databricks-tellr-app 0.4.3.dev41__tar.gz → 0.4.3.dev43__tar.gz

{databricks_tellr_app-0.4.3.dev41 → databricks_tellr_app-0.4.3.dev43}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: databricks-tellr-app
-Version: 0.4.3.dev41
+Version: 0.4.3.dev43
 Summary: Tellr application package for Databricks Apps
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown

{databricks_tellr_app-0.4.3.dev41 → databricks_tellr_app-0.4.3.dev43}/databricks_tellr_app.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: databricks-tellr-app
-Version: 0.4.3.dev41
+Version: 0.4.3.dev43
 Summary: Tellr application package for Databricks Apps
 Requires-Python: >=3.10
 Description-Content-Type: text/markdown

{databricks_tellr_app-0.4.3.dev41 → databricks_tellr_app-0.4.3.dev43}/databricks_tellr_app.egg-info/SOURCES.txt

@@ -124,7 +124,6 @@ setup.py
 ./src/services/pptx_prompts_defaults.py
 ./src/services/profile_service.py
 ./src/services/streaming_callback.py
-./src/services/svg_sanitizer.py
 ./src/services/validator.py
 ./src/services/converter_jail/__init__.py
 ./src/services/converter_jail/ast_guard.py
@@ -287,7 +286,6 @@ src/services/pptx_from_records.py
 src/services/pptx_prompts_defaults.py
 src/services/profile_service.py
 src/services/streaming_callback.py
-src/services/svg_sanitizer.py
 src/services/validator.py
 src/services/converter_jail/__init__.py
 src/services/converter_jail/ast_guard.py

{databricks_tellr_app-0.4.3.dev41 → databricks_tellr_app-0.4.3.dev43}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "databricks-tellr-app"
-version = "0.4.3.dev41"
+version = "0.4.3.dev43"
 description = "Tellr application package for Databricks Apps"
 readme = "README.md"
 requires-python = ">=3.10"

{databricks_tellr_app-0.4.3.dev41 → databricks_tellr_app-0.4.3.dev43}/src/api/routes/export.py

@@ -702,6 +702,7 @@ async def start_pptx_export_async(request: ExportPPTXRequest):
         enqueue_export_job,
         generate_job_id,
     )
+    from src.services.image_service import resolve_requesting_user
 
     import time
     start_time = time.time()
@@ -759,6 +760,9 @@ async def start_pptx_export_async(request: ExportPPTXRequest):
             job_id,
             {
                 "session_id": request.session_id,
+                # Acting user for {{image:ID}} ownership checks (F-CR-27): the
+                # worker runs outside this request's user context.
+                "user_identity": resolve_requesting_user(),
                 "chart_images_per_slide": chart_images_per_slide,
                 "title": slide_deck.get("title", "slides"),
                 "total_slides": total_slides,

{databricks_tellr_app-0.4.3.dev41 → databricks_tellr_app-0.4.3.dev43}/src/api/services/export_job_queue.py

@@ -212,7 +212,9 @@ async def process_export_job(job_id: str, payload: dict) -> None:
         from src.core.database import get_db_session
         ds_id = resolve_active_design_system_id(session_id)
         with get_db_session() as db:
-            substitute_deck_dict_images(slide_deck, db)
+            substitute_deck_dict_images(
+                slide_deck, db, requesting_user=payload.get("user_identity")
+            )
             substitute_deck_dict_ds_assets(slide_deck, db, design_system_id=ds_id)
 
         slides = slide_deck.get("slides", [])

{databricks_tellr_app-0.4.3.dev41 → databricks_tellr_app-0.4.3.dev43}/src/services/design_system_service.py

@@ -74,7 +74,6 @@ from src.database.models.design_system import (
     DesignSystemToken,
 )
 from src.services.design_system_compiler import recompute_compiled_style_content
-from src.services.svg_sanitizer import looks_like_svg, sanitize_svg
 
 logger = logging.getLogger(__name__)
 
@@ -2165,20 +2164,6 @@ def _collect_assets_and_files(
                 mime = sniffed
             else:
                 mime = _guess_mime(rel)
-                if mime == "image/svg+xml":
-                    try:
-                        data = sanitize_svg(data)
-                    except ValueError as exc:
-                        raise DesignSystemImportError(
-                            f"Bundle entry '{rel}' is not a valid SVG: {exc}"
-                        ) from exc
-                elif looks_like_svg(data):
-                    # Extension says raster/font/etc. but the bytes are SVG:
-                    # never trust the name to skip sanitization.
-                    raise DesignSystemImportError(
-                        f"Bundle entry '{rel}' contains SVG content but does not "
-                        "have an .svg extension"
-                    )
             width, height = _image_dimensions(data, mime)
             asset = DesignSystemAsset(
                 kind=kind,

{databricks_tellr_app-0.4.3.dev41 → databricks_tellr_app-0.4.3.dev43}/src/services/image_service.py

@@ -1,6 +1,7 @@
 """Image upload, thumbnail generation, and retrieval service."""
 import base64
 import logging
+import os
 import uuid
 from io import BytesIO
 from typing import List, Optional
@@ -11,7 +12,6 @@ from sqlalchemy.dialects.postgresql import JSONB
 from sqlalchemy.orm import Query, Session
 
 from src.database.models.image import ImageAsset
-from src.services.svg_sanitizer import looks_like_svg, sanitize_svg
 
 logger = logging.getLogger(__name__)
 
@@ -42,20 +42,6 @@ def upload_image(
     if len(file_content) > MAX_FILE_SIZE:
         raise ValueError(f"File too large: {len(file_content)} bytes (max {MAX_FILE_SIZE})")
 
-    # The client-declared MIME type is untrusted: decide on the CONTENT.
-    # SVG can carry script, so it is always sanitized; SVG bytes declared as
-    # any other type are rejected rather than stored under a raster label.
-    if mime_type == "image/svg+xml":
-        try:
-            file_content = sanitize_svg(file_content)
-        except ValueError as exc:
-            raise ValueError(f"Invalid SVG upload: {exc}") from exc
-    elif looks_like_svg(file_content):
-        raise ValueError(
-            f"File content is SVG but was declared as {mime_type}; "
-            "upload it as image/svg+xml."
-        )
-
     # Check for duplicate original_filename among active images (case-insensitive).
     # Ephemeral images (paste-to-chat) skip this check — they use throwaway names
     # and should never block each other.
@@ -101,10 +87,35 @@ def upload_image(
     return image
 
 
-def get_image_base64(db: Session, token: str) -> tuple[str, str]:
+def resolve_requesting_user() -> Optional[str]:
+    """Identity to check image ownership against on non-route fetch paths.
+
+    Mirrors ``routes.images._get_current_user`` (the identity ``upload_image``
+    stamps into ``uploaded_by``) without an extra API call: the dev/test
+    ``"system"`` fallback, otherwise the request-scoped username set by the auth
+    middleware / MCP auth. ``None`` when no identity is bound (fail-closed for
+    ephemeral images).
+    """
+    if os.getenv("ENVIRONMENT") in ("development", "test"):
+        return "system"
+    from src.core.user_context import get_current_user
+
+    return get_current_user()
+
+
+def get_image_base64(
+    db: Session, token: str, *, requesting_user: Optional[str]
+) -> tuple[str, str]:
     """
     Get full image as base64 string, keyed by the opaque token (SDR-4437 F-TM-7).
 
+    Applies the same privacy guard as the ``GET /api/images/{token}`` routes
+    (F-CR-27): chat-pasted ("ephemeral") images are private to their uploader,
+    so a token for another user's ephemeral image is reported not-found. Library
+    images stay open-read. ``requesting_user`` is mandatory and keyword-only so
+    every caller makes the acting identity explicit; ``None`` fails closed for
+    ephemeral images.
+
     Returns:
         Tuple of (base64_data, mime_type)
     """
@@ -112,7 +123,10 @@ def get_image_base64(db: Session, token: str) -> tuple[str, str]:
         ImageAsset.token == token,
         ImageAsset.is_active == True,
     ).first()
-    if not image:
+    if not image or (
+        image.category == "ephemeral"
+        and (requesting_user is None or image.uploaded_by != requesting_user)
+    ):
         raise ValueError(f"Image {token} not found")
 
     b64 = base64.b64encode(image.image_data).decode("utf-8")

{databricks_tellr_app-0.4.3.dev41 → databricks_tellr_app-0.4.3.dev43}/src/services/image_tools.py

@@ -10,6 +10,8 @@ from src.services import image_service
 
 logger = logging.getLogger(__name__)
 
+_NO_IMAGES_MESSAGE = "No images found matching your criteria."
+
 
 class SearchImagesInput(BaseModel):
     """Input schema for image search tool."""
@@ -41,12 +43,27 @@ def search_images(
     Returns:
         JSON list of matching images with id, filename, description, and tags
     """
+    # F-CR-27: chat-pasted ("ephemeral") images are private to their uploader and
+    # are never discoverable through the agent tool. The user's own pastes reach
+    # the agent via their message, not via search.
+    if category is not None and category.strip().lower() == "ephemeral":
+        logger.warning("search_images: refusing category='ephemeral' (private images)")
+        return json.dumps({"message": _NO_IMAGES_MESSAGE, "images": []})
+
+    # Scope the listing to the caller (same as GET /api/images). Fail closed when
+    # no identity is bound rather than listing every user's images.
+    requesting_user = image_service.resolve_requesting_user()
+    if requesting_user is None:
+        logger.warning("search_images: no requesting user bound; returning no images")
+        return json.dumps({"message": _NO_IMAGES_MESSAGE, "images": []})
+
     with get_db_session() as db:
         images = image_service.search_images(
             db=db,
             query=query,
             category=category,
             tags=tags,
+            uploaded_by=requesting_user,
         )
 
         # Return metadata only - NEVER base64
@@ -65,6 +82,6 @@ def search_images(
         ]
 
     if not results:
-        return json.dumps({"message": "No images found matching your criteria.", "images": []})
+        return json.dumps({"message": _NO_IMAGES_MESSAGE, "images": []})
 
     return json.dumps({"message": f"Found {len(results)} image(s).", "images": results})

{databricks_tellr_app-0.4.3.dev41 → databricks_tellr_app-0.4.3.dev43}/src/utils/image_utils.py

@@ -1,6 +1,7 @@
 """Image placeholder substitution for generated slides."""
 import logging
 import re
+from typing import Optional
 
 from sqlalchemy.orm import Session
 
@@ -12,20 +13,31 @@ logger = logging.getLogger(__name__)
 IMAGE_PLACEHOLDER_PATTERN = re.compile(r"\{\{image:([A-Za-z0-9_-]+)\}\}")
 
 
-def substitute_image_placeholders(html: str, db: Session) -> str:
+def substitute_image_placeholders(
+    html: str, db: Session, *, requesting_user: Optional[str] = None
+) -> str:
     """
     Replace {{image:ID}} placeholders with base64 data URIs.
 
     Called after agent generates HTML, before returning to frontend.
     Works in both HTML img src and CSS url() contexts.
+
+    Resolution is scoped to the acting user (F-CR-27): another user's ephemeral
+    image is left as an unresolved placeholder. ``requesting_user`` defaults to
+    the request-scoped identity (``image_service.resolve_requesting_user``);
+    background callers without a request context must pass it explicitly.
     """
     if not html or "{{image:" not in html:
         return html
+    if requesting_user is None:
+        requesting_user = image_service.resolve_requesting_user()
 
     def replace_match(match):
         image_token = match.group(1)
         try:
-            b64_data, mime_type = image_service.get_image_base64(db, image_token)
+            b64_data, mime_type = image_service.get_image_base64(
+                db, image_token, requesting_user=requesting_user
+            )
             return f"data:{mime_type};base64,{b64_data}"
         except Exception as e:
             logger.warning(f"Failed to resolve image placeholder {{{{image:{image_token}}}}}: {e}")
@@ -41,23 +53,32 @@ def substitute_image_placeholders(html: str, db: Session) -> str:
 _DECK_IMAGE_FIELDS = ("html_content", "css")
 
 
-def substitute_deck_dict_images(deck_dict: dict, db: Session) -> dict:
+def substitute_deck_dict_images(
+    deck_dict: dict, db: Session, *, requesting_user: Optional[str] = None
+) -> dict:
     """Substitute {{image:ID}} placeholders across a deck dict.
 
     Covers every field that can carry the placeholder: each slide's ``html``,
     the deck's ``html_content`` (full knitted HTML) and its top-level ``css``
     (backgrounds). Deck-level fields are resolved independently of the slides
     array, mirroring ``substitute_deck_dict_ds_assets`` — the css gap fixed
-    there existed here too.
+    there existed here too. ``requesting_user`` is as for
+    ``substitute_image_placeholders``.
     """
     if not deck_dict:
         return deck_dict
+    if requesting_user is None:
+        requesting_user = image_service.resolve_requesting_user()
     for slide in deck_dict.get("slides") or []:
         html = slide.get("html", "")
         if "{{image:" in html:
-            slide["html"] = substitute_image_placeholders(html, db)
+            slide["html"] = substitute_image_placeholders(
+                html, db, requesting_user=requesting_user
+            )
     for field in _DECK_IMAGE_FIELDS:
         value = deck_dict.get(field)
         if value and "{{image:" in value:
-            deck_dict[field] = substitute_image_placeholders(value, db)
+            deck_dict[field] = substitute_image_placeholders(
+                value, db, requesting_user=requesting_user
+            )
     return deck_dict