vision-agent 1.1.10__py3-none-any.whl → 1.1.12__py3-none-any.whl

vision_agent/tools/tools.py

@@ -19,7 +19,7 @@ import requests
 from IPython.display import display
 from PIL import Image, ImageDraw, ImageFont
 from pillow_heif import register_heif_opener  # type: ignore
-from pytube import YouTube  # type: ignore
+import yt_dlp  # type: ignore
 import pymupdf  # type: ignore
 from google import genai  # type: ignore
 from google.genai import types  # type: ignore
@@ -3174,7 +3174,6 @@ def extract_frames_and_timestamps(
         [{"frame": np.ndarray, "timestamp": 0.0}, ...]
     """
     if isinstance(fps, str):
-        # fps could be a string when it's passed in from a web endpoint deployment
         fps = float(fps)
 
     def reformat(
@@ -3194,23 +3193,20 @@ def extract_frames_and_timestamps(
         )
     ):
         with tempfile.TemporaryDirectory() as temp_dir:
-            yt = YouTube(str(video_uri))
-            # Download the highest resolution video
-            video = (
-                yt.streams.filter(progressive=True, file_extension="mp4")
-                .order_by("resolution")
-                .desc()
-                .first()
-            )
-            if not video:
-                raise Exception("No suitable video stream found")
-            video_file_path = video.download(output_path=temp_dir)
+            ydl_opts = {
+                "outtmpl": os.path.join(temp_dir, "%(title)s.%(ext)s"),
+                "format": "bestvideo[ext=mp4]+bestaudio[ext=m4a]/best[ext=mp4]/best",
+                "quiet": True,
+            }
+            with yt_dlp.YoutubeDL(ydl_opts) as ydl:
+                info = ydl.extract_info(str(video_uri), download=True)
+                video_file_path = ydl.prepare_filename(info)
 
             return reformat(extract_frames_from_video(video_file_path, fps))
+
     elif str(video_uri).startswith(("http", "https")):
         _, image_suffix = os.path.splitext(video_uri)
         with tempfile.NamedTemporaryFile(delete=False, suffix=image_suffix) as tmp_file:
-            # Download the video and save it to the temporary file
             with urllib.request.urlopen(str(video_uri)) as response:
                 tmp_file.write(response.read())
             return reformat(extract_frames_from_video(tmp_file.name, fps))

{vision_agent-1.1.10.dist-info → vision_agent-1.1.12.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: vision-agent
-Version: 1.1.10
+Version: 1.1.12
 Summary: Toolset for Vision Agent
 Project-URL: Homepage, https://landing.ai
 Project-URL: repository, https://github.com/landing-ai/vision-agent
@@ -28,7 +28,6 @@ Requires-Dist: pillow-heif<0.17,>=0.16.0
 Requires-Dist: pillow==10.*
 Requires-Dist: pydantic<3,>=2.0.0
 Requires-Dist: pymupdf<2,>=1.23.0
-Requires-Dist: pytube==15.0.0
 Requires-Dist: requests==2.*
 Requires-Dist: rich<14,>=13.7.1
 Requires-Dist: scikit-learn<2,>=1.5.2
@@ -37,6 +36,7 @@ Requires-Dist: tabulate<0.10,>=0.9.0
 Requires-Dist: tenacity<9,>=8.3.0
 Requires-Dist: tqdm<5.0.0,>=4.64.0
 Requires-Dist: typing-extensions==4.*
+Requires-Dist: yt-dlp>=2025.3.31
 Description-Content-Type: text/markdown
 
 <div align="center">

{vision_agent-1.1.10.dist-info → vision_agent-1.1.12.dist-info}/RECORD

@@ -30,7 +30,7 @@ vision_agent/tools/__init__.py,sha256=PRUka2eqHwPWJxwfpLj-O2Ab7hXG_dsE1Aov3TE6te
 vision_agent/tools/meta_tools.py,sha256=9iJilpGYEiXW0nYPTYAWHa7l23wGN8IM5KbE7mWDOT0,6798
 vision_agent/tools/planner_tools.py,sha256=iQWtTgXdomn0IWrbmvXXM-y8Q_RSEOxyP04HIRLrgWI,19576
 vision_agent/tools/prompts.py,sha256=V1z4YJLXZuUl_iZ5rY0M5hHc_2tmMEUKr0WocXKGt4E,1430
-vision_agent/tools/tools.py,sha256=6YlqUA2OWlCh4Vqp8vZztW2xdSeQTagR5jx44JCS4-o,131052
+vision_agent/tools/tools.py,sha256=K2QW-0esNBygtlRcyUwVRRhV1FV5UaznHYhFkFZkL-I,130851
 vision_agent/utils/__init__.py,sha256=mANUs_84VL-3gpZbXryvV2mWU623eWnRlJCSUHtMjuw,122
 vision_agent/utils/agent.py,sha256=2ifTP5QElItnr4YHOJR6L5P1PUzV0GhChTTqVxuVyQg,15153
 vision_agent/utils/exceptions.py,sha256=zis8smCbdEylBVZBTVfEUfAh7Rb7cWV3MSPambu6FsQ,1837
@@ -40,7 +40,7 @@ vision_agent/utils/tools.py,sha256=Days0dETPRQLSDamMKPnXFsc5g5IKX9QJcPPNmSHNdM,8
 vision_agent/utils/tools_doc.py,sha256=PKcXXbJktiuPi9q6Q1zXzFx24Dh229SNgWBDtZ2fQSQ,2730
 vision_agent/utils/video.py,sha256=rjsQ1sKKisaQ6AVjJz0zd_G4g-ovRweS_rs4JEhenoI,5340
 vision_agent/utils/video_tracking.py,sha256=DZLFpNCuzuPJQzbQoVNcp-m4dKxgiKdCNM5QTh_zURE,12245
-vision_agent-1.1.10.dist-info/METADATA,sha256=kPMiMa4Nc4-OP54nToXy0qRIS_bTfPdTnoCBZDD_I3I,12670
-vision_agent-1.1.10.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-vision_agent-1.1.10.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-vision_agent-1.1.10.dist-info/RECORD,,
+vision_agent-1.1.12.dist-info/METADATA,sha256=Yj5WJJN4ZoeZJ6o0_m7xJf8jgpaQVQ3jJGp8Tb0BsqI,12673
+vision_agent-1.1.12.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+vision_agent-1.1.12.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+vision_agent-1.1.12.dist-info/RECORD,,