@nshiab/simple-data-analysis-core 2.1.10 → 2.1.11

package/esm/helpers/printTable.d.ts

@@ -3,6 +3,8 @@
  * within headers and cells. Unlike `console.table()`, this function properly
  * handles multi-line content, making it ideal for displaying data with long
  * column names and text values.
+ * Widths count graphemes, keeping combining accents with their letters. Wide
+ * characters and emoji are not measured by their terminal display width.
  *
  * @example
  * ```typescript

package/esm/helpers/printTable.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"printTable.d.ts","sourceRoot":"","sources":["../../src/helpers/printTable.ts"],"names":[],"mappings":"AAGA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4CG;AACH,MAAM,CAAC,OAAO,UAAU,UAAU,CAChC,IAAI,EAAE;IAAE,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAAA;CAAE,EAAE,EAClC,OAAO,CAAC,EAAE;IACR,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,aAAa,CAAC,EAAE,MAAM,CAAC;CACxB,GACA,IAAI,CAgNN"}
+{"version":3,"file":"printTable.d.ts","sourceRoot":"","sources":["../../src/helpers/printTable.ts"],"names":[],"mappings":"AAIA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8CG;AACH,MAAM,CAAC,OAAO,UAAU,UAAU,CAChC,IAAI,EAAE;IAAE,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAAA;CAAE,EAAE,EAClC,OAAO,CAAC,EAAE;IACR,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,aAAa,CAAC,EAAE,MAAM,CAAC;CACxB,GACA,IAAI,CAqNN"}

package/esm/helpers/printTable.js

@@ -1,10 +1,13 @@
 import LogValue from "./LogValue.js";
+import splitGraphemes from "./splitGraphemes.js";
 import wrapString from "./wrapString.js";
 /**
  * Prints a formatted table to the console with support for word wrapping
  * within headers and cells. Unlike `console.table()`, this function properly
  * handles multi-line content, making it ideal for displaying data with long
  * column names and text values.
+ * Widths count graphemes, keeping combining accents with their letters. Wide
+ * characters and emoji are not measured by their terminal display width.
  *
  * @example
  * ```typescript
@@ -73,7 +76,7 @@ export default function printTable(data, options) {
     // First, seed widths from header names.
     const columnWidths = {};
     for (const col of columns) {
-        columnWidths[col] = Math.max(...col.split("\n").map((line) => line.length));
+        columnWidths[col] = Math.max(...col.split("\n").map((line) => splitGraphemes(line).length));
     }
     // Format every cell, track the longest raw line per column, and
     // store the formatted strings for the wrapping step that follows.
@@ -88,8 +91,9 @@ export default function printTable(data, options) {
             const lines = value.split("\n");
             let longestLine = 0;
             for (const line of lines) {
-                if (line.length > longestLine)
-                    longestLine = line.length;
+                const length = splitGraphemes(line).length;
+                if (length > longestLine)
+                    longestLine = length;
             }
             if (longestLine > columnWidths[col]) {
                 columnWidths[col] = longestLine;
@@ -103,7 +107,9 @@ export default function printTable(data, options) {
     }
     // Use the same newline and word wrapping rules for headers and cells.
     const wrapCell = (value, width) => {
-        const lines = value.split("\n").flatMap((line) => line.length <= width ? [line] : wrapString(line, width).split("\n"));
+        const lines = value.split("\n").flatMap((line) => splitGraphemes(line).length <= width
+            ? [line]
+            : wrapString(line, width).split("\n"));
         return lines.length > 0 ? lines : [""];
     };
     const wrappedHeaders = columns.map((col) => wrapCell(col, columnWidths[col]));
@@ -151,7 +157,7 @@ export default function printTable(data, options) {
     const createBottomBorder = () => createBorderLine("└", "┴", "┘");
     // Helper function to pad a string to a specific width
     const pad = (str, width) => {
-        return str + " ".repeat(Math.max(0, width - str.length));
+        return str + " ".repeat(Math.max(0, width - splitGraphemes(str).length));
     };
     // Print top border
     console.log(createTopBorder());

package/esm/helpers/splitGraphemes.d.ts

@@ -0,0 +1,2 @@
+export default function splitGraphemes(text: string): string[];
+//# sourceMappingURL=splitGraphemes.d.ts.map

package/esm/helpers/splitGraphemes.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"splitGraphemes.d.ts","sourceRoot":"","sources":["../../src/helpers/splitGraphemes.ts"],"names":[],"mappings":"AAIA,MAAM,CAAC,OAAO,UAAU,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,CAE7D"}

package/esm/helpers/splitGraphemes.js

@@ -0,0 +1,6 @@
+const segmenter = new Intl.Segmenter(undefined, { granularity: "grapheme" });
+// Keep combining characters attached to their base without normalizing text.
+// Grapheme counts are not terminal display widths for wide characters or emoji.
+export default function splitGraphemes(text) {
+    return Array.from(segmenter.segment(text), ({ segment }) => segment);
+}

package/esm/helpers/wrapString.d.ts

@@ -1,8 +1,10 @@
 /**
  * Wraps a string to a specified maximum width, attempting to break at word
  * boundaries when possible. If a single word exceeds the maximum width, it
- * will be broken at the character boundary. This function is primarily used
- * for preparing text to be displayed in console tables.
+ * will be broken at a grapheme boundary, keeping combining accents with their
+ * letters. This function is primarily used for preparing text to be displayed
+ * in console tables. Widths count graphemes, not terminal columns for wide
+ * characters or emoji.
  *
  * @example
  * ```typescript
@@ -26,10 +28,10 @@
  *
  * @param str - The string to wrap.
  *
- * @param maxWidth - The maximum width of each line.
+ * @param maxWidth - The maximum number of graphemes in each line.
  *
  * @param wordWrap - If true, attempts to break at word boundaries. If false,
- *   breaks at character boundaries. Defaults to `true`.
+ *   breaks at grapheme boundaries. Defaults to `true`.
  *
  * @returns The wrapped string with newline characters inserted at appropriate
  *   positions.

package/esm/helpers/wrapString.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"wrapString.d.ts","sourceRoot":"","sources":["../../src/helpers/wrapString.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAmCG;AACH,MAAM,CAAC,OAAO,UAAU,UAAU,CAChC,GAAG,EAAE,MAAM,EACX,QAAQ,EAAE,MAAM,EAChB,QAAQ,UAAO,GACd,MAAM,CAqDR"}
+{"version":3,"file":"wrapString.d.ts","sourceRoot":"","sources":["../../src/helpers/wrapString.ts"],"names":[],"mappings":"AAEA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAqCG;AACH,MAAM,CAAC,OAAO,UAAU,UAAU,CAChC,GAAG,EAAE,MAAM,EACX,QAAQ,EAAE,MAAM,EAChB,QAAQ,UAAO,GACd,MAAM,CA6DR"}

package/esm/helpers/wrapString.js

@@ -1,8 +1,11 @@
+import splitGraphemes from "./splitGraphemes.js";
 /**
  * Wraps a string to a specified maximum width, attempting to break at word
  * boundaries when possible. If a single word exceeds the maximum width, it
- * will be broken at the character boundary. This function is primarily used
- * for preparing text to be displayed in console tables.
+ * will be broken at a grapheme boundary, keeping combining accents with their
+ * letters. This function is primarily used for preparing text to be displayed
+ * in console tables. Widths count graphemes, not terminal columns for wide
+ * characters or emoji.
  *
  * @example
  * ```typescript
@@ -26,56 +29,64 @@
  *
  * @param str - The string to wrap.
  *
- * @param maxWidth - The maximum width of each line.
+ * @param maxWidth - The maximum number of graphemes in each line.
  *
  * @param wordWrap - If true, attempts to break at word boundaries. If false,
- *   breaks at character boundaries. Defaults to `true`.
+ *   breaks at grapheme boundaries. Defaults to `true`.
  *
  * @returns The wrapped string with newline characters inserted at appropriate
  *   positions.
  */
 export default function wrapString(str, maxWidth, wordWrap = true) {
-    if (str.length <= maxWidth) {
+    const graphemes = splitGraphemes(str);
+    if (graphemes.length <= maxWidth) {
         return str;
     }
     if (!wordWrap) {
-        // Simple character-based wrapping
+        // Simple grapheme-based wrapping
         const lines = [];
-        for (let i = 0; i < str.length; i += maxWidth) {
-            lines.push(str.slice(i, i + maxWidth));
+        for (let i = 0; i < graphemes.length; i += maxWidth) {
+            lines.push(graphemes.slice(i, i + maxWidth).join(""));
         }
         return lines.join("\n");
     }
     // Word-aware wrapping
     const lines = [];
     let currentLine = "";
+    let currentWidth = 0;
     const words = str.split(/(\s+)/); // Split but keep whitespace
     for (const word of words) {
+        const wordGraphemes = splitGraphemes(word);
+        const wordWidth = wordGraphemes.length;
         // If adding this word would exceed maxWidth
-        if (currentLine.length + word.length > maxWidth) {
+        if (currentWidth + wordWidth > maxWidth) {
             // If current line is not empty, save it
             if (currentLine.length > 0) {
                 lines.push(currentLine.trimEnd());
                 currentLine = "";
+                currentWidth = 0;
             }
             // If the word itself is longer than maxWidth, break it
-            if (word.length > maxWidth) {
-                for (let i = 0; i < word.length; i += maxWidth) {
-                    const chunk = word.slice(i, i + maxWidth);
-                    if (i + maxWidth < word.length) {
+            if (wordWidth > maxWidth) {
+                for (let i = 0; i < wordWidth; i += maxWidth) {
+                    const chunk = wordGraphemes.slice(i, i + maxWidth).join("");
+                    if (i + maxWidth < wordWidth) {
                         lines.push(chunk);
                     }
                     else {
                         currentLine = chunk;
+                        currentWidth = wordWidth - i;
                     }
                 }
             }
             else {
                 currentLine = word;
+                currentWidth = wordWidth;
             }
         }
         else {
             currentLine += word;
+            currentWidth += wordWidth;
         }
     }
     // Add any remaining text

package/llm.md

@@ -1,7 +1,7 @@
 # The Simple Data Analysis (Core) Library
 
 - Package: `@nshiab/simple-data-analysis-core`
-- Version: `2.1.10`
+- Version: `2.1.11`
 
 To install the library with Deno, use:
 

package/llms.txt

@@ -3,7 +3,7 @@
 > Lightweight DuckDB-powered TypeScript library for tabular, SQL, CSV, Parquet, and geospatial data analysis on Deno, Node.js, and Bun.
 
 - Package: `@nshiab/simple-data-analysis-core`
-- Version: `2.1.10`
+- Version: `2.1.11`
 
 Install with Deno: `deno add jsr:@nshiab/simple-data-analysis-core`
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@nshiab/simple-data-analysis-core",
-  "version": "2.1.10",
+  "version": "2.1.11",
   "description": "Lightweight DuckDB-powered TypeScript library for tabular, SQL, CSV, Parquet, and geospatial data analysis on Deno, Node.js, and Bun.",
   "keywords": [
     "data-analysis",

package/script/helpers/printTable.d.ts

@@ -3,6 +3,8 @@
  * within headers and cells. Unlike `console.table()`, this function properly
  * handles multi-line content, making it ideal for displaying data with long
  * column names and text values.
+ * Widths count graphemes, keeping combining accents with their letters. Wide
+ * characters and emoji are not measured by their terminal display width.
  *
  * @example
  * ```typescript

package/script/helpers/printTable.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"printTable.d.ts","sourceRoot":"","sources":["../../src/helpers/printTable.ts"],"names":[],"mappings":"AAGA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA4CG;AACH,MAAM,CAAC,OAAO,UAAU,UAAU,CAChC,IAAI,EAAE;IAAE,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAAA;CAAE,EAAE,EAClC,OAAO,CAAC,EAAE;IACR,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,aAAa,CAAC,EAAE,MAAM,CAAC;CACxB,GACA,IAAI,CAgNN"}
+{"version":3,"file":"printTable.d.ts","sourceRoot":"","sources":["../../src/helpers/printTable.ts"],"names":[],"mappings":"AAIA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GA8CG;AACH,MAAM,CAAC,OAAO,UAAU,UAAU,CAChC,IAAI,EAAE;IAAE,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAAA;CAAE,EAAE,EAClC,OAAO,CAAC,EAAE;IACR,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,cAAc,CAAC,EAAE,MAAM,CAAC;IACxB,aAAa,CAAC,EAAE,MAAM,CAAC;CACxB,GACA,IAAI,CAqNN"}

package/script/helpers/printTable.js

@@ -5,12 +5,15 @@ var __importDefault = (this && this.__importDefault) || function (mod) {
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.default = printTable;
 const LogValue_js_1 = __importDefault(require("./LogValue.js"));
+const splitGraphemes_js_1 = __importDefault(require("./splitGraphemes.js"));
 const wrapString_js_1 = __importDefault(require("./wrapString.js"));
 /**
  * Prints a formatted table to the console with support for word wrapping
  * within headers and cells. Unlike `console.table()`, this function properly
  * handles multi-line content, making it ideal for displaying data with long
  * column names and text values.
+ * Widths count graphemes, keeping combining accents with their letters. Wide
+ * characters and emoji are not measured by their terminal display width.
  *
  * @example
  * ```typescript
@@ -79,7 +82,7 @@ function printTable(data, options) {
     // First, seed widths from header names.
     const columnWidths = {};
     for (const col of columns) {
-        columnWidths[col] = Math.max(...col.split("\n").map((line) => line.length));
+        columnWidths[col] = Math.max(...col.split("\n").map((line) => (0, splitGraphemes_js_1.default)(line).length));
     }
     // Format every cell, track the longest raw line per column, and
     // store the formatted strings for the wrapping step that follows.
@@ -94,8 +97,9 @@ function printTable(data, options) {
             const lines = value.split("\n");
             let longestLine = 0;
             for (const line of lines) {
-                if (line.length > longestLine)
-                    longestLine = line.length;
+                const length = (0, splitGraphemes_js_1.default)(line).length;
+                if (length > longestLine)
+                    longestLine = length;
             }
             if (longestLine > columnWidths[col]) {
                 columnWidths[col] = longestLine;
@@ -109,7 +113,9 @@ function printTable(data, options) {
     }
     // Use the same newline and word wrapping rules for headers and cells.
     const wrapCell = (value, width) => {
-        const lines = value.split("\n").flatMap((line) => line.length <= width ? [line] : (0, wrapString_js_1.default)(line, width).split("\n"));
+        const lines = value.split("\n").flatMap((line) => (0, splitGraphemes_js_1.default)(line).length <= width
+            ? [line]
+            : (0, wrapString_js_1.default)(line, width).split("\n"));
         return lines.length > 0 ? lines : [""];
     };
     const wrappedHeaders = columns.map((col) => wrapCell(col, columnWidths[col]));
@@ -157,7 +163,7 @@ function printTable(data, options) {
     const createBottomBorder = () => createBorderLine("└", "┴", "┘");
     // Helper function to pad a string to a specific width
     const pad = (str, width) => {
-        return str + " ".repeat(Math.max(0, width - str.length));
+        return str + " ".repeat(Math.max(0, width - (0, splitGraphemes_js_1.default)(str).length));
     };
     // Print top border
     console.log(createTopBorder());

package/script/helpers/splitGraphemes.d.ts

@@ -0,0 +1,2 @@
+export default function splitGraphemes(text: string): string[];
+//# sourceMappingURL=splitGraphemes.d.ts.map

package/script/helpers/splitGraphemes.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"splitGraphemes.d.ts","sourceRoot":"","sources":["../../src/helpers/splitGraphemes.ts"],"names":[],"mappings":"AAIA,MAAM,CAAC,OAAO,UAAU,cAAc,CAAC,IAAI,EAAE,MAAM,GAAG,MAAM,EAAE,CAE7D"}

package/script/helpers/splitGraphemes.js

@@ -0,0 +1,9 @@
+"use strict";
+Object.defineProperty(exports, "__esModule", { value: true });
+exports.default = splitGraphemes;
+const segmenter = new Intl.Segmenter(undefined, { granularity: "grapheme" });
+// Keep combining characters attached to their base without normalizing text.
+// Grapheme counts are not terminal display widths for wide characters or emoji.
+function splitGraphemes(text) {
+    return Array.from(segmenter.segment(text), ({ segment }) => segment);
+}

package/script/helpers/wrapString.d.ts

@@ -1,8 +1,10 @@
 /**
  * Wraps a string to a specified maximum width, attempting to break at word
  * boundaries when possible. If a single word exceeds the maximum width, it
- * will be broken at the character boundary. This function is primarily used
- * for preparing text to be displayed in console tables.
+ * will be broken at a grapheme boundary, keeping combining accents with their
+ * letters. This function is primarily used for preparing text to be displayed
+ * in console tables. Widths count graphemes, not terminal columns for wide
+ * characters or emoji.
  *
  * @example
  * ```typescript
@@ -26,10 +28,10 @@
  *
  * @param str - The string to wrap.
  *
- * @param maxWidth - The maximum width of each line.
+ * @param maxWidth - The maximum number of graphemes in each line.
  *
  * @param wordWrap - If true, attempts to break at word boundaries. If false,
- *   breaks at character boundaries. Defaults to `true`.
+ *   breaks at grapheme boundaries. Defaults to `true`.
  *
  * @returns The wrapped string with newline characters inserted at appropriate
  *   positions.

package/script/helpers/wrapString.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"wrapString.d.ts","sourceRoot":"","sources":["../../src/helpers/wrapString.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAmCG;AACH,MAAM,CAAC,OAAO,UAAU,UAAU,CAChC,GAAG,EAAE,MAAM,EACX,QAAQ,EAAE,MAAM,EAChB,QAAQ,UAAO,GACd,MAAM,CAqDR"}
+{"version":3,"file":"wrapString.d.ts","sourceRoot":"","sources":["../../src/helpers/wrapString.ts"],"names":[],"mappings":"AAEA;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;GAqCG;AACH,MAAM,CAAC,OAAO,UAAU,UAAU,CAChC,GAAG,EAAE,MAAM,EACX,QAAQ,EAAE,MAAM,EAChB,QAAQ,UAAO,GACd,MAAM,CA6DR"}

package/script/helpers/wrapString.js

@@ -1,11 +1,17 @@
 "use strict";
+var __importDefault = (this && this.__importDefault) || function (mod) {
+    return (mod && mod.__esModule) ? mod : { "default": mod };
+};
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.default = wrapString;
+const splitGraphemes_js_1 = __importDefault(require("./splitGraphemes.js"));
 /**
  * Wraps a string to a specified maximum width, attempting to break at word
  * boundaries when possible. If a single word exceeds the maximum width, it
- * will be broken at the character boundary. This function is primarily used
- * for preparing text to be displayed in console tables.
+ * will be broken at a grapheme boundary, keeping combining accents with their
+ * letters. This function is primarily used for preparing text to be displayed
+ * in console tables. Widths count graphemes, not terminal columns for wide
+ * characters or emoji.
  *
  * @example
  * ```typescript
@@ -29,56 +35,64 @@ exports.default = wrapString;
  *
  * @param str - The string to wrap.
  *
- * @param maxWidth - The maximum width of each line.
+ * @param maxWidth - The maximum number of graphemes in each line.
  *
  * @param wordWrap - If true, attempts to break at word boundaries. If false,
- *   breaks at character boundaries. Defaults to `true`.
+ *   breaks at grapheme boundaries. Defaults to `true`.
  *
  * @returns The wrapped string with newline characters inserted at appropriate
  *   positions.
  */
 function wrapString(str, maxWidth, wordWrap = true) {
-    if (str.length <= maxWidth) {
+    const graphemes = (0, splitGraphemes_js_1.default)(str);
+    if (graphemes.length <= maxWidth) {
         return str;
     }
     if (!wordWrap) {
-        // Simple character-based wrapping
+        // Simple grapheme-based wrapping
         const lines = [];
-        for (let i = 0; i < str.length; i += maxWidth) {
-            lines.push(str.slice(i, i + maxWidth));
+        for (let i = 0; i < graphemes.length; i += maxWidth) {
+            lines.push(graphemes.slice(i, i + maxWidth).join(""));
         }
         return lines.join("\n");
     }
     // Word-aware wrapping
     const lines = [];
     let currentLine = "";
+    let currentWidth = 0;
     const words = str.split(/(\s+)/); // Split but keep whitespace
     for (const word of words) {
+        const wordGraphemes = (0, splitGraphemes_js_1.default)(word);
+        const wordWidth = wordGraphemes.length;
         // If adding this word would exceed maxWidth
-        if (currentLine.length + word.length > maxWidth) {
+        if (currentWidth + wordWidth > maxWidth) {
             // If current line is not empty, save it
             if (currentLine.length > 0) {
                 lines.push(currentLine.trimEnd());
                 currentLine = "";
+                currentWidth = 0;
             }
             // If the word itself is longer than maxWidth, break it
-            if (word.length > maxWidth) {
-                for (let i = 0; i < word.length; i += maxWidth) {
-                    const chunk = word.slice(i, i + maxWidth);
-                    if (i + maxWidth < word.length) {
+            if (wordWidth > maxWidth) {
+                for (let i = 0; i < wordWidth; i += maxWidth) {
+                    const chunk = wordGraphemes.slice(i, i + maxWidth).join("");
+                    if (i + maxWidth < wordWidth) {
                         lines.push(chunk);
                     }
                     else {
                         currentLine = chunk;
+                        currentWidth = wordWidth - i;
                     }
                 }
             }
             else {
                 currentLine = word;
+                currentWidth = wordWidth;
             }
         }
         else {
             currentLine += word;
+            currentWidth += wordWidth;
         }
     }
     // Add any remaining text