Spaces:
Sleeping
Sleeping
Update app.py
Browse files
app.py
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
def format_scene_breakdown(scenes):
|
| 2 |
rows = """
|
| 3 |
-
<table style='width:100%; border-collapse: collapse; background-color:#1a1a1a; color: #FFFFFF; border: 2px solid #FF8C00; font-size:
|
| 4 |
<tr style='background-color:#FF8C00; color: #000000;'>
|
| 5 |
-
<th style='padding: 8px; border: 1px solid #FF8C00;color: #000000;'
|
| 6 |
-
<th style='padding: 8px; border: 1px solid #FF8C00;color: #000000;'
|
| 7 |
</tr>
|
| 8 |
"""
|
| 9 |
pattern = re.compile(r"\*\*\[(.*?)\]\*\*:\s*(.*)")
|
|
@@ -16,8 +16,8 @@ def format_scene_breakdown(scenes):
|
|
| 16 |
description = match.group(2).strip()
|
| 17 |
rows += f"""
|
| 18 |
<tr style='background-color:#1a1a1a;'>
|
| 19 |
-
<td style='padding: 8px; border: 1px solid #444; color: #87CEEB; font-weight: bold;font-size:
|
| 20 |
-
<td style='padding: 8px; border: 1px solid #444; color: #87CEEB; font-weight: bold;font-size:
|
| 21 |
</tr>
|
| 22 |
"""
|
| 23 |
|
|
@@ -40,154 +40,167 @@ from xhtml2pdf import pisa
|
|
| 40 |
from io import BytesIO
|
| 41 |
|
| 42 |
|
| 43 |
-
cached_reports = {}
|
| 44 |
-
|
| 45 |
def generate_pdf_from_html(html_content):
|
| 46 |
-
"""Generate
|
| 47 |
try:
|
| 48 |
-
|
| 49 |
-
|
| 50 |
-
<html>
|
| 51 |
-
<head>
|
| 52 |
-
<meta charset='UTF-8'>
|
| 53 |
-
<style>
|
| 54 |
-
@page {{
|
| 55 |
-
size: A4;
|
| 56 |
-
margin: 0.7cm;
|
| 57 |
-
}}
|
| 58 |
-
body {{
|
| 59 |
-
font-family: 'Segoe UI', sans-serif;
|
| 60 |
-
font-size: 9px;
|
| 61 |
-
line-height: 1.3;
|
| 62 |
-
color: #000;
|
| 63 |
-
background-color: #fff;
|
| 64 |
-
}}
|
| 65 |
-
table {{
|
| 66 |
-
width: 100%;
|
| 67 |
-
border-collapse: collapse;
|
| 68 |
-
font-size: 9px;
|
| 69 |
-
margin-bottom: 10px;
|
| 70 |
-
}}
|
| 71 |
-
th, td {{
|
| 72 |
-
border: 1px solid #ccc;
|
| 73 |
-
padding: 2px;
|
| 74 |
-
vertical-align: top;
|
| 75 |
-
line-height: 1.1;
|
| 76 |
-
}}
|
| 77 |
-
h1, h2, h3 {{
|
| 78 |
-
font-size: 11px;
|
| 79 |
-
background-color: #D3D3D3;
|
| 80 |
-
color: #000;
|
| 81 |
-
padding: 4px;
|
| 82 |
-
margin: 4px 0;
|
| 83 |
-
}}
|
| 84 |
-
</style>
|
| 85 |
-
</head>
|
| 86 |
-
<body>
|
| 87 |
-
{html_content}
|
| 88 |
-
</body>
|
| 89 |
-
</html>
|
| 90 |
-
"""
|
| 91 |
-
|
| 92 |
-
result = BytesIO()
|
| 93 |
-
pisa_status = pisa.CreatePDF(pdf_html, dest=result)
|
| 94 |
-
if pisa_status.err:
|
| 95 |
-
print("❌ Pisa PDF creation error")
|
| 96 |
-
return None
|
| 97 |
-
result.seek(0)
|
| 98 |
-
return result
|
| 99 |
-
|
| 100 |
-
except Exception as e:
|
| 101 |
-
print(f"❌ PDF generation exception: {e}")
|
| 102 |
-
return None
|
| 103 |
-
|
| 104 |
-
def generate_pdf_from_html_debug(html_content):
|
| 105 |
-
"""Debug version to identify PDF generation issues"""
|
| 106 |
-
try:
|
| 107 |
-
print("🐛 Starting PDF generation...")
|
| 108 |
-
print(f"🐛 HTML content length: {len(html_content)} characters")
|
| 109 |
-
print(f"🐛 HTML preview: {html_content[:200]}...")
|
| 110 |
-
|
| 111 |
-
# Check if we have the required imports
|
| 112 |
-
try:
|
| 113 |
-
from xhtml2pdf import pisa
|
| 114 |
-
print("✅ pisa import successful")
|
| 115 |
-
except ImportError as e:
|
| 116 |
-
print(f"❌ pisa import failed: {e}")
|
| 117 |
-
return None
|
| 118 |
-
|
| 119 |
-
try:
|
| 120 |
-
import re
|
| 121 |
-
print("✅ re import successful")
|
| 122 |
-
except ImportError as e:
|
| 123 |
-
print(f"❌ re import failed: {e}")
|
| 124 |
-
return None
|
| 125 |
-
|
| 126 |
-
# Simple HTML cleanup (minimal for debugging)
|
| 127 |
simplified_html = html_content.replace(
|
| 128 |
"background: linear-gradient(135deg, #2d3748, #1a202c);",
|
| 129 |
-
"background-color: #
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 130 |
).replace(
|
| 131 |
"color: #FFFFFF;",
|
| 132 |
"color: #000000;"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 133 |
)
|
| 134 |
|
| 135 |
-
|
|
|
|
|
|
|
| 136 |
|
| 137 |
-
#
|
| 138 |
pdf_html = f"""
|
| 139 |
<!DOCTYPE html>
|
| 140 |
<html>
|
| 141 |
<head>
|
| 142 |
<meta charset="UTF-8">
|
| 143 |
<style>
|
|
|
|
|
|
|
|
|
|
|
|
|
| 144 |
body {{
|
| 145 |
font-family: Arial, sans-serif;
|
| 146 |
font-size: 12px;
|
|
|
|
| 147 |
color: #000000;
|
| 148 |
background-color: #ffffff;
|
| 149 |
}}
|
| 150 |
-
|
| 151 |
-
|
| 152 |
-
|
| 153 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 154 |
</style>
|
| 155 |
</head>
|
| 156 |
<body>
|
| 157 |
-
|
|
|
|
|
|
|
| 158 |
</body>
|
| 159 |
</html>
|
| 160 |
"""
|
| 161 |
|
| 162 |
-
print("🐛 PDF HTML template created")
|
| 163 |
-
|
| 164 |
-
from io import BytesIO
|
| 165 |
result = BytesIO()
|
| 166 |
-
|
| 167 |
-
print("🐛 Creating PDF with pisa...")
|
| 168 |
pisa_status = pisa.CreatePDF(pdf_html, dest=result)
|
| 169 |
-
|
| 170 |
-
print(f"🐛 Pisa status - err: {pisa_status.err}")
|
| 171 |
-
print(f"🐛 PDF buffer size: {len(result.getvalue())} bytes")
|
| 172 |
|
| 173 |
if pisa_status.err:
|
| 174 |
-
print(f"
|
| 175 |
-
return None
|
| 176 |
-
|
| 177 |
-
if len(result.getvalue()) == 0:
|
| 178 |
-
print("❌ PDF buffer is empty")
|
| 179 |
return None
|
| 180 |
|
| 181 |
result.seek(0)
|
| 182 |
-
print("✅ PDF generation successful!")
|
| 183 |
return result
|
| 184 |
|
| 185 |
except Exception as e:
|
| 186 |
-
print(f"
|
| 187 |
-
import traceback
|
| 188 |
-
traceback.print_exc()
|
| 189 |
return None
|
| 190 |
-
|
| 191 |
class YouTubeDownloader:
|
| 192 |
def __init__(self):
|
| 193 |
self.download_dir = tempfile.mkdtemp()
|
|
@@ -241,7 +254,7 @@ class YouTubeDownloader:
|
|
| 241 |
|
| 242 |
# Create enhanced prompt for Gemini
|
| 243 |
prompt = f"""
|
| 244 |
-
Analyze this YouTube video and create a detailed, scene-by-scene breakdown with precise timestamps and specific descriptions:
|
| 245 |
|
| 246 |
Title: {title}
|
| 247 |
Duration: {duration} seconds
|
|
@@ -430,19 +443,19 @@ class YouTubeDownloader:
|
|
| 430 |
text = (title + " " + description).lower()
|
| 431 |
|
| 432 |
if any(word in text for word in ['music', 'song', 'album', 'artist', 'band', 'lyrics']):
|
| 433 |
-
return "Music Video"
|
| 434 |
elif any(word in text for word in ['tutorial', 'how to', 'guide', 'learn', 'teaching']):
|
| 435 |
-
return "Tutorial/Educational"
|
| 436 |
elif any(word in text for word in ['funny', 'comedy', 'entertainment', 'vlog', 'challenge']):
|
| 437 |
-
return "Entertainment/Comedy"
|
| 438 |
elif any(word in text for word in ['news', 'breaking', 'report', 'update']):
|
| 439 |
-
return "News/Information"
|
| 440 |
elif any(word in text for word in ['review', 'unboxing', 'test', 'comparison']):
|
| 441 |
-
return "Review/Unboxing
|
| 442 |
elif any(word in text for word in ['commercial', 'ad', 'brand', 'product']):
|
| 443 |
-
return "Commercial/Advertisement"
|
| 444 |
else:
|
| 445 |
-
return "General Content"
|
| 446 |
|
| 447 |
def detect_background_music(self, video_info):
|
| 448 |
"""Detect background music style"""
|
|
@@ -450,15 +463,15 @@ class YouTubeDownloader:
|
|
| 450 |
description = video_info.get('description', '').lower()
|
| 451 |
|
| 452 |
if any(word in title for word in ['music', 'song', 'soundtrack']):
|
| 453 |
-
return "Original Music/Soundtrack - Primary audio content"
|
| 454 |
elif any(word in title for word in ['commercial', 'ad', 'brand']):
|
| 455 |
-
return "Upbeat Commercial Music - Designed to enhance brand appeal"
|
| 456 |
elif any(word in title for word in ['tutorial', 'how to', 'guide']):
|
| 457 |
-
return "Minimal/No Background Music - Focus on instruction"
|
| 458 |
elif any(word in title for word in ['vlog', 'daily', 'life']):
|
| 459 |
-
return "Ambient Background Music - Complementary to narration"
|
| 460 |
else:
|
| 461 |
-
return "Background Music - Complementing video mood and pacing"
|
| 462 |
|
| 463 |
def detect_influencer_status(self, video_info):
|
| 464 |
"""Detect influencer status"""
|
|
@@ -466,17 +479,17 @@ class YouTubeDownloader:
|
|
| 466 |
view_count = video_info.get('view_count', 0)
|
| 467 |
|
| 468 |
if subscriber_count > 10000000:
|
| 469 |
-
return "Mega Influencer (10M+ subscribers)"
|
| 470 |
elif subscriber_count > 1000000:
|
| 471 |
-
return "Major Influencer (1M+ subscribers)"
|
| 472 |
elif subscriber_count > 100000:
|
| 473 |
-
return "Mid-tier Influencer (100K+ subscribers)"
|
| 474 |
elif subscriber_count > 10000:
|
| 475 |
-
return "Micro Influencer (10K+ subscribers)"
|
| 476 |
elif view_count > 100000:
|
| 477 |
-
return "Viral Content Creator"
|
| 478 |
else:
|
| 479 |
-
return "Regular Content Creator"
|
| 480 |
|
| 481 |
def format_number(self, num):
|
| 482 |
if num is None or num == 0:
|
|
@@ -522,53 +535,62 @@ class YouTubeDownloader:
|
|
| 522 |
|
| 523 |
# Generate compact report with contrasting background
|
| 524 |
report = f"""
|
| 525 |
-
<div style='font-family:
|
| 526 |
-
|
| 527 |
-
|
| 528 |
-
|
| 529 |
-
|
| 530 |
-
</div>
|
| 531 |
-
<!-- INFO GRID (2 Columns) -->
|
| 532 |
-
<div style='display: grid; grid-template-columns: 1fr 1fr; gap: 8px; margin-bottom: 12px;'>
|
| 533 |
-
<!-- LEFT: BASIC INFO -->
|
| 534 |
-
<div style='background-color:rgb(211, 211, 211); border: 1px solid #CCC; border-left: 3px solid #FF6F00; border-radius: 4px; padding: 6px; max-width: 100%;'>
|
| 535 |
-
<h3 style='margin: 0 0 4px 0; font-size: 14px; background-color:#D3D3D3; color: #000000 !important; line-height: 1.2;'>Basic Information</h3>
|
| 536 |
-
<table style='width: 100%; font-size: 11px; color: #212121; border-spacing: 0;'>
|
| 537 |
-
<tr><td style='padding: 0px 1px;'><strong>Creator</strong></td><td style='padding: 1px 2px;'>{uploader[:20]}{'...' if len(uploader) > 20 else ''}</td></tr>
|
| 538 |
-
<tr><td style='padding: 0px 1px;'><strong>Date</strong></td><td style='padding: 1px 2px;'>{formatted_date}</td></tr>
|
| 539 |
-
<tr><td style='padding: 0px 1px;'><strong>Duration</strong></td><td style='padding: 1px 2px;'>{duration_str}</td></tr>
|
| 540 |
-
</table>
|
| 541 |
</div>
|
| 542 |
-
|
| 543 |
-
|
| 544 |
-
|
| 545 |
-
|
| 546 |
-
|
| 547 |
-
|
| 548 |
-
|
| 549 |
-
|
| 550 |
-
<tr><td style='
|
| 551 |
-
<tr><td style='padding:
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 552 |
</table>
|
| 553 |
</div>
|
| 554 |
-
|
| 555 |
-
|
| 556 |
-
|
| 557 |
-
<
|
| 558 |
-
|
| 559 |
-
<tr><td style='
|
| 560 |
-
<tr><td style='
|
|
|
|
|
|
|
|
|
|
|
|
|
| 561 |
</table>
|
| 562 |
</div>
|
| 563 |
</div>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 564 |
</div>
|
| 565 |
-
|
| 566 |
-
|
| 567 |
-
<h3 style='text-align: center; font-size: 16px; color: #000000 !important; margin-bottom: 6px; line-height: 1.2;'> Scene-by-Scene Breakdown</h3>
|
| 568 |
-
{scene_table_html}
|
| 569 |
-
</div>
|
| 570 |
-
</div>
|
| 571 |
-
"""
|
| 572 |
return report.strip()
|
| 573 |
|
| 574 |
def get_video_info(self, url, progress=gr.Progress(), cookiefile=None):
|
|
@@ -589,12 +611,6 @@ class YouTubeDownloader:
|
|
| 589 |
|
| 590 |
if cookiefile and os.path.exists(cookiefile):
|
| 591 |
ydl_opts['cookiefile'] = cookiefile
|
| 592 |
-
else:
|
| 593 |
-
print(f"⚠️ Cookie file not provided or not found: {cookiefile}")
|
| 594 |
-
|
| 595 |
-
# 🧪 Log yt_dlp options
|
| 596 |
-
print("🔍 yt_dlp options:")
|
| 597 |
-
print(json.dumps(ydl_opts, indent=2))
|
| 598 |
|
| 599 |
progress(0.5, desc="Extracting video metadata...")
|
| 600 |
|
|
@@ -606,19 +622,7 @@ class YouTubeDownloader:
|
|
| 606 |
return info, "✅ Video information extracted successfully"
|
| 607 |
|
| 608 |
except Exception as e:
|
| 609 |
-
|
| 610 |
-
print(f"❌ yt_dlp extraction error: {error_msg}")
|
| 611 |
-
|
| 612 |
-
if "Video unavailable" in error_msg or "This content isn’t available" in error_msg:
|
| 613 |
-
return None, (
|
| 614 |
-
"❌ This video is unavailable or restricted. "
|
| 615 |
-
"Please check if it's private, deleted, age-restricted, or try again with a valid cookies.txt file."
|
| 616 |
-
)
|
| 617 |
-
elif "cookies" in error_msg.lower():
|
| 618 |
-
return None, f"❌ Error: {str(e)}"
|
| 619 |
-
|
| 620 |
-
return None, f"❌ Unexpected extraction error: {str(e)}"
|
| 621 |
-
|
| 622 |
|
| 623 |
def download_video(self, url, quality="best", audio_only=False, progress=gr.Progress(), cookiefile=None):
|
| 624 |
"""Download video with progress tracking"""
|
|
@@ -733,13 +737,15 @@ def analyze_with_cookies(url, cookies_file, progress=gr.Progress()):
|
|
| 733 |
try:
|
| 734 |
progress(0.05, desc="Starting analysis...")
|
| 735 |
|
| 736 |
-
cookiefile =
|
|
|
|
|
|
|
|
|
|
| 737 |
info, msg = downloader.get_video_info(url, progress=progress, cookiefile=cookiefile)
|
| 738 |
|
| 739 |
if info:
|
| 740 |
progress(0.95, desc="Generating comprehensive report...")
|
| 741 |
formatted_info = downloader.format_video_info(info)
|
| 742 |
-
cached_reports[url] = formatted_info # Cache the result
|
| 743 |
progress(1.0, desc="✅ Complete!")
|
| 744 |
return formatted_info
|
| 745 |
else:
|
|
@@ -749,34 +755,202 @@ def analyze_with_cookies(url, cookies_file, progress=gr.Progress()):
|
|
| 749 |
return f"❌ System Error: {str(e)}"
|
| 750 |
|
| 751 |
|
| 752 |
-
def analyze_and_generate_pdf(url, cookies_file, progress=
|
| 753 |
-
"""Generate PDF from cached HTML only"""
|
| 754 |
try:
|
| 755 |
-
|
|
|
|
| 756 |
|
| 757 |
-
|
| 758 |
-
|
| 759 |
-
return
|
| 760 |
|
| 761 |
-
|
| 762 |
-
|
| 763 |
|
| 764 |
-
|
| 765 |
-
|
| 766 |
-
|
| 767 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 768 |
|
|
|
|
| 769 |
pdf_path = os.path.join(tempfile.gettempdir(), f"analysis_report_{uuid.uuid4().hex}.pdf")
|
| 770 |
with open(pdf_path, "wb") as f:
|
| 771 |
-
f.write(
|
| 772 |
|
| 773 |
-
|
| 774 |
-
|
|
|
|
| 775 |
|
| 776 |
except Exception as e:
|
| 777 |
-
print(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 778 |
return None
|
| 779 |
|
|
|
|
|
|
|
| 780 |
def download_with_cookies(url, quality, audio_only, cookies_file, progress=gr.Progress()):
|
| 781 |
"""Main download function"""
|
| 782 |
try:
|
|
@@ -810,22 +984,27 @@ def create_interface():
|
|
| 810 |
h3, .gr-group h3, .gradio-container h3 {
|
| 811 |
color: #87CEEB !important;
|
| 812 |
}
|
|
|
|
| 813 |
label, .gr-textbox label, .gr-file label, .gr-dropdown label, .gr-checkbox label {
|
| 814 |
color: #00008B !important;
|
| 815 |
font-weight: bold !important;
|
| 816 |
}
|
|
|
|
| 817 |
.gr-file .file-name {
|
| 818 |
color: #00008B !important;
|
| 819 |
font-weight: bold !important;
|
| 820 |
}
|
|
|
|
| 821 |
/* Make tab labels dark blue too */
|
| 822 |
.gr-tab-nav button {
|
| 823 |
color: #00008B !important;
|
| 824 |
}
|
|
|
|
| 825 |
.gr-tab-nav button.selected {
|
| 826 |
background-color: #FF8C00 !important;
|
| 827 |
color: #000000 !important;
|
| 828 |
}
|
|
|
|
| 829 |
/* Light blue text for API status */
|
| 830 |
.light-blue-text textarea {
|
| 831 |
color: #87CEEB !important;
|
|
@@ -835,12 +1014,14 @@ def create_interface():
|
|
| 835 |
background-color: #2a2a2a !important;
|
| 836 |
border: 2px dashed #444 !important;
|
| 837 |
}
|
|
|
|
| 838 |
.gr-group, .gr-form, .gr-row {
|
| 839 |
background-color: #1a1a1a !important;
|
| 840 |
border: 1px solid #444 !important;
|
| 841 |
border-radius: 10px;
|
| 842 |
padding: 15px;
|
| 843 |
}
|
|
|
|
| 844 |
""",
|
| 845 |
theme=gr.themes.Soft(),
|
| 846 |
title="📊 YouTube Video Analyzer & Downloader"
|
|
|
|
| 1 |
def format_scene_breakdown(scenes):
|
| 2 |
rows = """
|
| 3 |
+
<table style='width:100%; border-collapse: collapse; background-color:#1a1a1a; color: #FFFFFF; border: 2px solid #FF8C00; font-size: 16px;box-shadow: 0 4px 8px rgba(0,0,0,0.3);'>
|
| 4 |
<tr style='background-color:#FF8C00; color: #000000;'>
|
| 5 |
+
<th style='padding: 8px; border: 1px solid #FF8C00; color: #000000; font-weight: bold;'>⏱️ Timestamp</th>
|
| 6 |
+
<th style='padding: 8px; border: 1px solid #FF8C00; color: #000000; font-weight: bold;'>📝 Description</th>
|
| 7 |
</tr>
|
| 8 |
"""
|
| 9 |
pattern = re.compile(r"\*\*\[(.*?)\]\*\*:\s*(.*)")
|
|
|
|
| 16 |
description = match.group(2).strip()
|
| 17 |
rows += f"""
|
| 18 |
<tr style='background-color:#1a1a1a;'>
|
| 19 |
+
<td style='padding: 8px; border: 1px solid #444; color: #87CEEB; font-weight: bold;font-size: 16px;vertical-align: top;'>{timestamp}</td>
|
| 20 |
+
<td style='padding: 8px; border: 1px solid #444; color: #87CEEB; font-weight: bold;font-size: 16px;line-height: 1.4;'>{description}</td>
|
| 21 |
</tr>
|
| 22 |
"""
|
| 23 |
|
|
|
|
| 40 |
from io import BytesIO
|
| 41 |
|
| 42 |
|
|
|
|
|
|
|
| 43 |
def generate_pdf_from_html(html_content):
|
| 44 |
+
"""Generate PDF with simplified HTML that works better with xhtml2pdf"""
|
| 45 |
try:
|
| 46 |
+
# Create a simplified version of the HTML for PDF generation
|
| 47 |
+
# Remove complex CSS that xhtml2pdf can't handle
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 48 |
simplified_html = html_content.replace(
|
| 49 |
"background: linear-gradient(135deg, #2d3748, #1a202c);",
|
| 50 |
+
"background-color: #f5f5f5;"
|
| 51 |
+
).replace(
|
| 52 |
+
"background: linear-gradient(90deg, #FF8C00, #87CEEB);",
|
| 53 |
+
"background-color: #FF8C00;"
|
| 54 |
+
).replace(
|
| 55 |
+
"rgba(135, 206, 235, 0.1)",
|
| 56 |
+
"#f9f9f9"
|
| 57 |
+
).replace(
|
| 58 |
+
"rgba(0, 0, 0, 0.3)",
|
| 59 |
+
"#ffffff"
|
| 60 |
+
).replace(
|
| 61 |
+
"text-shadow: 2px 2px 4px rgba(0,0,0,0.5);",
|
| 62 |
+
""
|
| 63 |
+
).replace(
|
| 64 |
+
"box-shadow: 0 8px 32px rgba(255, 140, 0, 0.3);",
|
| 65 |
+
""
|
| 66 |
+
).replace(
|
| 67 |
+
"box-shadow: 0 4px 8px rgba(0,0,0,0.3);",
|
| 68 |
+
""
|
| 69 |
+
).replace(
|
| 70 |
+
"display: grid; grid-template-columns: 1fr 1fr 1fr; gap: 15px;",
|
| 71 |
+
"display: block;"
|
| 72 |
+
).replace(
|
| 73 |
+
"background-color:#1a1a1a;",
|
| 74 |
+
"background-color:#ffffff;"
|
| 75 |
).replace(
|
| 76 |
"color: #FFFFFF;",
|
| 77 |
"color: #000000;"
|
| 78 |
+
).replace(
|
| 79 |
+
"background-color:#FF8C00; color: #000000;",
|
| 80 |
+
"background-color:#FF8C00; color: #000000;"
|
| 81 |
+
).replace(
|
| 82 |
+
"color: #87CEEB;",
|
| 83 |
+
"color: #000080;"
|
| 84 |
+
).replace(
|
| 85 |
+
"border: 2px solid #FF8C00;",
|
| 86 |
+
"border: 1px solid #FF8C00;"
|
| 87 |
)
|
| 88 |
|
| 89 |
+
# Remove table styling that causes issues
|
| 90 |
+
simplified_html = re.sub(r"style='[^']*background-color:#1a1a1a[^']*'", "style='background-color:#ffffff;'", simplified_html)
|
| 91 |
+
simplified_html = re.sub(r"style='[^']*color: #87CEEB[^']*'", "style='color: #000080; padding: 8px;'", simplified_html)
|
| 92 |
|
| 93 |
+
# Wrap in a complete HTML document with PDF-friendly CSS
|
| 94 |
pdf_html = f"""
|
| 95 |
<!DOCTYPE html>
|
| 96 |
<html>
|
| 97 |
<head>
|
| 98 |
<meta charset="UTF-8">
|
| 99 |
<style>
|
| 100 |
+
@page {{
|
| 101 |
+
size: A4;
|
| 102 |
+
margin: 1cm;
|
| 103 |
+
}}
|
| 104 |
body {{
|
| 105 |
font-family: Arial, sans-serif;
|
| 106 |
font-size: 12px;
|
| 107 |
+
line-height: 1.4;
|
| 108 |
color: #000000;
|
| 109 |
background-color: #ffffff;
|
| 110 |
}}
|
| 111 |
+
.report-container {{
|
| 112 |
+
background-color: #ffffff;
|
| 113 |
+
padding: 15px;
|
| 114 |
+
border: 2px solid #FF8C00;
|
| 115 |
+
border-radius: 8px;
|
| 116 |
+
}}
|
| 117 |
+
.header {{
|
| 118 |
+
text-align: center;
|
| 119 |
+
color: #FF8C00;
|
| 120 |
+
font-size: 20px;
|
| 121 |
+
font-weight: bold;
|
| 122 |
+
margin-bottom: 15px;
|
| 123 |
+
border-bottom: 2px solid #FF8C00;
|
| 124 |
+
padding-bottom: 8px;
|
| 125 |
+
}}
|
| 126 |
+
.info-card {{
|
| 127 |
+
background-color: #f9f9f9;
|
| 128 |
+
padding: 12px;
|
| 129 |
+
margin: 8px 0;
|
| 130 |
+
border-left: 3px solid #87CEEB;
|
| 131 |
+
border-radius: 4px;
|
| 132 |
+
page-break-inside: avoid;
|
| 133 |
+
}}
|
| 134 |
+
.info-title {{
|
| 135 |
+
color: #000080;
|
| 136 |
+
font-size: 14px;
|
| 137 |
+
font-weight: bold;
|
| 138 |
+
margin-bottom: 8px;
|
| 139 |
+
}}
|
| 140 |
+
table {{
|
| 141 |
+
width: 100%;
|
| 142 |
+
border-collapse: collapse;
|
| 143 |
+
margin: 8px 0;
|
| 144 |
+
page-break-inside: avoid;
|
| 145 |
+
}}
|
| 146 |
+
th, td {{
|
| 147 |
+
padding: 6px 8px;
|
| 148 |
+
border: 1px solid #cccccc;
|
| 149 |
+
text-align: left;
|
| 150 |
+
vertical-align: top;
|
| 151 |
+
font-size: 11px;
|
| 152 |
+
}}
|
| 153 |
+
th {{
|
| 154 |
+
background-color: #FF8C00;
|
| 155 |
+
color: #000000;
|
| 156 |
+
font-weight: bold;
|
| 157 |
+
}}
|
| 158 |
+
tr:nth-child(even) {{
|
| 159 |
+
background-color: #f9f9f9;
|
| 160 |
+
}}
|
| 161 |
+
.scene-table {{
|
| 162 |
+
margin-top: 15px;
|
| 163 |
+
}}
|
| 164 |
+
.scene-header {{
|
| 165 |
+
color: #000080;
|
| 166 |
+
font-size: 16px;
|
| 167 |
+
font-weight: bold;
|
| 168 |
+
text-align: center;
|
| 169 |
+
margin-bottom: 10px;
|
| 170 |
+
}}
|
| 171 |
+
div[style*="display: grid"] {{
|
| 172 |
+
display: block !important;
|
| 173 |
+
}}
|
| 174 |
+
div[style*="grid-template-columns"] > div {{
|
| 175 |
+
display: block !important;
|
| 176 |
+
margin-bottom: 10px !important;
|
| 177 |
+
width: 100% !important;
|
| 178 |
+
}}
|
| 179 |
</style>
|
| 180 |
</head>
|
| 181 |
<body>
|
| 182 |
+
<div class="report-container">
|
| 183 |
+
{simplified_html}
|
| 184 |
+
</div>
|
| 185 |
</body>
|
| 186 |
</html>
|
| 187 |
"""
|
| 188 |
|
|
|
|
|
|
|
|
|
|
| 189 |
result = BytesIO()
|
|
|
|
|
|
|
| 190 |
pisa_status = pisa.CreatePDF(pdf_html, dest=result)
|
| 191 |
+
print("PDF buffer length:", len(result.getvalue()))
|
|
|
|
|
|
|
| 192 |
|
| 193 |
if pisa_status.err:
|
| 194 |
+
print(f"PDF generation error: {pisa_status.err}")
|
|
|
|
|
|
|
|
|
|
|
|
|
| 195 |
return None
|
| 196 |
|
| 197 |
result.seek(0)
|
|
|
|
| 198 |
return result
|
| 199 |
|
| 200 |
except Exception as e:
|
| 201 |
+
print(f"PDF generation exception: {e}")
|
|
|
|
|
|
|
| 202 |
return None
|
| 203 |
+
|
| 204 |
class YouTubeDownloader:
|
| 205 |
def __init__(self):
|
| 206 |
self.download_dir = tempfile.mkdtemp()
|
|
|
|
| 254 |
|
| 255 |
# Create enhanced prompt for Gemini
|
| 256 |
prompt = f"""
|
| 257 |
+
Analyze this YouTube video and create a highly detailed, scene-by-scene breakdown with precise timestamps and specific descriptions:
|
| 258 |
|
| 259 |
Title: {title}
|
| 260 |
Duration: {duration} seconds
|
|
|
|
| 443 |
text = (title + " " + description).lower()
|
| 444 |
|
| 445 |
if any(word in text for word in ['music', 'song', 'album', 'artist', 'band', 'lyrics']):
|
| 446 |
+
return "🎵 Music Video"
|
| 447 |
elif any(word in text for word in ['tutorial', 'how to', 'guide', 'learn', 'teaching']):
|
| 448 |
+
return "📚 Tutorial/Educational"
|
| 449 |
elif any(word in text for word in ['funny', 'comedy', 'entertainment', 'vlog', 'challenge']):
|
| 450 |
+
return "🎭 Entertainment/Comedy"
|
| 451 |
elif any(word in text for word in ['news', 'breaking', 'report', 'update']):
|
| 452 |
+
return "📰 News/Information"
|
| 453 |
elif any(word in text for word in ['review', 'unboxing', 'test', 'comparison']):
|
| 454 |
+
return "⭐ Review/Unboxing"
|
| 455 |
elif any(word in text for word in ['commercial', 'ad', 'brand', 'product']):
|
| 456 |
+
return "📺 Commercial/Advertisement"
|
| 457 |
else:
|
| 458 |
+
return "🎬 General Content"
|
| 459 |
|
| 460 |
def detect_background_music(self, video_info):
|
| 461 |
"""Detect background music style"""
|
|
|
|
| 463 |
description = video_info.get('description', '').lower()
|
| 464 |
|
| 465 |
if any(word in title for word in ['music', 'song', 'soundtrack']):
|
| 466 |
+
return "🎵 Original Music/Soundtrack - Primary audio content"
|
| 467 |
elif any(word in title for word in ['commercial', 'ad', 'brand']):
|
| 468 |
+
return "🎶 Upbeat Commercial Music - Designed to enhance brand appeal"
|
| 469 |
elif any(word in title for word in ['tutorial', 'how to', 'guide']):
|
| 470 |
+
return "🔇 Minimal/No Background Music - Focus on instruction"
|
| 471 |
elif any(word in title for word in ['vlog', 'daily', 'life']):
|
| 472 |
+
return "🎼 Ambient Background Music - Complementary to narration"
|
| 473 |
else:
|
| 474 |
+
return "🎵 Background Music - Complementing video mood and pacing"
|
| 475 |
|
| 476 |
def detect_influencer_status(self, video_info):
|
| 477 |
"""Detect influencer status"""
|
|
|
|
| 479 |
view_count = video_info.get('view_count', 0)
|
| 480 |
|
| 481 |
if subscriber_count > 10000000:
|
| 482 |
+
return "🌟 Mega Influencer (10M+ subscribers)"
|
| 483 |
elif subscriber_count > 1000000:
|
| 484 |
+
return "⭐ Major Influencer (1M+ subscribers)"
|
| 485 |
elif subscriber_count > 100000:
|
| 486 |
+
return "🎯 Mid-tier Influencer (100K+ subscribers)"
|
| 487 |
elif subscriber_count > 10000:
|
| 488 |
+
return "📈 Micro Influencer (10K+ subscribers)"
|
| 489 |
elif view_count > 100000:
|
| 490 |
+
return "🔥 Viral Content Creator"
|
| 491 |
else:
|
| 492 |
+
return "👤 Regular Content Creator"
|
| 493 |
|
| 494 |
def format_number(self, num):
|
| 495 |
if num is None or num == 0:
|
|
|
|
| 535 |
|
| 536 |
# Generate compact report with contrasting background
|
| 537 |
report = f"""
|
| 538 |
+
<div style='font-family: Arial, sans-serif; background: linear-gradient(135deg, #2d3748, #1a202c); padding: 20px; border-radius: 15px; border: 2px solid #FF8C00; box-shadow: 0 8px 32px rgba(255, 140, 0, 0.3);'>
|
| 539 |
+
|
| 540 |
+
<div style='text-align: center; margin-bottom: 20px;'>
|
| 541 |
+
<h2 style='color: #FF8C00; font-size: 24px; margin: 0; text-shadow: 2px 2px 4px rgba(0,0,0,0.5);'>🎬 YouTube Video Analysis Report</h2>
|
| 542 |
+
<div style='height: 3px; background: linear-gradient(90deg, #FF8C00, #87CEEB); margin: 10px 0; border-radius: 5px;'></div>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 543 |
</div>
|
| 544 |
+
|
| 545 |
+
<!-- Compact Information Grid -->
|
| 546 |
+
<div style='display: grid; grid-template-columns: 1fr 1fr 1fr; gap: 15px; margin-bottom: 20px;'>
|
| 547 |
+
|
| 548 |
+
<!-- Basic Information Card -->
|
| 549 |
+
<div style='background: rgba(135, 206, 235, 0.1); padding: 15px; border-radius: 10px; border-left: 4px solid #87CEEB;'>
|
| 550 |
+
<h3 style='color: #FF8C00; margin: 0 0 10px 0; font-size: 16px;'>📋 Basic Info</h3>
|
| 551 |
+
<table style='width: 100%; font-size: 14px;'>
|
| 552 |
+
<tr><td style='color: #87CEEB; font-weight: bold; padding: 4px 0;'>📹 Title:</td></tr>
|
| 553 |
+
<tr><td style='color: #87CEEB; font-weight: bold: padding: 4px 0 8px 0; word-wrap: wrap-word; white-space: normal; max-width: 200px;'>{title}</td></tr>
|
| 554 |
+
<tr><td style='color: #87CEEB; font-weight: bold; padding: 4px 0;'>👤 Creator:</td><td style='color: #87CEEB; padding: 2px 0;'>{uploader[:20]}{'...' if len(uploader) > 20 else ''}</td></tr>
|
| 555 |
+
<tr><td style='color: #87CEEB; font-weight: bold; padding: 4px 0;'>📅 Date:</td><td style='color: #87CEEB; padding: 2px 0;'>{formatted_date}</td></tr>
|
| 556 |
+
<tr><td style='color: #87CEEB; font-weight: bold; padding: 4px 0;'>⏱️ Duration:</td><td style='color: #87CEEB; padding: 2px 0;'>{duration_str}</td></tr>
|
| 557 |
+
</table>
|
| 558 |
+
</div>
|
| 559 |
+
|
| 560 |
+
<!-- Performance Metrics Card -->
|
| 561 |
+
<div style='background: rgba(135, 206, 235, 0.1); padding: 15px; border-radius: 10px; border-left: 4px solid #FF8C00;border: 1px solid #444'>
|
| 562 |
+
<h3 style='color: #FF8C00; margin: 0 0 10px 0; font-size: 16px;'>📊 Metrics</h3>
|
| 563 |
+
<table style='width: 100%; font-size: 12px;'>
|
| 564 |
+
<tr><td style='color: #87CEEB; font-weight: bold; padding: 4px 0;'>👀 Views:</td><td style='color: #87CEEB; padding: 4px 0;'>{self.format_number(view_count)}</td></tr>
|
| 565 |
+
<tr><td style='color: #87CEEB; font-weight: bold; padding: 4px 0;'>👍 Likes:</td><td style='color: #87CEEB; padding: 4px 0;'>{self.format_number(like_count)}</td></tr>
|
| 566 |
+
<tr><td style='color: #87CEEB; font-weight: bold; padding: 4px 0;'>💬 Comments:</td><td style='color: #87CEEB; padding: 4px 0;'>{self.format_number(comment_count)}</td></tr>
|
| 567 |
+
<tr><td style='color: #87CEEB; font-weight: bold; padding: 4px 0;'>📈 Engagement:</td><td style='color: #87CEEB; padding: 4px 0;'>{engagement_rate:.2f}%</td></tr>
|
| 568 |
</table>
|
| 569 |
</div>
|
| 570 |
+
|
| 571 |
+
<!-- Content Analysis Card -->
|
| 572 |
+
<div style='background:rgba(135, 206, 235, 0.1); padding: 15px; border-radius: 10px; border-left: 4px solid #87CEEB;border: 1px solid #444'>
|
| 573 |
+
<h3 style='color:#FF8C00; margin: 0 0 10px 0; font-size: 16px;'>🎯 Analysis</h3>
|
| 574 |
+
<table style='width: 100%; font-size: 12px;'>
|
| 575 |
+
<tr><td style='color: #87CEEB; font-weight: bold; padding: 4px 0;'>📂 Type:</td></tr>
|
| 576 |
+
<tr><td style='color: 87CEEB; padding: 4px 0 8px 0; word-break: break-word;'>{video_type}</td></tr>
|
| 577 |
+
<tr><td style='color: #87CEEB; font-weight: bold; padding: 4px 0;'>🎵 Music:</td></tr>
|
| 578 |
+
<tr><td style='color: #87CEEB; padding: 4px 0 8px 0; word-break: break-word;'>{background_music[:30]}{'...' if len(background_music) > 30 else ''}</td></tr>
|
| 579 |
+
<tr><td style='color: #87CEEB; font-weight: bold; padding: 4px 0;'>👑 Status:</td></tr>
|
| 580 |
+
<tr><td style='color: #87CEEB; padding: 4px 0; word-break: break-word;'>{influencer_status[:25]}{'...' if len(influencer_status) > 25 else ''}</td></tr>
|
| 581 |
</table>
|
| 582 |
</div>
|
| 583 |
</div>
|
| 584 |
+
|
| 585 |
+
<!-- Scene Breakdown Section -->
|
| 586 |
+
<div style='background: rgba(0, 0, 0, 0.3); padding: 15px; border-radius: 10px; border: 1px solid #444;'>
|
| 587 |
+
<h3 style='color: #87CEEB; margin: 0 0 15px 0; font-size: 18px; text-align: center;'>🎬 Scene-by-Scene Breakdown</h3>
|
| 588 |
+
{scene_table_html}
|
| 589 |
+
</div>
|
| 590 |
+
|
| 591 |
</div>
|
| 592 |
+
"""
|
| 593 |
+
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 594 |
return report.strip()
|
| 595 |
|
| 596 |
def get_video_info(self, url, progress=gr.Progress(), cookiefile=None):
|
|
|
|
| 611 |
|
| 612 |
if cookiefile and os.path.exists(cookiefile):
|
| 613 |
ydl_opts['cookiefile'] = cookiefile
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 614 |
|
| 615 |
progress(0.5, desc="Extracting video metadata...")
|
| 616 |
|
|
|
|
| 622 |
return info, "✅ Video information extracted successfully"
|
| 623 |
|
| 624 |
except Exception as e:
|
| 625 |
+
return None, f"❌ Error: {str(e)}"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 626 |
|
| 627 |
def download_video(self, url, quality="best", audio_only=False, progress=gr.Progress(), cookiefile=None):
|
| 628 |
"""Download video with progress tracking"""
|
|
|
|
| 737 |
try:
|
| 738 |
progress(0.05, desc="Starting analysis...")
|
| 739 |
|
| 740 |
+
cookiefile = None
|
| 741 |
+
if cookies_file and os.path.exists(cookies_file):
|
| 742 |
+
cookiefile = cookies_file
|
| 743 |
+
|
| 744 |
info, msg = downloader.get_video_info(url, progress=progress, cookiefile=cookiefile)
|
| 745 |
|
| 746 |
if info:
|
| 747 |
progress(0.95, desc="Generating comprehensive report...")
|
| 748 |
formatted_info = downloader.format_video_info(info)
|
|
|
|
| 749 |
progress(1.0, desc="✅ Complete!")
|
| 750 |
return formatted_info
|
| 751 |
else:
|
|
|
|
| 755 |
return f"❌ System Error: {str(e)}"
|
| 756 |
|
| 757 |
|
| 758 |
+
def analyze_and_generate_pdf(url, cookies_file, progress=gr.Progress()):
|
|
|
|
| 759 |
try:
|
| 760 |
+
progress(0.1, desc="Extracting video info...")
|
| 761 |
+
cookiefile = cookies_file if cookies_file and os.path.exists(cookies_file) else None
|
| 762 |
|
| 763 |
+
info, _ = downloader.get_video_info(url, progress=progress, cookiefile=cookiefile)
|
| 764 |
+
if not info:
|
| 765 |
+
return "❌ Failed to extract video info"
|
| 766 |
|
| 767 |
+
progress(0.6, desc="Generating HTML report...")
|
| 768 |
+
report_html = downloader.format_video_info(info)
|
| 769 |
|
| 770 |
+
progress(0.8, desc="Creating PDF...")
|
| 771 |
+
|
| 772 |
+
# Generate PDF from HTML
|
| 773 |
+
pdf_data = BytesIO()
|
| 774 |
+
result = pisa.CreatePDF(report_html, dest=pdf_data)
|
| 775 |
+
|
| 776 |
+
if result.err:
|
| 777 |
+
return "❌ PDF generation failed"
|
| 778 |
|
| 779 |
+
pdf_data.seek(0)
|
| 780 |
pdf_path = os.path.join(tempfile.gettempdir(), f"analysis_report_{uuid.uuid4().hex}.pdf")
|
| 781 |
with open(pdf_path, "wb") as f:
|
| 782 |
+
f.write(pdf_data.read())
|
| 783 |
|
| 784 |
+
progress(1.0, desc="✅ PDF ready!")
|
| 785 |
+
print("✅ PDF generated at:", pdf_path)
|
| 786 |
+
return pdf_path # ✅ Return direct path like in test code
|
| 787 |
|
| 788 |
except Exception as e:
|
| 789 |
+
print("❌ Exception:", e)
|
| 790 |
+
return f"❌ Error: {str(e)}"
|
| 791 |
+
|
| 792 |
+
def generate_pdf_from_html(html_content):
|
| 793 |
+
"""Generate PDF with simplified HTML that works better with xhtml2pdf"""
|
| 794 |
+
try:
|
| 795 |
+
# Create a simplified version of the HTML for PDF generation
|
| 796 |
+
# Remove complex CSS that xhtml2pdf can't handle
|
| 797 |
+
simplified_html = html_content.replace(
|
| 798 |
+
"background: linear-gradient(135deg, #2d3748, #1a202c);",
|
| 799 |
+
"background-color: #f5f5f5;"
|
| 800 |
+
).replace(
|
| 801 |
+
"background: linear-gradient(90deg, #FF8C00, #87CEEB);",
|
| 802 |
+
"background-color: #FF8C00;"
|
| 803 |
+
).replace(
|
| 804 |
+
"rgba(135, 206, 235, 0.1)",
|
| 805 |
+
"#f9f9f9"
|
| 806 |
+
).replace(
|
| 807 |
+
"rgba(0, 0, 0, 0.3)",
|
| 808 |
+
"#ffffff"
|
| 809 |
+
).replace(
|
| 810 |
+
"text-shadow: 2px 2px 4px rgba(0,0,0,0.5);",
|
| 811 |
+
""
|
| 812 |
+
).replace(
|
| 813 |
+
"box-shadow: 0 8px 32px rgba(255, 140, 0, 0.3);",
|
| 814 |
+
""
|
| 815 |
+
).replace(
|
| 816 |
+
"box-shadow: 0 4px 8px rgba(0,0,0,0.3);",
|
| 817 |
+
""
|
| 818 |
+
).replace(
|
| 819 |
+
"display: grid; grid-template-columns: 1fr 1fr 1fr; gap: 15px;",
|
| 820 |
+
"display: block;"
|
| 821 |
+
).replace(
|
| 822 |
+
"background-color:#1a1a1a;",
|
| 823 |
+
"background-color:#ffffff;"
|
| 824 |
+
).replace(
|
| 825 |
+
"color: #FFFFFF;",
|
| 826 |
+
"color: #000000;"
|
| 827 |
+
).replace(
|
| 828 |
+
"background-color:#FF8C00; color: #000000;",
|
| 829 |
+
"background-color:#FF8C00; color: #000000;"
|
| 830 |
+
).replace(
|
| 831 |
+
"color: #87CEEB;",
|
| 832 |
+
"color: #000080;"
|
| 833 |
+
).replace(
|
| 834 |
+
"border: 2px solid #FF8C00;",
|
| 835 |
+
"border: 1px solid #FF8C00;"
|
| 836 |
+
)
|
| 837 |
+
|
| 838 |
+
# Remove table styling that causes issues
|
| 839 |
+
simplified_html = re.sub(r"style='[^']*background-color:#1a1a1a[^']*'", "style='background-color:#ffffff;'", simplified_html)
|
| 840 |
+
simplified_html = re.sub(r"style='[^']*color: #87CEEB[^']*'", "style='color: #000080; padding: 8px;'", simplified_html)
|
| 841 |
+
|
| 842 |
+
# Wrap in a complete HTML document with PDF-friendly CSS
|
| 843 |
+
pdf_html = f"""
|
| 844 |
+
<!DOCTYPE html>
|
| 845 |
+
<html>
|
| 846 |
+
<head>
|
| 847 |
+
<meta charset="UTF-8">
|
| 848 |
+
<style>
|
| 849 |
+
@page {{
|
| 850 |
+
size: A4;
|
| 851 |
+
margin: 1cm;
|
| 852 |
+
}}
|
| 853 |
+
body {{
|
| 854 |
+
font-family: Arial, sans-serif;
|
| 855 |
+
font-size: 12px;
|
| 856 |
+
line-height: 1.4;
|
| 857 |
+
color: #000000;
|
| 858 |
+
background-color: #ffffff;
|
| 859 |
+
}}
|
| 860 |
+
.report-container {{
|
| 861 |
+
background-color: #ffffff;
|
| 862 |
+
padding: 15px;
|
| 863 |
+
border: 2px solid #FF8C00;
|
| 864 |
+
border-radius: 8px;
|
| 865 |
+
}}
|
| 866 |
+
.header {{
|
| 867 |
+
text-align: center;
|
| 868 |
+
color: #FF8C00;
|
| 869 |
+
font-size: 20px;
|
| 870 |
+
font-weight: bold;
|
| 871 |
+
margin-bottom: 15px;
|
| 872 |
+
border-bottom: 2px solid #FF8C00;
|
| 873 |
+
padding-bottom: 8px;
|
| 874 |
+
}}
|
| 875 |
+
.info-card {{
|
| 876 |
+
background-color: #f9f9f9;
|
| 877 |
+
padding: 12px;
|
| 878 |
+
margin: 8px 0;
|
| 879 |
+
border-left: 3px solid #87CEEB;
|
| 880 |
+
border-radius: 4px;
|
| 881 |
+
page-break-inside: avoid;
|
| 882 |
+
}}
|
| 883 |
+
.info-title {{
|
| 884 |
+
color: #000080;
|
| 885 |
+
font-size: 14px;
|
| 886 |
+
font-weight: bold;
|
| 887 |
+
margin-bottom: 8px;
|
| 888 |
+
}}
|
| 889 |
+
table {{
|
| 890 |
+
width: 100%;
|
| 891 |
+
border-collapse: collapse;
|
| 892 |
+
margin: 8px 0;
|
| 893 |
+
page-break-inside: avoid;
|
| 894 |
+
}}
|
| 895 |
+
th, td {{
|
| 896 |
+
padding: 6px 8px;
|
| 897 |
+
border: 1px solid #cccccc;
|
| 898 |
+
text-align: left;
|
| 899 |
+
vertical-align: top;
|
| 900 |
+
font-size: 11px;
|
| 901 |
+
}}
|
| 902 |
+
th {{
|
| 903 |
+
background-color: #FF8C00;
|
| 904 |
+
color: #000000;
|
| 905 |
+
font-weight: bold;
|
| 906 |
+
}}
|
| 907 |
+
tr:nth-child(even) {{
|
| 908 |
+
background-color: #f9f9f9;
|
| 909 |
+
}}
|
| 910 |
+
.scene-table {{
|
| 911 |
+
margin-top: 15px;
|
| 912 |
+
}}
|
| 913 |
+
.scene-header {{
|
| 914 |
+
color: #000080;
|
| 915 |
+
font-size: 16px;
|
| 916 |
+
font-weight: bold;
|
| 917 |
+
text-align: center;
|
| 918 |
+
margin-bottom: 10px;
|
| 919 |
+
}}
|
| 920 |
+
div[style*="display: grid"] {{
|
| 921 |
+
display: block !important;
|
| 922 |
+
}}
|
| 923 |
+
div[style*="grid-template-columns"] > div {{
|
| 924 |
+
display: block !important;
|
| 925 |
+
margin-bottom: 10px !important;
|
| 926 |
+
width: 100% !important;
|
| 927 |
+
}}
|
| 928 |
+
</style>
|
| 929 |
+
</head>
|
| 930 |
+
<body>
|
| 931 |
+
<div class="report-container">
|
| 932 |
+
{simplified_html}
|
| 933 |
+
</div>
|
| 934 |
+
</body>
|
| 935 |
+
</html>
|
| 936 |
+
"""
|
| 937 |
+
|
| 938 |
+
result = BytesIO()
|
| 939 |
+
pisa_status = pisa.CreatePDF(pdf_html, dest=result)
|
| 940 |
+
|
| 941 |
+
if pisa_status.err:
|
| 942 |
+
print(f"PDF generation error: {pisa_status.err}")
|
| 943 |
+
return None
|
| 944 |
+
|
| 945 |
+
result.seek(0)
|
| 946 |
+
return result
|
| 947 |
+
|
| 948 |
+
except Exception as e:
|
| 949 |
+
print(f"PDF generation exception: {e}")
|
| 950 |
return None
|
| 951 |
|
| 952 |
+
|
| 953 |
+
|
| 954 |
def download_with_cookies(url, quality, audio_only, cookies_file, progress=gr.Progress()):
|
| 955 |
"""Main download function"""
|
| 956 |
try:
|
|
|
|
| 984 |
h3, .gr-group h3, .gradio-container h3 {
|
| 985 |
color: #87CEEB !important;
|
| 986 |
}
|
| 987 |
+
|
| 988 |
label, .gr-textbox label, .gr-file label, .gr-dropdown label, .gr-checkbox label {
|
| 989 |
color: #00008B !important;
|
| 990 |
font-weight: bold !important;
|
| 991 |
}
|
| 992 |
+
|
| 993 |
.gr-file .file-name {
|
| 994 |
color: #00008B !important;
|
| 995 |
font-weight: bold !important;
|
| 996 |
}
|
| 997 |
+
|
| 998 |
/* Make tab labels dark blue too */
|
| 999 |
.gr-tab-nav button {
|
| 1000 |
color: #00008B !important;
|
| 1001 |
}
|
| 1002 |
+
|
| 1003 |
.gr-tab-nav button.selected {
|
| 1004 |
background-color: #FF8C00 !important;
|
| 1005 |
color: #000000 !important;
|
| 1006 |
}
|
| 1007 |
+
|
| 1008 |
/* Light blue text for API status */
|
| 1009 |
.light-blue-text textarea {
|
| 1010 |
color: #87CEEB !important;
|
|
|
|
| 1014 |
background-color: #2a2a2a !important;
|
| 1015 |
border: 2px dashed #444 !important;
|
| 1016 |
}
|
| 1017 |
+
|
| 1018 |
.gr-group, .gr-form, .gr-row {
|
| 1019 |
background-color: #1a1a1a !important;
|
| 1020 |
border: 1px solid #444 !important;
|
| 1021 |
border-radius: 10px;
|
| 1022 |
padding: 15px;
|
| 1023 |
}
|
| 1024 |
+
|
| 1025 |
""",
|
| 1026 |
theme=gr.themes.Soft(),
|
| 1027 |
title="📊 YouTube Video Analyzer & Downloader"
|