Release 0.71.4

RELEASE.rst

@@ -1,6 +1,13 @@
 Release Notes
 =============
 
+Version 0.71.4
+--------------
+
+- Exclude MicroMasters FIN programs from ETL ingestion (#3458)
+- feat: add surrogate-key in the header for Fastly (#3393)
+- fix: pick cover image of video if no image is inserted in article content (#3433)
+
 Version 0.71.3 (Released June 10, 2026)
 --------------
 

frontends/api/src/hooks/website_content/index.ts

@@ -42,7 +42,12 @@ const useWebsiteContentCreate = () => {
     mutationFn: (
       data: Omit<
         WebsiteContent,
-        "id" | "user" | "created_on" | "updated_on" | "publish_date"
+        | "id"
+        | "user"
+        | "created_on"
+        | "updated_on"
+        | "publish_date"
+        | "cover_image"
       >,
     ) =>
       websiteContentApi

frontends/api/src/test-utils/factories/websiteContent.ts

@@ -6,6 +6,7 @@ import type { WebsiteContent } from "../../generated/v1"
 const websiteContent: Factory<WebsiteContent> = (overrides = {}) => ({
   id: faker.number.int(),
   title: faker.lorem.sentence(),
+  cover_image: faker.image.url(),
   content: {
     type: "doc",
     content: [

frontends/main/src/common/websiteContentUtils.ts

@@ -94,19 +94,30 @@ export function extractArticleContent(
   const banner = topLevel.find((n) => n.type === "banner")
   const headingNode = banner?.content?.find((n) => n.type === "heading")
   const paragraphNode = banner?.content?.find((n) => n.type === "paragraph")
-
-  const imageNode = topLevel.find((n) => n.type === "imageWithCaption")
-  const imageAttrs = imageNode?.attrs
-  const image =
-    imageAttrs?.src && typeof imageAttrs.src === "string"
-      ? {
-          src: imageAttrs.src,
-          alt: typeof imageAttrs.alt === "string" ? imageAttrs.alt : null,
-          caption:
-            typeof imageAttrs.caption === "string" ? imageAttrs.caption : null,
-        }
-      : null
-
+  let image = null
+  if (article?.cover_image) {
+    image = {
+      src: article?.cover_image || "",
+      alt: "",
+      caption: "",
+    }
+  }
+  if (image === null) {
+    const imageNode = topLevel.find((n) => n.type === "imageWithCaption")
+    const imageAttrs = imageNode?.attrs
+
+    image =
+      imageAttrs?.src && typeof imageAttrs.src === "string"
+        ? {
+            src: imageAttrs.src,
+            alt: typeof imageAttrs.alt === "string" ? imageAttrs.alt : null,
+            caption:
+              typeof imageAttrs.caption === "string"
+                ? imageAttrs.caption
+                : null,
+          }
+        : null
+  }
   return {
     heading: extractText(headingNode?.content) || null,
     paragraph: nodesToHtml(paragraphNode?.content) || null,

frontends/main/src/page-components/TiptapEditor/extensions/node/ByLineInfoBar/ByLineInfoBar.tsx

@@ -30,12 +30,11 @@ const StyledWrapper = styled.div(({ theme }) => ({
   border: `1px solid ${theme.custom.colors.lightGray2}`,
 }))
 
-const InnerContainer = styled(Container, {
-  shouldForwardProp: (prop) => prop !== "noAuthor",
-})<{ noAuthor?: boolean }>(({ noAuthor }) => ({
+const InnerContainer = styled(Container)(() => ({
   display: "flex",
-  justifyContent: noAuthor ? "flex-end" : "space-between",
+  justifyContent: "space-between",
   alignItems: "center",
+  gap: "8px",
   "&&": {
     maxWidth: "890px",
   },
@@ -126,38 +125,38 @@ export const ByLineInfoBarContent = ({
         onClose={() => setShareOpen(false)}
         pageUrl={`${NEXT_PUBLIC_ORIGIN}/${article?.content_type === "article" ? "articles" : "news"}/${article?.slug}`}
       />
-      <InnerContainer noAuthor={!displayAuthorName && !isEditable}>
-        {(displayAuthorName || isEditable) && (
-          <InfoContainer>
-            {isEditable ? (
-              <AuthorInput
-                name="authorName"
-                label=""
-                size="small"
-                placeholder="Add author name(s)"
-                value={authorName || ""}
-                onChange={(e) => onAuthorNameChange?.(e.target.value)}
-              />
-            ) : (
-              <NameText>By {displayAuthorName}</NameText>
-            )}
-            {readTime ? <InfoText>{readTime} min read</InfoText> : null}
-            {readTime && publishedDate ? (
-              <Separator className="byline-info-bar__separator" />
-            ) : null}
-            <InfoText>
-              {publishedDate
-                ? new Date(publishedDate).toLocaleDateString("en-US", {
-                    month: "short",
-                    day: "numeric",
-                    year: "numeric",
-                  })
-                : isEditable
-                  ? null
-                  : "Draft"}
-            </InfoText>
-          </InfoContainer>
-        )}
+      <InnerContainer>
+        <InfoContainer>
+          {isEditable ? (
+            <AuthorInput
+              name="authorName"
+              label=""
+              size="small"
+              placeholder="Add author name(s)"
+              value={authorName || ""}
+              onChange={(e) => onAuthorNameChange?.(e.target.value)}
+            />
+          ) : (
+            <>
+              {displayAuthorName && <NameText>By {displayAuthorName}</NameText>}
+            </>
+          )}
+          {readTime ? <InfoText>{readTime} min read</InfoText> : null}
+          {readTime && publishedDate ? (
+            <Separator className="byline-info-bar__separator" />
+          ) : null}
+          <InfoText>
+            {publishedDate
+              ? new Date(publishedDate).toLocaleDateString("en-US", {
+                  month: "short",
+                  day: "numeric",
+                  year: "numeric",
+                })
+              : isEditable
+                ? null
+                : "Draft"}
+          </InfoText>
+        </InfoContainer>
         <div ref={shareButtonRef}>
           <ActionButton
             size="small"

frontends/main/src/proxy.test.ts

@@ -1,5 +1,5 @@
 import { NextRequest } from "next/server"
-import { isPageRoute, proxy } from "./proxy"
+import { isPageRoute, mitxonlineSurrogateKey, proxy } from "./proxy"
 
 describe("isPageRoute", () => {
   test.each([
@@ -33,19 +33,136 @@ describe("isPageRoute", () => {
   })
 })
 
+describe("mitxonlineSurrogateKey", () => {
+  describe("course pages — /courses/:readable_id", () => {
+    it("returns mitxonline:course key", () => {
+      expect(
+        mitxonlineSurrogateKey("/courses/course-v1:MITx+6.00.1x+3T2019"),
+      ).toBe("mitxonline:course:course-v1:MITx+6.00.1x+3T2019")
+    })
+
+    it("handles URL-encoded readable_id", () => {
+      const encoded = encodeURIComponent("course-v1:MITx+6.00.1x+3T2019")
+      expect(mitxonlineSurrogateKey(`/courses/${encoded}`)).toBe(
+        "mitxonline:course:course-v1:MITx+6.00.1x+3T2019",
+      )
+    })
+  })
+
+  describe("program pages — /programs/:readable_id and /courses/p/:readable_id", () => {
+    it("returns mitxonline:program key for /programs/:readable_id", () => {
+      expect(mitxonlineSurrogateKey("/programs/program-v1:MITx+SDS")).toBe(
+        "mitxonline:program:program-v1:MITx+SDS",
+      )
+    })
+
+    it("returns mitxonline:program key for /courses/p/:readable_id (ProgramAsCoursePage)", () => {
+      // /courses/p/ renders ProgramAsCoursePage — the readable_id belongs to a
+      // program, so the surrogate key must use the program namespace so that
+      // MITxOnline's program-save signal purges this page correctly.
+      expect(mitxonlineSurrogateKey("/courses/p/program-v1:MITx+SDS")).toBe(
+        "mitxonline:program:program-v1:MITx+SDS",
+      )
+    })
+  })
+
+  describe("non-product pages", () => {
+    it.each(["/", "/search", "/about", "/courses", "/programs"])(
+      "returns null for %s",
+      (pathname) => {
+        expect(mitxonlineSurrogateKey(pathname)).toBeNull()
+      },
+    )
+  })
+
+  describe("hostile URL segments — regression: Headers.set() crash on control characters", () => {
+    it("returns null without throwing on malformed percent-encoding in course path", () => {
+      // %GG is not valid percent-encoding — decodeURIComponent would throw URIError
+      expect(() => mitxonlineSurrogateKey("/courses/%GG")).not.toThrow()
+      expect(mitxonlineSurrogateKey("/courses/%GG")).toBeNull()
+    })
+
+    it("returns null without throwing on malformed percent-encoding in program path", () => {
+      expect(() => mitxonlineSurrogateKey("/programs/%GG")).not.toThrow()
+      expect(mitxonlineSurrogateKey("/programs/%GG")).toBeNull()
+    })
+
+    it("returns null on CRLF in course path (%0D%0A decodes to \\r\\n)", () => {
+      // Without safeDecodeSegment, passing this to Headers.set() would throw TypeError
+      expect(() =>
+        mitxonlineSurrogateKey("/courses/foo%0D%0AX-Injected%3A+yes"),
+      ).not.toThrow()
+      expect(
+        mitxonlineSurrogateKey("/courses/foo%0D%0AX-Injected%3A+yes"),
+      ).toBeNull()
+    })
+
+    it("returns null on CRLF in program path", () => {
+      expect(() =>
+        mitxonlineSurrogateKey("/programs/foo%0D%0AX-Injected%3A+yes"),
+      ).not.toThrow()
+      expect(
+        mitxonlineSurrogateKey("/programs/foo%0D%0AX-Injected%3A+yes"),
+      ).toBeNull()
+    })
+
+    it("returns null on null byte in course path", () => {
+      expect(() => mitxonlineSurrogateKey("/courses/foo%00bar")).not.toThrow()
+      expect(mitxonlineSurrogateKey("/courses/foo%00bar")).toBeNull()
+    })
+
+    it("returns null on null byte in program path", () => {
+      expect(() => mitxonlineSurrogateKey("/programs/foo%00bar")).not.toThrow()
+      expect(mitxonlineSurrogateKey("/programs/foo%00bar")).toBeNull()
+    })
+  })
+})
+
 describe("proxy", () => {
   const makeRequest = (pathname: string) =>
     new NextRequest(new URL(pathname, "https://learn.mit.edu"))
 
-  test("tags page routes with both Cache-Control and Surrogate-Key", () => {
-    const response = proxy(makeRequest("/courses/course-v1:MITxT+5.601x"))
+  test("tags generic page routes with Cache-Control and html-pages Surrogate-Key", () => {
+    const response = proxy(makeRequest("/about"))
     expect(response.headers.get("Surrogate-Key")).toBe("html-pages")
     expect(response.headers.get("Cache-Control")).toContain("s-maxage=")
   })
 
+  test("appends per-item surrogate key for MITxOnline course pages", () => {
+    const response = proxy(makeRequest("/courses/course-v1:MITxT+5.601x"))
+    expect(response.headers.get("Surrogate-Key")).toBe(
+      "html-pages mitxonline:course:course-v1:MITxT+5.601x",
+    )
+    expect(response.headers.get("Cache-Control")).toContain("s-maxage=")
+  })
+
+  test("appends per-item surrogate key for MITxOnline program pages (/programs/)", () => {
+    const response = proxy(makeRequest("/programs/program-v1:MITxT+18.01x"))
+    expect(response.headers.get("Surrogate-Key")).toBe(
+      "html-pages mitxonline:program:program-v1:MITxT+18.01x",
+    )
+  })
+
+  test("appends mitxonline:program surrogate key for /courses/p/ (ProgramAsCoursePage)", () => {
+    const response = proxy(makeRequest("/courses/p/program-v1:MITxT+18.01x"))
+    expect(response.headers.get("Surrogate-Key")).toBe(
+      "html-pages mitxonline:program:program-v1:MITxT+18.01x",
+    )
+  })
+
   test("leaves non-page routes untagged", () => {
     const response = proxy(makeRequest("/healthcheck"))
     expect(response.headers.get("Surrogate-Key")).toBeNull()
     expect(response.headers.get("Cache-Control")).toBeNull()
   })
+
+  test("does not throw on CRLF in course path and falls back to html-pages only", () => {
+    // Regression: without safeDecodeSegment, Headers.set() would throw TypeError
+    // when the decoded segment contains \r\n (%0D%0A).
+    expect(() =>
+      proxy(makeRequest("/courses/foo%0D%0AX-Injected%3A+yes")),
+    ).not.toThrow()
+    const response = proxy(makeRequest("/courses/foo%0D%0AX-Injected%3A+yes"))
+    expect(response.headers.get("Surrogate-Key")).toBe("html-pages")
+  })
 })