Fix/handle report meta info

apps/reports/ai_features/extraction.py

@@ -15,9 +15,8 @@
 
 
 @dataclass
-class PdfExtraction:
+class BaseExtraction:
     report: Report
-    data: bytes
 
     llm_handler: OllamaHandler = field(init=False)
     llm_chat_model: ChatOllama | ChatOpenAI = field(init=False)
@@ -31,6 +30,34 @@ def __post_init__(self):
         except Exception as e:
             raise e
 
+    def handle_meta_info(self):
+        """Extract meta information of the report."""
+        title = self.report.title
+        description = self.report.description
+        if title and title.strip():
+            DocumentExtraction.objects.create(
+                report=self.report,
+                status=DocumentExtractionStatus.SUCCESS,
+                text=title,
+                page_number=None,
+                chunk_type=DocumentExtraction.ExtractionType.TITLE,
+                embedding=self.llm_embedding_model.embed_query(title),
+            )
+        if description and description.strip():
+            DocumentExtraction.objects.create(
+                report=self.report,
+                status=DocumentExtractionStatus.SUCCESS,
+                text=description,
+                page_number=None,
+                chunk_type=DocumentExtraction.ExtractionType.DESCRIPTION,
+                embedding=self.llm_embedding_model.embed_query(description),
+            )
+
+
+@dataclass
+class PdfExtraction(BaseExtraction):
+    data: bytes
+
     def img_to_base64(self, data: fitz.Pixmap) -> str:
         img_bytes = data.tobytes("png")
         return base64.b64encode(img_bytes).decode("utf-8")
@@ -39,6 +66,9 @@ def pdf_to_images(self, zoom: float = 2.0):
         page_summaries = []
         doc = fitz.open(stream=self.data, filetype="pdf")
 
+        # Get the title and description
+        self.handle_meta_info()
+
         for page_idx in range(len(doc)):
             page = doc[page_idx]
 
@@ -108,3 +138,8 @@ def pdf_to_images(self, zoom: float = 2.0):
             )
         else:
             logger.warning("The key doc_summary is missing in the output")
+
+
+@dataclass
+class HeaderExtraction(BaseExtraction):
+    """Get the basic meta information extraction."""

apps/reports/ai_features/search_docs.py

@@ -35,6 +35,8 @@ def __post_init__(self):
             DocumentExtraction.ExtractionType.KEYWORDS: 0.9,
             DocumentExtraction.ExtractionType.TABLE: 0.3,
             DocumentExtraction.ExtractionType.CHART: 0.3,
+            DocumentExtraction.ExtractionType.TITLE: 0.2,
+            DocumentExtraction.ExtractionType.DESCRIPTION: 0.2,
         }
 
     def generate_query_embedding(self) -> list[float]:

apps/reports/extraction.py

@@ -2,7 +2,7 @@
 
 import logging
 
-from apps.reports.models import Report
+from apps.reports.models import Report, ReportContentType
 from apps.reports.tasks.task import handle_documents
 
 logger = logging.getLogger(__name__)
@@ -15,13 +15,12 @@ async def trigger_document_extraction(report: Report) -> None:
     The AI tool is expected to update extracted_contents, search_text, summary,
     and status directly in the database once processing completes.
     """
-    if not report.file.name:
+    if report.content_type == ReportContentType.FILE and not report.file.name:
         logger.warning("Report file is missing")
         return
 
     logger.info(
-        "Triggering document extraction for report_id=%s file_path=%s",
+        "Triggering document extraction for report_id=%s",
         str(report.pk),
-        report.file.name,
     )
-    handle_documents.apply_async(args=[report.pk])  # pyright: ignore[reportFunctionMemberAccess]
+    handle_documents.apply_async(args=(report.pk, report.content_type))  # pyright: ignore[reportFunctionMemberAccess]

apps/reports/migrations/0011_alter_documentextraction_chunk_type.py

@@ -0,0 +1,19 @@
+# Generated by Django 6.0.3 on 2026-06-30 09:57
+
+import django_choices_field.fields
+from django.db import migrations
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('reports', '0010_alter_report_thematic_area'),
+    ]
+
+    operations = [
+        migrations.AlterField(
+            model_name='documentextraction',
+            name='chunk_type',
+            field=django_choices_field.fields.IntegerChoicesField(choices=[(1, 'Extracted Content'), (2, 'Document Summary'), (3, 'Page Summary'), (4, 'Keywords'), (5, 'Table'), (6, 'Chart'), (7, 'Title'), (8, 'Description')], default=2),
+        ),
+    ]

apps/reports/models.py

@@ -166,6 +166,8 @@ class ExtractionType(models.IntegerChoices):
         KEYWORDS = 4, "Keywords"
         TABLE = 5, "Table"
         CHART = 6, "Chart"
+        TITLE = 7, "Title"
+        DESCRIPTION = 8, "Description"
 
     report = models.ForeignKey(
         Report,

apps/reports/tasks/task.py

@@ -3,25 +3,33 @@
 from celery import shared_task
 from django.core.files.storage import default_storage
 
-from apps.reports.ai_features.extraction import PdfExtraction
-from apps.reports.models import Report
+from apps.reports.ai_features.extraction import HeaderExtraction, PdfExtraction
+from apps.reports.models import Report, ReportContentType
 
 logger = logging.getLogger(__name__)
 
 
 @shared_task
-def handle_documents(report_id: int) -> None:
+def handle_documents(report_id: int, report_type: ReportContentType) -> None:
     report = Report.objects.filter(pk=report_id).first()
 
-    if report is None or report.file.name is None:
-        logger.warning("Report with id %s is set to None", report_id)
+    if report is None:
+        logger.warning("Report with id %s not found", report_id)
         return
 
-    with default_storage.open(report.file.name, "rb") as f:
-        pdf_bytes = f.read()
+    if report_type == ReportContentType.FILE and report.file.name is None:
+        logger.warning("Report with id %s has no file", report_id)
+        return
 
-    ext = PdfExtraction(
-        report=report,
-        data=pdf_bytes,
-    )
-    ext.pdf_to_images()
+    if report.file and (file_name := report.file.name) and report_type == ReportContentType.FILE:
+        with default_storage.open(file_name, "rb") as f:
+            pdf_bytes = f.read()
+
+        ext = PdfExtraction(
+            report=report,
+            data=pdf_bytes,
+        )
+        ext.pdf_to_images()
+    elif report_type == ReportContentType.IFRAME:
+        ext = HeaderExtraction(report=report)
+        ext.handle_meta_info()

apps/reports/tests/mutations_test.py

@@ -111,23 +111,26 @@ def setUpClass(cls):
 
     def test_create_iframe_report(self):
         self.force_login(self.user)
-        content = self.query_check(
-            self.Mutation.CREATE_REPORT,
-            variables={
-                "data": {
-                    "title": "Test Report",
-                    "contentType": Report.ContentType.IFRAME.name,
-                    "iframeUrl": "https://app.powerbi.com/embed/123",
-                    "thematicArea": str(self.thematic_area.pk),
+        with patch("apps.reports.graphql.mutations.trigger_document_extraction") as mock_trigger:
+            mock_trigger.return_value = None
+            content = self.query_check(
+                self.Mutation.CREATE_REPORT,
+                variables={
+                    "data": {
+                        "title": "Test Report",
+                        "contentType": Report.ContentType.IFRAME.name,
+                        "iframeUrl": "https://app.powerbi.com/embed/123",
+                        "thematicArea": str(self.thematic_area.pk),
+                    },
                 },
-            },
-        )
+            )
         resp = content["data"]["createReport"]
         assert resp["ok"] is True
         assert resp["errors"] is None
         assert resp["result"]["title"] == "Test Report"
         assert resp["result"]["contentType"] == Report.ContentType.IFRAME.name
         assert resp["result"]["visibility"] == Report.Visibility.PUBLIC.name
+        mock_trigger.assert_called_once()
 
     def test_create_report_requires_auth(self):
         self.logout()

schema.graphql

@@ -462,6 +462,8 @@ enum ExtractionType {
   KEYWORDS
   TABLE
   CHART
+  TITLE
+  DESCRIPTION
 }
 
 input GalleryAlbumCreateInput {