fix(instagram): explain unavailable posts

Author: rosethornbush

apps/api/src/index.ts

@@ -124,11 +124,22 @@ const app = new Hono<{ Bindings: CloudflareBindings }>()
             EMBED_USER_AGENT: c.env.EMBED_USER_AGENT,
           });
         } catch (cause) {
-          const problem = createProblem(EmbedlyErrors.PlatformFetchFailed, {
+          const errorContext = getErrorContext(cause);
+          let event = EmbedlyErrors.PlatformFetchFailed;
+          if (platform === "Instagram") {
+            if (errorContext.upstream_reason === "instagram.age_restricted") {
+              event = EmbedlyErrors.InstagramAgeRestricted;
+            } else if (errorContext.upstream_reason === "instagram.post_unavailable") {
+              event = EmbedlyErrors.InstagramPostUnavailable;
+            } else if (errorContext.upstream_reason === "instagram.media_unavailable") {
+              event = EmbedlyErrors.InstagramMediaUnavailable;
+            }
+          }
+          const problem = createProblem(event, {
             request_id: requestId,
-            context: { ...logContext, ...getErrorContext(cause) },
+            context: { ...logContext, ...errorContext },
           });
-          Object.assign(logContext, getErrorContext(cause), {
+          Object.assign(logContext, errorContext, {
             outcome: "error",
             status_code: problem.status,
             error_type: problem.type,

packages/logging/src/main.ts

@@ -8,6 +8,7 @@ export interface ErrorContext {
   error_message?: string;
   upstream_status?: LogValue;
   upstream_message?: LogValue;
+  upstream_reason?: LogValue;
 }
 
 export interface EmbedlyEvent {
@@ -62,6 +63,25 @@ export const EmbedlyErrors = {
     detail: "Embedly could not fetch that post from the platform.",
     status: 502,
   }),
+  InstagramAgeRestricted: defineError({
+    type: "instagram.age_restricted",
+    title: "Age-restricted content.",
+    detail: "Instagram requires login to view this post. Embedly can only access public posts.",
+    status: 403,
+  }),
+  InstagramPostUnavailable: defineError({
+    type: "instagram.post_unavailable",
+    title: "Post unavailable.",
+    detail: "Instagram isn't making this post available without login.",
+    status: 404,
+  }),
+  InstagramMediaUnavailable: defineError({
+    type: "platform.fetch_failed",
+    title: "Couldn't load this post.",
+    detail:
+      "Instagram didn't provide public media for this post. It may require login or be unavailable.",
+    status: 502,
+  }),
   PlatformTransformFailed: defineError({
     type: "platform.transform_failed",
     title: "Failed to build post data.",
@@ -250,6 +270,7 @@ export function getErrorContext(error: unknown): ErrorContext {
     if ("code" in error) context.upstream_status = toLogValue(error.code);
     if ("status" in error) context.upstream_status = toLogValue(error.status);
     if ("message" in error) context.upstream_message = toLogValue(error.message);
+    if ("reason" in error) context.upstream_reason = toLogValue(error.reason);
     return context;
   }
 

packages/platforms/src/platforms/instagram.ts

@@ -101,24 +101,36 @@ export const Instagram: Platform<"Instagram", InstagramMedia, {}> = {
         (text) => text.includes("RelayPrefetchedStreamCache") && text.includes(PRELOADER_PREFIX),
       );
 
-    if (!script) {
-      throw {
-        code: 500,
-        message: "Instagram page structure changed: missing data script",
-      };
-    }
-
     let media: InstagramMedia | undefined | null;
-    try {
-      media = parseRelayMedia(script);
-    } catch {
-      throw { code: 500, message: "Failed to parse Instagram data" };
+    if (script) {
+      try {
+        media = parseRelayMedia(script);
+      } catch {
+        throw { code: 500, message: "Failed to parse Instagram data" };
+      }
     }
 
     if (!media) {
+      $("script, style").remove();
+      const text = $("body")
+        .text()
+        .replace(/\s+/g, " ")
+        .replace(/\u2019/g, "'");
+      let reason = "instagram.media_unavailable";
+      let code = 500;
+      if (text.includes("Age-restricted content")) {
+        reason = "instagram.age_restricted";
+        code = 403;
+      } else if (text.includes("Post isn't available")) {
+        reason = "instagram.post_unavailable";
+        code = 404;
+      }
       throw {
-        code: 500,
-        message: "Instagram page structure changed: missing media data",
+        code,
+        reason,
+        message: script
+          ? "Instagram page structure changed: missing media data"
+          : "Instagram page structure changed: missing data script",
       };
     }