story/HOP-63

hospexplorer/ask/admin.py

@@ -5,8 +5,22 @@
 from django.contrib.auth.admin import UserAdmin
 from django.contrib.auth.models import User
 from django.db import transaction
-
-from ask.models import Conversation, TermsAcceptance, QARecord, SimWorkflow, WebsiteResource, PDFResource
+from django.http import HttpResponseRedirect
+from django.shortcuts import render
+from django.urls import path, reverse
+
+from ask.models import (
+    Conversation,
+    TermsAcceptance,
+    QARecord,
+    SimWorkflow,
+    WebsiteResource,
+    PDFResource,
+    DocumentType,
+    DocumentAuthorInstitution,
+    InstitutionType,
+)
+from ask.admin_csv import import_names_csv
 from ask.kb_connector import delete_kb_document
 from ask.tasks import run_kb_resource_upload
 
@@ -181,16 +195,98 @@ def delete_queryset(self, request, queryset):
                 return
 
 
+class LookupCSVImportMixin:
+    """Adds an Import CSV button + upload view to a lookup ModelAdmin.
+
+    CSV is single-column name. Duplicates are skipped, header row optional.
+    """
+
+    change_list_template = "admin/ask/lookup_change_list.html"
+
+    def get_urls(self):
+        urls = super().get_urls()
+        info = (self.model._meta.app_label, self.model._meta.model_name)
+        return [
+            path(
+                "import-csv/",
+                self.admin_site.admin_view(self.import_csv_view),
+                name=f"{info[0]}_{info[1]}_import_csv",
+            ),
+        ] + urls
+
+    def import_csv_view(self, request):
+        info = (self.model._meta.app_label, self.model._meta.model_name)
+        changelist_url = reverse(f"admin:{info[0]}_{info[1]}_changelist")
+
+        if request.method == "POST":
+            file_obj = request.FILES.get("csv_file")
+            if file_obj is None:
+                self.message_user(request, "No file provided.", level="error")
+            elif not file_obj.name.lower().endswith(".csv"):
+                self.message_user(request, "File must have a .csv extension.", level="error")
+            else:
+                try:
+                    created, skipped = import_names_csv(self.model, file_obj)
+                except Exception as e:
+                    logger.exception("CSV import failed for %s", self.model.__name__)
+                    self.message_user(request, f"Import failed: {e}", level="error")
+                else:
+                    self.message_user(
+                        request,
+                        f"Imported {created} new {self.model._meta.verbose_name_plural} "
+                        f"(skipped {skipped} duplicate or empty rows).",
+                    )
+            return HttpResponseRedirect(changelist_url)
+
+        context = {
+            **self.admin_site.each_context(request),
+            "title": f"Import {self.model._meta.verbose_name_plural} from CSV",
+            "opts": self.model._meta,
+            "changelist_url": changelist_url,
+        }
+        return render(request, "admin/ask/lookup_csv_import.html", context)
+
+
+@admin.register(DocumentType)
+class DocumentTypeAdmin(LookupCSVImportMixin, admin.ModelAdmin):
+    list_display = ("name",)
+    search_fields = ("name",)
+
+
+@admin.register(DocumentAuthorInstitution)
+class DocumentAuthorInstitutionAdmin(LookupCSVImportMixin, admin.ModelAdmin):
+    list_display = ("name",)
+    search_fields = ("name",)
+
+
+@admin.register(InstitutionType)
+class InstitutionTypeAdmin(LookupCSVImportMixin, admin.ModelAdmin):
+    list_display = ("name",)
+    search_fields = ("name",)
+
+
 @admin.register(WebsiteResource)
 class WebsiteResourceAdmin(KBDeleteAdminMixin, admin.ModelAdmin):
     list_display = ("title", "url", "creator", "status", "modified_at")
     list_filter = ("status",)
     search_fields = ("title", "url")
     readonly_fields = ("created_at", "modified_at", "creator", "modifier", "mcp_kb_document_id", "status", "status_message")
+    fieldsets = (
+        (None, {"fields": ("title", "description", "url")}),
+        ("Metadata", {"fields": (
+            "date_published", "date_published_precision",
+            "document_type", "document_author_institution", "institution_type",
+        )}),
+        ("Status", {"fields": (
+            "status", "status_message", "mcp_kb_document_id",
+            "created_at", "modified_at", "creator", "modifier",
+        )}),
+    )
     help_texts = {
         "title": "A short name to identify this website resource.",
         "description": "Optional details about what this website covers.",
         "url": "The URL the LLM will use as context when answering questions.",
+        "date_published_precision": "Granularity of the date above (year / month / day). Leave blank if unknown.",
     }
 
     def get_form(self, request, obj=None, **kwargs):
@@ -230,10 +326,22 @@ class PDFResourceAdmin(KBDeleteAdminMixin, admin.ModelAdmin):
     list_filter = ("status",)
     search_fields = ("title",)
     readonly_fields = ("created_at", "modified_at", "creator", "modifier", "mcp_kb_document_id", "status", "status_message")
+    fieldsets = (
+        (None, {"fields": ("title", "description", "file")}),
+        ("Metadata", {"fields": (
+            "date_published", "date_published_precision",
+            "document_type", "document_author_institution", "institution_type",
+        )}),
+        ("Status", {"fields": (
+            "status", "status_message", "mcp_kb_document_id",
+            "created_at", "modified_at", "creator", "modifier",
+        )}),
+    )
     help_texts = {
         "title": "A short name to identify this PDF resource.",
         "description": "Optional details about what this PDF covers.",
         "file": "The PDF file the LLM will use as context when answering questions.",
+        "date_published_precision": "Granularity of the date above (year / month / day). Leave blank if unknown.",
     }
 
     def get_form(self, request, obj=None, **kwargs):

hospexplorer/ask/admin_csv.py

@@ -0,0 +1,26 @@
+import csv
+import io
+
+
+def import_names_csv(model, file_obj):
+    """Import a one-column CSV into a model with a ``name`` field.
+
+    Returns ``(created, skipped)``. Blank rows, a leading header row of ``name``,
+    and rows whose name already exists in the table are all counted as skipped.
+    """
+    text = file_obj.read().decode("utf-8-sig", errors="replace")
+    reader = csv.reader(io.StringIO(text))
+
+    created = 0
+    skipped = 0
+    for row in reader:
+        name = row[0].strip() if row else ""
+        if not name or name.lower() == "name":
+            skipped += 1
+            continue
+        _, was_created = model.objects.get_or_create(name=name)
+        if was_created:
+            created += 1
+        else:
+            skipped += 1
+    return created, skipped

hospexplorer/ask/kb_connector.py

@@ -1,3 +1,4 @@
+import json
 import logging
 
 import httpx
@@ -30,22 +31,24 @@ def list_kb_documents(page=1, page_size=10):
     return response.json()
 
 
-def add_website_to_kb(url):
+def add_website_to_kb(url, metadata=None):
     """Send a website URL to the MCP KB server for ingestion.
 
     Calls POST /docs/website/add?url={url} on the MCP KB server.
-    The KB server fetches the page, chunks it, generates embeddings,
-    and stores it for semantic search.
+    ``metadata`` (if provided) is sent as a JSON body ``{"metadata": ...}`` so
+    the KB server can store it on the Document row.
     """
     headers = {
         "Authorization": f"Bearer {settings.KB_MCP_JWT_TOKEN}",
+        "Content-Type": "application/json",
     }
     endpoint = f"{settings.KB_MCP_HOST}/docs/website/add"
 
     with httpx.Client() as client:
         response = client.post(
             endpoint,
             params={"url": url},
+            json={"metadata": metadata} if metadata is not None else {},
             headers=headers,
             timeout=settings.KB_MCP_TIMEOUT,
         )
@@ -54,12 +57,12 @@ def add_website_to_kb(url):
     return response.json()
 
 
-def add_pdf_to_kb(file_bytes, filename, title, url=None):
+def add_pdf_to_kb(file_bytes, filename, title, url=None, metadata=None):
     """Upload a PDF to the MCP KB server for ingestion.
 
     Calls POST /docs/pdf/add on the MCP KB server with multipart form data.
-    The KB server extracts text, chunks it, generates embeddings,
-    and stores it for semantic search.
+    metadata (if provided) is JSON-encoded into a metadata form field so
+    it can travel alongside the file.
     """
     headers = {
         "Authorization": f"Bearer {settings.KB_MCP_JWT_TOKEN}",
@@ -70,6 +73,8 @@ def add_pdf_to_kb(file_bytes, filename, title, url=None):
     data = {"title": title}
     if url:
         data["url"] = url
+    if metadata is not None:
+        data["metadata"] = json.dumps(metadata)
 
     with httpx.Client() as client:
         response = client.post(

hospexplorer/ask/migrations/0013_documentauthorinstitution_documenttype_and_more.py

@@ -0,0 +1,94 @@
+# Generated by Django 6.0.2 on 2026-05-11 23:30
+
+import django.db.models.deletion
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ('ask', '0012_pdfresource_status_pdfresource_status_message_and_more'),
+    ]
+
+    operations = [
+        migrations.CreateModel(
+            name='DocumentAuthorInstitution',
+            fields=[
+                ('id', models.BigAutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('name', models.CharField(max_length=255, unique=True)),
+            ],
+            options={
+                'ordering': ['name'],
+            },
+        ),
+        migrations.CreateModel(
+            name='DocumentType',
+            fields=[
+                ('id', models.BigAutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('name', models.CharField(max_length=255, unique=True)),
+            ],
+            options={
+                'ordering': ['name'],
+            },
+        ),
+        migrations.CreateModel(
+            name='InstitutionType',
+            fields=[
+                ('id', models.BigAutoField(auto_created=True, primary_key=True, serialize=False, verbose_name='ID')),
+                ('name', models.CharField(max_length=255, unique=True)),
+            ],
+            options={
+                'ordering': ['name'],
+            },
+        ),
+        migrations.AddField(
+            model_name='pdfresource',
+            name='date_published',
+            field=models.DateField(blank=True, null=True),
+        ),
+        migrations.AddField(
+            model_name='pdfresource',
+            name='date_published_precision',
+            field=models.CharField(blank=True, choices=[('year', 'Year'), ('month', 'Month'), ('day', 'Day')], default='', max_length=10),
+        ),
+        migrations.AddField(
+            model_name='websiteresource',
+            name='date_published',
+            field=models.DateField(blank=True, null=True),
+        ),
+        migrations.AddField(
+            model_name='websiteresource',
+            name='date_published_precision',
+            field=models.CharField(blank=True, choices=[('year', 'Year'), ('month', 'Month'), ('day', 'Day')], default='', max_length=10),
+        ),
+        migrations.AddField(
+            model_name='pdfresource',
+            name='document_author_institution',
+            field=models.ForeignKey(blank=True, null=True, on_delete=django.db.models.deletion.SET_NULL, related_name='%(class)s_resources', to='ask.documentauthorinstitution'),
+        ),
+        migrations.AddField(
+            model_name='websiteresource',
+            name='document_author_institution',
+            field=models.ForeignKey(blank=True, null=True, on_delete=django.db.models.deletion.SET_NULL, related_name='%(class)s_resources', to='ask.documentauthorinstitution'),
+        ),
+        migrations.AddField(
+            model_name='pdfresource',
+            name='document_type',
+            field=models.ForeignKey(blank=True, null=True, on_delete=django.db.models.deletion.SET_NULL, related_name='%(class)s_resources', to='ask.documenttype'),
+        ),
+        migrations.AddField(
+            model_name='websiteresource',
+            name='document_type',
+            field=models.ForeignKey(blank=True, null=True, on_delete=django.db.models.deletion.SET_NULL, related_name='%(class)s_resources', to='ask.documenttype'),
+        ),
+        migrations.AddField(
+            model_name='pdfresource',
+            name='institution_type',
+            field=models.ForeignKey(blank=True, null=True, on_delete=django.db.models.deletion.SET_NULL, related_name='%(class)s_resources', to='ask.institutiontype'),
+        ),
+        migrations.AddField(
+            model_name='websiteresource',
+            name='institution_type',
+            field=models.ForeignKey(blank=True, null=True, on_delete=django.db.models.deletion.SET_NULL, related_name='%(class)s_resources', to='ask.institutiontype'),
+        ),
+    ]