@cj-tech-master/excelts 3.1.0-canary.20260102071304.d80b29d → 4.0.0-canary.20260103013222.0e5d37f

package/dist/browser/modules/csv/csv-core.js

@@ -0,0 +1,723 @@
+/**
+ * CSV Parser and Formatter - RFC 4180 compliant
+ *
+ * A lightweight, cross-platform CSV implementation that works in both
+ * Node.js and Browser environments with zero dependencies.
+ *
+ * High-performance RFC 4180 compliant CSV parser and formatter.
+ *
+ * @see https://tools.ietf.org/html/rfc4180
+ */
+import { formatNumberForCsv } from "./csv-number.js";
+// =============================================================================
+// Helper Functions
+// =============================================================================
+/**
+ * Escape special regex characters
+ */
+function escapeRegex(str) {
+    return str.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
+}
+/**
+ * Check if a transform function is synchronous (1 argument) vs async (2 arguments)
+ */
+export function isSyncTransform(transform) {
+    return transform.length === 1;
+}
+/**
+ * Check if a validate function is synchronous (1 argument) vs async (2 arguments)
+ */
+export function isSyncValidate(validate) {
+    return validate.length === 1;
+}
+/**
+ * Check if headers are unique
+ */
+function validateUniqueHeaders(headers) {
+    const seen = new Set();
+    const duplicates = [];
+    for (const header of headers) {
+        if (header !== null && header !== undefined) {
+            if (seen.has(header)) {
+                duplicates.push(header);
+            }
+            seen.add(header);
+        }
+    }
+    if (duplicates.length > 0) {
+        throw new Error(`Duplicate headers found ${JSON.stringify(duplicates)}`);
+    }
+}
+// =============================================================================
+// Parse Functions
+// =============================================================================
+/**
+ * Parse a CSV string into rows of fields
+ */
+export function parseCsv(input, options = {}) {
+    const { delimiter = ",", quote: quoteOption = '"', escape: escapeOption = '"', skipEmptyLines = false, ignoreEmpty = false, trim = false, ltrim = false, rtrim = false, headers = false, renameHeaders = false, comment, maxRows, skipLines = 0, skipRows = 0, strictColumnHandling = false, discardUnmappedColumns = false, transform, validate } = options;
+    const shouldSkipEmpty = skipEmptyLines || ignoreEmpty;
+    // Handle quote: null/false to disable quoting
+    const quoteEnabled = quoteOption !== null && quoteOption !== false;
+    const quote = quoteEnabled ? String(quoteOption) : "";
+    const escape = escapeOption !== null && escapeOption !== false ? String(escapeOption) : "";
+    const rows = [];
+    const invalidRows = [];
+    let currentRow = [];
+    let currentField = "";
+    let inQuotes = false;
+    let i = 0;
+    let lineNumber = 0;
+    let dataRowCount = 0;
+    let skippedDataRows = 0;
+    // Header handling
+    let headerRow = null;
+    let headersLength = 0;
+    let useHeaders = false;
+    let headerRowProcessed = false;
+    // Determine header mode
+    if (headers === true) {
+        useHeaders = true;
+    }
+    else if (Array.isArray(headers)) {
+        headerRow = headers;
+        headersLength = headers.filter(h => h !== null && h !== undefined).length;
+        validateUniqueHeaders(headers);
+        useHeaders = true;
+        if (!renameHeaders) {
+            headerRowProcessed = true; // We already have headers, don't wait for first row
+        }
+    }
+    else if (typeof headers === "function") {
+        useHeaders = true;
+    }
+    // Pre-compute trim function to avoid repeated condition checks
+    const trimField = trim || (ltrim && rtrim)
+        ? (s) => s.trim()
+        : ltrim
+            ? (s) => s.trimStart()
+            : rtrim
+                ? (s) => s.trimEnd()
+                : (s) => s;
+    const processRow = (row) => {
+        // If we have headers defined and this is the first data row (for headers: true)
+        // or we need to validate headers from a function
+        if (useHeaders && !headerRowProcessed) {
+            // First row is headers
+            if (typeof headers === "function") {
+                const transformed = headers(row);
+                validateUniqueHeaders(transformed);
+                headerRow = transformed;
+            }
+            else if (!Array.isArray(headers)) {
+                validateUniqueHeaders(row);
+                headerRow = row;
+            }
+            headersLength = headerRow.filter(h => h !== null && h !== undefined).length;
+            headerRowProcessed = true;
+            // If renameHeaders and custom headers provided, discard this row
+            if (renameHeaders) {
+                return false;
+            }
+            // For headers: true, don't add header row to data
+            if (headers === true || typeof headers === "function") {
+                return false;
+            }
+            return false;
+        }
+        // Skip data rows
+        if (skippedDataRows < skipRows) {
+            skippedDataRows++;
+            return false;
+        }
+        // Column validation when using headers
+        if (headerRow && headerRow.length > 0) {
+            const expectedCols = headersLength;
+            const actualCols = row.length;
+            if (actualCols > expectedCols) {
+                if (strictColumnHandling && !discardUnmappedColumns) {
+                    // Mark as invalid but continue
+                    invalidRows.push({
+                        row,
+                        reason: `Column header mismatch expected: ${expectedCols} columns got: ${actualCols}`
+                    });
+                    return false;
+                }
+                else {
+                    // Default: trim extra columns
+                    row.length = headerRow.length;
+                }
+            }
+            else if (actualCols < expectedCols) {
+                if (strictColumnHandling) {
+                    invalidRows.push({
+                        row,
+                        reason: `Column header mismatch expected: ${expectedCols} columns got: ${actualCols}`
+                    });
+                    return false;
+                }
+                // Pad with empty strings
+                while (row.length < headerRow.length) {
+                    row.push("");
+                }
+            }
+        }
+        return true;
+    };
+    const len = input.length;
+    while (i < len) {
+        const char = input[i];
+        if (inQuotes && quoteEnabled) {
+            // Inside quoted field
+            if (escape && char === escape && input[i + 1] === quote) {
+                // Escaped quote
+                currentField += quote;
+                i += 2;
+            }
+            else if (char === quote) {
+                // End of quoted field
+                inQuotes = false;
+                i++;
+            }
+            else if (char === "\r") {
+                // Normalize CRLF to LF inside quoted fields
+                if (input[i + 1] === "\n") {
+                    i++; // Skip \r, will add \n on next iteration
+                }
+                else {
+                    currentField += "\n"; // Convert standalone \r to \n
+                    i++;
+                }
+            }
+            else {
+                currentField += char;
+                i++;
+            }
+        }
+        else {
+            // Outside quoted field
+            if (quoteEnabled && char === quote && currentField === "") {
+                // Start of quoted field
+                inQuotes = true;
+                i++;
+            }
+            else if (char === delimiter) {
+                // Field separator
+                currentRow.push(trimField(currentField));
+                currentField = "";
+                i++;
+            }
+            else if (char === "\n" || char === "\r") {
+                // End of row - handle \r\n, \r, and \n
+                if (char === "\r" && input[i + 1] === "\n") {
+                    i++; // Skip the \n in \r\n
+                }
+                currentRow.push(trimField(currentField));
+                currentField = "";
+                lineNumber++;
+                // Skip lines at beginning
+                if (lineNumber <= skipLines) {
+                    currentRow = [];
+                    i++;
+                    continue;
+                }
+                // Skip comment lines
+                if (comment && currentRow[0]?.startsWith(comment)) {
+                    currentRow = [];
+                    i++;
+                    continue;
+                }
+                // Skip empty lines
+                const isEmpty = currentRow.length === 1 && currentRow[0] === "";
+                if (shouldSkipEmpty && isEmpty) {
+                    currentRow = [];
+                    i++;
+                    continue;
+                }
+                // Process row (handles headers, validation)
+                if (processRow(currentRow)) {
+                    rows.push(currentRow);
+                    dataRowCount++;
+                }
+                currentRow = [];
+                i++;
+                // Check max rows - after resetting currentRow
+                if (maxRows !== undefined && dataRowCount >= maxRows) {
+                    break;
+                }
+            }
+            else {
+                currentField += char;
+                i++;
+            }
+        }
+    }
+    // Handle last field/row
+    if (currentField !== "" || currentRow.length > 0) {
+        currentRow.push(trimField(currentField));
+        // Skip lines at beginning
+        if (lineNumber >= skipLines) {
+            // Skip comment lines
+            if (!(comment && currentRow[0]?.startsWith(comment))) {
+                // Skip empty lines
+                const isEmpty = currentRow.length === 1 && currentRow[0] === "";
+                if (!(shouldSkipEmpty && isEmpty)) {
+                    if (!(maxRows !== undefined && dataRowCount >= maxRows)) {
+                        if (processRow(currentRow)) {
+                            rows.push(currentRow);
+                        }
+                    }
+                }
+            }
+        }
+    }
+    // Convert to objects if headers enabled
+    if (useHeaders && headerRow) {
+        let dataRows = rows.map(row => {
+            const obj = {};
+            headerRow.forEach((header, index) => {
+                if (header !== null && header !== undefined) {
+                    obj[header] = row[index] ?? "";
+                }
+            });
+            return obj;
+        });
+        // Apply transform if provided
+        if (transform) {
+            dataRows = dataRows
+                .map(row => transform(row))
+                .filter((row) => row !== null && row !== undefined);
+        }
+        // Apply validate if provided
+        if (validate) {
+            const validatedRows = [];
+            for (const row of dataRows) {
+                const result = validate(row);
+                if (typeof result === "boolean") {
+                    if (result) {
+                        validatedRows.push(row);
+                    }
+                    else {
+                        invalidRows.push({ row: Object.values(row), reason: "Validation failed" });
+                    }
+                }
+                else {
+                    if (result.isValid) {
+                        validatedRows.push(row);
+                    }
+                    else {
+                        invalidRows.push({
+                            row: Object.values(row),
+                            reason: result.reason || "Validation failed"
+                        });
+                    }
+                }
+            }
+            dataRows = validatedRows;
+        }
+        if ((strictColumnHandling || validate) && invalidRows.length > 0) {
+            return {
+                headers: headerRow.filter((h) => h !== null && h !== undefined),
+                rows: dataRows,
+                invalidRows
+            };
+        }
+        return {
+            headers: headerRow.filter((h) => h !== null && h !== undefined),
+            rows: dataRows
+        };
+    }
+    // For array mode (no headers), apply transform and validate
+    let resultRows = rows;
+    if (transform) {
+        resultRows = resultRows
+            .map(row => transform(row))
+            .filter((row) => row !== null && row !== undefined);
+    }
+    if (validate) {
+        const validatedRows = [];
+        const arrayInvalidRows = [];
+        for (const row of resultRows) {
+            const result = validate(row);
+            if (typeof result === "boolean") {
+                if (result) {
+                    validatedRows.push(row);
+                }
+                else {
+                    arrayInvalidRows.push({ row, reason: "Validation failed" });
+                }
+            }
+            else {
+                if (result.isValid) {
+                    validatedRows.push(row);
+                }
+                else {
+                    arrayInvalidRows.push({ row, reason: result.reason || "Validation failed" });
+                }
+            }
+        }
+        resultRows = validatedRows;
+        if (arrayInvalidRows.length > 0) {
+            return {
+                rows: resultRows,
+                invalidRows: arrayInvalidRows
+            }; // Return with invalidRows for array mode too
+        }
+    }
+    return resultRows;
+}
+// =============================================================================
+// Format Functions
+// =============================================================================
+/**
+ * Format data as a CSV string
+ */
+export function formatCsv(data, options = {}) {
+    const { delimiter = ",", quote: quoteOption = '"', escape: escapeOption, rowDelimiter = "\n", alwaysQuote = false, quoteColumns = false, quoteHeaders = false, headers, writeHeaders: writeHeadersOption, writeBOM = false, includeEndRowDelimiter = false, alwaysWriteHeaders = false, transform, decimalSeparator = "." } = options;
+    // Determine if headers should be written (default: true when headers is provided)
+    const shouldWriteHeaders = writeHeadersOption ?? true;
+    // If quote is false or null, disable quoting entirely
+    const quoteEnabled = quoteOption !== false && quoteOption !== null;
+    const quote = quoteEnabled ? String(quoteOption) : "";
+    const escape = escapeOption !== undefined && escapeOption !== false && escapeOption !== null
+        ? String(escapeOption)
+        : quote;
+    // Pre-compile regex patterns for performance
+    const needsQuoteRegex = quoteEnabled
+        ? new RegExp(`[${escapeRegex(delimiter)}${escapeRegex(quote)}\r\n]`)
+        : null;
+    const escapeQuoteRegex = quoteEnabled ? new RegExp(escapeRegex(quote), "g") : null;
+    const escapedQuote = escape + quote;
+    const lines = [];
+    const shouldQuoteColumn = (index, header, isHeader = false) => {
+        const quoteConfig = isHeader ? quoteHeaders : quoteColumns;
+        if (typeof quoteConfig === "boolean") {
+            return quoteConfig;
+        }
+        if (Array.isArray(quoteConfig)) {
+            return quoteConfig[index] === true;
+        }
+        if (typeof quoteConfig === "object" && header) {
+            return quoteConfig[header] === true;
+        }
+        return false;
+    };
+    const formatField = (value, index, header, isHeader = false) => {
+        if (value === null || value === undefined) {
+            return "";
+        }
+        const str = typeof value === "number"
+            ? formatNumberForCsv(value, decimalSeparator)
+            : String(value);
+        // If quoting is disabled, return raw string
+        if (!quoteEnabled) {
+            return str;
+        }
+        // Check if quoting is needed
+        const forceQuote = alwaysQuote || shouldQuoteColumn(index, header, isHeader);
+        const needsQuote = forceQuote || needsQuoteRegex.test(str);
+        if (needsQuote) {
+            // Escape quotes using pre-compiled regex
+            const escaped = str.replace(escapeQuoteRegex, escapedQuote);
+            return quote + escaped + quote;
+        }
+        return str;
+    };
+    const formatRow = (row, rowHeaders, isHeader = false) => {
+        return row
+            .map((value, index) => formatField(value, index, rowHeaders?.[index], isHeader))
+            .join(delimiter);
+    };
+    // Determine headers
+    let keys = null;
+    // Helper to apply transform if provided (sync only)
+    const applyTransform = (row) => {
+        if (transform) {
+            // Check if it's a sync transform (1 argument)
+            if (transform.length === 1) {
+                return transform(row);
+            }
+            // For async transform in sync context, just return the row unchanged
+            // Async transforms should use streaming API
+            return row;
+        }
+        return row;
+    };
+    // Handle array of objects
+    if (data.length > 0 && !Array.isArray(data[0])) {
+        const objects = data;
+        keys = headers === true ? Object.keys(objects[0]) : Array.isArray(headers) ? headers : null;
+        if (keys && shouldWriteHeaders) {
+            // Add header row
+            lines.push(formatRow(keys, keys, true));
+        }
+        // Add data rows
+        for (const obj of objects) {
+            const transformedObj = applyTransform(obj);
+            if (transformedObj === null || transformedObj === undefined) {
+                continue; // Skip row if transform returns null
+            }
+            const row = keys ? keys.map(key => transformedObj[key]) : Object.values(transformedObj);
+            lines.push(formatRow(row, keys ?? undefined));
+        }
+    }
+    else if (data.length > 0) {
+        // Handle 2D array with data
+        const arrays = data;
+        // Add custom headers if provided
+        if (Array.isArray(headers)) {
+            keys = headers;
+            if (shouldWriteHeaders) {
+                lines.push(formatRow(headers, headers, true));
+            }
+        }
+        for (const row of arrays) {
+            const transformedRow = applyTransform(row);
+            if (transformedRow === null || transformedRow === undefined) {
+                continue; // Skip row if transform returns null
+            }
+            lines.push(formatRow(transformedRow, keys ?? undefined));
+        }
+    }
+    else if (alwaysWriteHeaders && Array.isArray(headers) && shouldWriteHeaders) {
+        // Handle empty data with alwaysWriteHeaders
+        lines.push(formatRow(headers, headers, true));
+    }
+    let result = lines.join(rowDelimiter);
+    // Add trailing row delimiter
+    if (result.length > 0 && includeEndRowDelimiter) {
+        result += rowDelimiter;
+    }
+    // Add BOM for UTF-8
+    if (writeBOM) {
+        result = "\uFEFF" + result;
+    }
+    return result;
+}
+// =============================================================================
+// Streaming Parser
+// =============================================================================
+/**
+ * Async CSV parser that yields rows one at a time
+ */
+export async function* parseCsvStream(input, options = {}) {
+    const { delimiter = ",", quote: quoteOption = '"', escape: escapeOption = '"', skipEmptyLines = false, ignoreEmpty = false, trim = false, ltrim = false, rtrim = false, headers = false, renameHeaders = false, comment, maxRows, skipLines = 0, skipRows = 0, strictColumnHandling = false, discardUnmappedColumns = false } = options;
+    const shouldSkipEmpty = skipEmptyLines || ignoreEmpty;
+    // Handle quote: null/false to disable quoting
+    const quoteEnabled = quoteOption !== null && quoteOption !== false;
+    const quote = quoteEnabled ? String(quoteOption) : "";
+    const escape = escapeOption !== null && escapeOption !== false ? String(escapeOption) : "";
+    let headerRow = null;
+    let headersLength = 0;
+    let useHeaders = false;
+    let headerRowProcessed = false;
+    let buffer = "";
+    let currentRow = [];
+    let currentField = "";
+    let inQuotes = false;
+    let lineNumber = 0;
+    let dataRowCount = 0;
+    let skippedDataRows = 0;
+    // Determine header mode
+    if (headers === true) {
+        useHeaders = true;
+    }
+    else if (Array.isArray(headers)) {
+        headerRow = headers;
+        headersLength = headers.filter(h => h !== null && h !== undefined).length;
+        validateUniqueHeaders(headers);
+        useHeaders = true;
+        if (!renameHeaders) {
+            headerRowProcessed = true;
+        }
+    }
+    else if (typeof headers === "function") {
+        useHeaders = true;
+    }
+    // Pre-compute trim function to avoid repeated condition checks
+    const trimField = trim || (ltrim && rtrim)
+        ? (s) => s.trim()
+        : ltrim
+            ? (s) => s.trimStart()
+            : rtrim
+                ? (s) => s.trimEnd()
+                : (s) => s;
+    const processRow = (row) => {
+        // Header handling
+        if (useHeaders && !headerRowProcessed) {
+            if (typeof headers === "function") {
+                const transformed = headers(row);
+                validateUniqueHeaders(transformed);
+                headerRow = transformed;
+            }
+            else if (!Array.isArray(headers)) {
+                validateUniqueHeaders(row);
+                headerRow = row;
+            }
+            headersLength = headerRow.filter(h => h !== null && h !== undefined).length;
+            headerRowProcessed = true;
+            if (renameHeaders) {
+                return { valid: false, row: null };
+            }
+            if (headers === true || typeof headers === "function") {
+                return { valid: false, row: null };
+            }
+            return { valid: false, row: null };
+        }
+        // Skip data rows
+        if (skippedDataRows < skipRows) {
+            skippedDataRows++;
+            return { valid: false, row: null };
+        }
+        // Column validation
+        if (headerRow && headerRow.length > 0) {
+            const expectedCols = headersLength;
+            const actualCols = row.length;
+            if (actualCols > expectedCols) {
+                if (strictColumnHandling && !discardUnmappedColumns) {
+                    return { valid: false, row: null };
+                }
+                else {
+                    // Default: trim extra columns
+                    row.length = headerRow.length;
+                }
+            }
+            else if (actualCols < expectedCols) {
+                if (strictColumnHandling) {
+                    return { valid: false, row: null };
+                }
+                while (row.length < headerRow.length) {
+                    row.push("");
+                }
+            }
+        }
+        // Convert to object if using headers
+        if (useHeaders && headerRow) {
+            const obj = {};
+            headerRow.forEach((header, index) => {
+                if (header !== null && header !== undefined) {
+                    obj[header] = row[index] ?? "";
+                }
+            });
+            return { valid: true, row: obj };
+        }
+        return { valid: true, row };
+    };
+    const processBuffer = function* () {
+        let i = 0;
+        const len = buffer.length;
+        while (i < len) {
+            const char = buffer[i];
+            if (inQuotes && quoteEnabled) {
+                if (escape && char === escape && buffer[i + 1] === quote) {
+                    currentField += quote;
+                    i += 2;
+                }
+                else if (char === quote) {
+                    inQuotes = false;
+                    i++;
+                }
+                else if (i === len - 1) {
+                    // Need more data for quoted field
+                    buffer = buffer.slice(i);
+                    return;
+                }
+                else if (char === "\r") {
+                    // Normalize CRLF to LF inside quoted fields
+                    if (buffer[i + 1] === "\n") {
+                        i++; // Skip \r, will add \n on next iteration
+                    }
+                    else {
+                        currentField += "\n"; // Convert standalone \r to \n
+                        i++;
+                    }
+                }
+                else {
+                    currentField += char;
+                    i++;
+                }
+            }
+            else {
+                if (quoteEnabled && char === quote && currentField === "") {
+                    inQuotes = true;
+                    i++;
+                }
+                else if (char === delimiter) {
+                    currentRow.push(trimField(currentField));
+                    currentField = "";
+                    i++;
+                }
+                else if (char === "\n" || char === "\r") {
+                    if (char === "\r" && buffer[i + 1] === "\n") {
+                        i++;
+                    }
+                    currentRow.push(trimField(currentField));
+                    currentField = "";
+                    lineNumber++;
+                    if (lineNumber <= skipLines) {
+                        currentRow = [];
+                        i++;
+                        continue;
+                    }
+                    if (comment && currentRow[0]?.startsWith(comment)) {
+                        currentRow = [];
+                        i++;
+                        continue;
+                    }
+                    const isEmpty = currentRow.length === 1 && currentRow[0] === "";
+                    if (shouldSkipEmpty && isEmpty) {
+                        currentRow = [];
+                        i++;
+                        continue;
+                    }
+                    const result = processRow(currentRow);
+                    if (result.valid && result.row) {
+                        dataRowCount++;
+                        if (maxRows !== undefined && dataRowCount > maxRows) {
+                            return;
+                        }
+                        yield result.row;
+                    }
+                    currentRow = [];
+                    i++;
+                }
+                else {
+                    currentField += char;
+                    i++;
+                }
+            }
+        }
+        buffer = "";
+    };
+    // Handle string input
+    if (typeof input === "string") {
+        buffer = input;
+        yield* processBuffer();
+        // Handle last row
+        if (currentField !== "" || currentRow.length > 0) {
+            currentRow.push(trimField(currentField));
+            if (!(maxRows !== undefined && dataRowCount >= maxRows)) {
+                const result = processRow(currentRow);
+                if (result.valid && result.row) {
+                    yield result.row;
+                }
+            }
+        }
+        return;
+    }
+    // Handle async iterable
+    for await (const chunk of input) {
+        buffer += chunk;
+        yield* processBuffer();
+    }
+    // Handle last row
+    if (currentField !== "" || currentRow.length > 0) {
+        currentRow.push(trimField(currentField));
+        if (!(maxRows !== undefined && dataRowCount >= maxRows)) {
+            const result = processRow(currentRow);
+            if (result.valid && result.row) {
+                yield result.row;
+            }
+        }
+    }
+}