vision-agent 0.2.238__py3-none-any.whl → 0.2.240__py3-none-any.whl

vision_agent/tools/tools.py

@@ -8,7 +8,7 @@ from base64 import b64encode
 from concurrent.futures import ThreadPoolExecutor, as_completed
 from importlib import resources
 from pathlib import Path
-from typing import Any, Callable, Dict, List, Optional, Tuple, Union, cast
+from typing import IO, Any, Callable, Dict, List, Optional, Tuple, Union, cast
 from uuid import UUID
 
 import cv2
@@ -2797,16 +2797,17 @@ def save_video(
         ):
             raise ValueError("A frame is not a valid NumPy array with shape (H, W, C)")
 
+    output_file: IO[bytes]
     if output_video_path is None:
-        output_video_path = tempfile.NamedTemporaryFile(
-            delete=False, suffix=".mp4"
-        ).name
+        output_file = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4")
     else:
         Path(output_video_path).parent.mkdir(parents=True, exist_ok=True)
+        output_file = open(output_video_path, "wb")
 
-    output_video_path = video_writer(frames, fps, output_video_path)
-    _save_video_to_result(output_video_path)
-    return output_video_path
+    with output_file as file:
+        video_writer(frames, fps, file=file)
+    _save_video_to_result(output_file.name)
+    return output_file.name
 
 
 def _save_video_to_result(video_uri: str) -> None:

vision_agent/utils/video.py

@@ -1,8 +1,7 @@
-import base64
 import logging
 import tempfile
 from functools import lru_cache
-from typing import List, Optional, Tuple
+from typing import IO, List, Optional, Tuple
 
 import av  # type: ignore
 import cv2
@@ -15,37 +14,6 @@ _DEFAULT_VIDEO_FPS = 24
 _DEFAULT_INPUT_FPS = 1.0
 
 
-def play_video(video_base64: str) -> None:
-    """Play a video file"""
-    video_data = base64.b64decode(video_base64)
-    with tempfile.NamedTemporaryFile(delete=False, suffix=".mp4") as temp_video:
-        temp_video.write(video_data)
-        temp_video_path = temp_video.name
-
-        cap = cv2.VideoCapture(temp_video_path)
-        if not cap.isOpened():
-            _LOGGER.error("Error: Could not open video.")
-            return
-
-        # Display the first frame and wait for any key press to start the video
-        ret, frame = cap.read()
-        if ret:
-            cv2.imshow("Video Player", frame)
-            _LOGGER.info(f"Press any key to start playing the video: {temp_video_path}")
-            cv2.waitKey(0)  # Wait for any key press
-
-        while cap.isOpened():
-            ret, frame = cap.read()
-            if not ret:
-                break
-            cv2.imshow("Video Player", frame)
-            # Press 'q' to exit the video
-            if cv2.waitKey(200) & 0xFF == ord("q"):
-                break
-        cap.release()
-        cv2.destroyAllWindows()
-
-
 def _resize_frame(frame: np.ndarray) -> np.ndarray:
     height, width = frame.shape[:2]
     new_width = width - (width % 2)
@@ -56,33 +24,32 @@ def _resize_frame(frame: np.ndarray) -> np.ndarray:
 def video_writer(
     frames: List[np.ndarray],
     fps: float = _DEFAULT_INPUT_FPS,
-    filename: Optional[str] = None,
+    file: Optional[IO[bytes]] = None,
 ) -> str:
     if isinstance(fps, str):
         # fps could be a string when it's passed in from a web endpoint deployment
         fps = float(fps)
-    if filename is None:
-        filename = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4").name
-    container = av.open(filename, mode="w")
-    stream = container.add_stream("h264", rate=fps)
-    height, width = frames[0].shape[:2]
-    stream.height = height - (height % 2)
-    stream.width = width - (width % 2)
-    stream.pix_fmt = "yuv420p"
-    stream.options = {"crf": "10"}
-    for frame in frames:
-        # Remove the alpha channel (convert RGBA to RGB)
-        frame_rgb = frame[:, :, :3]
-        # Resize the frame to make dimensions divisible by 2
-        frame_rgb = _resize_frame(frame_rgb)
-        av_frame = av.VideoFrame.from_ndarray(frame_rgb, format="rgb24")
-        for packet in stream.encode(av_frame):
+    if file is None:
+        file = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4")
+    with av.open(file, "w") as container:
+        stream = container.add_stream("h264", rate=fps)
+        height, width = frames[0].shape[:2]
+        stream.height = height - (height % 2)
+        stream.width = width - (width % 2)
+        stream.pix_fmt = "yuv420p"
+        stream.options = {"crf": "10"}
+        for frame in frames:
+            # Remove the alpha channel (convert RGBA to RGB)
+            frame_rgb = frame[:, :, :3]
+            # Resize the frame to make dimensions divisible by 2
+            frame_rgb = _resize_frame(frame_rgb)
+            av_frame = av.VideoFrame.from_ndarray(frame_rgb, format="rgb24")
+            for packet in stream.encode(av_frame):
+                container.mux(packet)
+
+        for packet in stream.encode():
             container.mux(packet)
-
-    for packet in stream.encode():
-        container.mux(packet)
-    container.close()
-    return filename
+    return file.name
 
 
 def frames_to_bytes(
@@ -98,11 +65,10 @@ def frames_to_bytes(
     if isinstance(fps, str):
         # fps could be a string when it's passed in from a web endpoint deployment
         fps = float(fps)
-    with tempfile.NamedTemporaryFile(delete=True, suffix=file_ext) as temp_file:
-        video_writer(frames, fps, temp_file.name)
-
-        with open(temp_file.name, "rb") as f:
-            buffer_bytes = f.read()
+    with tempfile.NamedTemporaryFile(delete=True, suffix=file_ext) as f:
+        video_writer(frames, fps, f)
+        f.seek(0)
+        buffer_bytes = f.read()
     return buffer_bytes
 
 

{vision_agent-0.2.238.dist-info → vision_agent-0.2.240.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: vision-agent
-Version: 0.2.238
+Version: 0.2.240
 Summary: Toolset for Vision Agent
 Author: Landing AI
 Author-email: dev@landing.ai

{vision_agent-0.2.238.dist-info → vision_agent-0.2.240.dist-info}/RECORD

@@ -38,7 +38,7 @@ vision_agent/tools/__init__.py,sha256=T-MPNBVbvWtfo71hobaZsdYzQ52oyymolk_OAb2Pq_
 vision_agent/tools/meta_tools.py,sha256=-heMwGkx0hX_9zUp1dgBqsJpVnl6Y6tErMsjFy0dwLM,28652
 vision_agent/tools/planner_tools.py,sha256=orBTdJQz2NKoLuX9WE6XixaYuG305xz0UBYvZOiuquQ,19474
 vision_agent/tools/prompts.py,sha256=V1z4YJLXZuUl_iZ5rY0M5hHc_2tmMEUKr0WocXKGt4E,1430
-vision_agent/tools/tools.py,sha256=-xg5Msq5ZtHgaISpHnbq5rJ5MIERwfH6wPHg6KpaYjg,111457
+vision_agent/tools/tools.py,sha256=8J-SYpyUeqMDajF7kp2aiTeBBQrJEWGVdEsQLPAc-OM,111511
 vision_agent/utils/__init__.py,sha256=mANUs_84VL-3gpZbXryvV2mWU623eWnRlJCSUHtMjuw,122
 vision_agent/utils/agent.py,sha256=QGKcbzpAjcVj0958bXYLv07-d2i1GU7-bXVG7bTGRMA,14619
 vision_agent/utils/exceptions.py,sha256=booSPSuoULF7OXRr_YbC4dtKt6gM_HyiFQHBuaW86C4,2052
@@ -47,9 +47,9 @@ vision_agent/utils/image_utils.py,sha256=bJM2mEvB6E__M9pxi74yQYzAiZ7mu3KE2ptyVrp
 vision_agent/utils/tools.py,sha256=USZL0MKsiJgqA8RFiYRTcj_Kn2FVYKLHK4wIk0gP1Ow,7694
 vision_agent/utils/tools_doc.py,sha256=yFue6KSXoa_Z1ngCdBEc4SdPZOWF1rVLeaHu02I8Wis,2523
 vision_agent/utils/type_defs.py,sha256=BE12s3JNQy36QvauXHjwyeffVh5enfcvd4vTzSwvEZI,1384
-vision_agent/utils/video.py,sha256=Dt9_pqGgr63gmpurzisnpF6d9tr65-zxS1CccXdVuxk,6458
+vision_agent/utils/video.py,sha256=rjsQ1sKKisaQ6AVjJz0zd_G4g-ovRweS_rs4JEhenoI,5340
 vision_agent/utils/video_tracking.py,sha256=GM9qfeawqhmZVWoKrzw5-NETd4gEo7ImMfWtBnhC3bw,12086
-vision_agent-0.2.238.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-vision_agent-0.2.238.dist-info/METADATA,sha256=VnupHm4Iav889sO4JPGeWYM7902KwPKaJYem81_EDCk,5712
-vision_agent-0.2.238.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
-vision_agent-0.2.238.dist-info/RECORD,,
+vision_agent-0.2.240.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+vision_agent-0.2.240.dist-info/METADATA,sha256=l9FlzNIT3ncQNxkIlTTUsB1aaL-7u2b1OtvYcRv0AIE,5712
+vision_agent-0.2.240.dist-info/WHEEL,sha256=7Z8_27uaHI_UZAc4Uox4PpBhQ9Y5_modZXWMxtUi4NU,88
+vision_agent-0.2.240.dist-info/RECORD,,