Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
@@ -9,6 +9,7 @@ from PIL import Image, ImageDraw, ImageFont
|
|
9 |
import ffmpeg
|
10 |
import textwrap
|
11 |
import random
|
|
|
12 |
|
13 |
# OpenAI LLM
|
14 |
llm = ChatOpenAI(model="gpt-3.5-turbo", temperature=0.3)
|
@@ -29,31 +30,23 @@ def extract_main_content(url):
|
|
29 |
paras = [p.get_text() for p in soup.find_all("p") if len(p.get_text()) > 60]
|
30 |
return "\n".join(paras[:20]) or None
|
31 |
|
32 |
-
#
|
33 |
-
|
34 |
-
|
35 |
-
|
36 |
-
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
43 |
-
|
44 |
-
return Image.open(emoji_path).resize((48, 48)).convert("RGBA")
|
45 |
-
else:
|
46 |
-
fallback = Image.new("RGBA", (48, 48), (255, 0, 0, 0))
|
47 |
-
draw = ImageDraw.Draw(fallback)
|
48 |
-
draw.ellipse((4, 4, 44, 44), fill=(255, 69, 0, 255))
|
49 |
-
return fallback
|
50 |
-
|
51 |
-
# Create image slides from text chunks with emojis and logo
|
52 |
def create_slides(text, duration, output_folder, max_lines=6):
|
53 |
-
font_path = "/usr/share/fonts/truetype/dejavu/DejaVuSans.ttf"
|
54 |
font = ImageFont.truetype(font_path, 48)
|
55 |
-
logo_path =
|
56 |
-
|
57 |
|
58 |
chunks = textwrap.wrap(text, width=36)
|
59 |
slides = ["\n".join(chunks[i:i+max_lines]) for i in range(0, len(chunks), max_lines)]
|
@@ -61,9 +54,8 @@ def create_slides(text, duration, output_folder, max_lines=6):
|
|
61 |
slide_paths = []
|
62 |
|
63 |
for i, slide_text in enumerate(slides):
|
64 |
-
|
65 |
-
draw = ImageDraw.Draw(
|
66 |
-
|
67 |
lines = slide_text.split("\n")
|
68 |
total_height = sum([font.getbbox(line)[3] - font.getbbox(line)[1] for line in lines]) + (len(lines)-1)*20
|
69 |
y = max((720 - total_height) // 2, 20)
|
@@ -72,18 +64,16 @@ def create_slides(text, duration, output_folder, max_lines=6):
|
|
72 |
w = font.getbbox(line)[2] - font.getbbox(line)[0]
|
73 |
text_x = (1280 - w) // 2
|
74 |
draw.text((text_x, y), line, font=font, fill="white")
|
75 |
-
img.paste(emoji_img, (text_x - 60, y), emoji_img)
|
76 |
-
img.paste(emoji_img, (text_x + w + 12, y), emoji_img)
|
77 |
y += font.getbbox(line)[3] - font.getbbox(line)[1] + 20
|
78 |
|
79 |
logo = Image.open(logo_path).convert("RGBA")
|
80 |
-
logo_width = min(180, int(0.15 *
|
81 |
logo_height = int(logo.size[1] * (logo_width / logo.size[0]))
|
82 |
logo = logo.resize((logo_width, logo_height))
|
83 |
-
|
84 |
|
85 |
frame_path = os.path.join(output_folder, f"slide_{i}.png")
|
86 |
-
|
87 |
slide_paths.append((frame_path, per_slide_time))
|
88 |
|
89 |
return slide_paths
|
@@ -134,8 +124,8 @@ iface = gr.Interface(
|
|
134 |
gr.Textbox(label="Summary"),
|
135 |
gr.Video(label="Generated AV Summary")
|
136 |
],
|
137 |
-
title="🎞️ AV Summary Generator (
|
138 |
-
description="Generates a 5/10 sec video summary from article URL with clean text,
|
139 |
)
|
140 |
|
141 |
if __name__ == '__main__':
|
|
|
9 |
import ffmpeg
|
10 |
import textwrap
|
11 |
import random
|
12 |
+
from urllib.request import urlretrieve
|
13 |
|
14 |
# OpenAI LLM
|
15 |
llm = ChatOpenAI(model="gpt-3.5-turbo", temperature=0.3)
|
|
|
30 |
paras = [p.get_text() for p in soup.find_all("p") if len(p.get_text()) > 60]
|
31 |
return "\n".join(paras[:20]) or None
|
32 |
|
33 |
+
# Load external image assets like logo or graphics from URL
|
34 |
+
ASSETS = {
|
35 |
+
"logo": "CSHARP logo.png", # Free C# Corner-style logo
|
36 |
+
"bg": "https://img.freepik.com/free-photo/blue-abstract-gradient-wave-wallpaper_53876-102605.jpg"
|
37 |
+
}
|
38 |
+
|
39 |
+
def download_asset(url):
|
40 |
+
local_path = tempfile.NamedTemporaryFile(delete=False, suffix=".png").name
|
41 |
+
urlretrieve(url, local_path)
|
42 |
+
return local_path
|
43 |
+
|
44 |
+
# Create image slides from text chunks
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
45 |
def create_slides(text, duration, output_folder, max_lines=6):
|
46 |
+
font_path = "/usr/share/fonts/truetype/dejavu/DejaVuSans-Bold.ttf"
|
47 |
font = ImageFont.truetype(font_path, 48)
|
48 |
+
logo_path = download_asset(ASSETS["logo"])
|
49 |
+
bg_path = download_asset(ASSETS["bg"])
|
50 |
|
51 |
chunks = textwrap.wrap(text, width=36)
|
52 |
slides = ["\n".join(chunks[i:i+max_lines]) for i in range(0, len(chunks), max_lines)]
|
|
|
54 |
slide_paths = []
|
55 |
|
56 |
for i, slide_text in enumerate(slides):
|
57 |
+
bg = Image.open(bg_path).resize((1280, 720)).convert("RGBA")
|
58 |
+
draw = ImageDraw.Draw(bg)
|
|
|
59 |
lines = slide_text.split("\n")
|
60 |
total_height = sum([font.getbbox(line)[3] - font.getbbox(line)[1] for line in lines]) + (len(lines)-1)*20
|
61 |
y = max((720 - total_height) // 2, 20)
|
|
|
64 |
w = font.getbbox(line)[2] - font.getbbox(line)[0]
|
65 |
text_x = (1280 - w) // 2
|
66 |
draw.text((text_x, y), line, font=font, fill="white")
|
|
|
|
|
67 |
y += font.getbbox(line)[3] - font.getbbox(line)[1] + 20
|
68 |
|
69 |
logo = Image.open(logo_path).convert("RGBA")
|
70 |
+
logo_width = min(180, int(0.15 * bg.width))
|
71 |
logo_height = int(logo.size[1] * (logo_width / logo.size[0]))
|
72 |
logo = logo.resize((logo_width, logo_height))
|
73 |
+
bg.paste(logo, (bg.width - logo_width - 30, bg.height - logo_height - 30), logo)
|
74 |
|
75 |
frame_path = os.path.join(output_folder, f"slide_{i}.png")
|
76 |
+
bg.convert("RGB").save(frame_path)
|
77 |
slide_paths.append((frame_path, per_slide_time))
|
78 |
|
79 |
return slide_paths
|
|
|
124 |
gr.Textbox(label="Summary"),
|
125 |
gr.Video(label="Generated AV Summary")
|
126 |
],
|
127 |
+
title="🎞️ AV Summary Generator (Enhanced Slides with Background & Logo)",
|
128 |
+
description="Generates a 5/10 sec video summary from article URL with clean text, background visuals, logo, and better visuals."
|
129 |
)
|
130 |
|
131 |
if __name__ == '__main__':
|