officeparser 7.0.3 → 7.2.0

package/dist/generators/MarkdownGenerator.d.ts

@@ -37,7 +37,7 @@ export declare class MarkdownGenerator extends BaseGenerator<'md'> {
      *
      * @returns A Markdown string
      */
-    generate(): Promise<ConversionResult>;
+    generate(): Promise<ConversionResult<'md'>>;
     /**
      * Recursively processes nodes and builds output.
      * Overridden to provide AST optimization (merging adjacent text nodes).

package/dist/generators/MarkdownGenerator.js

@@ -136,7 +136,9 @@ class MarkdownGenerator extends BaseGenerator_js_1.BaseGenerator {
                     else if (this.config.generateIds) {
                         id = ` {#${this.slugify(this.getNodeText(node))}}`;
                     }
-                    const anchors = remainingAnchors.map(aid => `<a name="${aid}"></a>`).join('');
+                    const anchors = this.config.mdConfig.fallbackToHtml
+                        ? remainingAnchors.map(aid => `<a name="${aid}"></a>`).join('')
+                        : '';
                     let content = `${prefix}${childrenOutput}${id}`;
                     // Alignment fallback via HTML div/p
                     if (this.config.mdConfig.fallbackToHtml && meta?.alignment && meta.alignment !== 'left') {

package/dist/generators/PdfGenerator.d.ts

@@ -9,7 +9,7 @@ import { BaseGenerator } from './BaseGenerator.js';
  */
 export declare class PdfGenerator extends BaseGenerator<'pdf'> {
     constructor(ast: OfficeParserAST, config?: GeneratorConfig<'pdf'>);
-    generate(): Promise<ConversionResult>;
+    generate(): Promise<ConversionResult<'pdf'>>;
     /**
      * Node.js implementation using Puppeteer.
      * Uses dynamic import to avoid bundling puppeteer into the library core.

package/dist/generators/PdfGenerator.js

@@ -3,6 +3,7 @@ Object.defineProperty(exports, "__esModule", { value: true });
 exports.PdfGenerator = void 0;
 const types_js_1 = require("../types.js");
 const envUtils_js_1 = require("../utils/envUtils.js");
+const errorUtils_js_1 = require("../utils/errorUtils.js");
 const BaseGenerator_js_1 = require("./BaseGenerator.js");
 const HtmlGenerator_js_1 = require("./HtmlGenerator.js");
 /**
@@ -22,12 +23,6 @@ class PdfGenerator extends BaseGenerator_js_1.BaseGenerator {
         const htmlGenerator = new HtmlGenerator_js_1.HtmlGenerator(this.ast, {
             ...this.config,
             htmlConfig: { ...this.config.htmlConfig, standalone: true },
-            mdConfig: undefined,
-            chunksConfig: undefined,
-            csvConfig: undefined,
-            textConfig: undefined,
-            pdfConfig: undefined,
-            rtfConfig: undefined,
         });
         const htmlResult = await htmlGenerator.generate();
         const html = typeof htmlResult.value === 'string' ? htmlResult.value : '';
@@ -44,6 +39,24 @@ class PdfGenerator extends BaseGenerator_js_1.BaseGenerator {
      * Uses dynamic import to avoid bundling puppeteer into the library core.
      */
     async generateInNode(html) {
+        const signal = this.config.abortSignal;
+        if (signal?.aborted) {
+            throw (0, errorUtils_js_1.getAbortError)();
+        }
+        let browser;
+        const onAbort = async () => {
+            if (browser) {
+                try {
+                    await browser.close();
+                }
+                catch (e) {
+                    // ignore
+                }
+            }
+        };
+        if (signal) {
+            signal.addEventListener('abort', onAbort);
+        }
         try {
             // Dynamic import for peer dependency
             // @ts-ignore
@@ -70,11 +83,16 @@ class PdfGenerator extends BaseGenerator_js_1.BaseGenerator {
             // to ensure transparency about the environment state. Programmatically fixing this
             // would require force-downloading a ~300MB arm64 browser binary or switching to 
             // a system-installed Chrome, both of which are too intrusive for a library.
-            const browser = await puppeteer.launch(launchOptions);
+            browser = await puppeteer.launch(launchOptions);
             const page = await browser.newPage();
+            const pdfConfig = this.config.pdfConfig;
+            const timeout = pdfConfig.timeout;
+            if (timeout !== undefined && timeout > 0) {
+                page.setDefaultTimeout(timeout);
+                page.setDefaultNavigationTimeout(timeout);
+            }
             // Set content and wait for network/assets to load
             await page.setContent(html, { waitUntil: 'networkidle0' });
-            const pdfConfig = this.config.pdfConfig;
             const pdfBuffer = await page.pdf({
                 format: pdfConfig.format,
                 width: pdfConfig.width,
@@ -88,18 +106,41 @@ class PdfGenerator extends BaseGenerator_js_1.BaseGenerator {
                 footerTemplate: pdfConfig.footerTemplate,
             });
             await browser.close();
+            browser = null;
             return {
                 value: new Uint8Array(pdfBuffer),
                 messages: this.messages
             };
         }
         catch (err) {
-            this.warn(types_js_1.OfficeWarningType.DEPENDENCY_LOAD_FAILED, `puppeteer. Please install it with 'npm install puppeteer'. Error: ${err.message}`);
+            if (browser) {
+                try {
+                    await browser.close();
+                }
+                catch (e) {
+                    // ignore
+                }
+                browser = null;
+            }
+            if (signal?.aborted) {
+                throw (0, errorUtils_js_1.getAbortError)();
+            }
+            if (err.message && (err.message.includes('timeout') || err.message.includes('Timeout'))) {
+                this.warn(types_js_1.OfficeWarningType.PAGE_LOAD_FAILED, `PDF generation timed out: ${err.message}`);
+            }
+            else {
+                this.warn(types_js_1.OfficeWarningType.DEPENDENCY_LOAD_FAILED, `puppeteer. Please install it with 'npm install puppeteer'. Error: ${err.message}`);
+            }
             return {
-                value: '',
+                value: new Uint8Array(),
                 messages: this.messages
             };
         }
+        finally {
+            if (signal) {
+                signal.removeEventListener('abort', onAbort);
+            }
+        }
     }
     /**
      * Browser implementation using hidden iframe and native print.

package/dist/generators/RtfGenerator.d.ts

@@ -7,9 +7,10 @@ export declare class RtfGenerator extends BaseGenerator<'rtf'> {
     private colorTable;
     private inTable;
     constructor(ast: OfficeParserAST, config?: GeneratorConfig<'rtf'>);
-    generate(): Promise<ConversionResult>;
+    generate(): Promise<ConversionResult<'rtf'>>;
     protected processNodeRecursive(node: OfficeContentNode, processor: (node: OfficeContentNode, childrenOutput: string) => Promise<string>): Promise<string>;
     private renderBody;
     private getColorIndex;
+    private isLightColor;
     private escapeRtf;
 }

package/dist/generators/RtfGenerator.js

@@ -15,7 +15,7 @@ class RtfGenerator extends BaseGenerator_js_1.BaseGenerator {
         this.colorTable = [];
         // We first process all nodes to collect colors and analyze structure
         const bodyContent = await this.renderBody(this.ast);
-        let output = '{\\rtf1\\ansi\\deff0\n';
+        let output = '{\\rtf1\\ansi\\uc1\\deff0\n';
         // 1. Info Group (Metadata)
         if (this.config.renderMetadata && this.ast.metadata) {
             output += '{\\info';
@@ -100,16 +100,34 @@ class RtfGenerator extends BaseGenerator_js_1.BaseGenerator {
                             suffix = '\\strike0 ' + suffix;
                         }
                         if (f.color) {
-                            const idx = this.getColorIndex(f.color);
-                            prefix += `\\cf${idx + 1} `;
+                            // RTF default background is white. Ensure light text without a dark background remains readable.
+                            const isTextLight = this.isLightColor(f.color);
+                            const isBgLight = !f.backgroundColor || this.isLightColor(f.backgroundColor);
+                            if (!(isTextLight && isBgLight)) {
+                                const idx = this.getColorIndex(f.color);
+                                prefix += `\\cf${idx + 1} `;
+                            }
                         }
                         if (f.backgroundColor) {
                             const idx = this.getColorIndex(f.backgroundColor);
                             prefix += `\\highlight${idx + 1} `;
                         }
                         if (f.size) {
-                            const pt = parseInt(f.size);
-                            prefix += `\\fs${pt * 2} `;
+                            let pt = 12; // default
+                            const val = parseFloat(f.size);
+                            if (!isNaN(val)) {
+                                if (f.size.includes('in'))
+                                    pt = val * 72;
+                                else if (f.size.includes('cm'))
+                                    pt = val * 28.3465;
+                                else if (f.size.includes('mm'))
+                                    pt = val * 2.83465;
+                                else if (f.size.includes('px'))
+                                    pt = val * 0.75;
+                                else
+                                    pt = val;
+                            }
+                            prefix += `\\fs${Math.round(pt * 2)} `;
                         }
                         text = `{\\f0 ${prefix}${text}${suffix}}`;
                     }
@@ -214,13 +232,32 @@ class RtfGenerator extends BaseGenerator_js_1.BaseGenerator {
         }
         return idx;
     }
+    isLightColor(hex) {
+        if (!hex || hex.length !== 7 || !hex.startsWith('#'))
+            return false;
+        const r = parseInt(hex.substring(1, 3), 16);
+        const g = parseInt(hex.substring(3, 5), 16);
+        const b = parseInt(hex.substring(5, 7), 16);
+        if (isNaN(r) || isNaN(g) || isNaN(b))
+            return false;
+        // Simple luminance calculation
+        const luminance = (0.299 * r + 0.587 * g + 0.114 * b) / 255;
+        return luminance > 0.8;
+    }
     escapeRtf(text) {
         return text
             .replace(/\\/g, '\\\\')
             .replace(/{/g, '\\{')
             .replace(/}/g, '\\}')
             .replace(/[^\x00-\x7F]/g, (match) => {
-            return `\\u${match.charCodeAt(0)}?`;
+            let code = match.charCodeAt(0);
+            if (code < 256) {
+                return `\\'${code.toString(16).padStart(2, '0')}`;
+            }
+            if (code > 32767) {
+                code -= 65536;
+            }
+            return `{\\uc0\\u${code}}`;
         });
     }
 }

package/dist/generators/TextGenerator.d.ts

@@ -8,6 +8,6 @@ export declare class TextGenerator extends BaseGenerator<'text'> {
     /**
      * Generates plain text by concatenating text content from nodes.
      */
-    generate(): Promise<ConversionResult>;
+    generate(): Promise<ConversionResult<'text'>>;
     private renderTable;
 }