Adding new video actions, not quite working the way I like though

Author: cxbxmxcx

logs/logs.txt

@@ -1,13 +1,11 @@
 Running on local URL:  http://127.0.0.1:7860
 
 To create a public link, set `share=True` in `launch()`.
-action: create_image(args={"prompt":"a scary monster surrounded by happy children carrying a sign with a positive affirmation","model":"dall-e-3","size":"1024x1024","quality":"standard"}) -> a_scary_monster_surrounded_by_happy_children_carry.png
-assistant > Moviepy - Building video assistants_working_folder\monster_zoom.mp4.
-Moviepy - Writing video assistants_working_folder\monster_zoom.mp4
+assistant > action: create_image(args={"prompt":"A minimalistic office space with neutral colors, a simple desk, chair, and computer. The room is tidy and devoid of any personal touches or decorations. Lighting is plain and soft.","model":"dall-e-3","size":"1024x1024"}) -> A_minimalistic_office_space_with_neutral_colors_a_.png
+assistant > Moviepy - Building video assistants_working_folder\boring_office_with_text_effect.mp4.
+Moviepy - Writing video assistants_working_folder\boring_office_with_text_effect.mp4
 
 Moviepy - Done !
-Moviepy - video ready assistants_working_folder\monster_zoom.mp4
-action: zoom_to(args={"image_filename":"a_scary_monster_surrounded_by_happy_children_carry.png","output_filename":"monster_zoom.mp4","duration":"10","zoom_factor_from":"2.0","zoom_factor_to":"1.0","zoom_start_position":"(0.5, 0.2)","zoom_end_position":"(0.5, 0.5)"}) -> monster_zoom.mp4
-assistant > MoviePy - Building file assistants_working_folder\monster_zoom.gif with imageio.
-action: create_gif_from_clip(args={"clip_filename":"monster_zoom.mp4","size":"512x512","fps":"15"}) -> GIF created: monster_zoom.gif
+Moviepy - video ready assistants_working_folder\boring_office_with_text_effect.mp4
+action: add_text_effect_to_image(args={"image_filename":"A_minimalistic_office_space_with_neutral_colors_a_.png","output_filename":"boring_office_with_text_effect.mp4","text":"Boring","text_position":"center","text_effect":"cascade"}) -> boring_office_with_text_effect.mp4
 assistant > 

playground/assistant_actions/video_actions.py

@@ -1,9 +1,128 @@
 import cv2
 import re
-from moviepy.editor import VideoClip, concatenate_videoclips, VideoFileClip
+from moviepy.editor import (
+    VideoClip,
+    concatenate_videoclips,
+    VideoFileClip,
+    ImageClip,
+    TextClip,
+    CompositeVideoClip,
+)
 from playground.actions_manager import agent_action
 import os
 from playground.global_values import GlobalValues
+import numpy as np
+from moviepy.video.tools.segmenting import findObjects
+
+# Helper function for rotation matrix
+rotMatrix = lambda a: np.array([[np.cos(a), np.sin(a)], [-np.sin(a), np.cos(a)]])
+
+
+# Define movement functions for text effect
+def vortex(screenpos, i, nletters):
+    d = lambda t: 1.0 / (0.3 + t**8)  # damping
+    a = i * np.pi / nletters  # angle of the movement
+    v = rotMatrix(a).dot([-1, 0])
+    if i % 2:
+        v[1] = -v[1]
+    return lambda t: screenpos + 400 * d(t) * rotMatrix(0.5 * d(t) * a).dot(v)
+
+
+def cascade(screenpos, i, nletters):
+    v = np.array([0, -1])
+    d = lambda t: 1 if t < 0 else abs(np.sinc(t) / (1 + t**4))
+    return lambda t: screenpos + v * 400 * d(t - 0.15 * i)
+
+
+def arrive(screenpos, i, nletters):
+    v = np.array([-1, 0])
+    d = lambda t: max(0, 3 - 3 * t)
+    return lambda t: screenpos - 400 * v * d(t - 0.2 * i)
+
+
+def vortexout(screenpos, i, nletters):
+    d = lambda t: max(0, t)  # damping
+    a = i * np.pi / nletters  # angle of the movement
+    v = rotMatrix(a).dot([-1, 0])
+    if i % 2:
+        v[1] = -v[1]
+    return lambda t: screenpos + 400 * d(t - 0.1 * i) * rotMatrix(-0.2 * d(t) * a).dot(
+        v
+    )
+
+
+@agent_action
+def add_text_effect_to_image(
+    image_filename, output_filename, text, text_position="center", text_effect="vortex"
+):
+    """
+    Adds a text effect to an image and creates a video.
+
+    This function takes an image and overlays a text effect onto it, creating a video
+    with a specified duration. The text effect includes animations such as vortex, cascade,
+    arrive, and vortexout. The position of the text can also be specified.
+
+    Parameters:
+    image_filename (str): The filename of the input image.
+    output_filename (str): The filename of the output video.
+    text (str): The text to be displayed with effects.
+    text_position (str): The position of the text in the video ('bottom', 'center', 'top').
+    text_effect (str): The text effect to apply ('vortexout', 'arrive', 'cascade', 'vortex').
+
+    Returns:
+    str: The filename of the output video.
+    """
+    # Load the image
+    image_path = os.path.join(GlobalValues.ASSISTANTS_WORKING_FOLDER, image_filename)
+    image_clip = ImageClip(image_path)
+    screensize = image_clip.size
+
+    # Create the text clip
+    txt_clip = TextClip(text, color="white", font="Amiri-Bold", kerning=5, fontsize=100)
+    cvc = CompositeVideoClip([txt_clip.set_pos("center")], size=screensize)
+
+    # Locate and separate each letter
+    letters = findObjects(cvc)  # a list of ImageClips
+
+    # Function to move letters
+    def moveLetters(letters, funcpos):
+        return [
+            letter.set_pos(funcpos(letter.screenpos, i, len(letters)))
+            for i, letter in enumerate(letters)
+        ]
+
+    # Select the text effect function and set duration based on the effect
+    effect_function = {
+        "vortex": (vortex, 5),
+        "cascade": (cascade, 5),
+        "arrive": (arrive, 5),
+        "vortexout": (vortexout, 5),
+    }.get(text_effect, (vortex, 5))
+
+    effect_func, duration = effect_function
+
+    # Create a clip with the selected text effect
+    effect_clip = CompositeVideoClip(
+        moveLetters(letters, effect_func), size=screensize
+    ).set_duration(duration)
+
+    # Position the text
+    position_dict = {
+        "bottom": ("center", "bottom"),
+        "center": ("center", "center"),
+        "top": ("center", "top"),
+    }
+    text_position = position_dict.get(text_position, ("center", "center"))
+
+    # Overlay the text effect on the image
+    final_clip = CompositeVideoClip(
+        [image_clip.set_duration(duration), effect_clip.set_pos(text_position)]
+    )
+
+    # Write the result to a file
+    output_path = os.path.join(GlobalValues.ASSISTANTS_WORKING_FOLDER, output_filename)
+    final_clip.write_videofile(output_path, fps=25)
+    return output_filename
 
 
 def safe_eval(expr):

playground/assistant_actions/youtube_search.py

@@ -8,16 +8,35 @@
 
 
 class YoutubeSearch:
-    def __init__(self, search_terms: str, max_results=None):
+    def __init__(self, search_terms: str, max_results=None, publish_time="today"):
         self.search_terms = search_terms
         self.max_results = int(max_results)
+        self.filter = self._get_filter(publish_time)
         self.videos = self._search()
 
+        # Upload Date:
+        # EgQIAhAB: Today
+        # EgQIAxAB: This week
+        # EgQIBBAB: This month
+        # EgQIBRAB: This year
+
+    def _get_filter(self, publish_time):
+        if publish_time == "today":
+            return "EgQIAhAB"
+        elif publish_time == "this week":
+            return "EgQIAxAB"
+        elif publish_time == "this month":
+            return "EgQIBBAB"
+        elif publish_time == "this year":
+            return "EgQIBRAB"
+        else:
+            return "EgQIAhAB"
+
     def _search(self):
         encoded_search = urllib.parse.quote_plus(self.search_terms)
         BASE_URL = "https://youtube.com"
         # This filter shows only videos uploaded in the last hour, sorted by relevance
-        url = f"{BASE_URL}/results?search_query={encoded_search}&sp=EgQIARAB"
+        url = f"{BASE_URL}/results?search_query={encoded_search}&sp={self.filter}"
         response = requests.get(url).text
         while "ytInitialData" not in response:
             response = requests.get(url).text
@@ -97,9 +116,32 @@ def to_json(self, clear_cache=True):
 
 
 @agent_action
-def search_youtube_videos(query: str, max_results=10):
-    """Searches for videos on YouTube based on the query string and returns the video titles and IDs."""
-    results = YoutubeSearch(query, max_results=max_results)
+def search_youtube_videos(query: str, max_results=10, publish_time="this year"):
+    """
+    Search for YouTube videos based on a query and filter by publish time.
+
+    This function uses the YouTube Search API to find videos matching the specified query.
+    The results can be limited by the maximum number of results and filtered by the
+    publish time (e.g., today, this week, this month, this year).
+
+    Args:
+        query (str): The search query string.
+        max_results (int, optional): The maximum number of search results to return. Default is 10.
+        publish_time (str, optional): The time filter for the search results. Options include "today",
+                                      "this week", "this month", and "this year". Default is "today".
+
+    Returns:
+        list: A list of dictionaries containing video titles and their corresponding IDs.
+              Each dictionary has the following keys:
+              - "title" (str): The title of the video.
+              - "id" (str): The YouTube ID of the video.
+
+    Example:
+        >>> search_youtube_videos("create a YouTube channel", max_results=5, publish_time="this week")
+        [{'title': 'How to Create a YouTube Channel in 2024 (Step-by-Step Tutorial)', 'id': 'abcd1234'},
+         {'title': 'Creating a YouTube Channel: The Ultimate Guide', 'id': 'efgh5678'}]
+    """
+    results = YoutubeSearch(query, max_results=max_results, publish_time=publish_time)
     videos = results.videos
     return [{"title": video["title"], "id": video["id"]} for video in videos]
 

playground/assistants_utils.py

@@ -176,4 +176,4 @@ def submit_tool_outputs(self, tool_outputs, run_id):
                 for text in stream.text_deltas:
                     self.output_queue.put(("text", text))
         except Exception as e:
-            self.output_queue.put(("text", str(e)))
+            self.output_queue.put(("text", f"Error in tool outputs: {str(e)}"))

requirements.txt

@@ -10,3 +10,4 @@ python-pptx
 markdown
 moviepy
 opencv-python
+scipy