Update app.py

app.py

@@ -1,5 +1,5 @@
 import gradio as gr
-import tempfile, requests, os
+import tempfile, requests, os, subprocess
 from langchain.chains import LLMChain
 from langchain.prompts import PromptTemplate
 from langchain.chat_models import ChatOpenAI
@@ -8,7 +8,6 @@ from bs4 import BeautifulSoup
 from PIL import Image, ImageDraw, ImageFont
 import ffmpeg
 import textwrap
-import subprocess
 
 # OpenAI LLM
 llm = ChatOpenAI(model="gpt-3.5-turbo", temperature=0.3)
@@ -40,7 +39,7 @@ def get_uploaded_logo():
 # Create image slides from text chunks
 def create_slides(text, duration, output_folder, max_lines=6):
     font_path = "/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf"
-    font = ImageFont.truetype(font_path, 48)
+    font = ImageFont.truetype(font_path, 64)
     logo_path = get_uploaded_logo()
 
     chunks = textwrap.wrap(text, width=40)
@@ -85,18 +84,26 @@ def url_to_av_summary(url, duration):
     frame_dir = tempfile.mkdtemp()
     slides = create_slides(summary, duration, frame_dir)
 
-    slide_inputs = []
-    for path, t in slides:
-        slide_inputs.extend(["-t", str(t), "-loop", "1", "-i", path])
+    concat_txt_path = os.path.join(frame_dir, "slides.txt")
+    with open(concat_txt_path, "w") as f:
+        for path, t in slides:
+            f.write(f"file '{path}'\n")
+            f.write(f"duration {t}\n")
+        f.write(f"file '{slides[-1][0]}'\n")
 
-    filter_complex = "".join([f"[{i}:v]" for i in range(len(slides))]) + f"concat=n={len(slides)}:v=1[outv]"
     concat_img = os.path.join(frame_dir, "video_input.mp4")
-
-    cmd = ["ffmpeg", "-y"] + slide_inputs + ["-filter_complex", filter_complex, "-map", "[outv]", concat_img]
-    subprocess.run(cmd, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
+    subprocess.run([
+        "ffmpeg", "-y", "-f", "concat", "-safe", "0", "-i", concat_txt_path,
+        "-vsync", "vfr", "-pix_fmt", "yuv420p", concat_img
+    ], stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL)
 
     final_video = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4").name
-    ffmpeg.input(concat_img).output(audio_path, final_video, vcodec='libx264', acodec='aac', pix_fmt='yuv420p', shortest=None).run(overwrite_output=True, quiet=True)
+    video_input = ffmpeg.input(concat_img)
+    audio_input = ffmpeg.input(audio_path)
+
+    ffmpeg.output(video_input, audio_input, final_video,
+                  vcodec='libx264', acodec='aac', pix_fmt='yuv420p', shortest=None
+    ).run(overwrite_output=True, quiet=True)
 
     return summary, final_video