XLSX support with ExcelJS (#248)

* XLSX support with ExcelJS

* Prettier

* Change range option to nested arrays

General code clean up

* Tests and bug fixes

* Respect header row order when resolving conflicts

* Fil/xlsx (#249)

* document xlsx (minimalist, we'll work on the notebook first)

* fix coverage reporter

(avoids a crash on my computer; solution found at tapjs/tapjs#624)

* unknown sheet name

* simplify rows naming

* NN is always called on string (cell specifier such as "AA99")

* test name

* more range specifiers

* Column only range test case

* sheetNames is enumerable

* One more test to check for empty columns

Prettier + use default/base tap reporter

* Add Node 16 to the test matrix

* Revert reporter to classic for Node 16

* Don't fail matrix quickly in actions

* More coverage.

* Example of .xlsx in README

* Remove Excel from Workbook naming

* Fix dates

* Fix for sharedFormula

* Coerce errors to NaN

* Properly escape html

* Make sheetNames read-only

* Require colons in range specifiers

* Include row numbers

* Use only string form ranges

* Coerce range specifiers to strings

* Update README.md

Co-authored-by: Mike Bostock <mbostock@gmail.com>

* Apply suggestions from code review

Co-authored-by: Mike Bostock <mbostock@gmail.com>

* Simplify hyperlinks

* Prettier

* Pass options through

* Rename helper functions for clarity, range tests

* Simpler

* Consistent comment format

* Consistent regexes

* Fix hyperlinks for certain cases

Co-authored-by: Philippe Rivière <fil@rezo.net>
Co-authored-by: Mike Bostock <mbostock@gmail.com>

.github/workflows/nodejs.yml

@@ -4,23 +4,23 @@ on: [push]
 
 jobs:
   build:
-
     strategy:
+      fail-fast: false
       matrix:
         os: [ubuntu-latest]
-        node-version: [12.x, 14.x]
+        node-version: [12.x, 14.x, 16.x]
 
     runs-on: ${{ matrix.os }}
 
     steps:
-    - uses: actions/checkout@v1
-    - name: Use Node.js ${{ matrix.node-version }}
-      uses: actions/setup-node@v1
-      with:
-        node-version: ${{ matrix.node-version }}
-    - name: yarn install and test
-      run: |
-        yarn install --frozen-lockfile
-        yarn test
-      env:
-        CI: true
+      - uses: actions/checkout@v1
+      - name: Use Node.js ${{ matrix.node-version }}
+        uses: actions/setup-node@v1
+        with:
+          node-version: ${{ matrix.node-version }}
+      - name: yarn install and test
+        run: |
+          yarn install --frozen-lockfile
+          yarn test
+        env:
+          CI: true

README.md

@@ -379,6 +379,15 @@ Returns a promise to the file loaded as a [SQLite database client](https://obser
 const db = await FileAttachment("chinook.db").sqlite();
 ```
 
+<a href="#attachment_xlsx" name="attachment_xlsx">#</a> *attachment*.<b>xlsx</b>() [<>](https://github.com/observablehq/stdlib/blob/master/src/xlsx.js "Source")
+
+Returns a promise to the file loaded as a [Workbook](https://observablehq.com/@observablehq/xlsx).
+
+```js
+const workbook = await FileAttachment("profit-and-loss.xlsx").xlsx();
+const sheet = workbook.sheet("Sheet1", {range: "B4:AF234", headers: true});
+```
+
 <a href="#attachment_xml" name="attachment_xml">#</a> *attachment*.<b>xml</b>() [<>](https://github.com/observablehq/stdlib/blob/master/src/fileAttachment.js "Source")
 
 Returns a promise to an [XMLDocument](https://developer.mozilla.org/en-US/docs/Web/API/XMLDocument) containing the contents of the file.

package.json

@@ -13,7 +13,7 @@
     "url": "https://github.com/observablehq/stdlib.git"
   },
   "scripts": {
-    "test": "tap 'test/**/*-test.js'",
+    "test": "tap 'test/**/*-test.js' --reporter classic",
     "prepublishOnly": "rollup -c",
     "postpublish": "git push && git push --tags"
   },

src/dependencies.js

@@ -16,3 +16,4 @@ export const vegaliteApi = dependency("vega-lite-api", "5.0.0", "build/vega-lite
 export const arrow = dependency("apache-arrow", "4.0.1", "Arrow.es2015.min.js");
 export const arquero = dependency("arquero", "4.8.4", "dist/arquero.min.js");
 export const topojson = dependency("topojson-client", "3.1.0", "dist/topojson-client.min.js");
+export const exceljs = dependency("exceljs", "4.3.0", "dist/exceljs.min.js");

src/fileAttachment.js

@@ -1,7 +1,8 @@
 import {autoType, csvParse, csvParseRows, tsvParse, tsvParseRows} from "d3-dsv";
 import {require as requireDefault} from "d3-require";
-import {arrow, jszip} from "./dependencies.js";
+import {arrow, jszip, exceljs} from "./dependencies.js";
 import {SQLiteDatabaseClient} from "./sqlite.js";
+import {Workbook} from "./xlsx.js";
 
 async function remote_fetch(file) {
   const response = await fetch(await file.url());
@@ -70,6 +71,10 @@ class AbstractFile {
   async html() {
     return this.xml("text/html");
   }
+  async xlsx() {
+    const [ExcelJS, buffer] = await Promise.all([requireDefault(exceljs.resolve()), this.arrayBuffer()]);
+    return new Workbook(await new ExcelJS.Workbook().xlsx.load(buffer));
+  }
 }
 
 class FileAttachment extends AbstractFile {

src/xlsx.js

@@ -0,0 +1,104 @@
+export class Workbook {
+  constructor(workbook) {
+    Object.defineProperties(this, {
+      _: {value: workbook},
+      sheetNames: {
+        value: workbook.worksheets.map((s) => s.name),
+        enumerable: true,
+      },
+    });
+  }
+  sheet(name, options) {
+    const sname =
+      typeof name === "number"
+        ? this.sheetNames[name]
+        : this.sheetNames.includes((name += ""))
+        ? name
+        : null;
+    if (sname == null) throw new Error(`Sheet not found: ${name}`);
+    const sheet = this._.getWorksheet(sname);
+    return extract(sheet, options);
+  }
+}
+
+function extract(sheet, {range, headers = false} = {}) {
+  let [[c0, r0], [c1, r1]] = parseRange(range, sheet);
+  const headerRow = headers && sheet._rows[r0++];
+  let names = new Set(["#"]);
+  for (let n = c0; n <= c1; n++) {
+    let name = (headerRow ? valueOf(headerRow._cells[n]) : null) || toColumn(n);
+    while (names.has(name)) name += "_";
+    names.add(name);
+  }
+  names = new Array(c0).concat(Array.from(names));
+
+  const output = new Array(r1 - r0 + 1);
+  for (let r = r0; r <= r1; r++) {
+    const row = (output[r - r0] = Object.defineProperty({}, "#", {
+      value: r + 1,
+    }));
+    const _row = sheet._rows[r];
+    if (_row && _row.hasValues)
+      for (let c = c0; c <= c1; c++) {
+        const value = valueOf(_row._cells[c]);
+        if (value != null) row[names[c + 1]] = value;
+      }
+  }
+
+  output.columns = names.filter(() => true); // Filter sparse columns
+  return output;
+}
+
+function valueOf(cell) {
+  if (!cell) return;
+  const {value} = cell;
+  if (value && value instanceof Date) return value;
+  if (value && typeof value === "object") {
+    if (value.formula || value.sharedFormula)
+      return value.result && value.result.error ? NaN : value.result;
+    if (value.richText) return value.richText.map((d) => d.text).join("");
+    if (value.text) {
+      let {text} = value;
+      if (text.richText) text = text.richText.map((d) => d.text).join("");
+      return value.hyperlink && value.hyperlink !== text
+        ? `${value.hyperlink} ${text}`
+        : text;
+    }
+    return value;
+  }
+  return value;
+}
+
+function parseRange(specifier = ":", {columnCount, rowCount}) {
+  specifier += "";
+  if (!specifier.match(/^[A-Z]*\d*:[A-Z]*\d*$/))
+    throw new Error("Malformed range specifier");
+  const [[c0 = 0, r0 = 0], [c1 = columnCount - 1, r1 = rowCount - 1]] =
+    specifier.split(":").map(fromCellReference);
+  return [
+    [c0, r0],
+    [c1, r1],
+  ];
+}
+
+// Returns the default column name for a zero-based column index.
+// For example: 0 -> "A", 1 -> "B", 25 -> "Z", 26 -> "AA", 27 -> "AB".
+function toColumn(c) {
+  let sc = "";
+  c++;
+  do {
+    sc = String.fromCharCode(64 + (c % 26 || 26)) + sc;
+  } while ((c = Math.floor((c - 1) / 26)));
+  return sc;
+}
+
+// Returns the zero-based indexes from a cell reference.
+// For example: "A1" -> [0, 0], "B2" -> [1, 1], "AA10" -> [26, 9].
+function fromCellReference(s) {
+  const [, sc, sr] = s.match(/^([A-Z]*)(\d*)$/);
+  let c = 0;
+  if (sc)
+    for (let i = 0; i < sc.length; i++)
+      c += Math.pow(26, sc.length - i - 1) * (sc.charCodeAt(i) - 64);
+  return [c ? c - 1 : undefined, sr ? +sr - 1 : undefined];
+}