officeparser 6.1.1 → 7.0.1

package/dist/parsers/PowerPointParser.js

@@ -24,6 +24,8 @@
  */
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.parsePowerPoint = 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");
@@ -704,10 +706,10 @@ const parsePowerPoint = 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);
                     }
                 }
             }
@@ -729,7 +731,7 @@ const parsePowerPoint = async (buffer, config) => {
                 attachment.chartData = chartData;
             }
             catch (e) {
-                (0, errorUtils_js_1.logWarning)(`Failed to extract text from chart ${chart.path}:`, config, e);
+                (0, errorUtils_js_1.logWarning)(types_js_1.OfficeWarningType.CHART_DATA_EXTRACTION_FAILED, config, chart.path, e);
             }
         }
         // Loop through nodes to find images and charts and link their text and chartData
@@ -745,7 +747,7 @@ const parsePowerPoint = async (buffer, config) => {
                                 node.text = attachment.ocrText;
                         }
                         if (node.type === 'chart') {
-                            node.text = attachment.chartData?.rawTexts.join(config.newlineDelimiter || '\n');
+                            node.text = attachment.chartData?.rawTexts.join(config.newlineDelimiter);
                         }
                     }
                 }
@@ -764,24 +766,19 @@ const parsePowerPoint = async (buffer, config) => {
             return aIsNote - bIsNote;
         });
     }
-    return {
-        type: 'pptx',
-        metadata: metadata,
-        content: content,
-        attachments: attachments,
-        toText: () => content.map(c => {
-            // Recursive text extraction
-            const getText = (node) => {
-                let t = '';
-                if (node.children) {
-                    t += node.children.map(getText).filter(t => t != '').join(!node.children[0]?.children ? '' : config.newlineDelimiter ?? '\n');
-                }
-                else
-                    t += node.text || '';
-                return t;
-            };
-            return getText(c);
-        }).filter(t => t != '').join(config.newlineDelimiter ?? '\n')
-    };
+    const toTextSync = () => content.map(c => {
+        // Recursive text extraction
+        const getText = (node) => {
+            let t = '';
+            if (node.children) {
+                t += node.children.map(getText).filter(t => t != '').join(!node.children[0]?.children ? '' : config.newlineDelimiter);
+            }
+            else
+                t += node.text || '';
+            return t;
+        };
+        return getText(c);
+    }).filter(t => t != '').join(config.newlineDelimiter);
+    return (0, astUtils_js_1.createAST)('pptx', metadata, content, attachments, config, toTextSync);
 };
 exports.parsePowerPoint = parsePowerPoint;

package/dist/parsers/RtfParser.d.ts

@@ -39,7 +39,7 @@
  * @see https://www.biblioscape.com/rtf15_spec.htm RTF 1.5 Specification
  * @see https://latex2rtf.sourceforge.net/RTF-Spec-1.2.pdf RTF 1.2 Specification
  */
-import { OfficeParserAST, OfficeParserConfig } from '../types.js';
+import { FullOfficeParserConfig, OfficeParserAST } from '../types.js';
 /**
  * Represents an RTF group (content enclosed in braces).
  * Groups create formatting scopes and can contain other groups, control words, or text.
@@ -184,4 +184,4 @@ export declare class SimpleRtfParser {
  * @param config The parser configuration.
  * @returns The parsed AST.
  */
-export declare const parseRtf: (buffer: Buffer, config: OfficeParserConfig) => Promise<OfficeParserAST>;
+export declare const parseRtf: (buffer: Buffer, config: FullOfficeParserConfig) => Promise<OfficeParserAST>;