merge: ui/csv-export

src/lib/csv.test.ts

@@ -0,0 +1,108 @@
+import { describe, expect, it } from "vitest";
+import { defaultCsvFilename, toCsv } from "./csv";
+
+describe("toCsv", () => {
+  it("emits a header row with column labels", () => {
+    const out = toCsv<{ id: string }>([], [
+      { header: "Claim ID", value: (r) => r.id },
+    ]);
+    expect(out).toBe("Claim ID\r\n");
+  });
+
+  it("serializes primitive cell values verbatim", () => {
+    const out = toCsv(
+      [{ a: "hello", b: 42, c: 3.14 }],
+      [
+        { header: "A", value: (r) => r.a },
+        { header: "B", value: (r) => r.b },
+        { header: "C", value: (r) => r.c },
+      ],
+    );
+    // 42 and 3.14 both serialize without quotes, no locale formatting.
+    expect(out).toBe("A,B,C\r\nhello,42,3.14\r\n");
+  });
+
+  it("quotes fields containing commas", () => {
+    const out = toCsv([{ name: "Smith, John" }], [
+      { header: "Name", value: (r) => r.name },
+    ]);
+    expect(out).toBe('Name\r\n"Smith, John"\r\n');
+  });
+
+  it("escapes embedded quotes by doubling them", () => {
+    const out = toCsv([{ note: 'she said "hi"' }], [
+      { header: "Note", value: (r) => r.note },
+    ]);
+    expect(out).toBe('Note\r\n"she said ""hi"""\r\n');
+  });
+
+  it("quotes fields containing newlines", () => {
+    const out = toCsv([{ note: "line1\nline2" }], [
+      { header: "Note", value: (r) => r.note },
+    ]);
+    expect(out).toBe('Note\r\n"line1\nline2"\r\n');
+  });
+
+  it("quotes fields containing carriage returns", () => {
+    const out = toCsv([{ note: "line1\rline2" }], [
+      { header: "Note", value: (r) => r.note },
+    ]);
+    expect(out).toBe('Note\r\n"line1\rline2"\r\n');
+  });
+
+  it("renders null and undefined as empty cells", () => {
+    const out = toCsv(
+      [{ a: null, b: undefined, c: "kept" }],
+      [
+        { header: "A", value: (r) => r.a },
+        { header: "B", value: (r) => r.b },
+        { header: "C", value: (r) => r.c },
+      ],
+    );
+    expect(out).toBe("A,B,C\r\n,,kept\r\n");
+  });
+
+  it("renders an empty-string cell as a bare empty field", () => {
+    const out = toCsv([{ a: "" }], [{ header: "A", value: (r) => r.a }]);
+    expect(out).toBe("A\r\n\r\n");
+  });
+
+  it("serializes multiple rows separated by CRLF", () => {
+    const out = toCsv(
+      [{ id: 1 }, { id: 2 }, { id: 3 }],
+      [{ header: "id", value: (r) => r.id }],
+    );
+    expect(out).toBe("id\r\n1\r\n2\r\n3\r\n");
+  });
+
+  it("quotes a header that itself contains a comma", () => {
+    const out = toCsv([], [{ header: "Last, First", value: () => "" }]);
+    expect(out).toBe('"Last, First"\r\n');
+  });
+
+  it("forces a string around non-string inputs (booleans, dates)", () => {
+    const out = toCsv(
+      [{ on: true as const, at: new Date("2024-01-02T03:04:05.000Z") }],
+      [
+        { header: "On", value: (r) => String(r.on) },
+        { header: "At", value: (r) => r.at.toISOString() },
+      ],
+    );
+    expect(out).toBe("On,At\r\ntrue,2024-01-02T03:04:05.000Z\r\n");
+  });
+
+  it("returns only the header when rows is empty (no trailing body CRLF doubled)", () => {
+    const out = toCsv<{ id: string }>([], [
+      { header: "ID", value: (r) => r.id },
+      { header: "Name", value: () => "x" },
+    ]);
+    expect(out).toBe("ID,Name\r\n");
+  });
+});
+
+describe("defaultCsvFilename", () => {
+  it("formats a prefix + ISO date suffix as .csv", () => {
+    const fixed = new Date("2024-03-05T22:00:00.000Z");
+    expect(defaultCsvFilename("claims", fixed)).toBe("claims-2024-03-05.csv");
+  });
+});

src/lib/csv.ts

@@ -0,0 +1,98 @@
+/**
+ * Minimal RFC 4180 CSV serialization + browser-side download helper.
+ *
+ * The serializer handles the four awkward cases that show up in real data:
+ *   - commas inside a field (quote it)
+ *   - quotes inside a field (double them, then quote it)
+ *   - newlines inside a field (quote it)
+ *   - null / undefined / empty values (render as empty cell)
+ *
+ * Anything that isn't a primitive is coerced via String(); numeric inputs are
+ * emitted verbatim (no locale formatting — the caller is responsible for
+ * `toFixed(2)` and friends).
+ *
+ * `downloadCsv` is intentionally synchronous: the browser serializes the blob
+ * in the same tick and we revoke the object URL on the next microtask so the
+ * download dialog still gets a live URL. No setTimeout needed.
+ */
+
+export type CsvColumn<T> = {
+  /** Display label written to the header row. */
+  header: string;
+  /** Cell extractor. Return primitives only. */
+  value: (row: T) => string | number | null | undefined;
+};
+
+const NEEDS_QUOTING = /[",\r\n]/;
+
+function escapeField(raw: string | number | null | undefined): string {
+  if (raw === null || raw === undefined) return "";
+  const s = typeof raw === "string" ? raw : String(raw);
+  if (NEEDS_QUOTING.test(s)) {
+    return `"${s.replace(/"/g, '""')}"`;
+  }
+  return s;
+}
+
+/**
+ * Serialize an array of rows to an RFC 4180 CSV string.
+ *
+ * Lines are joined with `\r\n` per the spec. The output ends with a trailing
+ * CRLF so the last row is terminated identically to the rest — Excel and
+ * Numbers both prefer this for auto-detecting row count.
+ */
+export function toCsv<T>(
+  rows: T[],
+  columns: CsvColumn<T>[],
+): string {
+  const header = columns.map((c) => escapeField(c.header)).join(",");
+  if (rows.length === 0) {
+    // No body rows: emit header + CRLF only. The two CRLFs that would
+    // appear from "empty body + terminator" are dropped — a trailing blank
+    // line confuses Excel's row-count heuristics.
+    return `${header}\r\n`;
+  }
+  const body = rows
+    .map((row) => columns.map((c) => escapeField(c.value(row))).join(","))
+    .join("\r\n");
+  // RFC 4180: every record (including the header) is CRLF-terminated.
+  // A row that serializes to "" still produces a CRLF in body plus the
+  // terminator CRLF, yielding two CRLFs after the header.
+  return `${header}\r\n${body}\r\n`;
+}
+
+/**
+ * Default filename for a CSV download: `{prefix}-{YYYY-MM-DD}.csv` using the
+ * local date. We slice an ISO string instead of constructing the date with
+ * `getFullYear` etc. because the ISO format matches the rest of the codebase.
+ */
+export function defaultCsvFilename(prefix: string, now: Date = new Date()): string {
+  const isoDay = now.toISOString().slice(0, 10);
+  return `${prefix}-${isoDay}.csv`;
+}
+
+/**
+ * Trigger a browser download of `csvText` as a file. Creates an in-memory
+ * Blob, attaches it to a hidden `<a download>`, clicks the link, then revokes
+ * the object URL on the next tick so the download still has a live handle.
+ *
+ * Safe to call from SSR/Node — guards against a missing `document`.
+ */
+export function downloadCsv(filename: string, csvText: string): void {
+  if (typeof document === "undefined" || typeof URL === "undefined") return;
+  // Prepend BOM so Excel auto-detects UTF-8 on Windows. Tools that already
+  // understand UTF-8 (Numbers, Google Sheets, every CSV parser in this repo)
+  // ignore the leading bytes silently.
+  const blob = new Blob(["\ufeff", csvText], { type: "text/csv;charset=utf-8" });
+  const url = URL.createObjectURL(blob);
+  const a = document.createElement("a");
+  a.href = url;
+  a.download = filename;
+  // Some browsers require the link to be in the DOM before .click() works.
+  a.style.display = "none";
+  document.body.appendChild(a);
+  a.click();
+  document.body.removeChild(a);
+  // Defer revoke so the browser has the URL handle when the download kicks off.
+  queueMicrotask(() => URL.revokeObjectURL(url));
+}