Update app.py

app.py

@@ -52,11 +52,13 @@ def create_slides(text, duration, output_folder, max_lines=6):
         draw = ImageDraw.Draw(img)
 
         lines = slide_text.split("\n")
-        total_height = sum([draw.textbbox((0, 0), line, font=font)[3] for line in lines]) + (len(lines)-1)*10
+        total_height = sum([draw.textbbox((0, 0), line, font=font)[3] - draw.textbbox((0, 0), line, font=font)[1] for line in lines]) + (len(lines)-1)*10
         y = (720 - total_height) // 2
 
         for line in lines:
-            w, h = draw.textbbox((0, 0), line, font=font)[2:]
+            bbox = draw.textbbox((0, 0), line, font=font)
+            w = bbox[2] - bbox[0]
+            h = bbox[3] - bbox[1]
             draw.text(((1280 - w) // 2, y), line, font=font, fill="white")
             y += h + 10
 
@@ -82,27 +84,20 @@ def url_to_av_summary(url, duration):
     frame_dir = tempfile.mkdtemp()
     slides = create_slides(summary, duration, frame_dir)
 
-    concat_file = os.path.join(frame_dir, "frames.txt")
-    with open(concat_file, "w") as f:
-        for path, t in slides:
-            f.write(f"file '{path}'\n")
-            f.write(f"duration {t}\n")
+    slide_inputs = []
+    for path, t in slides:
+        slide_inputs.extend(["-t", str(t), "-loop", "1", "-i", path])
 
+    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")
-    ffmpeg.input(concat_file, format='concat', safe=0).output(
-        concat_img,
-        r=1,
-        vcodec='libx264', pix_fmt='yuv420p'
-    ).run(overwrite_output=True, quiet=True)
-
-    video_path = tempfile.NamedTemporaryFile(delete=False, suffix=".mp4").name
-    ffmpeg.input(concat_img).output(
-        video_path,
-        i=audio_path,
-        vcodec='libx264', acodec='aac', pix_fmt='yuv420p', shortest=None
-    ).run(overwrite_output=True, quiet=True)
-
-    return summary, video_path
+
+    cmd = ["ffmpeg", "-y"] + slide_inputs + ["-filter_complex", filter_complex, "-map", "[outv]", concat_img]
+    subprocess.run(cmd, 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)
+
+    return summary, final_video
 
 iface = gr.Interface(
     fn=url_to_av_summary,