df-script 1.5.0 → 1.7.0

package/dist/functions/read_json.js

@@ -1,21 +0,0 @@
-"use strict";
-Object.defineProperty(exports, "__esModule", { value: true });
-exports.read_json = read_json;
-const dataframe_1 = require("../dataframe/dataframe");
-const utils_1 = require("../utils");
-const exceptions_1 = require("../exceptions");
-/**
- * Parses JSON content (JSON or NDJSON) and loads it into a new DataFrame.
- *
- * @param content The JSON or NDJSON content string.
- * @param options Parse and configuration options.
- * @returns A new DataFrame instance populated with the parsed records.
- */
-function read_json(content, { format = "json", trimBeforeParse = true, schema, ...parseOpts } = {}) {
-    const parsed = (0, utils_1.safeJsonParse)(content, { format, trimBeforeParse, ...parseOpts });
-    if (parsed === content) {
-        throw new exceptions_1.DataFrameError(`Invalid JSON input: must be a valid, non-empty JSON ${format} string.`);
-    }
-    const parsedData = Array.isArray(parsed) ? parsed : ((0, utils_1.isObj)(parsed) ? [parsed] : []);
-    return new dataframe_1.DataFrame(parsedData, schema);
-}

package/dist/functions/transpose.d.ts

@@ -1,7 +0,0 @@
-import { DataFrame } from "../dataframe/dataframe";
-import type { RowRecord } from "../types";
-import type { TransposeOptions } from "../dataframe/types";
-/**
- * Transposes the rows and columns of a DataFrame.
- */
-export declare function transpose<T extends RowRecord = any>(df: DataFrame<T>, { include_header: includeHeader, header_name: headerName, column_names: colNamesOpt }?: TransposeOptions): DataFrame<any>;

package/dist/functions/transpose.js

@@ -1,84 +0,0 @@
-"use strict";
-Object.defineProperty(exports, "__esModule", { value: true });
-exports.transpose = transpose;
-const dataframe_1 = require("../dataframe/dataframe");
-const datatypes_1 = require("../datatypes");
-const utils_1 = require("../dataframe/utils");
-const exceptions_1 = require("../exceptions");
-/**
- * Transposes the rows and columns of a DataFrame.
- */
-function transpose(df, { include_header: includeHeader = false, header_name: headerName = "column", column_names: colNamesOpt } = {}) {
-    const height = df.height;
-    if (height === 0) {
-        const cols = {};
-        const schema = {};
-        if (includeHeader) {
-            cols[headerName] = (0, utils_1.coerceColumn)([], datatypes_1.DataTypeRegistry.Utf8, 0);
-            schema[headerName] = datatypes_1.DataTypeRegistry.Utf8;
-        }
-        return dataframe_1.DataFrame._createDirect(cols, schema, 0);
-    }
-    let dataColumns = df.columns;
-    if (typeof colNamesOpt === "string") {
-        (0, exceptions_1.assertColumnExists)(colNamesOpt, df._columns, "column_names");
-        dataColumns = dataColumns.filter(c => c !== colNamesOpt);
-    }
-    let newColNames = [];
-    if (typeof colNamesOpt === "string") {
-        const keyCol = df._columns[colNamesOpt];
-        newColNames = new Array(height);
-        for (let i = 0; i < height; i++) {
-            const val = keyCol[i];
-            if (val == null) {
-                throw new exceptions_1.DataFrameError(`Transpose column_names column "${colNamesOpt}" contains null/undefined at index ${i}`);
-            }
-            newColNames[i] = String(val);
-        }
-    }
-    else if (colNamesOpt != null && typeof colNamesOpt !== "string" && Symbol.iterator in Object(colNamesOpt)) {
-        const colNamesArr = Array.from(colNamesOpt);
-        if (colNamesArr.length !== height) {
-            throw new exceptions_1.DataFrameError(`column_names length (${colNamesArr.length}) must match the height of the DataFrame (${height})`);
-        }
-        newColNames = new Array(height);
-        for (let i = 0; i < height; i++) {
-            newColNames[i] = String(colNamesArr[i]);
-        }
-    }
-    else {
-        newColNames = new Array(height);
-        for (let i = 0; i < height; i++) {
-            newColNames[i] = `column_${i}`;
-        }
-    }
-    const uniqueNames = new Set();
-    if (includeHeader) {
-        uniqueNames.add(headerName);
-    }
-    for (let i = 0; i < newColNames.length; i++) {
-        const name = newColNames[i];
-        if (uniqueNames.has(name)) {
-            throw new exceptions_1.DataFrameError(`Duplicate column name in transposed DataFrame: "${name}"`);
-        }
-        uniqueNames.add(name);
-    }
-    const numDataCols = dataColumns.length;
-    const newCols = {};
-    const newSchema = {};
-    if (includeHeader) {
-        newCols[headerName] = (0, utils_1.coerceColumn)(dataColumns, datatypes_1.DataTypeRegistry.Utf8, numDataCols);
-        newSchema[headerName] = datatypes_1.DataTypeRegistry.Utf8;
-    }
-    for (let i = 0; i < height; i++) {
-        const colName = newColNames[i];
-        const rawVals = new Array(numDataCols);
-        for (let j = 0; j < numDataCols; j++) {
-            rawVals[j] = df._columns[dataColumns[j]][i];
-        }
-        const type = (0, utils_1.inferColumnType)(rawVals);
-        newCols[colName] = (0, utils_1.coerceColumn)(rawVals, type, numDataCols);
-        newSchema[colName] = type;
-    }
-    return dataframe_1.DataFrame._createDirect(newCols, newSchema, numDataCols);
-}

package/dist/types.js

@@ -1,2 +0,0 @@
-"use strict";
-Object.defineProperty(exports, "__esModule", { value: true });

package/dist/utils/boolean.d.ts

@@ -1 +0,0 @@
-export declare function tryParseBoolean(v: unknown): boolean | undefined;

package/dist/utils/boolean.js

@@ -1,18 +0,0 @@
-"use strict";
-Object.defineProperty(exports, "__esModule", { value: true });
-exports.tryParseBoolean = tryParseBoolean;
-const boolMap = {
-    true: true, "1": true, yes: true, y: true, on: true,
-    false: false, "0": false, no: false, n: false, off: false
-};
-function tryParseBoolean(v) {
-    if (typeof v === "boolean")
-        return v;
-    if (v === 1)
-        return true;
-    if (v === 0)
-        return false;
-    if (typeof v !== "string")
-        return undefined;
-    return boolMap[v.trim().toLowerCase()];
-}

package/dist/utils/csv.js

@@ -1,313 +0,0 @@
-"use strict";
-Object.defineProperty(exports, "__esModule", { value: true });
-exports.formatCsvValue = formatCsvValue;
-exports.stringifyCSV = stringifyCSV;
-exports.parseCSV = parseCSV;
-exports.inferAndCoerceCSVColumn = inferAndCoerceCSVColumn;
-const date_1 = require("./date");
-const json_1 = require("./json");
-const number_1 = require("./number");
-const constants_1 = require("../dataframe/constants");
-const datatypes_1 = require("../datatypes");
-// Removed stringifyCsvObject
-function formatCsvValueInternal(options = {}) {
-    const nullValue = options.nullValue !== undefined ? options.nullValue : "";
-    const formatNum = (0, number_1.formatNumber)(options.numericFormatOptions);
-    const replacerOptions = options.replacerOptions || {};
-    const formatDate = (v) => {
-        if (options.datetimeFormat)
-            return (0, date_1.strftime)(v, options.datetimeFormat);
-        if (options.dateFormat)
-            return (0, date_1.strftime)(v, options.dateFormat);
-        if (options.timeFormat)
-            return (0, date_1.strftime)(v, options.timeFormat);
-        return v.toISOString();
-    };
-    const replacer = (0, json_1.createSafeJsonReplacer)({
-        formatDate,
-        ...replacerOptions
-    });
-    const hasCustomBigInt = typeof replacerOptions.onBigInt === "function";
-    return (val) => {
-        if (val === null || val === undefined || typeof val === "symbol" || typeof val === "function") {
-            return { str: nullValue, isNumeric: false };
-        }
-        if (typeof val === "bigint" && !hasCustomBigInt && !replacerOptions.voidBigIntReplacement) {
-            return { str: formatNum(val), isNumeric: true };
-        }
-        const processed = replacer.call(null, "", val);
-        if (processed === null || processed === undefined || typeof processed === "symbol" || typeof processed === "function") {
-            return { str: nullValue, isNumeric: false };
-        }
-        if (typeof processed === "string")
-            return { str: processed, isNumeric: false };
-        if (typeof processed === "number" || typeof processed === "bigint") {
-            return { str: formatNum(processed), isNumeric: true };
-        }
-        if (typeof processed === "boolean")
-            return { str: processed ? "true" : "false", isNumeric: false };
-        if (processed instanceof Number)
-            return { str: formatNum(processed.valueOf()), isNumeric: true };
-        if (processed instanceof String)
-            return { str: processed.valueOf(), isNumeric: false };
-        if (processed instanceof Boolean)
-            return { str: processed.valueOf() ? "true" : "false", isNumeric: false };
-        if (typeof processed === "object") {
-            // Instantiate a fresh replacer specifically for nested stringification
-            // to avoid state carry-over (such as visited circular reference sets)
-            return {
-                str: JSON.stringify(processed, (0, json_1.createSafeJsonReplacer)({
-                    formatDate,
-                    ...replacerOptions
-                })),
-                isNumeric: false
-            };
-        }
-        return { str: String(processed), isNumeric: false };
-    };
-}
-function formatCsvValue(options = {}) {
-    const formatted = formatCsvValueInternal(options);
-    return (val) => formatted(val).str;
-}
-function stringifyCSV(columns, height, options = {}) {
-    const { separator = ",", quoteChar = '"', includeHeader = true, lineTerminator = constants_1.NEWLINE, quoteStyle = "necessary", includeBom = false, onRow, ...formatOptions } = options;
-    const keys = Object.keys(columns);
-    const numKeys = keys.length;
-    const lines = [];
-    let isFirstRow = true;
-    const formatValue = formatCsvValueInternal(formatOptions);
-    const escapeAndQuote = (val, isHeader = false) => {
-        let strVal;
-        let isNumeric = false;
-        if (isHeader) {
-            strVal = String(val);
-        }
-        else {
-            const formatted = formatValue(val);
-            strVal = formatted.str;
-            isNumeric = formatted.isNumeric;
-        }
-        if (quoteStyle === "never") {
-            return strVal;
-        }
-        if (quoteStyle === "always") {
-            const escaped = strVal.split(quoteChar).join(quoteChar + quoteChar);
-            return quoteChar + escaped + quoteChar;
-        }
-        if (quoteStyle === "non_numeric") {
-            if (!isNumeric && val != null) {
-                const escaped = strVal.split(quoteChar).join(quoteChar + quoteChar);
-                return quoteChar + escaped + quoteChar;
-            }
-        }
-        // Default: "necessary"
-        const needsQuoting = strVal.includes(separator) ||
-            strVal.includes(quoteChar) ||
-            strVal.includes(constants_1.NEWLINE) ||
-            strVal.includes(constants_1.CARRIAGE_RETURN);
-        if (needsQuoting) {
-            const escaped = strVal.split(quoteChar).join(quoteChar + quoteChar);
-            return quoteChar + escaped + quoteChar;
-        }
-        return strVal;
-    };
-    const outputLine = (line) => {
-        if (isFirstRow) {
-            const initial = includeBom ? constants_1.UTF8_BOM + line : line;
-            if (onRow) {
-                onRow(initial);
-            }
-            else {
-                lines.push(initial);
-            }
-            isFirstRow = false;
-        }
-        else {
-            if (onRow) {
-                onRow(lineTerminator + line);
-            }
-            else {
-                lines.push(line);
-            }
-        }
-    };
-    // Headers
-    if (includeHeader) {
-        const headerRow = new Array(numKeys);
-        for (let i = 0; i < numKeys; i++) {
-            headerRow[i] = escapeAndQuote(keys[i], true);
-        }
-        outputLine(headerRow.join(separator));
-    }
-    // Rows
-    for (let r = 0; r < height; r++) {
-        const row = new Array(numKeys);
-        for (let i = 0; i < numKeys; i++) {
-            row[i] = escapeAndQuote(columns[keys[i]][r], false);
-        }
-        outputLine(row.join(separator));
-    }
-    return onRow ? (includeBom ? constants_1.UTF8_BOM : "") : lines.join(lineTerminator);
-}
-function parseCSV(content, options = {}) {
-    const separator = options.separator || ",";
-    const quoteChar = options.quoteChar || '"';
-    let csvContent = content;
-    if (csvContent.startsWith(constants_1.UTF8_BOM)) {
-        csvContent = csvContent.substring(constants_1.UTF8_BOM.length);
-    }
-    const rows = [];
-    let currentRow = [];
-    let currentCell = "";
-    let inQuotes = false;
-    let hasAnyData = false;
-    let lastCharWasSeparator = false;
-    const len = csvContent.length;
-    for (let i = 0; i < len; i++) {
-        const char = csvContent[i];
-        if (inQuotes) {
-            hasAnyData = true;
-            lastCharWasSeparator = false;
-            if (char === quoteChar) {
-                if (i + 1 < len && csvContent[i + 1] === quoteChar) {
-                    currentCell += quoteChar;
-                    i++; // Skip escaped quote
-                }
-                else {
-                    inQuotes = false;
-                }
-            }
-            else {
-                currentCell += char;
-            }
-        }
-        else {
-            if (char === quoteChar) {
-                hasAnyData = true;
-                lastCharWasSeparator = false;
-                inQuotes = true;
-            }
-            else if (char === separator) {
-                currentRow.push(currentCell);
-                currentCell = "";
-                lastCharWasSeparator = true;
-            }
-            else {
-                let isLineBreak = false;
-                if (char === constants_1.CARRIAGE_RETURN) {
-                    isLineBreak = true;
-                    if (i + 1 < len && csvContent[i + 1] === constants_1.NEWLINE) {
-                        i++;
-                    }
-                }
-                else if (char === constants_1.NEWLINE) {
-                    isLineBreak = true;
-                }
-                if (isLineBreak) {
-                    if (hasAnyData || currentRow.length > 0 || currentCell !== "" || lastCharWasSeparator) {
-                        currentRow.push(currentCell);
-                        rows.push(currentRow);
-                        currentRow = [];
-                        currentCell = "";
-                        hasAnyData = false;
-                        lastCharWasSeparator = false;
-                    }
-                }
-                else {
-                    hasAnyData = true;
-                    lastCharWasSeparator = false;
-                    currentCell += char;
-                }
-            }
-        }
-    }
-    if (hasAnyData || currentRow.length > 0 || currentCell !== "" || lastCharWasSeparator) {
-        currentRow.push(currentCell);
-        rows.push(currentRow);
-    }
-    return rows;
-}
-function inferAndCoerceCSVColumn(values, options = {}) {
-    const nullValues = new Set(options.nullValues ?? ["", "NA", "null", "NaN"]);
-    const len = values.length;
-    let isAllBoolean = true;
-    let isAllNumber = true;
-    let isAllBigInt = true;
-    let isAllDate = true;
-    let hasValidData = false;
-    for (let i = 0; i < len; i++) {
-        const val = values[i].trim();
-        if (nullValues.has(val))
-            continue;
-        hasValidData = true;
-        if (isAllBoolean) {
-            const lower = val.toLowerCase();
-            if (lower !== "true" && lower !== "false" && lower !== "1" && lower !== "0") {
-                isAllBoolean = false;
-            }
-        }
-        if (isAllBigInt) {
-            if ((0, number_1.toValidBigInt)(val, { truncate: false }) === null) {
-                isAllBigInt = false;
-            }
-        }
-        if (isAllNumber) {
-            if ((0, number_1.toValidNumber)(val, { allowNonFiniteNumbers: true }) === null) {
-                isAllNumber = false;
-            }
-        }
-        if (isAllDate) {
-            if ((0, date_1.toValidDate)(val) === null) {
-                isAllDate = false;
-            }
-        }
-        // Fast exit if it's strictly a string column
-        if (!isAllBoolean && !isAllNumber && !isAllBigInt && !isAllDate) {
-            break;
-        }
-    }
-    const out = new Array(len);
-    if (!hasValidData || (!isAllBoolean && !isAllNumber && !isAllBigInt && !isAllDate)) {
-        for (let i = 0; i < len; i++) {
-            const val = values[i];
-            out[i] = nullValues.has(val.trim()) ? null : val;
-        }
-        return { type: datatypes_1.Utf8, values: out };
-    }
-    if (isAllBoolean) {
-        for (let i = 0; i < len; i++) {
-            const val = values[i].trim();
-            if (nullValues.has(val)) {
-                out[i] = null;
-            }
-            else {
-                const lower = val.toLowerCase();
-                out[i] = (lower === "true" || lower === "1");
-            }
-        }
-        return { type: datatypes_1.Boolean, values: out };
-    }
-    if (isAllBigInt) {
-        for (let i = 0; i < len; i++) {
-            const val = values[i].trim();
-            out[i] = nullValues.has(val) ? null : (0, number_1.toValidBigInt)(val, { truncate: false });
-        }
-        return { type: datatypes_1.Int64, values: out };
-    }
-    if (isAllNumber) {
-        for (let i = 0; i < len; i++) {
-            const val = values[i].trim();
-            out[i] = nullValues.has(val) ? null : (0, number_1.toValidNumber)(val, { allowNonFiniteNumbers: true });
-        }
-        return { type: datatypes_1.Float64, values: out };
-    }
-    if (isAllDate) {
-        for (let i = 0; i < len; i++) {
-            const val = values[i].trim();
-            out[i] = nullValues.has(val) ? null : (0, date_1.toValidDate)(val);
-        }
-        return { type: datatypes_1.Datetime, values: out };
-    }
-    return { type: datatypes_1.Utf8, values: out };
-}