vastgrid-js 0.0.0-stage → 0.2.0

package/src/workbook.ts

@@ -0,0 +1,48 @@
+/**
+ * Excel workbooks as tables: a sheet is read with `vastgrid-spec/workbook` into lines of CSV kept as a `Blob`, which
+ * `readFile` of text.ts then reads as it reads a CSV file, within `maxBytes`, with the columns and types that the
+ * DuckDB engine reads the sheet with.
+ */
+import type { Column } from "vastgrid-spec";
+import { blobSource, convertSheet, type ReadOptions, WorkbookError } from "vastgrid-spec/workbook";
+
+export { readOptionsOf } from "vastgrid-spec/workbook";
+
+import { type CheckedTable, type FileTable, TableProblem, type TextFile } from "./table.js";
+import { readFile } from "./text.js";
+
+/** The platform's `Blob`, which keeps the sheet's lines out of the heap where it can (in a page, on disk). */
+declare const Blob: new (parts: (string | TextFile)[]) => TextFile;
+
+/** Reads the sheet of a workbook that `read` chooses, the first visible one by default. */
+export async function readSheet(table: FileTable, read: ReadOptions): Promise<CheckedTable> {
+  const { file } = table;
+  if (
+    file === null ||
+    typeof file !== "object" ||
+    typeof file.size !== "number" ||
+    typeof (file as { slice?: unknown }).slice !== "function"
+  ) {
+    throw new TableProblem("/file", "file is a Blob or a File.");
+  }
+  const parts: TextFile[] = [];
+  let sheet: Awaited<ReturnType<typeof convertSheet>>;
+  try {
+    sheet = await convertSheet(
+      blobSource(file),
+      (text) => {
+        parts.push(new Blob([text]));
+      },
+      read,
+    );
+  } catch (error) {
+    if (error instanceof WorkbookError) throw new TableProblem("/file", error.message);
+    throw error;
+  }
+  const fields: (Column | null)[] = new Array(sheet.fields).fill(null);
+  for (const column of sheet.columns) fields[column.field] = { name: column.name, type: column.type };
+  const lines: FileTable = { file: new Blob(parts), format: "csv", delimiter: ",", header: false };
+  if (table.maxBytes !== undefined) lines.maxBytes = table.maxBytes;
+  const checked = await readFile(lines, undefined, fields);
+  return { ...checked, workbook: { sheets: sheet.sheets, sheet: sheet.sheet, notes: sheet.notes } };
+}