From 006e6d0e442df9c4932ca83b4c66a7719adf0fbe Mon Sep 17 00:00:00 2001
From: Minor Gordon <minorg@server.fake>
Date: Wed, 31 Jan 2024 16:48:14 -0500
Subject: [PATCH] handle 403 on lookup file

---
 __tests__/SchemaDotOrgDataSet.test.ts | 12 ++++++++++
 package.json                          |  2 +-
 src/SchemaDotOrgDataSet.ts            | 34 ++++++++++++++++-----------
 3 files changed, 33 insertions(+), 15 deletions(-)

diff --git a/__tests__/SchemaDotOrgDataSet.test.ts b/__tests__/SchemaDotOrgDataSet.test.ts
index be3f47c..52afa39 100644
--- a/__tests__/SchemaDotOrgDataSet.test.ts
+++ b/__tests__/SchemaDotOrgDataSet.test.ts
@@ -137,4 +137,16 @@ describe("SchemaDotOrgDataSet", () => {
       await administrativeAreaClassSubset.samplePagesByIri();
     expect(Object.keys(samplePagesByIri)).toHaveLength(1);
   });
+
+  it("handles 4xx on a class lookup file", async () => {
+    if (process.env.CI) {
+      return;
+    }
+
+    const paintingClassSubset = classSubsets.find(
+      (classSubset) => classSubset.className === "Painting"
+    );
+    expect(paintingClassSubset).not.toBeUndefined();
+    expect(await paintingClassSubset?.payLevelDomainSubsets()).toHaveLength(0);
+  });
 });
diff --git a/package.json b/package.json
index 574c0e3..8ba55e8 100644
--- a/package.json
+++ b/package.json
@@ -63,5 +63,5 @@
   },
   "type": "module",
   "types": "dist/index.d.ts",
-  "version": "3.0.4"
+  "version": "3.0.5"
 }
diff --git a/src/SchemaDotOrgDataSet.ts b/src/SchemaDotOrgDataSet.ts
index 453ca48..d3ad7b5 100644
--- a/src/SchemaDotOrgDataSet.ts
+++ b/src/SchemaDotOrgDataSet.ts
@@ -267,12 +267,6 @@ namespace SchemaDotOrgDataSet {
       this.showProgress = showProgress;
     }
 
-    private async lookupCsvString(): Promise<string> {
-      return (
-        await streamToBuffer(await this.httpClient.get(this.lookupFileUrl))
-      ).toString("utf8");
-    }
-
     async *dataset() {
       for (let fileI = 0; fileI < this.numberOfFiles; fileI++) {
         for await (const quad of parseNQuadsStream(
@@ -308,14 +302,26 @@ namespace SchemaDotOrgDataSet {
           return {};
       }
 
-      const pldDataFileNames = Papa.parse(await this.lookupCsvString(), {
-        header: true,
-      }).data.reduce((map: Record<string, string>, row: any) => {
-        if (row["pld"].length > 0) {
-          map[row["pld"]] = row["file_lookup"];
-        }
-        return map;
-      }, {});
+      let pldDataFileNames: Record<string, string>;
+      try {
+        pldDataFileNames = Papa.parse(
+          (
+            await streamToBuffer(await this.httpClient.get(this.lookupFileUrl))
+          ).toString("utf8"),
+          {
+            header: true,
+          }
+        ).data.reduce((map: Record<string, string>, row: any) => {
+          if (row["pld"].length > 0) {
+            map[row["pld"]] = row["file_lookup"];
+          }
+          return map;
+        }, {});
+      } catch (e) {
+        // The 2022-12 Painting lookup returns 403
+        logger.error("error getting and parsing %s: %s", this.lookupFileUrl, e);
+        return {};
+      }
 
       return (
         Papa.parse(