Import from Another LMS

LearnHouse converts course exports from other learning platforms into native LearnHouse courses: pages built with the course editor, graded assignments with auto-graded tasks, quiz blocks, videos, PDFs and SCORM activities. Supported formats:

The format is detected automatically. A zip holding several exports (a publisher test bank with one archive per chapter, for example) is combined into one course. You can import from the dashboard, from the command line, or through the API.

From the dashboard

1. Go to Courses in the dashboard and click Import Course.
2. Choose Another LMS.
3. Pick the Source platform (or leave Auto-detect), choose how Quizzes are converted, drop the export file and click Upload and Convert.
4. Wait for the conversion. Large backups with many videos can take a few minutes.
5. Review the report: what was created, what was skipped and why. Pick the courses to import and set the import options (Course Name Prefix, Set as private, Set as unpublished), then click the import button.

The report lists every source item that has no LearnHouse equivalent under Skipped items (discussions, LTI tools, files missing from the export…). Warnings only appear when something in the export could not be read or converted, so a healthy export shows none.

Imported courses are private and unpublished by default, so you can review them before learners see them.

What gets converted

Quizzes

Choose one of two modes:

• Graded assignments (default): every quiz becomes an assignment. Questions are mapped to task types and keep their answer keys, attempts, due dates and pass mark. Assignments with only objective questions are auto-graded; a question without an answer key (a survey question, for example) makes the assignment graded by hand.
• Interactive quiz blocks in pages: quizzes become pages with quiz blocks and flashcards, for self-checks without grades. Practice quizzes and ungraded problems always use this mode. Question images and tables are shown above their quiz block, and questions with no interactive equivalent keep their text.

Consecutive multiple choice questions are grouped into a single quiz task. Images in a question become the task’s reference file. When answers are pictures (graphs, diagrams) or several questions in one task have images, a “figures” page is added right before the assignment and the picture answers are labelled “Option A”, “Option B” and so on. A quiz with no question LearnHouse can grade (for example only STACK or drag-onto-image questions) is kept as a page with its questions.

What is not migrated

The report lists every skipped item. Not migrated:

• Learners, enrollments, grades and submissions (exports usually leave them out; course content only).
• Forums and discussions (their text is kept as a page), chats, wikis that only exist as links, LTI tools and H5P content.
• Hot spot, drag-onto-image, STACK and script-graded questions (their text is kept, they are not graded). CodeRunner questions become code tasks; tests that call the learner’s code from test code are listed in the prompt instead, and problems graded by an external code queue arrive without tests.
• Availability and release conditions, completion rules and rubrics. Point values are kept as task weights (the heaviest task is worth 100).
• Due dates that have already passed: they would close the assignment to learners, so they are removed (the report counts them). Set new ones after the import.
• Time limits, question shuffling and quiz passwords, which LearnHouse assignments do not have.
• Random question draws import every question of the pool; calculated questions use their first set of values; A/B test variants show their first group.

Media in formats LearnHouse cannot play (such as .mov or .bmp) is converted when ffmpeg and Pillow are available; otherwise the original file is kept as a downloadable Library file.

From the command line

The converter ships in the LearnHouse image as learnhouse-lms-import:

# What is in this export?
docker exec -it learnhouse learnhouse-lms-import inspect /data/backup.mbz
 
# Convert to a LearnHouse course package (import it from the dashboard: Import Course › LearnHouse Courses)
docker exec -it learnhouse learnhouse-lms-import convert /data/backup.mbz -o /data/course.zip --report /data/report.json
 
# Convert and import into an instance in one step, with an API token
learnhouse-lms-import push backup.mbz --url https://learn.example.com --token lh_... --org-id 1

From a source checkout, run it from apps/api: uv run python cli.py lms-import --help.

push reads LEARNHOUSE_URL, LEARNHOUSE_API_TOKEN and LEARNHOUSE_ORG_ID from the environment when the options are not given.

Through the API

# 1. Upload the export; conversion starts in the background
curl -X POST "https://learn.example.com/api/v1/courses/import/lms/analyze?org_id=1&source=auto&quiz_mode=assignment" \
  -H "Authorization: Bearer lh_..." -F "package=@backup.mbz"
# → {"temp_id": "…", "status": "processing"}
 
# 2. Poll until status is "ready" (or "failed", with an error)
curl "https://learn.example.com/api/v1/courses/import/lms/analyze/{temp_id}?org_id=1" -H "Authorization: Bearer lh_..."
# → {"status": "ready", "source": "moodle", "courses": [...], "report": {"counts": {...}, "warnings": [...], "skipped": [...]}}
 
# 3. Import, exactly like a LearnHouse package
curl -X POST "https://learn.example.com/api/v1/courses/import?org_id=1" -H "Authorization: Bearer lh_..." \
  -H "Content-Type: application/json" \
  -d '{"temp_id": "…", "course_uuids": ["course_…"], "set_private": true, "set_unpublished": true}'

GET /api/v1/courses/import/lms/sources lists the supported formats. All three calls need permission to create courses in the organization; a conversion is only visible to the user or token that uploaded it.