vision-agent 1.1.10__tar.gz → 1.1.12__tar.gz

{vision_agent-1.1.10 → vision_agent-1.1.12}/.gitignore

@@ -97,3 +97,4 @@ local/
 vision-agent-benchmark/
 vision_agent/tools/suggestion.py
 vision_agent/agent/visual_design_patterns.py
+*/node_modules

{vision_agent-1.1.10 → vision_agent-1.1.12}/PKG-INFO

@@ -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 → vision_agent-1.1.12}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "vision-agent"
-version = "1.1.10"
+version = "1.1.12"
 description = "Toolset for Vision Agent"
 authors = [{ name = "Landing AI", email = "dev@landing.ai" }]
 requires-python = ">=3.9,<4.0"
@@ -28,7 +28,6 @@ dependencies = [
     "ipykernel>=6.29.4,<7",
     "tenacity>=8.3.0,<9",
     "pillow-heif>=0.16.0,<0.17",
-    "pytube==15.0.0",
     "anthropic>=0.31.0,<0.32",
     "pydantic>=2.0.0,<3",
     "av>=11.0.0,<12",
@@ -39,6 +38,7 @@ dependencies = [
     "dotenv>=0.9.9,<0.10",
     "pymupdf>=1.23.0,<2",
     "google-genai>=1.0.0,<2",
+    "yt-dlp>=2025.3.31",
 ]
 
 [project.urls]

{vision_agent-1.1.10 → vision_agent-1.1.12}/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))