fix: full publisher name in the timeseries download when there is no short one

scripts/build-timeseries-files.ts

@@ -26,7 +26,11 @@ import {
   getTechnologyDefinition,
   UnknownTaxonomyError,
 } from "../src/utils/timeseriesTaxonomy.ts";
-import { jsonToCsv, sectorSegmentColumns } from "./timeseries-csv.ts";
+import {
+  jsonToCsv,
+  publisherLabel,
+  sectorSegmentColumns,
+} from "./timeseries-csv.ts";
 
 const STRICT = process.env.TS_INDEX_STRICT === "1";
 const DEBUG = process.env.TS_INDEX_DEBUG === "1";
@@ -363,7 +367,7 @@ async function main() {
           .join(", ");
         const optionalBlock = optionalParts ? ` (${optionalParts})` : "";
         // Final string
-        const source = `${pub.publisher.short}, ${pub.year}, ${pub.title.full}${optionalBlock}`;
+        const source = `${publisherLabel(pub.publisher)}, ${pub.year}, ${pub.title.full}${optionalBlock}`;
 
         // Look up definitions from centralized taxonomy.
         // These WILL throw if unknown, which is what we want in strict mode.
@@ -378,7 +382,7 @@ async function main() {
           : undefined;
 
         const outRaw = {
-          publisher: parsed.publication.publisher.short,
+          publisher: publisherLabel(pub.publisher),
           publicationName: parsed.publication.title.full,
           publicationYear: parsed.publication.year,
           pathwayName: parsed.pathwayName,

scripts/timeseries-csv.test.ts

@@ -1,5 +1,10 @@
 import { describe, it, expect } from "vitest";
-import { csvCell, jsonToCsv, sectorSegmentColumns } from "./timeseries-csv.ts";
+import {
+  csvCell,
+  jsonToCsv,
+  publisherLabel,
+  sectorSegmentColumns,
+} from "./timeseries-csv.ts";
 
 describe("sectorSegmentColumns", () => {
   it("lists the row's segments and their definitions in the same order", () => {
@@ -47,3 +52,19 @@ describe("CSV writing", () => {
     expect(row).toBe('"Southeast Asia","A; B",1');
   });
 });
+
+describe("publisherLabel", () => {
+  it("uses the short name when there is one", () => {
+    expect(
+      publisherLabel({ short: "IEA", full: "International Energy Agency" }),
+    ).toBe("IEA");
+  });
+
+  it("falls back to the full name when there is no short one", () => {
+    // TransitionZero's files carry a full name only.
+    expect(publisherLabel({ full: "TransitionZero" })).toBe("TransitionZero");
+    expect(publisherLabel({ short: " ", full: "TransitionZero" })).toBe(
+      "TransitionZero",
+    );
+  });
+});

scripts/timeseries-csv.ts

@@ -40,6 +40,20 @@ export function sectorSegmentColumns(
   };
 }
 
+/**
+ * The publisher as the download names it: the short name, or the full name
+ * where the publication has none. Some publishers carry only a full name
+ * (TransitionZero), and reading `short` alone wrote an empty `publisher` and a
+ * `source` starting with "undefined". The prep repo's review workbook uses
+ * the same rule, so the two match.
+ */
+export function publisherLabel(publisher: {
+  short?: string | null;
+  full: string;
+}): string {
+  return publisher.short?.trim() || publisher.full;
+}
+
 /**
  * One CSV cell. Strings are always quoted; a list is joined first, so it can
  * never spill its commas into the next columns.