formatferry 1.0.1 → 1.0.2

package/dist/src/shared/adapters/pdf-adapter.d.ts

@@ -0,0 +1,94 @@
+/**
+ * PDF Adapter Module
+ *
+ * Converts PDF files to text/HTML using PDF.js for processing through
+ * the existing ConversionEngine pipeline.
+ *
+ * Requirements: 3.1, 3.2, 3.3, 3.4, 3.5, 3.6, 3.7, 3.8
+ */
+/** Page information extracted from PDF */
+export interface PdfPageInfo {
+    /** Page number (1-indexed) */
+    pageNumber: number;
+    /** Text content extracted from the page */
+    textContent: string;
+    /** Whether the page contains images */
+    hasImages: boolean;
+    /** Viewport width for multi-column detection */
+    viewportWidth?: number;
+}
+/** PDF metadata extracted during parsing */
+export interface PdfMetadata {
+    /** Document title (if available) */
+    title?: string;
+    /** Document author (if available) */
+    author?: string;
+    /** Total number of pages */
+    pageCount: number;
+    /** Whether the PDF is text-based (has text layer) */
+    isTextBased: boolean;
+}
+/** Result of parsing a PDF file */
+export interface PdfParseResult {
+    /** HTML content with page markers */
+    html: string;
+    /** Information about each page */
+    pages: PdfPageInfo[];
+    /** Document metadata */
+    metadata: PdfMetadata;
+    /** Whether the PDF is text-based */
+    isTextBased: boolean;
+    /** Any warnings generated during parsing */
+    warnings: string[];
+}
+/** Error thrown when PDF parsing fails */
+export declare class PdfParseError extends Error {
+    readonly code: 'CORRUPTED' | 'PASSWORD_PROTECTED' | 'IMAGE_ONLY' | 'PARSE_ERROR';
+    readonly originalError?: Error | undefined;
+    constructor(message: string, code: 'CORRUPTED' | 'PASSWORD_PROTECTED' | 'IMAGE_ONLY' | 'PARSE_ERROR', originalError?: Error | undefined);
+}
+/**
+ * PDF Adapter class
+ *
+ * Handles conversion of PDF files to HTML using PDF.js.
+ * Extracts text content and adds page markers between pages.
+ */
+export declare class PdfAdapter {
+    private pdfjs;
+    /**
+     * Load the PDF.js library if not already loaded
+     */
+    private ensurePdfjsLoaded;
+    /**
+     * Parse a PDF file and convert it to HTML with page markers
+     *
+     * @param arrayBuffer - The PDF file content as an ArrayBuffer
+     * @param onProgress - Optional progress callback (0-1)
+     * @returns Promise resolving to the parse result with HTML and metadata
+     * @throws PdfParseError if the file cannot be parsed
+     */
+    parseToText(arrayBuffer: ArrayBuffer, onProgress?: (progress: number) => void): Promise<PdfParseResult>;
+    /**
+     * Convert parsed pages to HTML with page markers
+     *
+     * Requirement 3.5: Insert page markers between pages
+     *
+     * Note: We use a <div> element with data-page-marker attribute instead of
+     * HTML comments because DOMPurify strips comments by default. The marker
+     * element survives sanitization and is converted to the comment format
+     * after sanitization in the file processor pipeline.
+     */
+    private pagesToHtml;
+    /**
+     * Get the number of pages in a PDF without full parsing
+     *
+     * @param arrayBuffer - The PDF file content as an ArrayBuffer
+     * @returns Promise resolving to the page count
+     */
+    getPageCount(arrayBuffer: ArrayBuffer): Promise<number>;
+}
+/**
+ * Singleton instance for convenience
+ */
+export declare const pdfAdapter: PdfAdapter;
+//# sourceMappingURL=pdf-adapter.d.ts.map

package/dist/src/shared/adapters/pdf-adapter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdf-adapter.d.ts","sourceRoot":"","sources":["../../../../../../src/shared/adapters/pdf-adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAIH,0CAA0C;AAC1C,MAAM,WAAW,WAAW;IAC1B,8BAA8B;IAC9B,UAAU,EAAE,MAAM,CAAC;IACnB,2CAA2C;IAC3C,WAAW,EAAE,MAAM,CAAC;IACpB,uCAAuC;IACvC,SAAS,EAAE,OAAO,CAAC;IACnB,gDAAgD;IAChD,aAAa,CAAC,EAAE,MAAM,CAAC;CACxB;AAED,4CAA4C;AAC5C,MAAM,WAAW,WAAW;IAC1B,oCAAoC;IACpC,KAAK,CAAC,EAAE,MAAM,CAAC;IACf,qCAAqC;IACrC,MAAM,CAAC,EAAE,MAAM,CAAC;IAChB,4BAA4B;IAC5B,SAAS,EAAE,MAAM,CAAC;IAClB,qDAAqD;IACrD,WAAW,EAAE,OAAO,CAAC;CACtB;AAED,mCAAmC;AACnC,MAAM,WAAW,cAAc;IAC7B,qCAAqC;IACrC,IAAI,EAAE,MAAM,CAAC;IACb,kCAAkC;IAClC,KAAK,EAAE,WAAW,EAAE,CAAC;IACrB,wBAAwB;IACxB,QAAQ,EAAE,WAAW,CAAC;IACtB,oCAAoC;IACpC,WAAW,EAAE,OAAO,CAAC;IACrB,4CAA4C;IAC5C,QAAQ,EAAE,MAAM,EAAE,CAAC;CACpB;AAED,0CAA0C;AAC1C,qBAAa,aAAc,SAAQ,KAAK;aAGpB,IAAI,EAAE,WAAW,GAAG,oBAAoB,GAAG,YAAY,GAAG,aAAa;aACvE,aAAa,CAAC,EAAE,KAAK;gBAFrC,OAAO,EAAE,MAAM,EACC,IAAI,EAAE,WAAW,GAAG,oBAAoB,GAAG,YAAY,GAAG,aAAa,EACvE,aAAa,CAAC,EAAE,KAAK,YAAA;CAKxC;AA+DD;;;;;GAKG;AACH,qBAAa,UAAU;IACrB,OAAO,CAAC,KAAK,CAAyB;IAEtC;;OAEG;YACW,iBAAiB;IAO/B;;;;;;;OAOG;IACG,WAAW,CACf,WAAW,EAAE,WAAW,EACxB,UAAU,CAAC,EAAE,CAAC,QAAQ,EAAE,MAAM,KAAK,IAAI,GACtC,OAAO,CAAC,cAAc,CAAC;IA6O1B;;;;;;;;;OASG;IACH,OAAO,CAAC,WAAW;IAgCnB;;;;;OAKG;IACG,YAAY,CAAC,WAAW,EAAE,WAAW,GAAG,OAAO,CAAC,MAAM,CAAC;CAoB9D;AAED;;GAEG;AACH,eAAO,MAAM,UAAU,YAAmB,CAAC"}

package/dist/src/shared/adapters/pdf-adapter.js

@@ -0,0 +1,322 @@
+/**
+ * PDF Adapter Module
+ *
+ * Converts PDF files to text/HTML using PDF.js for processing through
+ * the existing ConversionEngine pipeline.
+ *
+ * Requirements: 3.1, 3.2, 3.3, 3.4, 3.5, 3.6, 3.7, 3.8
+ */
+import { loadLibrary } from '@/shared/dynamic-loader';
+/** Error thrown when PDF parsing fails */
+export class PdfParseError extends Error {
+    code;
+    originalError;
+    constructor(message, code, originalError) {
+        super(message);
+        this.code = code;
+        this.originalError = originalError;
+        this.name = 'PdfParseError';
+    }
+}
+/** Threshold for considering a PDF as image-only (minimum text characters) */
+const MIN_TEXT_CHARS_THRESHOLD = 10;
+/** Threshold for detecting multi-column layout (x-position variance) */
+const MULTI_COLUMN_X_VARIANCE_THRESHOLD = 100;
+/**
+ * Escape HTML special characters
+ */
+function escapeHtml(str) {
+    return str
+        .replace(/&/g, '&')
+        .replace(/</g, '&lt;')
+        .replace(/>/g, '&gt;')
+        .replace(/"/g, '&quot;')
+        .replace(/'/g, '&#039;');
+}
+/**
+ * PDF Adapter class
+ *
+ * Handles conversion of PDF files to HTML using PDF.js.
+ * Extracts text content and adds page markers between pages.
+ */
+export class PdfAdapter {
+    pdfjs = null;
+    /**
+     * Load the PDF.js library if not already loaded
+     */
+    async ensurePdfjsLoaded() {
+        if (!this.pdfjs) {
+            this.pdfjs = await loadLibrary('pdfjs');
+        }
+        return this.pdfjs;
+    }
+    /**
+     * Parse a PDF file and convert it to HTML with page markers
+     *
+     * @param arrayBuffer - The PDF file content as an ArrayBuffer
+     * @param onProgress - Optional progress callback (0-1)
+     * @returns Promise resolving to the parse result with HTML and metadata
+     * @throws PdfParseError if the file cannot be parsed
+     */
+    async parseToText(arrayBuffer, onProgress) {
+        const pdfjs = await this.ensurePdfjsLoaded();
+        const warnings = [];
+        const pages = [];
+        let loadingTask = null;
+        try {
+            // Load the PDF document with JavaScript disabled (Req 3.13)
+            loadingTask = pdfjs.getDocument({
+                data: arrayBuffer,
+                disableAutoFetch: true,
+                disableStream: true,
+                // Disable JavaScript execution in PDF
+                isEvalSupported: false,
+            });
+            let pdf;
+            try {
+                pdf = await loadingTask.promise;
+            }
+            catch (error) {
+                // Check for password-protected PDF (Req 3.8)
+                const errorMessage = error instanceof Error ? error.message.toLowerCase() : String(error).toLowerCase();
+                if (errorMessage.includes('password') ||
+                    errorMessage.includes('encrypted') ||
+                    errorMessage.includes('need a password')) {
+                    throw new PdfParseError('This PDF is password-protected and cannot be processed.', 'PASSWORD_PROTECTED', error instanceof Error ? error : undefined);
+                }
+                throw error;
+            }
+            const numPages = pdf.numPages;
+            let totalTextLength = 0;
+            let hasMultiColumnWarning = false;
+            // Extract metadata
+            const metadata = {
+                pageCount: numPages,
+                isTextBased: false,
+            };
+            try {
+                const pdfMetadata = await pdf.getMetadata();
+                if (pdfMetadata.info) {
+                    metadata.title = pdfMetadata.info.Title;
+                    metadata.author = pdfMetadata.info.Author;
+                }
+            }
+            catch {
+                // Metadata extraction failed, continue without it
+            }
+            // Process each page
+            for (let pageNum = 1; pageNum <= numPages; pageNum++) {
+                onProgress?.(pageNum / numPages);
+                const page = await pdf.getPage(pageNum);
+                const viewport = page.getViewport({ scale: 1.0 });
+                // Get text content
+                const textContent = await page.getTextContent();
+                // Extract text items and check for multi-column layout
+                const textItems = [];
+                const xPositions = [];
+                for (const item of textContent.items) {
+                    if ('str' in item && typeof item.str === 'string') {
+                        textItems.push(item);
+                        // Track x-positions for multi-column detection
+                        if (item.transform && item.transform.length >= 5) {
+                            xPositions.push(item.transform[4]);
+                        }
+                    }
+                }
+                // Detect multi-column layout (Req 3.3)
+                if (!hasMultiColumnWarning && xPositions.length > 10) {
+                    const uniqueXPositions = [...new Set(xPositions.map(x => Math.round(x / 10) * 10))];
+                    if (uniqueXPositions.length >= 3) {
+                        const xRange = Math.max(...xPositions) - Math.min(...xPositions);
+                        if (xRange > viewport.width * 0.5 && uniqueXPositions.length >= 3) {
+                            // Check if there are distinct column groups
+                            const sortedX = [...uniqueXPositions].sort((a, b) => a - b);
+                            let hasGap = false;
+                            for (let i = 1; i < sortedX.length; i++) {
+                                if (sortedX[i] - sortedX[i - 1] > MULTI_COLUMN_X_VARIANCE_THRESHOLD) {
+                                    hasGap = true;
+                                    break;
+                                }
+                            }
+                            if (hasGap) {
+                                warnings.push('Multi-column PDFs may not preserve layout accurately');
+                                hasMultiColumnWarning = true;
+                            }
+                        }
+                    }
+                }
+                // Build text content for the page
+                let pageText = '';
+                let lastY = null;
+                for (const item of textItems) {
+                    const text = item.str;
+                    if (!text)
+                        continue;
+                    // Check for line breaks based on y-position changes
+                    if (item.transform && item.transform.length >= 6) {
+                        const currentY = item.transform[5];
+                        if (lastY !== null && Math.abs(currentY - lastY) > 5) {
+                            // New line detected
+                            pageText += '\n';
+                        }
+                        else if (pageText && !pageText.endsWith(' ') && !pageText.endsWith('\n')) {
+                            // Add space between items on same line
+                            pageText += ' ';
+                        }
+                        lastY = currentY;
+                    }
+                    else if (pageText && !pageText.endsWith(' ') && !pageText.endsWith('\n')) {
+                        pageText += ' ';
+                    }
+                    pageText += text;
+                }
+                // Clean up the text
+                pageText = pageText
+                    .replace(/\n{3,}/g, '\n\n') // Collapse multiple newlines
+                    .replace(/[ \t]+/g, ' ') // Collapse multiple spaces
+                    .trim();
+                totalTextLength += pageText.length;
+                // Check for images in the page (for hasImages flag)
+                let hasImages = false;
+                try {
+                    const operatorList = await page.getOperatorList();
+                    // OPS.paintImageXObject = 85, OPS.paintInlineImageXObject = 86
+                    hasImages = operatorList.fnArray.some(op => op === 85 || op === 86);
+                }
+                catch {
+                    // Operator list extraction failed, assume no images
+                }
+                pages.push({
+                    pageNumber: pageNum,
+                    textContent: pageText,
+                    hasImages,
+                    viewportWidth: viewport.width,
+                });
+            }
+            // Check if PDF is image-only (Req 3.7)
+            // A PDF is considered image-only if:
+            // 1. It has very little text AND contains images, OR
+            // 2. It has absolutely no text at all
+            // This avoids false positives for legitimate short-text PDFs (receipts, forms)
+            const hasAnyImages = pages.some(p => p.hasImages);
+            const hasNoText = totalTextLength === 0;
+            const hasMinimalTextWithImages = totalTextLength < MIN_TEXT_CHARS_THRESHOLD && hasAnyImages;
+            if (hasNoText || hasMinimalTextWithImages) {
+                throw new PdfParseError('This PDF contains only images. Text extraction requires a text layer. OCR is not supported.', 'IMAGE_ONLY');
+            }
+            metadata.isTextBased = true;
+            // Convert pages to HTML with page markers (Req 3.5)
+            const html = this.pagesToHtml(pages);
+            return {
+                html,
+                pages,
+                metadata,
+                isTextBased: true,
+                warnings,
+            };
+        }
+        catch (error) {
+            // Handle PdfParseError as-is
+            if (error instanceof PdfParseError) {
+                throw error;
+            }
+            // Handle specific PDF.js errors
+            const errorMessage = error instanceof Error ? error.message.toLowerCase() : String(error).toLowerCase();
+            if (errorMessage.includes('password') ||
+                errorMessage.includes('encrypted')) {
+                throw new PdfParseError('This PDF is password-protected and cannot be processed.', 'PASSWORD_PROTECTED', error instanceof Error ? error : undefined);
+            }
+            if (errorMessage.includes('invalid pdf') ||
+                errorMessage.includes('not a pdf') ||
+                errorMessage.includes('missing pdf') ||
+                errorMessage.includes('corrupted') ||
+                errorMessage.includes('unexpected') ||
+                errorMessage.includes('invalid xref') ||
+                errorMessage.includes('bad xref')) {
+                throw new PdfParseError('Unable to read this PDF. It may be corrupted or in an unsupported format.', 'CORRUPTED', error instanceof Error ? error : undefined);
+            }
+            // Generic parse error
+            throw new PdfParseError(`Failed to parse PDF file: ${error instanceof Error ? error.message : String(error)}`, 'PARSE_ERROR', error instanceof Error ? error : undefined);
+        }
+        finally {
+            // Clean up loading task
+            if (loadingTask) {
+                try {
+                    loadingTask.destroy();
+                }
+                catch {
+                    // Ignore cleanup errors
+                }
+            }
+        }
+    }
+    /**
+     * Convert parsed pages to HTML with page markers
+     *
+     * Requirement 3.5: Insert page markers between pages
+     *
+     * Note: We use a <div> element with data-page-marker attribute instead of
+     * HTML comments because DOMPurify strips comments by default. The marker
+     * element survives sanitization and is converted to the comment format
+     * after sanitization in the file processor pipeline.
+     */
+    pagesToHtml(pages) {
+        if (pages.length === 0) {
+            return '';
+        }
+        return pages.map((page, index) => {
+            // Add page marker for pages after the first (Req 3.5)
+            // Use a div with data attribute that survives DOMPurify sanitization
+            const pageMarker = index > 0
+                ? `<div data-page-marker="${page.pageNumber}" class="pdf-page-break">--- Page ${page.pageNumber} ---</div>\n`
+                : '';
+            // Convert text content to HTML paragraphs
+            const paragraphs = page.textContent
+                .split('\n\n')
+                .filter(p => p.trim())
+                .map(p => {
+                const trimmed = p.trim();
+                // Try to detect headings (lines that are short and don't end with punctuation)
+                if (trimmed.length < 100 && !trimmed.match(/[.,:;!?]$/)) {
+                    // Could be a heading - wrap in strong for emphasis
+                    // The ConversionEngine will handle proper heading detection
+                    return `<p><strong>${escapeHtml(trimmed)}</strong></p>`;
+                }
+                return `<p>${escapeHtml(trimmed)}</p>`;
+            })
+                .join('\n');
+            return pageMarker + paragraphs;
+        }).join('\n\n');
+    }
+    /**
+     * Get the number of pages in a PDF without full parsing
+     *
+     * @param arrayBuffer - The PDF file content as an ArrayBuffer
+     * @returns Promise resolving to the page count
+     */
+    async getPageCount(arrayBuffer) {
+        const pdfjs = await this.ensurePdfjsLoaded();
+        const loadingTask = pdfjs.getDocument({
+            data: arrayBuffer,
+            disableAutoFetch: true,
+            disableStream: true,
+        });
+        try {
+            const pdf = await loadingTask.promise;
+            return pdf.numPages;
+        }
+        finally {
+            try {
+                loadingTask.destroy();
+            }
+            catch {
+                // Ignore cleanup errors
+            }
+        }
+    }
+}
+/**
+ * Singleton instance for convenience
+ */
+export const pdfAdapter = new PdfAdapter();
+//# sourceMappingURL=pdf-adapter.js.map

package/dist/src/shared/adapters/pdf-adapter.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pdf-adapter.js","sourceRoot":"","sources":["../../../../../../src/shared/adapters/pdf-adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,EAAE,WAAW,EAAiB,MAAM,yBAAyB,CAAC;AAwCrE,0CAA0C;AAC1C,MAAM,OAAO,aAAc,SAAQ,KAAK;IAGpB;IACA;IAHlB,YACE,OAAe,EACC,IAAuE,EACvE,aAAqB;QAErC,KAAK,CAAC,OAAO,CAAC,CAAC;QAHC,SAAI,GAAJ,IAAI,CAAmE;QACvE,kBAAa,GAAb,aAAa,CAAQ;QAGrC,IAAI,CAAC,IAAI,GAAG,eAAe,CAAC;IAC9B,CAAC;CACF;AA6CD,8EAA8E;AAC9E,MAAM,wBAAwB,GAAG,EAAE,CAAC;AAEpC,wEAAwE;AACxE,MAAM,iCAAiC,GAAG,GAAG,CAAC;AAE9C;;GAEG;AACH,SAAS,UAAU,CAAC,GAAW;IAC7B,OAAO,GAAG;SACP,OAAO,CAAC,IAAI,EAAE,OAAO,CAAC;SACtB,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC;SACrB,OAAO,CAAC,IAAI,EAAE,MAAM,CAAC;SACrB,OAAO,CAAC,IAAI,EAAE,QAAQ,CAAC;SACvB,OAAO,CAAC,IAAI,EAAE,QAAQ,CAAC,CAAC;AAC7B,CAAC;AAED;;;;;GAKG;AACH,MAAM,OAAO,UAAU;IACb,KAAK,GAAoB,IAAI,CAAC;IAEtC;;OAEG;IACK,KAAK,CAAC,iBAAiB;QAC7B,IAAI,CAAC,IAAI,CAAC,KAAK,EAAE,CAAC;YAChB,IAAI,CAAC,KAAK,GAAG,MAAM,WAAW,CAAC,OAAO,CAAC,CAAC;QAC1C,CAAC;QACD,OAAO,IAAI,CAAC,KAAK,CAAC;IACpB,CAAC;IAED;;;;;;;OAOG;IACH,KAAK,CAAC,WAAW,CACf,WAAwB,EACxB,UAAuC;QAEvC,MAAM,KAAK,GAAG,MAAM,IAAI,CAAC,iBAAiB,EAAE,CAAC;QAC7C,MAAM,QAAQ,GAAa,EAAE,CAAC;QAC9B,MAAM,KAAK,GAAkB,EAAE,CAAC;QAChC,IAAI,WAAW,GAAkC,IAAI,CAAC;QAEtD,IAAI,CAAC;YACH,4DAA4D;YAC5D,WAAW,GAAG,KAAK,CAAC,WAAW,CAAC;gBAC9B,IAAI,EAAE,WAAW;gBACjB,gBAAgB,EAAE,IAAI;gBACtB,aAAa,EAAE,IAAI;gBACnB,sCAAsC;gBACtC,eAAe,EAAE,KAAK;aACvB,CAAsC,CAAC;YAExC,IAAI,GAAqB,CAAC;YAC1B,IAAI,CAAC;gBACH,GAAG,GAAG,MAAM,WAAW,CAAC,OAAO,CAAC;YAClC,CAAC;YAAC,OAAO,KAAK,EAAE,CAAC;gBACf,6CAA6C;gBAC7C,MAAM,YAAY,GAAG,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,WAAW,EAAE,CAAC;gBACxG,IACE,YAAY,CAAC,QAAQ,CAAC,UAAU,CAAC;oBACjC,YAAY,CAAC,QAAQ,CAAC,WAAW,CAAC;oBAClC,YAAY,CAAC,QAAQ,CAAC,iBAAiB,CAAC,EACxC,CAAC;oBACD,MAAM,IAAI,aAAa,CACrB,yDAAyD,EACzD,oBAAoB,EACpB,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,SAAS,CAC3C,CAAC;gBACJ,CAAC;gBACD,MAAM,KAAK,CAAC;YACd,CAAC;YAED,MAAM,QAAQ,GAAG,GAAG,CAAC,QAAQ,CAAC;YAC9B,IAAI,eAAe,GAAG,CAAC,CAAC;YACxB,IAAI,qBAAqB,GAAG,KAAK,CAAC;YAElC,mBAAmB;YACnB,MAAM,QAAQ,GAAgB;gBAC5B,SAAS,EAAE,QAAQ;gBACnB,WAAW,EAAE,KAAK;aACnB,CAAC;YAEF,IAAI,CAAC;gBACH,MAAM,WAAW,GAAG,MAAM,GAAG,CAAC,WAAW,EAAE,CAAC;gBAC5C,IAAI,WAAW,CAAC,IAAI,EAAE,CAAC;oBACrB,QAAQ,CAAC,KAAK,GAAG,WAAW,CAAC,IAAI,CAAC,KAA2B,CAAC;oBAC9D,QAAQ,CAAC,MAAM,GAAG,WAAW,CAAC,IAAI,CAAC,MAA4B,CAAC;gBAClE,CAAC;YACH,CAAC;YAAC,MAAM,CAAC;gBACP,kDAAkD;YACpD,CAAC;YAED,oBAAoB;YACpB,KAAK,IAAI,OAAO,GAAG,CAAC,EAAE,OAAO,IAAI,QAAQ,EAAE,OAAO,EAAE,EAAE,CAAC;gBACrD,UAAU,EAAE,CAAC,OAAO,GAAG,QAAQ,CAAC,CAAC;gBAEjC,MAAM,IAAI,GAAG,MAAM,GAAG,CAAC,OAAO,CAAC,OAAO,CAAC,CAAC;gBACxC,MAAM,QAAQ,GAAG,IAAI,CAAC,WAAW,CAAC,EAAE,KAAK,EAAE,GAAG,EAAE,CAAC,CAAC;gBAElD,mBAAmB;gBACnB,MAAM,WAAW,GAAG,MAAM,IAAI,CAAC,cAAc,EAAE,CAAC;gBAEhD,uDAAuD;gBACvD,MAAM,SAAS,GAAe,EAAE,CAAC;gBACjC,MAAM,UAAU,GAAa,EAAE,CAAC;gBAEhC,KAAK,MAAM,IAAI,IAAI,WAAW,CAAC,KAAK,EAAE,CAAC;oBACrC,IAAI,KAAK,IAAI,IAAI,IAAI,OAAO,IAAI,CAAC,GAAG,KAAK,QAAQ,EAAE,CAAC;wBAClD,SAAS,CAAC,IAAI,CAAC,IAAgB,CAAC,CAAC;wBACjC,+CAA+C;wBAC/C,IAAI,IAAI,CAAC,SAAS,IAAI,IAAI,CAAC,SAAS,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;4BACjD,UAAU,CAAC,IAAI,CAAC,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC,CAAC;wBACrC,CAAC;oBACH,CAAC;gBACH,CAAC;gBAED,uCAAuC;gBACvC,IAAI,CAAC,qBAAqB,IAAI,UAAU,CAAC,MAAM,GAAG,EAAE,EAAE,CAAC;oBACrD,MAAM,gBAAgB,GAAG,CAAC,GAAG,IAAI,GAAG,CAAC,UAAU,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,CAAC,IAAI,CAAC,KAAK,CAAC,CAAC,GAAG,EAAE,CAAC,GAAG,EAAE,CAAC,CAAC,CAAC,CAAC;oBACpF,IAAI,gBAAgB,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;wBACjC,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,UAAU,CAAC,GAAG,IAAI,CAAC,GAAG,CAAC,GAAG,UAAU,CAAC,CAAC;wBACjE,IAAI,MAAM,GAAG,QAAQ,CAAC,KAAK,GAAG,GAAG,IAAI,gBAAgB,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;4BAClE,4CAA4C;4BAC5C,MAAM,OAAO,GAAG,CAAC,GAAG,gBAAgB,CAAC,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,GAAG,CAAC,CAAC,CAAC;4BAC5D,IAAI,MAAM,GAAG,KAAK,CAAC;4BACnB,KAAK,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,OAAO,CAAC,MAAM,EAAE,CAAC,EAAE,EAAE,CAAC;gCACxC,IAAI,OAAO,CAAC,CAAC,CAAC,GAAG,OAAO,CAAC,CAAC,GAAG,CAAC,CAAC,GAAG,iCAAiC,EAAE,CAAC;oCACpE,MAAM,GAAG,IAAI,CAAC;oCACd,MAAM;gCACR,CAAC;4BACH,CAAC;4BACD,IAAI,MAAM,EAAE,CAAC;gCACX,QAAQ,CAAC,IAAI,CAAC,sDAAsD,CAAC,CAAC;gCACtE,qBAAqB,GAAG,IAAI,CAAC;4BAC/B,CAAC;wBACH,CAAC;oBACH,CAAC;gBACH,CAAC;gBAED,kCAAkC;gBAClC,IAAI,QAAQ,GAAG,EAAE,CAAC;gBAClB,IAAI,KAAK,GAAkB,IAAI,CAAC;gBAEhC,KAAK,MAAM,IAAI,IAAI,SAAS,EAAE,CAAC;oBAC7B,MAAM,IAAI,GAAG,IAAI,CAAC,GAAG,CAAC;oBACtB,IAAI,CAAC,IAAI;wBAAE,SAAS;oBAEpB,oDAAoD;oBACpD,IAAI,IAAI,CAAC,SAAS,IAAI,IAAI,CAAC,SAAS,CAAC,MAAM,IAAI,CAAC,EAAE,CAAC;wBACjD,MAAM,QAAQ,GAAG,IAAI,CAAC,SAAS,CAAC,CAAC,CAAC,CAAC;wBACnC,IAAI,KAAK,KAAK,IAAI,IAAI,IAAI,CAAC,GAAG,CAAC,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,EAAE,CAAC;4BACrD,oBAAoB;4BACpB,QAAQ,IAAI,IAAI,CAAC;wBACnB,CAAC;6BAAM,IAAI,QAAQ,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;4BAC3E,uCAAuC;4BACvC,QAAQ,IAAI,GAAG,CAAC;wBAClB,CAAC;wBACD,KAAK,GAAG,QAAQ,CAAC;oBACnB,CAAC;yBAAM,IAAI,QAAQ,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,GAAG,CAAC,IAAI,CAAC,QAAQ,CAAC,QAAQ,CAAC,IAAI,CAAC,EAAE,CAAC;wBAC3E,QAAQ,IAAI,GAAG,CAAC;oBAClB,CAAC;oBAED,QAAQ,IAAI,IAAI,CAAC;gBACnB,CAAC;gBAED,oBAAoB;gBACpB,QAAQ,GAAG,QAAQ;qBAChB,OAAO,CAAC,SAAS,EAAE,MAAM,CAAC,CAAE,6BAA6B;qBACzD,OAAO,CAAC,SAAS,EAAE,GAAG,CAAC,CAAM,2BAA2B;qBACxD,IAAI,EAAE,CAAC;gBAEV,eAAe,IAAI,QAAQ,CAAC,MAAM,CAAC;gBAEnC,oDAAoD;gBACpD,IAAI,SAAS,GAAG,KAAK,CAAC;gBACtB,IAAI,CAAC;oBACH,MAAM,YAAY,GAAG,MAAM,IAAI,CAAC,eAAe,EAAE,CAAC;oBAClD,+DAA+D;oBAC/D,SAAS,GAAG,YAAY,CAAC,OAAO,CAAC,IAAI,CAAC,EAAE,CAAC,EAAE,CAAC,EAAE,KAAK,EAAE,IAAI,EAAE,KAAK,EAAE,CAAC,CAAC;gBACtE,CAAC;gBAAC,MAAM,CAAC;oBACP,oDAAoD;gBACtD,CAAC;gBAED,KAAK,CAAC,IAAI,CAAC;oBACT,UAAU,EAAE,OAAO;oBACnB,WAAW,EAAE,QAAQ;oBACrB,SAAS;oBACT,aAAa,EAAE,QAAQ,CAAC,KAAK;iBAC9B,CAAC,CAAC;YACL,CAAC;YAED,uCAAuC;YACvC,qCAAqC;YACrC,qDAAqD;YACrD,sCAAsC;YACtC,+EAA+E;YAC/E,MAAM,YAAY,GAAG,KAAK,CAAC,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,SAAS,CAAC,CAAC;YAClD,MAAM,SAAS,GAAG,eAAe,KAAK,CAAC,CAAC;YACxC,MAAM,wBAAwB,GAAG,eAAe,GAAG,wBAAwB,IAAI,YAAY,CAAC;YAE5F,IAAI,SAAS,IAAI,wBAAwB,EAAE,CAAC;gBAC1C,MAAM,IAAI,aAAa,CACrB,6FAA6F,EAC7F,YAAY,CACb,CAAC;YACJ,CAAC;YAED,QAAQ,CAAC,WAAW,GAAG,IAAI,CAAC;YAE5B,oDAAoD;YACpD,MAAM,IAAI,GAAG,IAAI,CAAC,WAAW,CAAC,KAAK,CAAC,CAAC;YAErC,OAAO;gBACL,IAAI;gBACJ,KAAK;gBACL,QAAQ;gBACR,WAAW,EAAE,IAAI;gBACjB,QAAQ;aACT,CAAC;QACJ,CAAC;QAAC,OAAO,KAAK,EAAE,CAAC;YACf,6BAA6B;YAC7B,IAAI,KAAK,YAAY,aAAa,EAAE,CAAC;gBACnC,MAAM,KAAK,CAAC;YACd,CAAC;YAED,gCAAgC;YAChC,MAAM,YAAY,GAAG,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,WAAW,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,WAAW,EAAE,CAAC;YAExG,IACE,YAAY,CAAC,QAAQ,CAAC,UAAU,CAAC;gBACjC,YAAY,CAAC,QAAQ,CAAC,WAAW,CAAC,EAClC,CAAC;gBACD,MAAM,IAAI,aAAa,CACrB,yDAAyD,EACzD,oBAAoB,EACpB,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,SAAS,CAC3C,CAAC;YACJ,CAAC;YAED,IACE,YAAY,CAAC,QAAQ,CAAC,aAAa,CAAC;gBACpC,YAAY,CAAC,QAAQ,CAAC,WAAW,CAAC;gBAClC,YAAY,CAAC,QAAQ,CAAC,aAAa,CAAC;gBACpC,YAAY,CAAC,QAAQ,CAAC,WAAW,CAAC;gBAClC,YAAY,CAAC,QAAQ,CAAC,YAAY,CAAC;gBACnC,YAAY,CAAC,QAAQ,CAAC,cAAc,CAAC;gBACrC,YAAY,CAAC,QAAQ,CAAC,UAAU,CAAC,EACjC,CAAC;gBACD,MAAM,IAAI,aAAa,CACrB,2EAA2E,EAC3E,WAAW,EACX,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,SAAS,CAC3C,CAAC;YACJ,CAAC;YAED,sBAAsB;YACtB,MAAM,IAAI,aAAa,CACrB,6BAA6B,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,EAAE,EACrF,aAAa,EACb,KAAK,YAAY,KAAK,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,SAAS,CAC3C,CAAC;QACJ,CAAC;gBAAS,CAAC;YACT,wBAAwB;YACxB,IAAI,WAAW,EAAE,CAAC;gBAChB,IAAI,CAAC;oBACH,WAAW,CAAC,OAAO,EAAE,CAAC;gBACxB,CAAC;gBAAC,MAAM,CAAC;oBACP,wBAAwB;gBAC1B,CAAC;YACH,CAAC;QACH,CAAC;IACH,CAAC;IAED;;;;;;;;;OASG;IACK,WAAW,CAAC,KAAoB;QACtC,IAAI,KAAK,CAAC,MAAM,KAAK,CAAC,EAAE,CAAC;YACvB,OAAO,EAAE,CAAC;QACZ,CAAC;QAED,OAAO,KAAK,CAAC,GAAG,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,EAAE;YAC/B,sDAAsD;YACtD,qEAAqE;YACrE,MAAM,UAAU,GAAG,KAAK,GAAG,CAAC;gBAC1B,CAAC,CAAC,0BAA0B,IAAI,CAAC,UAAU,qCAAqC,IAAI,CAAC,UAAU,cAAc;gBAC7G,CAAC,CAAC,EAAE,CAAC;YAEP,0CAA0C;YAC1C,MAAM,UAAU,GAAG,IAAI,CAAC,WAAW;iBAChC,KAAK,CAAC,MAAM,CAAC;iBACb,MAAM,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC;iBACrB,GAAG,CAAC,CAAC,CAAC,EAAE;gBACP,MAAM,OAAO,GAAG,CAAC,CAAC,IAAI,EAAE,CAAC;gBACzB,+EAA+E;gBAC/E,IAAI,OAAO,CAAC,MAAM,GAAG,GAAG,IAAI,CAAC,OAAO,CAAC,KAAK,CAAC,WAAW,CAAC,EAAE,CAAC;oBACxD,mDAAmD;oBACnD,4DAA4D;oBAC5D,OAAO,cAAc,UAAU,CAAC,OAAO,CAAC,eAAe,CAAC;gBAC1D,CAAC;gBACD,OAAO,MAAM,UAAU,CAAC,OAAO,CAAC,MAAM,CAAC;YACzC,CAAC,CAAC;iBACD,IAAI,CAAC,IAAI,CAAC,CAAC;YAEd,OAAO,UAAU,GAAG,UAAU,CAAC;QACjC,CAAC,CAAC,CAAC,IAAI,CAAC,MAAM,CAAC,CAAC;IAClB,CAAC;IAED;;;;;OAKG;IACH,KAAK,CAAC,YAAY,CAAC,WAAwB;QACzC,MAAM,KAAK,GAAG,MAAM,IAAI,CAAC,iBAAiB,EAAE,CAAC;QAE7C,MAAM,WAAW,GAAG,KAAK,CAAC,WAAW,CAAC;YACpC,IAAI,EAAE,WAAW;YACjB,gBAAgB,EAAE,IAAI;YACtB,aAAa,EAAE,IAAI;SACpB,CAAsC,CAAC;QAExC,IAAI,CAAC;YACH,MAAM,GAAG,GAAG,MAAM,WAAW,CAAC,OAAO,CAAC;YACtC,OAAO,GAAG,CAAC,QAAQ,CAAC;QACtB,CAAC;gBAAS,CAAC;YACT,IAAI,CAAC;gBACH,WAAW,CAAC,OAAO,EAAE,CAAC;YACxB,CAAC;YAAC,MAAM,CAAC;gBACP,wBAAwB;YAC1B,CAAC;QACH,CAAC;IACH,CAAC;CACF;AAED;;GAEG;AACH,MAAM,CAAC,MAAM,UAAU,GAAG,IAAI,UAAU,EAAE,CAAC"}

package/dist/src/shared/adapters/spreadsheet-adapter.d.ts

@@ -0,0 +1,98 @@
+/**
+ * Spreadsheet Adapter Module
+ *
+ * Converts XLSX and CSV files to HTML using SheetJS for processing through
+ * the existing ConversionEngine pipeline.
+ *
+ * Requirements: 2.1, 2.2, 2.7, 2.8, 2.9, 2.14, 2.15, 2.16
+ */
+/** Sheet information extracted from spreadsheet */
+export interface SheetInfo {
+    /** Sheet name */
+    name: string;
+    /** Number of rows in the sheet */
+    rowCount: number;
+    /** Number of columns in the sheet */
+    columnCount: number;
+}
+/** Result of parsing a spreadsheet file */
+export interface SpreadsheetParseResult {
+    /** HTML table representation of the spreadsheet */
+    html: string;
+    /** Information about all sheets in the workbook */
+    sheets: SheetInfo[];
+    /** Index of the currently active/converted sheet */
+    activeSheetIndex: number;
+    /** Any warnings generated during parsing */
+    warnings: string[];
+}
+/** Error thrown when spreadsheet parsing fails */
+export declare class SpreadsheetParseError extends Error {
+    readonly code: 'CORRUPTED' | 'INVALID_FORMAT' | 'PARSE_ERROR' | 'EMPTY_FILE';
+    readonly originalError?: Error | undefined;
+    constructor(message: string, code: 'CORRUPTED' | 'INVALID_FORMAT' | 'PARSE_ERROR' | 'EMPTY_FILE', originalError?: Error | undefined);
+}
+/**
+ * Spreadsheet Adapter class
+ *
+ * Handles conversion of XLSX and CSV files to HTML using SheetJS.
+ * The HTML output is then processed through the ConversionEngine pipeline.
+ */
+export declare class SpreadsheetAdapter {
+    private xlsx;
+    private currentWorkbook;
+    /**
+     * Load the SheetJS library if not already loaded
+     */
+    private ensureXlsxLoaded;
+    /**
+     * Parse a spreadsheet file and convert it to HTML
+     *
+     * @param arrayBuffer - The file content as an ArrayBuffer
+     * @param format - The file format ('xlsx' or 'csv')
+     * @param sheetIndex - Optional sheet index to convert (default: 0)
+     * @returns Promise resolving to the parse result with HTML and metadata
+     * @throws SpreadsheetParseError if the file cannot be parsed
+     */
+    parseToHtml(arrayBuffer: ArrayBuffer, format?: 'xlsx' | 'csv', sheetIndex?: number): Promise<SpreadsheetParseResult>;
+    /**
+     * Convert a worksheet to HTML table format
+     *
+     * Requirements:
+     * - 2.7: Handle merged cells by repeating content
+     * - 2.9: Preserve empty rows/columns in table structure
+     * - 2.14: Use pipe (|) syntax for Markdown tables (via HTML table)
+     */
+    private worksheetToHtml;
+    /**
+     * Build a map of merged cells with their values repeated
+     *
+     * Requirement 2.7: Handle merged cells by repeating content
+     */
+    private buildMergedCellsMap;
+    /**
+     * Get the names of all sheets in the current workbook
+     *
+     * Requirement 2.4: Display sheet selector with sheet names
+     */
+    getSheetNames(): string[];
+    /**
+     * Get the number of sheets in the current workbook
+     */
+    getSheetCount(): number;
+    /**
+     * Re-parse the current workbook with a different sheet selected
+     *
+     * Requirement 2.5: Re-convert when user selects different sheet
+     */
+    selectSheet(sheetIndex: number): Promise<SpreadsheetParseResult>;
+    /**
+     * Clear the current workbook from memory
+     */
+    clearWorkbook(): void;
+}
+/**
+ * Singleton instance for convenience
+ */
+export declare const spreadsheetAdapter: SpreadsheetAdapter;
+//# sourceMappingURL=spreadsheet-adapter.d.ts.map

package/dist/src/shared/adapters/spreadsheet-adapter.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"spreadsheet-adapter.d.ts","sourceRoot":"","sources":["../../../../../../src/shared/adapters/spreadsheet-adapter.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAIH,mDAAmD;AACnD,MAAM,WAAW,SAAS;IACxB,iBAAiB;IACjB,IAAI,EAAE,MAAM,CAAC;IACb,kCAAkC;IAClC,QAAQ,EAAE,MAAM,CAAC;IACjB,qCAAqC;IACrC,WAAW,EAAE,MAAM,CAAC;CACrB;AAED,2CAA2C;AAC3C,MAAM,WAAW,sBAAsB;IACrC,mDAAmD;IACnD,IAAI,EAAE,MAAM,CAAC;IACb,mDAAmD;IACnD,MAAM,EAAE,SAAS,EAAE,CAAC;IACpB,oDAAoD;IACpD,gBAAgB,EAAE,MAAM,CAAC;IACzB,4CAA4C;IAC5C,QAAQ,EAAE,MAAM,EAAE,CAAC;CACpB;AAED,kDAAkD;AAClD,qBAAa,qBAAsB,SAAQ,KAAK;aAG5B,IAAI,EAAE,WAAW,GAAG,gBAAgB,GAAG,aAAa,GAAG,YAAY;aACnE,aAAa,CAAC,EAAE,KAAK;gBAFrC,OAAO,EAAE,MAAM,EACC,IAAI,EAAE,WAAW,GAAG,gBAAgB,GAAG,aAAa,GAAG,YAAY,EACnE,aAAa,CAAC,EAAE,KAAK,YAAA;CAKxC;AAyND;;;;;GAKG;AACH,qBAAa,kBAAkB;IAC7B,OAAO,CAAC,IAAI,CAAwB;IACpC,OAAO,CAAC,eAAe,CAA6B;IAEpD;;OAEG;YACW,gBAAgB;IAO9B;;;;;;;;OAQG;IACG,WAAW,CACf,WAAW,EAAE,WAAW,EACxB,MAAM,GAAE,MAAM,GAAG,KAAc,EAC/B,UAAU,GAAE,MAAU,GACrB,OAAO,CAAC,sBAAsB,CAAC;IAsGlC;;;;;;;OAOG;IACH,OAAO,CAAC,eAAe;IAkDvB;;;;OAIG;IACH,OAAO,CAAC,mBAAmB;IA6B3B;;;;OAIG;IACH,aAAa,IAAI,MAAM,EAAE;IASzB;;OAEG;IACH,aAAa,IAAI,MAAM;IAIvB;;;;OAIG;IACG,WAAW,CAAC,UAAU,EAAE,MAAM,GAAG,OAAO,CAAC,sBAAsB,CAAC;IA4CtE;;OAEG;IACH,aAAa,IAAI,IAAI;CAGtB;AAED;;GAEG;AACH,eAAO,MAAM,kBAAkB,oBAA2B,CAAC"}