Release 0.40.1

.github/workflows/ci.yml

@@ -47,12 +47,16 @@ jobs:
         with:
           python-version-file: "pyproject.toml"
           cache: "poetry"
-
+      - name: Install poetry with pip
+        run: python -m pip install poetry
       - name: Validate lockfile
         run: poetry check --lock
-
+      - name: Set Poetry Python
+        run: poetry env use python3.12
       - name: Install dependencies
-        run: poetry install --no-interaction
+        run: |
+          source $(poetry env info --path)/bin/activate
+          poetry install --no-interaction
 
       - name: Create test local state
         run: ./scripts/test/stub-data.sh

RELEASE.rst

@@ -1,6 +1,14 @@
 Release Notes
 =============
 
+Version 0.40.1
+--------------
+
+- fix program collection org filter bug (#2435)
+- Optimize memory footprint of pdf problem transcription task (#2433)
+- Replace the social media image (#2434)
+- Generate summaries for new video transcripts (#2428)
+
 Version 0.40.0 (Released August 13, 2025)
 --------------
 

frontends/main/src/app-pages/DashboardPage/OrganizationContent.tsx

@@ -194,6 +194,7 @@ const OrgProgramCollectionDisplay: React.FC<{
               key={item.programId}
               program={item.program}
               enrollments={enrollments}
+              orgId={orgId}
             />
           ) : null,
         )}
@@ -206,7 +207,7 @@ const OrgProgramDisplay: React.FC<{
   program: DashboardProgram
   courseRunEnrollments?: CourseRunEnrollment[]
   programLoading: boolean
-  orgId?: number
+  orgId: number
 }> = ({ program, courseRunEnrollments, programLoading, orgId }) => {
   const courses = useQuery(
     coursesQueries.coursesList({ id: program.courseIds, org_id: orgId }),
@@ -254,14 +255,17 @@ const OrgProgramDisplay: React.FC<{
 const ProgramCollectionItem: React.FC<{
   program: DashboardProgram
   enrollments?: CourseRunEnrollment[]
-}> = ({ program, enrollments }) => {
-  return <ProgramCard program={program} enrollments={enrollments} />
+  orgId: number
+}> = ({ program, enrollments, orgId }) => {
+  return (
+    <ProgramCard program={program} enrollments={enrollments} orgId={orgId} />
+  )
 }
 
 const ProgramCard: React.FC<{
   program: DashboardProgram
   enrollments?: CourseRunEnrollment[]
-  orgId?: number
+  orgId: number
 }> = ({ program, enrollments, orgId }) => {
   const courses = useQuery(
     coursesQueries.coursesList({

frontends/main/src/common/metadata.ts

@@ -98,8 +98,8 @@ export const standardizeMetadata = ({
           images: [
             {
               url: image,
-              width: image === DEFAULT_OG_IMAGE ? "" : 967,
-              height: image === DEFAULT_OG_IMAGE ? "" : 511,
+              width: image === DEFAULT_OG_IMAGE ? 967 : "",
+              height: image === DEFAULT_OG_IMAGE ? 511 : "",
               alt: imageAlt,
             },
           ],

learning_resources/constants.py

@@ -147,6 +147,7 @@ class LearningResourceRelationTypes(TextChoices):
     ".json",
     ".md",
     ".pdf",
+    ".tex",
     ".ppt",
     ".pptx",
     ".rtf",

learning_resources/content_summarizer.py

@@ -215,11 +215,15 @@ def _get_llm(self, model=None, temperature=0.0, max_tokens=1000) -> ChatLiteLLM:
         if not settings.LITELLM_CUSTOM_PROVIDER:
             raise ValueError("The 'LITELLM_CUSTOM_PROVIDER' setting must be set.")  # noqa: EM101, TRY003
 
+        if not settings.LITELLM_API_BASE:
+            raise ValueError("The 'LITELLM_API_BASE' setting must be set.")  # noqa: EM101, TRY003
+
         return ChatLiteLLM(
             model=model,
             temperature=temperature,
             max_tokens=max_tokens,
             custom_llm_provider=settings.LITELLM_CUSTOM_PROVIDER,
+            api_base=settings.LITELLM_API_BASE,
         )
 
     def _generate_summary(self, content: str, llm_model: str) -> str:

learning_resources/content_summarizer_test.py

@@ -21,6 +21,11 @@
 pytestmark = pytest.mark.django_db
 
 
+@pytest.fixture(autouse=True)
+def setup_settings(settings):
+    settings.LITELLM_API_BASE = "https://test/api/"
+
+
 @pytest.fixture
 def mock_summarize_single_content_file(mocker):
     """Fixture for mocking the process single file method"""

learning_resources/etl/canvas.py

@@ -9,10 +9,10 @@
 from pathlib import Path
 from tempfile import TemporaryDirectory
 
+import pypdfium2 as pdfium
 from defusedxml import ElementTree
 from django.conf import settings
 from litellm import completion
-from pdf2image import convert_from_path
 from PIL import Image
 
 from learning_resources.constants import (
@@ -71,7 +71,7 @@ def sync_canvas_archive(bucket, key: str, overwrite):
             run.checksum = checksum
             run.save()
 
-    return resource_readable_id, run
+    return resource_readable_id
 
 
 def _course_url(course_archive_path) -> str:
@@ -323,7 +323,7 @@ def extract_resources_by_identifierref(manifest_xml: str) -> dict:
     return dict(resources_dict)
 
 
-def pdf_to_base64_images(pdf_path, dpi=200, fmt="JPEG", max_size=2000, quality=85):
+def pdf_to_base64_images(pdf_path, fmt="JPEG", max_size=2000, quality=85):
     """
     Convert a PDF file to a list of base64 encoded images (one per page).
     Resizes images to reduce file size while keeping good OCR quality.
@@ -338,26 +338,24 @@ def pdf_to_base64_images(pdf_path, dpi=200, fmt="JPEG", max_size=2000, quality=8
     Returns:
         list: List of base64 encoded strings (one per page)
     """
-    images = convert_from_path(pdf_path, dpi=dpi)
-    base64_images = []
 
-    for image in images:
+    pdf = pdfium.PdfDocument(pdf_path)
+    for page_index in range(len(pdf)):
+        page = pdf.get_page(page_index)
+        image = page.render(scale=2).to_pil()
+        page.close()
         # Resize the image if it's too large (preserving aspect ratio)
         if max(image.size) > max_size:
             image.thumbnail((max_size, max_size), Image.Resampling.LANCZOS)
-
         buffered = BytesIO()
-
         # Save with optimized settings
         if fmt.upper() == "JPEG":
             image.save(buffered, format="JPEG", quality=quality, optimize=True)
         else:  # PNG
             image.save(buffered, format="PNG", optimize=True)
-
         img_str = base64.b64encode(buffered.getvalue()).decode("utf-8")
-        base64_images.append(img_str)
-
-    return base64_images
+        yield img_str
+    pdf.close()
 
 
 def _pdf_to_markdown(pdf_path):

learning_resources/migrations/0094_alter_contentsummarizerconfiguration_allowed_extensions.py

@@ -0,0 +1,67 @@
+# Generated by Django 4.2.23 on 2025-08-14 15:20
+
+import django.contrib.postgres.fields
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+    dependencies = [
+        ("learning_resources", "0093_tutorproblem_view_group"),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name="contentsummarizerconfiguration",
+            name="allowed_extensions",
+            field=django.contrib.postgres.fields.ArrayField(
+                base_field=models.CharField(
+                    choices=[
+                        (".csv", ".csv"),
+                        (".doc", ".doc"),
+                        (".docx", ".docx"),
+                        (".htm", ".htm"),
+                        (".html", ".html"),
+                        (".json", ".json"),
+                        (".m", ".m"),
+                        (".mat", ".mat"),
+                        (".md", ".md"),
+                        (".pdf", ".pdf"),
+                        (".ppt", ".ppt"),
+                        (".pptx", ".pptx"),
+                        (".ps", ".ps"),
+                        (".py", ".py"),
+                        (".r", ".r"),
+                        (".rtf", ".rtf"),
+                        (".sjson", ".sjson"),
+                        (".srt", ".srt"),
+                        (".txt", ".txt"),
+                        (".vtt", ".vtt"),
+                        (".xls", ".xls"),
+                        (".xlsx", ".xlsx"),
+                        (".xml", ".xml"),
+                        (".doc", ".doc"),
+                        (".docx", ".docx"),
+                        (".htm", ".htm"),
+                        (".html", ".html"),
+                        (".json", ".json"),
+                        (".md", ".md"),
+                        (".pdf", ".pdf"),
+                        (".tex", ".tex"),
+                        (".ppt", ".ppt"),
+                        (".pptx", ".pptx"),
+                        (".rtf", ".rtf"),
+                        (".sjson", ".sjson"),
+                        (".srt", ".srt"),
+                        (".txt", ".txt"),
+                        (".vtt", ".vtt"),
+                        (".xml", ".xml"),
+                    ],
+                    max_length=128,
+                ),
+                blank=True,
+                default=list,
+                null=True,
+                size=None,
+            ),
+        ),
+    ]

learning_resources/tasks.py

@@ -511,7 +511,7 @@ def sync_canvas_courses(canvas_course_ids, overwrite):
     for archive in latest_archives.values():
         key = archive.key
         log.info("Ingesting canvas course %s", key)
-        resource_readable_id, canvas_run = ingest_canvas_course(
+        resource_readable_id = ingest_canvas_course(
             key,
             overwrite=overwrite,
         )

learning_resources/tasks_test.py

@@ -658,7 +658,7 @@ def test_sync_canvas_courses(settings, mocker, django_assert_num_queries, canvas
     # Patch ingest_canvas_course to return the readable_ids for the two non-stale courses
     mock_ingest_course = mocker.patch(
         "learning_resources.tasks.ingest_canvas_course",
-        side_effect=[("course1", lr1.runs.first()), ("course2", lr2.runs.first())],
+        side_effect=["course1", "course2"],
     )
     sync_canvas_courses(canvas_course_ids=canvas_ids, overwrite=False)
 

main/settings.py

@@ -34,7 +34,7 @@
 from main.settings_pluggy import *  # noqa: F403
 from openapi.settings_spectacular import open_spectacular_settings
 
-VERSION = "0.40.0"
+VERSION = "0.40.1"
 
 log = logging.getLogger()
 

pyproject.toml

@@ -34,6 +34,8 @@ django-cors-headers = "^4.0.0"
 django-filter = "^2.4.0"
 django-guardian = "^3.0.0"
 django-health-check = { git = "https://github.com/revsys/django-health-check", rev="b0500d14c338040984f02ee34ffbe6643b005084" }  # pragma: allowlist secret
+
+
 django-imagekit = "^5.0.0"
 django-ipware = "^7.0.0"
 django-json-widget = "^2.0.0"
@@ -111,7 +113,9 @@ uwsgi = "^2.0.29"
 uwsgitop = "^0.12"
 wrapt = "^1.14.1"
 youtube-transcript-api = "^1.0.0"
-pdf2image = "^1.17.0"
+pypdfium2 = "^4.30.0"
+
+
 
 [tool.poetry.group.dev.dependencies]
 bpython = "^0.25"

vector_search/conftest.py

@@ -31,6 +31,7 @@ def _use_dummy_encoder(settings):
 def _use_test_qdrant_settings(settings, mocker):
     settings.QDRANT_HOST = "https://test"
     settings.QDRANT_BASE_COLLECTION_NAME = "test"
+    settings.LITELLM_API_BASE = "https://test/api/"
     settings.CONTENT_FILE_EMBEDDING_CHUNK_OVERLAP = 0
     settings.CONTENT_FILE_EMBEDDING_SEMANTIC_CHUNKING_ENABLED = False
     mock_qdrant = mocker.patch("qdrant_client.QdrantClient")

vector_search/utils.py

@@ -524,11 +524,15 @@ def embed_learning_resources(ids, resource_type, overwrite):
     else:
         serialized_resources = list(serialize_bulk_content_files(ids))
         # TODO: Pass actual Ids when we want scheduled content file summarization  # noqa: FIX002, TD002, TD003 E501
-        # Currently we only want to summarize content that already has a summary
+        # Currently we only want to summarize content that either already has a summary
+        # OR is in a course where atleast one other content file has a summary
         existing_summary_content_ids = [
             resource["id"]
             for resource in serialized_resources
             if resource.get("summary")
+            or ContentFile.objects.filter(run__id=resource.get("run_id"))
+            .exclude(summary="")
+            .exists()
         ]
         ContentSummarizer().summarize_content_files_by_ids(
             existing_summary_content_ids, overwrite