Downloader fixes with videos

2026-05-16 22:51:38 +02:00
parent ea9d4dc2dc
commit 522a8b22f8
2 changed files with 87 additions and 20 deletions
--- a/internal/canvas/client.go
+++ b/internal/canvas/client.go
@@ -240,7 +240,7 @@ func (c *Client) DownloadModules(courseRoot string) {
 					indent := strings.Repeat("  ", len(subHeaderStack)+1)
 					fmt.Printf("%s- Found direct video link: %s\n", indent, item.Title)

-					panopto.DownloadExternalPanoptoURL(c.HTTPClient, c.AccessToken, item.HTMLURL, item.ExternalURL, targetDir, item.Title)
+					panopto.DownloadVideo(c.HTTPClient, c.AccessToken, c.CourseID, targetDir, item.ExternalURL, item.Title)
 				}

 			case "Page":
--- a/internal/panopto/downloader.go
+++ b/internal/panopto/downloader.go
@@ -384,14 +384,23 @@ func DownloadMainRecordings(httpClient *http.Client, accessToken, courseID, root
 // moduleItemURL – item.URL  (https://canvas.vub.be/api/v1/.../module_item_redirect/<id>)
 // panoptoURL    – item.ExternalURL  (https://vub.cloud.panopto.eu/Panopto/Pages/...)
 func DownloadExternalPanoptoURL(httpClient *http.Client, accessToken, moduleItemURL, panoptoURL, modDir, title string) {
+	fmt.Printf("    [dbg] moduleItemURL: %s\n", moduleItemURL)
+	fmt.Printf("    [dbg] panoptoURL:    %s\n", panoptoURL)
+
 	jar, _ := cookiejar.New(nil)
-	// Follow all redirects automatically so Login.aspx → CookieCheck.aspx sets cookies.
-	webClient := &http.Client{Jar: jar}
+	// Manual redirect following so we can track cross-domain hops correctly.
+	noRedirectClient := &http.Client{
+		Jar: jar,
+		CheckRedirect: func(req *http.Request, via []*http.Request) error {
+			return http.ErrUseLastResponse
+		},
+	}

 	// Step 1: exchange the module item URL for a Canvas web session URL.
 	returnTo := moduleItemURL + "?display=borderless"
-	bridgeReq, _ := http.NewRequest("GET",
-		config.BaseURL+"/login/session_token?return_to="+url.QueryEscape(returnTo), nil)
+	sessionTokenURL := config.BaseURL + "/login/session_token?return_to=" + url.QueryEscape(returnTo)
+	fmt.Printf("    [dbg] session_token URL: %s\n", sessionTokenURL)
+	bridgeReq, _ := http.NewRequest("GET", sessionTokenURL, nil)
 	bridgeReq.Header.Set("Authorization", "Bearer "+accessToken)
 	bridgeReq.Header.Set("User-Agent", config.UserAgent)
 	bResp, err := httpClient.Do(bridgeReq)
@@ -399,26 +408,51 @@ func DownloadExternalPanoptoURL(httpClient *http.Client, accessToken, moduleItem
 		fmt.Printf("    [!] session_token request failed: %v\n", err)
 		return
 	}
+	rawBody, _ := io.ReadAll(bResp.Body)
+	bResp.Body.Close()
+	fmt.Printf("    [dbg] session_token response (%d): %s\n", bResp.StatusCode, string(rawBody))
+
 	var bridgeData struct {
 		SessionURL string `json:"session_url"`
 	}
-	json.NewDecoder(bResp.Body).Decode(&bridgeData)
-	bResp.Body.Close()
+	json.Unmarshal(rawBody, &bridgeData)

 	if bridgeData.SessionURL == "" {
 		fmt.Printf("    [!] No session_url returned (skipping %s)\n", title)
 		return
 	}
+	fmt.Printf("    [dbg] session_url: %s\n", bridgeData.SessionURL)

 	// Step 2: GET session URL → Canvas shows OAuth2 confirm page for Panopto.
-	formResp, err := webClient.Get(bridgeData.SessionURL)
-	if err != nil {
-		fmt.Printf("    [!] Failed to load OAuth confirm page: %v\n", err)
-		return
+	// We follow redirects manually so cross-domain hops (canvas→panopto) work correctly.
+	currentURL := bridgeData.SessionURL
+	var formHTML string
+	var formFinalURL string
+	for hop := 0; hop < 10; hop++ {
+		hopReq, _ := http.NewRequest("GET", currentURL, nil)
+		hopReq.Header.Set("User-Agent", config.UserAgent)
+		hopResp, err := noRedirectClient.Do(hopReq)
+		if err != nil {
+			fmt.Printf("    [!] Failed hop %d to %s: %v\n", hop, currentURL, err)
+			return
+		}
+		body, _ := io.ReadAll(hopResp.Body)
+		hopResp.Body.Close()
+		fmt.Printf("    [dbg] hop %d: status=%d url=%s body=%d\n", hop, hopResp.StatusCode, currentURL, len(body))
+		if hopResp.StatusCode == 301 || hopResp.StatusCode == 302 || hopResp.StatusCode == 303 {
+			loc, _ := hopResp.Location()
+			if loc == nil {
+				fmt.Printf("    [!] Redirect with no Location header\n")
+				return
+			}
+			currentURL = loc.String()
+			continue
+		}
+		formHTML = string(body)
+		formFinalURL = currentURL
+		break
 	}
-	formHTMLBytes, _ := io.ReadAll(formResp.Body)
-	formResp.Body.Close()
-	formHTML := string(formHTMLBytes)
+	fmt.Printf("    [dbg] OAuth page final URL: %s, length: %d\n", formFinalURL, len(formHTML))

 	if strings.Contains(formHTML, "U hebt geen toegang") || strings.Contains(formHTML, "You do not have access") {
 		fmt.Printf("    [!] Access denied. Skipping %s\n", title)
@@ -427,25 +461,58 @@ func DownloadExternalPanoptoURL(httpClient *http.Client, accessToken, moduleItem

 	// Step 3: POST /login/oauth2/accept → Canvas redirects to Panopto Login.aspx?code=...
 	// webClient follows all hops automatically, ending with Panopto setting cookies.
-	action := utils.ResolveAction(bridgeData.SessionURL, formHTML)
+	action := utils.ResolveAction(formFinalURL, formHTML)
 	formData := utils.ExtractFormFields(formHTML)
+	fmt.Printf("    [dbg] POST action: %s\n", action)

 	postReq, _ := http.NewRequest("POST", action, strings.NewReader(formData.Encode()))
 	postReq.Header.Set("Content-Type", "application/x-www-form-urlencoded")
 	postReq.Header.Set("User-Agent", config.UserAgent)
 	postReq.Header.Set("Origin", config.BaseURL)
-	postReq.Header.Set("Referer", bridgeData.SessionURL)
-	postResp, err := webClient.Do(postReq)
+	postReq.Header.Set("Referer", formFinalURL)
+	postResp, err := noRedirectClient.Do(postReq)
 	if err != nil {
 		fmt.Printf("    [!] OAuth accept POST failed: %v\n", err)
 		return
 	}
-	io.ReadAll(postResp.Body)
+	postBody, _ := io.ReadAll(postResp.Body)
 	postResp.Body.Close()
+	fmt.Printf("    [dbg] POST response status: %d, body len: %d\n", postResp.StatusCode, len(postBody))

-	// webClient.Jar now holds Panopto session cookies from the CookieCheck chain.
+	// Now follow the redirect chain from the POST (canvas → panopto Login.aspx → CookieCheck).
+	if postResp.StatusCode == 302 || postResp.StatusCode == 303 {
+		currentURL2 := ""
+		if loc, _ := postResp.Location(); loc != nil {
+			currentURL2 = loc.String()
+		}
+		for hop := 0; hop < 10 && currentURL2 != ""; hop++ {
+			hopReq, _ := http.NewRequest("GET", currentURL2, nil)
+			hopReq.Header.Set("User-Agent", config.UserAgent)
+			hopResp, err := noRedirectClient.Do(hopReq)
+			if err != nil {
+				fmt.Printf("    [dbg] post-redirect hop %d error: %v\n", hop, err)
+				break
+			}
+			body2, _ := io.ReadAll(hopResp.Body)
+			hopResp.Body.Close()
+			fmt.Printf("    [dbg] post-redirect hop %d: status=%d url=%s body=%d\n", hop, hopResp.StatusCode, currentURL2, len(body2))
+			if hopResp.StatusCode == 301 || hopResp.StatusCode == 302 || hopResp.StatusCode == 303 {
+				if loc, _ := hopResp.Location(); loc != nil {
+					currentURL2 = loc.String()
+					continue
+				}
+			}
+			break
+		}
+	}
+
+	// noRedirectClient.Jar now holds Panopto session cookies from the CookieCheck chain.
 	panoptoDomain, _ := url.Parse("https://vub.cloud.panopto.eu")
-	cookies := webClient.Jar.Cookies(panoptoDomain)
+	cookies := noRedirectClient.Jar.Cookies(panoptoDomain)
+	fmt.Printf("    [dbg] Panopto cookies: %d\n", len(cookies))
+	for _, c := range cookies {
+		fmt.Printf("    [dbg]   cookie: %s=%s\n", c.Name, c.Value[:min(20, len(c.Value))])
+	}
 	if len(cookies) == 0 {
 		fmt.Printf("    [!] No Panopto cookies after auth – falling back for: %s\n", title)
 		DownloadVideo(httpClient, accessToken, "", modDir, panoptoURL, title)