodeId) {
+                    if (!metadata.anchorIds)
+                        metadata.anchorIds = [];
+                    metadata.anchorIds.push(nodeId);
+                }
                 const pNode = {
                     type,
                     text: pContent.text,
                     children: pContent.children,
-                    metadata: {
-                        ...(pContent.alignment ? { alignment: pContent.alignment } : {}),
-                        ...(pContent.style ? { style: pContent.style } : {})
-                    }
+                    metadata
                 };
                 if (type === 'heading' && pNode.metadata) {
                     pNode.metadata.level = pNode.metadata.level || 1;
@@ -640,42 +710,65 @@ const parseOpenOffice = async (buffer, config) => {
                     pNode.rawContent = (0, xmlUtils_js_1.getRawContent)(node, sourceXml, config);
                 }
                 targetArray.push(pNode);
+                lastWasList = false;
             }
             else if (node.tagName === "text:h") {
                 const level = parseInt(node.getAttribute("text:outline-level") || "1");
                 const hContent = parseParagraphContent(node, paragraphStyleMap, styleMap, config, sourceXml);
+                const metadata = {
+                    level,
+                    ...(hContent.alignment ? { alignment: hContent.alignment } : {}),
+                    ...(hContent.style ? { style: hContent.style } : {}),
+                    ...(hContent.anchorIds?.length ? { anchorIds: hContent.anchorIds } : {})
+                };
+                const nodeId = node.getAttribute("xml:id") || node.getAttribute("text:id");
+                if (nodeId) {
+                    if (!metadata.anchorIds)
+                        metadata.anchorIds = [];
+                    metadata.anchorIds.push(nodeId);
+                }
                 const hNode = {
                     type: 'heading',
                     text: hContent.text,
                     children: hContent.children,
-                    metadata: {
-                        level,
-                        ...(hContent.alignment ? { alignment: hContent.alignment } : {}),
-                        ...(hContent.style ? { style: hContent.style } : {})
-                    }
+                    metadata
                 };
                 if (config.includeRawContent) {
                     hNode.rawContent = (0, xmlUtils_js_1.getRawContent)(node, sourceXml, config);
                 }
                 targetArray.push(hNode);
+                lastWasList = false;
             }
             else if (node.tagName === "table:table") {
                 // Parse table with proper structure
                 const tableNode = parseTable(node, paragraphStyleMap, styleMap, config, sourceXml);
+                if (asSheet) {
+                    tableNode.type = 'sheet';
+                    const sheetName = node.getAttribute("table:name");
+                    if (sheetName) {
+                        tableNode.metadata = { ...tableNode.metadata, sheetName };
+                    }
+                }
+                const tableId = node.getAttribute("xml:id") || node.getAttribute("table:name");
+                if (tableId) {
+                    if (!tableNode.metadata)
+                        tableNode.metadata = {};
+                    tableNode.metadata.anchorIds = tableNode.metadata.anchorIds || [];
+                    tableNode.metadata.anchorIds.push(tableId);
+                }
                 if (config.includeRawContent) {
                     tableNode.rawContent = (0, xmlUtils_js_1.getRawContent)(node, sourceXml, config);
                 }
                 targetArray.push(tableNode);
+                lastWasList = false;
             }
             else if (node.tagName === "text:list") {
                 // Parse list structure with proper listId tracking
                 const listItems = (0, xmlUtils_js_1.getDirectChildren)(node, "text:list-item");
-                // Get list style name to use as listId (or generate one)
-                const listStyleName = node.getAttribute("text:style-name") || node.getAttribute("xml:id");
-                const listId = listStyleName || `list-${targetArray.length}`;
                 // Determine list type by checking the list style definition
                 let listType = 'unordered';
                 let isVisible = false;
+                const listStyleName = node.getAttribute("text:style-name") || node.getAttribute("xml:id");
                 let styleNameToCheck = listStyleName;
                 // If no style name, check parent list for inherited style
                 if (!styleNameToCheck) {
@@ -689,9 +782,8 @@ const parseOpenOffice = async (buffer, config) => {
                         parentNode = parentNode.parentNode;
                     }
                 }
-                // Try to find list style in automatic styles to determine type and visibility
+                // Try to find list style in automatic styles or styles.xml to determine type and visibility
                 if (styleNameToCheck) {
-                    const automaticStyles = (0, xmlUtils_js_1.getFirstElementByTagName)((0, xmlUtils_js_1.parseXmlString)(mainContentFile?.content.toString() || ''), "office:automatic-styles");
                     if (automaticStyles) {
                         const listStyles = (0, xmlUtils_js_1.getElementsByTagName)(automaticStyles, "text:list-style");
                         for (const listStyle of listStyles) {
@@ -708,32 +800,32 @@ const parseOpenOffice = async (buffer, config) => {
                                     listType = 'unordered';
                                     isVisible = bulletLevels.some(l => !!l.getAttribute("text:bullet-char"));
                                 }
-                                if (imageLevels.length > 0)
+                                else if (imageLevels.length > 0) {
+                                    listType = 'unordered';
                                     isVisible = true;
+                                }
                                 break;
                             }
                         }
                     }
-                    // Also check in styles.xml if still unordered and hidden
-                    if (stylesFile && !isVisible) {
-                        const stylesXml = (0, xmlUtils_js_1.parseXmlString)(stylesFile.content.toString());
-                        const listStyles = (0, xmlUtils_js_1.getElementsByTagName)(stylesXml, "text:list-style");
-                        for (const listStyle of listStyles) {
-                            if (listStyle.getAttribute("style:name") === styleNameToCheck) {
-                                const bulletLevels = (0, xmlUtils_js_1.getElementsByTagName)(listStyle, "text:list-level-style-bullet");
-                                const numberLevels = (0, xmlUtils_js_1.getElementsByTagName)(listStyle, "text:list-level-style-number");
-                                const imageLevels = (0, xmlUtils_js_1.getElementsByTagName)(listStyle, "text:list-level-style-image");
-                                if (numberLevels.length > 0) {
-                                    listType = 'ordered';
-                                    isVisible = numberLevels.some(l => !!l.getAttribute("style:num-format"));
-                                }
-                                else if (bulletLevels.length > 0) {
-                                    listType = 'unordered';
-                                    isVisible = bulletLevels.some(l => !!l.getAttribute("text:bullet-char"));
+                    if (!isVisible && stylesDom) {
+                        const officeStyles = (0, xmlUtils_js_1.getFirstElementByTagName)(stylesDom, "office:styles");
+                        if (officeStyles) {
+                            const listStyles = (0, xmlUtils_js_1.getElementsByTagName)(officeStyles, "text:list-style");
+                            for (const listStyle of listStyles) {
+                                if (listStyle.getAttribute("style:name") === styleNameToCheck) {
+                                    const bulletLevels = (0, xmlUtils_js_1.getElementsByTagName)(listStyle, "text:list-level-style-bullet");
+                                    const numberLevels = (0, xmlUtils_js_1.getElementsByTagName)(listStyle, "text:list-level-style-number");
+                                    if (numberLevels.length > 0) {
+                                        listType = 'ordered';
+                                        isVisible = numberLevels.some(l => !!l.getAttribute("style:num-format"));
+                                    }
+                                    else if (bulletLevels.length > 0) {
+                                        listType = 'unordered';
+                                        isVisible = bulletLevels.some(l => !!l.getAttribute("text:bullet-char"));
+                                    }
+                                    break;
                                 }
-                                if (imageLevels.length > 0)
-                                    isVisible = true;
-                                break;
                             }
                         }
                     }
@@ -741,6 +833,7 @@ const parseOpenOffice = async (buffer, config) => {
                 // If the list is not visible, it's likely a layout list used by Impress.
                 // We should traverse its items and treat their content as regular nodes.
                 if (!isVisible) {
+                    lastWasList = false;
                     for (let i = 0; i < listItems.length; i++) {
                         const item = listItems[i];
                         if (item.childNodes) {
@@ -754,6 +847,24 @@ const parseOpenOffice = async (buffer, config) => {
                     }
                     return;
                 }
+                // List Continuity Logic:
+                // If this list follows another list of the same type and style, or we are in ODP and it's sequential,
+                // we should reuse the previous listId to maintain numbering.
+                const isODP = fileType === 'odp';
+                const sameStyle = styleNameToCheck && styleNameToCheck === lastListStyle;
+                const sameType = listType === lastListType;
+                let listId;
+                if (lastWasList && (sameStyle || (isODP && sameType))) {
+                    listId = currentListId;
+                }
+                else {
+                    // New list
+                    listId = styleNameToCheck || `list-${++listIdCounter}`;
+                    currentListId = listId;
+                    lastListType = listType;
+                    lastListStyle = styleNameToCheck;
+                }
+                lastWasList = true;
                 // Calculate indentation level by counting parent text:list elements
                 let indentation = 0;
                 let parent = node.parentNode;
@@ -774,59 +885,57 @@ const parseOpenOffice = async (buffer, config) => {
                 // Process each list item
                 for (let i = 0; i < listItems.length; i++) {
                     const item = listItems[i];
-                    // Increment item index for this list/level
-                    listCounters[listId][indentKey]++;
-                    const itemIndex = listCounters[listId][indentKey];
-                    // Reset deeper levels when we encounter an item at this level
-                    for (let k = indentation + 1; k < 10; k++) {
-                        if (listCounters[listId][k.toString()] !== undefined) {
-                            listCounters[listId][k.toString()] = -1;
-                        }
-                    }
+                    let hasIndexedThisItem = false;
                     // Iterate over direct children of list item (paragraphs, headings, nested lists)
                     if (item.childNodes) {
                         for (let j = 0; j < item.childNodes.length; j++) {
                             const child = item.childNodes[j];
                             if ((0, xmlUtils_js_1.isElement)(child)) { // Element
                                 const element = child;
-                                if (element.tagName === "text:p") {
-                                    const pContent = parseParagraphContent(element, paragraphStyleMap, styleMap, config, sourceXml);
-                                    const listNode = {
-                                        type: 'list',
-                                        text: pContent.text,
-                                        children: pContent.children,
-                                        metadata: {
-                                            listType,
-                                            indentation,
-                                            itemIndex,
-                                            listId,
-                                            alignment: pContent.alignment || 'left',
-                                            style: pContent.style
+                                if (element.tagName === "text:p" || element.tagName === "text:h") {
+                                    if (!hasIndexedThisItem) {
+                                        listCounters[listId][indentKey]++;
+                                        hasIndexedThisItem = true;
+                                        for (let k = indentation + 1; k < 10; k++) {
+                                            if (listCounters[listId][k.toString()] !== undefined) {
+                                                listCounters[listId][k.toString()] = -1;
+                                            }
                                         }
-                                    };
-                                    if (config.includeRawContent)
-                                        listNode.rawContent = (0, xmlUtils_js_1.getRawContent)(element, sourceXml, config);
-                                    targetArray.push(listNode);
-                                }
-                                else if (element.tagName === "text:h") {
-                                    const level = parseInt(element.getAttribute("text:outline-level") || "1");
-                                    const hContent = parseParagraphContent(element, paragraphStyleMap, styleMap, config, sourceXml);
-                                    const listNode = {
-                                        type: 'list',
-                                        text: hContent.text,
-                                        children: hContent.children,
-                                        metadata: {
-                                            listType,
-                                            indentation,
-                                            itemIndex,
-                                            listId,
-                                            ...(hContent.alignment ? { alignment: hContent.alignment } : {}),
-                                            style: hContent.style
+                                    }
+                                    const itemIndex = listCounters[listId][indentKey];
+                                    const pContent = parseParagraphContent(element, paragraphStyleMap, styleMap, config, sourceXml);
+                                    const segments = splitParagraphByBreaks(pContent);
+                                    for (let k = 0; k < segments.length; k++) {
+                                        const segment = segments[k];
+                                        if (!segment.text.trim() && segment.children.length === 0)
+                                            continue;
+                                        const isFirst = k === 0;
+                                        const nodeType = isFirst ? 'list' : 'paragraph';
+                                        const node = {
+                                            type: nodeType,
+                                            text: segment.text,
+                                            children: segment.children,
+                                            metadata: isFirst ? {
+                                                listType,
+                                                indentation,
+                                                itemIndex,
+                                                listId,
+                                                alignment: pContent.alignment || 'left',
+                                                style: pContent.style
+                                            } : {
+                                                alignment: pContent.alignment || 'left',
+                                                style: pContent.style
+                                            }
+                                        };
+                                        // Special case for headings in lists
+                                        if (isFirst && element.tagName === "text:h") {
+                                            const level = parseInt(element.getAttribute("text:outline-level") || "1");
+                                            node.metadata.level = level;
                                         }
-                                    };
-                                    if (config.includeRawContent)
-                                        listNode.rawContent = (0, xmlUtils_js_1.getRawContent)(element, sourceXml, config);
-                                    targetArray.push(listNode);
+                                        if (config.includeRawContent)
+                                            node.rawContent = (0, xmlUtils_js_1.getRawContent)(element, sourceXml, config);
+                                        targetArray.push(node);
+                                    }
                                 }
                                 else if (element.tagName === "text:list") {
                                     // Recursive call for nested list
@@ -871,14 +980,19 @@ const parseOpenOffice = async (buffer, config) => {
                         const parts = imageHref.split('/');
                         imageHref = parts[parts.length - 1];
                     }
+                    const metadata = {
+                        attachmentName: imageHref,
+                        ...(altText ? { altText } : {})
+                    };
+                    const frameId = node.getAttribute("xml:id") || node.getAttribute("draw:name");
+                    if (frameId) {
+                        metadata.anchorIds = [frameId];
+                    }
                     const imageNode = {
                         type: 'image',
                         text: '',
                         children: [],
-                        metadata: {
-                            attachmentName: imageHref,
-                            ...(altText ? { altText } : {})
-                        }
+                        metadata
                     };
                     if (config.includeRawContent) {
                         imageNode.rawContent = (0, xmlUtils_js_1.getRawContent)(node, sourceXml, config);
@@ -1211,10 +1325,10 @@ const parseOpenOffice = async (buffer, config) => {
             if (config.ocr) {
                 if (attachment.mimeType.startsWith('image/')) {
                     try {
-                        attachment.ocrText = (await (0, ocrUtils_js_1.performOcr)(media.content, { language: config.ocrLanguage, ...config.ocrConfig })).trim();
+                        attachment.ocrText = (await (0, ocrUtils_js_1.performOcr)(media.content, { ...config.ocrConfig })).trim();
                     }
                     catch (e) {
-                        (0, errorUtils_js_1.logWarning)(`OCR failed for ${attachment.name}:`, config, e);
+                        (0, errorUtils_js_1.logWarning)(types_js_1.OfficeWarningType.OCR_FAILED, config, attachment.name, e);
                     }
                 }
             }
@@ -1351,7 +1465,7 @@ const parseOpenOffice = async (buffer, config) => {
                             node.text = attachment.ocrText;
                         }
                         if (attachment.chartData && node.type === 'chart') {
-                            node.text = attachment.chartData.rawTexts.join(config.newlineDelimiter ?? '\n');
+                            node.text = attachment.chartData.rawTexts.join(config.newlineDelimiter);
                         }
                     }
                 }
@@ -1385,32 +1499,27 @@ const parseOpenOffice = async (buffer, config) => {
     if (config.putNotesAtLast && notes.length > 0) {
         content.push(...notes);
     }
-    return {
-        type: fileType,
-        metadata: {
-            ...metadata,
-            styleMap: combinedStyleMap
-        },
-        content: content,
-        attachments: attachments,
-        toText: () => content.map(c => {
-            const getText = (node) => {
-                let t = '';
-                if (node.children && node.children.length > 0) {
-                    // Check if children have their own children (container vs leaf)
-                    // If children are leaf nodes (text/image), join with empty string
-                    // If children are container nodes (paragraphs/rows), join with newline
-                    const hasGrandChildren = node.children.some(child => child.children && child.children.length > 0);
-                    const separator = hasGrandChildren ? (config.newlineDelimiter ?? '\n') : '';
-                    t += node.children.map(getText).filter(t => t != '').join(separator);
-                }
-                else {
-                    t += node.text || '';
-                }
-                return t;
-            };
-            return getText(c);
-        }).filter(t => t != '').join(config.newlineDelimiter ?? '\n')
-    };
+    const toTextSync = () => content.map(c => {
+        const getText = (node) => {
+            let t = '';
+            if (node.children && node.children.length > 0) {
+                // Check if children have their own children (container vs leaf)
+                // If children are leaf nodes (text/image), join with empty string
+                // If children are container nodes (paragraphs/rows), join with newline
+                const hasGrandChildren = node.children.some(child => child.children && child.children.length > 0);
+                const separator = hasGrandChildren ? config.newlineDelimiter : '';
+                t += node.children.map(getText).filter(t => t != '').join(separator);
+            }
+            else {
+                t += node.text || '';
+            }
+            return t;
+        };
+        return getText(c);
+    }).filter(t => t != '').join(config.newlineDelimiter);
+    return (0, astUtils_js_1.createAST)(fileType, {
+        ...metadata,
+        styleMap: combinedStyleMap
+    }, content, attachments, config, toTextSync);
 };
 exports.parseOpenOffice = parseOpenOffice;

package/dist/parsers/PdfParser.d.ts

@@ -56,7 +56,7 @@
  * @see https://mozilla.github.io/pdf.js/ PDF.js documentation
  * @see https://www.adobe.com/devnet/pdf/pdf_reference.html PDF Reference
  */
-import { OfficeParserAST, OfficeParserConfig } from '../types.js';
+import { FullOfficeParserConfig, OfficeParserAST } from '../types.js';
 /**
  * Parses a PDF file and extracts content.
  *
@@ -64,4 +64,4 @@ import { OfficeParserAST, OfficeParserConfig } from '../types.js';
  * @param config - Parser configuration
  * @returns Promise resolving to the parsed AST
  */
-export declare const parsePdf: (buffer: Buffer, config: OfficeParserConfig) => Promise<OfficeParserAST>;
+export declare const parsePdf: (buffer: Buffer, config: FullOfficeParserConfig) => Promise<OfficeParserAST>;