Merge pull request #20 from video-db/fix/vision-interface

Fix/vision and search interface

Author: ashish-spext

requirements-dev.txt

@@ -1,4 +1,4 @@
 ruff==0.1.7
 pytest==7.4.3
-twine==4.0.2
+twine==5.1.1
 wheel==0.42.0

setup.py

@@ -23,6 +23,7 @@
     version=about["__version__"],
     author=about["__author__"],
     author_email=about["__email__"],
+    license=about["__license__"],
     description="VideoDB Python SDK",
     long_description=long_description,
     long_description_content_type="text/markdown",

videodb/__about__.py

@@ -1,8 +1,9 @@
 """ About information for videodb sdk"""
 
 
-__version__ = "0.2.0"
+__version__ = "0.2.1"
 __title__ = "videodb"
 __author__ = "videodb"
 __email__ = "[email protected]"
 __url__ = "https://github.com/video-db/videodb-python"
+__license__ = "Apache License 2.0"

videodb/__init__.py

@@ -7,6 +7,7 @@
 from videodb._utils._video import play_stream
 from videodb._constants import (
     VIDEO_DB_API,
+    IndexType,
     SceneExtractionType,
     MediaType,
     SearchType,
@@ -30,6 +31,7 @@
     "VideodbError",
     "AuthenticationError",
     "InvalidRequestError",
+    "IndexType",
     "SearchError",
     "play_stream",
     "MediaType",

videodb/_constants.py

@@ -18,12 +18,12 @@ class SearchType:
 
 
 class IndexType:
-    semantic = "semantic"
+    spoken_word = "spoken_word"
     scene = "scene"
 
 
 class SceneExtractionType:
-    scene_based = "scene"
+    shot_based = "shot"
     time_based = "time"
 
 
@@ -58,6 +58,8 @@ class ApiPath:
     invoices = "invoices"
     scenes = "scenes"
     scene = "scene"
+    frame = "frame"
+    describe = "describe"
 
 
 class Status:

videodb/collection.py

@@ -10,6 +10,7 @@
 )
 from videodb._constants import (
     ApiPath,
+    IndexType,
     SearchType,
 )
 from videodb.video import Video
@@ -100,6 +101,7 @@ def search(
         self,
         query: str,
         search_type: Optional[str] = SearchType.semantic,
+        index_type: Optional[str] = IndexType.spoken_word,
         result_threshold: Optional[int] = None,
         score_threshold: Optional[float] = None,
         dynamic_score_percentage: Optional[float] = None,
@@ -108,6 +110,8 @@ def search(
         return search.search_inside_collection(
             collection_id=self.id,
             query=query,
+            search_type=search_type,
+            index_type=index_type,
             result_threshold=result_threshold,
             score_threshold=score_threshold,
             dynamic_score_percentage=dynamic_score_percentage,

videodb/image.py

@@ -61,3 +61,11 @@ def to_json(self):
             "frame_time": self.frame_time,
             "description": self.description,
         }
+
+    def describe(self, prompt: str = None, model_name=None):
+        description_data = self._connection.post(
+            path=f"{ApiPath.video}/{self.video_id}/{ApiPath.frame}/{self.id}/{ApiPath.describe}",
+            data={"prompt": prompt, "model_name": model_name},
+        )
+        self.description = description_data.get("description", None)
+        return self.description

videodb/scene.py

@@ -14,13 +14,15 @@ def __init__(
         description: str,
         id: str = None,
         frames: List[Frame] = [],
+        connection=None,
     ):
         self.id = id
         self.video_id = video_id
         self.start = start
         self.end = end
         self.frames: List[Frame] = frames
         self.description = description
+        self._connection = connection
 
     def __repr__(self) -> str:
         return (
@@ -43,6 +45,16 @@ def to_json(self):
             "description": self.description,
         }
 
+    def describe(self, prompt: str = None, model_name=None) -> None:
+        if self._connection is None:
+            raise ValueError("Connection is required to describe a scene")
+        description_data = self._connection.post(
+            path=f"{ApiPath.video}/{self.video_id}/{ApiPath.scene}/{self.id}/{ApiPath.describe}",
+            data={"prompt": prompt, "model_name": model_name},
+        )
+        self.description = description_data.get("description", None)
+        return self.description
+
 
 class SceneCollection:
     def __init__(

videodb/search.py

@@ -1,6 +1,7 @@
 from abc import ABC, abstractmethod
 from videodb._utils._video import play_stream
 from videodb._constants import (
+    IndexType,
     SearchType,
     ApiPath,
     SemanticSearchDefaultValues,
@@ -109,6 +110,8 @@ def search_inside_video(
         self,
         video_id: str,
         query: str,
+        search_type: str,
+        index_type: str,
         result_threshold: Optional[int] = None,
         score_threshold: Optional[float] = None,
         dynamic_score_percentage: Optional[float] = None,
@@ -117,7 +120,8 @@ def search_inside_video(
         search_data = self._connection.post(
             path=f"{ApiPath.video}/{video_id}/{ApiPath.search}",
             data={
-                "index_type": SearchType.semantic,
+                "search_type": search_type,
+                "index_type": index_type,
                 "query": query,
                 "score_threshold": score_threshold
                 or SemanticSearchDefaultValues.score_threshold,
@@ -133,6 +137,8 @@ def search_inside_collection(
         self,
         collection_id: str,
         query: str,
+        search_type: str,
+        index_type: str,
         result_threshold: Optional[int] = None,
         score_threshold: Optional[float] = None,
         dynamic_score_percentage: Optional[float] = None,
@@ -141,7 +147,8 @@ def search_inside_collection(
         search_data = self._connection.post(
             path=f"{ApiPath.collection}/{collection_id}/{ApiPath.search}",
             data={
-                "index_type": SearchType.semantic,
+                "search_type": search_type,
+                "index_type": index_type,
                 "query": query,
                 "score_threshold": score_threshold
                 or SemanticSearchDefaultValues.score_threshold,
@@ -162,6 +169,8 @@ def search_inside_video(
         self,
         video_id: str,
         query: str,
+        search_type: str,
+        index_type: str,
         result_threshold: Optional[int] = None,
         score_threshold: Optional[float] = None,
         dynamic_score_percentage: Optional[float] = None,
@@ -170,7 +179,8 @@ def search_inside_video(
         search_data = self._connection.post(
             path=f"{ApiPath.video}/{video_id}/{ApiPath.search}",
             data={
-                "index_type": SearchType.keyword,
+                "search_type": search_type,
+                "index_type": index_type,
                 "query": query,
                 "score_threshold": score_threshold,
                 "result_threshold": result_threshold,
@@ -190,6 +200,8 @@ def search_inside_video(
         self,
         video_id: str,
         query: str,
+        search_type: str,
+        index_type: str,
         result_threshold: Optional[int] = None,
         score_threshold: Optional[float] = None,
         dynamic_score_percentage: Optional[float] = None,
@@ -198,7 +210,8 @@ def search_inside_video(
         search_data = self._connection.post(
             path=f"{ApiPath.video}/{video_id}/{ApiPath.search}",
             data={
-                "index_type": SearchType.scene,
+                "search_type": search_type,
+                "index_type": IndexType.scene,
                 "query": query,
                 "score_threshold": score_threshold,
                 "result_threshold": result_threshold,

videodb/video.py

@@ -49,6 +49,7 @@ def search(
         self,
         query: str,
         search_type: Optional[str] = SearchType.semantic,
+        index_type: Optional[str] = IndexType.spoken_word,
         result_threshold: Optional[int] = None,
         score_threshold: Optional[float] = None,
         dynamic_score_percentage: Optional[float] = None,
@@ -58,6 +59,8 @@ def search(
         return search.search_inside_video(
             video_id=self.id,
             query=query,
+            search_type=search_type,
+            index_type=index_type,
             result_threshold=result_threshold,
             score_threshold=score_threshold,
             dynamic_score_percentage=dynamic_score_percentage,
@@ -152,7 +155,7 @@ def index_spoken_words(
         self._connection.post(
             path=f"{ApiPath.video}/{self.id}/{ApiPath.index}",
             data={
-                "index_type": IndexType.semantic,
+                "index_type": IndexType.spoken_word,
                 "language_code": language_code,
                 "force": force,
                 "callback_url": callback_url,
@@ -194,6 +197,7 @@ def _format_scene_collection(self, scene_collection_data: dict) -> SceneCollecti
                 description=scene.get("description"),
                 id=scene.get("scene_id"),
                 frames=frames,
+                connection=self._connection,
             )
             scenes.append(scene)
 
@@ -207,11 +211,11 @@ def _format_scene_collection(self, scene_collection_data: dict) -> SceneCollecti
 
     def extract_scenes(
         self,
-        extraction_type: SceneExtractionType = SceneExtractionType.scene_based,
+        extraction_type: SceneExtractionType = SceneExtractionType.shot_based,
         extraction_config: dict = {},
         force: bool = False,
         callback_url: str = None,
-    ):
+    ) -> Optional[SceneCollection]:
         scenes_data = self._connection.post(
             path=f"{ApiPath.video}/{self.id}/{ApiPath.scenes}",
             data={
@@ -225,10 +229,14 @@ def extract_scenes(
             return None
         return self._format_scene_collection(scenes_data.get("scene_collection"))
 
-    def get_scene_collection(self, collection_id: str):
+    def get_scene_collection(self, collection_id: str) -> Optional[SceneCollection]:
+        if not collection_id:
+            raise ValueError("collection_id is required")
         scenes_data = self._connection.get(
             path=f"{ApiPath.video}/{self.id}/{ApiPath.scenes}/{collection_id}"
         )
+        if not scenes_data:
+            return None
         return self._format_scene_collection(scenes_data.get("scene_collection"))
 
     def list_scene_collection(self):
@@ -238,29 +246,31 @@ def list_scene_collection(self):
         return scene_collections_data.get("scene_collections", [])
 
     def delete_scene_collection(self, collection_id: str) -> None:
+        if not collection_id:
+            raise ValueError("collection_id is required")
         self._connection.delete(
             path=f"{ApiPath.video}/{self.id}/{ApiPath.scenes}/{collection_id}"
         )
 
     def index_scenes(
         self,
-        extraction_type: SceneExtractionType = SceneExtractionType.scene_based,
+        extraction_type: SceneExtractionType = SceneExtractionType.shot_based,
         extraction_config: Dict = {},
         prompt: Optional[str] = None,
-        model: Optional[str] = None,
+        model_name: Optional[str] = None,
         model_config: Optional[Dict] = None,
         name: Optional[str] = None,
         scenes: Optional[List[Scene]] = None,
         force: Optional[bool] = False,
         callback_url: Optional[str] = None,
-    ) -> Optional[List]:
+    ) -> Optional[str]:
         scenes_data = self._connection.post(
             path=f"{ApiPath.video}/{self.id}/{ApiPath.index}/{ApiPath.scene}",
             data={
                 "extraction_type": extraction_type,
                 "extraction_config": extraction_config,
                 "prompt": prompt,
-                "model": model,
+                "model_name": model_name,
                 "model_config": model_config,
                 "name": name,
                 "force": force,
@@ -270,7 +280,7 @@ def index_scenes(
         )
         if not scenes_data:
             return None
-        return scenes_data.get("scene_index_records", [])
+        return scenes_data.get("scene_index_id")
 
     def list_scene_index(self) -> List:
         index_data = self._connection.get(
@@ -287,6 +297,8 @@ def get_scene_index(self, scene_index_id: str) -> Optional[List]:
         return index_data.get("scene_index_records", [])
 
     def delete_scene_index(self, scene_index_id: str) -> None:
+        if not scene_index_id:
+            raise ValueError("scene_index_id is required")
         self._connection.delete(
             path=f"{ApiPath.video}/{self.id}/{ApiPath.index}/{ApiPath.scene}/{scene_index_id}"
         )