-
Notifications
You must be signed in to change notification settings - Fork 0
/
shortMaker.py
397 lines (320 loc) · 14.5 KB
/
shortMaker.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
from pytube import YouTube
import cv2
import pandas as pd
import subprocess
from openai import OpenAI
import numpy as np
import json
import math
import pdb
import os
import cv2
import numpy as np
from moviepy.editor import VideoFileClip, CompositeVideoClip, TextClip
import json
from tqdm.notebook import tqdm
import openai
import logging
import json
from youtube_transcript_api import YouTubeTranscriptApi
import sys
import yt_dlp
import textwrap
video_folder = 'videos/'
output_folder = 'reels/'
api_key = '' # Replace with your actual OpenAI API key
client = OpenAI(
api_key=api_key
)
net = cv2.dnn.readNetFromCaffe('deploy.prototxt', 'res10_300x300_ssd_iter_140000.caffemodel')
zoom_factor = 2.0 # Adjust this value to zoom in (e.g., 1.2) or out (e.g., 0.8)
face_change_threshold = 400
os.makedirs(output_folder, exist_ok=True)
def get_video_title(yt_id):
url = f'https://www.youtube.com/watch?v={yt_id}'
def list_formats(url):
ydl_opts = {
'listformats': True,
}
with yt_dlp.YoutubeDL(ydl_opts) as ydl:
info = ydl.extract_info(url, download=False)
return info
info = list_formats(url)
video_title = info.get('title', 'video').replace('/', '_').replace('\\', '_').replace(' ', '_')
return video_title
def convert_transcript_to_subtitles(transcript):
subtitles = []
for entry in transcript:
subtitle = {
"text": entry["text"],
"start": entry["start"],
"end": entry["start"] + entry["duration"]
}
subtitles.append(subtitle)
return subtitles
def segment_video(response, input_file, video_id):
try:
output_folder = "videos"
# Create the output folder if it doesn't exist
if not os.path.exists(output_folder):
os.makedirs(output_folder)
for i, segment in enumerate(response):
start_time = math.floor(float(segment.get("start_time", 0)))
end_time = math.ceil(float(segment.get("end_time", 0))) + 2
output_file = os.path.join(output_folder, f"{video_id}_{str(i).zfill(3)}.mp4")
with VideoFileClip(input_file) as video:
subclip = video.subclip(start_time, end_time)
subclip.write_videofile(output_file, codec="libx264", audio_codec="aac")
segment["output_file"] = output_file
# csv_file_path = 'parsed_content.csv'
# df = pd.read_csv(csv_file_path)
# new_row = pd.DataFrame([{
# 'start_time': segment['start_time'],
# 'end_time': segment['end_time'],
# 'description': segment['description'],
# 'duration': segment['duration'],
# 'output_file': segment['output_file'],
# 'reel': '',
# 'posted_twitter': False,
# 'posted_instagram': False
# }])
# df = pd.concat([df, new_row], ignore_index=True)
# df.to_csv(csv_file_path, index=False)
except Exception as e:
print(e)
pass
def get_transcript(video_id):
# Get the transcript for the given YouTube video ID
transcript = YouTubeTranscriptApi.get_transcript(video_id)
# Format the transcript for feeding into GPT-4
formatted_transcript = ''
for entry in transcript:
start_time = "{:.2f}".format(entry['start'])
end_time = "{:.2f}".format(entry['start'] + entry['duration'])
text = entry['text']
formatted_transcript += f"{start_time} --> {end_time} : {text}\n"
return transcript
def distance(box1, box2):
# Calculate Euclidean distance between two boxes
(x1, y1, x2, y2) = box1
(a1, b1, a2, b2) = box2
return np.sqrt((x1 - a1)**2 + (y1 - b1)**2 + (x2 - a2)**2 + (y2 - b2)**2)
def get_subtitles_for_segment(output_file, subtitle_objects):
for obj in subtitle_objects:
try:
if obj['output_file'] == output_file:
return obj['subtitles']
except Exception as e:
print(e)
pass
return []
def analyze_transcript(transcripts, video_id, chunk_size=5000, overlap=50):
response_obj = '''[
{
"start_time": 97.19,
"end_time": 127.43,
"description": "One liner of the section with hashtags which is tweet long. This will be displade on the video header",
"duration": 36
},
{
"start_time": 169.58,
"end_time": 199.10,
"description": "One liner of the section with hashtags which is tweet long. This will be displade on the video header",
"duration": 33
}
]'''
# Combine all the texts in the transcripts into one big string
all_text = ' '.join([segment['text'] for segment in transcripts])
# Split the text into words and make chunks of words
words = all_text.split(' ')
chunks = []
index = 0
title = get_video_title(video_id)
while index < len(words):
start = max(0, index - overlap)
chunk = " ".join(words[start: index + chunk_size])
chunks.append(chunk)
index += chunk_size
logging.info(f'Transcript has been chunked into {len(chunks)} chunks')
# Process each chunk separately and collect responses
responses = []
for idx, chunk in enumerate(chunks, start=1):
logging.info(f'Processing chunk {idx}/{len(chunks)}')
prompt = f"""This is a transcript of a video/podcast.
Please identify the most viral sections from this part of the video,
make sure they are more than 120 seconds in duration and not more than 250 seconds,
and provide extremely accurate timestamps.
Ensure you capture the entire question and answer for each section.
Respond only in this format: {response_obj}.
Make sure that the return format is JSON and syntactically correct.
I just want JSON as Response (nothing else).
\n\nHere is the Transcription:\n{chunk}"""
messages = [
{"role": "system", "content": "You are a ViralGPT helpful assistant. You are master at reading YouTube transcripts and identifying the most interesting parts and viral content from the podcasts."},
{"role": "user", "content": prompt}
]
logging.info(f'Sending chunk {idx}/{len(chunks)} to the model')
response = client.chat.completions.create(
model="gpt-4o",
messages=messages,
max_tokens=512,
n=1,
stop=None
)
logging.info(f'Received response for chunk {idx}/{len(chunks)}')
response_content = response.choices[0].message.content
responses.append(response_content)
logging.info(f'Added response for chunk {idx} to responses')
# Combine all responses into a single response
combined_response = ' '.join(responses)
logging.info('Completed processing all chunks')
return responses
def convert_transcript_to_subtitles(transcript):
subtitles = []
for entry in transcript:
subtitle = {
"text": entry["text"],
"start": entry["start"],
"end": entry["start"] + entry["duration"]
}
subtitles.append(subtitle)
return subtitles
def create_subtitles_for_segments(transcript, parsed_content):
subtitle_objects = []
for segment in parsed_content:
try:
subtitles = []
for entry in transcript:
if segment['start_time'] <= entry['start'] < segment['end_time']:
subtitle = {
"text": entry['text'],
"start": entry['start'] - segment['start_time'],
"end": entry['start'] + entry['duration'] - segment['start_time']
}
subtitles.append(subtitle)
subtitle_object = {
'output_file': segment['output_file'],
'subtitles': subtitles
}
subtitle_objects.append(subtitle_object)
except Exception as e:
print(e)
pass
return subtitle_objects
import re
def remove_hashtags(text):
# Regular expression to find hashtags
return re.sub(r'#\S+', '', text).strip()
def get_shorts(video_id, parsed_content, width=1080, height=1920, logo_path='logo.png'):
try:
video_files = [f for f in os.listdir(video_folder) if f.endswith(".mp4") and video_id in f]
for one_parsed_content in tqdm(parsed_content, desc="Processing videos"):
file_name = one_parsed_content['output_file'].split('/')[-1]
input_video_path = os.path.join(video_folder, file_name)
temp_video_path = os.path.join(output_folder, f'{video_id}_temp_' + file_name)
output_video_path = os.path.join(output_folder, file_name)
# Load the video
cap = cv2.VideoCapture(input_video_path)
original_frame_rate = cap.get(cv2.CAP_PROP_FPS)
fourcc = cv2.VideoWriter_fourcc(*'mp4v')
out = cv2.VideoWriter(temp_video_path, fourcc, original_frame_rate, (width, height))
crop_box = None # To store the current crop region
frame_count = int(cap.get(cv2.CAP_PROP_FRAME_COUNT))
for _ in tqdm(range(frame_count), desc=f"Processing frames for {file_name}"):
ret, frame = cap.read()
if not ret:
break
(h, w) = frame.shape[:2]
blob = cv2.dnn.blobFromImage(cv2.resize(frame, (300, 300)), 1.0, (300, 300), (104.0, 177.0, 123.0))
net.setInput(blob)
detections = net.forward()
detected_face = None
for i in range(0, detections.shape[2]):
confidence = detections[0, 0, i, 2]
if confidence > 0.5: # confidence threshold
box = detections[0, 0, i, 3:7] * np.array([w, h, w, h])
(startX, startY, endX, endY) = box.astype("int")
# Adjust the bounding box based on zoom factor
box_w = (endX - startX) * zoom_factor
box_h = (endY - startY) * zoom_factor
center_x = startX + (endX - startX) // 2
center_y = startY + (endY - startY) // 2
startX = int(center_x - box_w // 2)
startY = int(center_y - box_h // 2)
endX = int(center_x + box_w // 2)
endY = int(center_y + box_h // 2)
# Ensure the coordinates are within the frame bounds
startX = max(0, startX)
startY = max(0, startY)
endX = min(w, endX)
endY = min(h, endY)
detected_face = (startX, startY, endX, endY)
break
if detected_face:
if crop_box is None or distance(detected_face, crop_box) > face_change_threshold:
crop_box = detected_face
if crop_box:
(startX, startY, endX, endY) = crop_box
# Crop the frame around the current region
cropped_frame = frame[startY:endY, startX:endX]
# Calculate the scaling factor to fit the frame
scale_w = width / (endX - startX)
scale_h = height / (endY - startY)
scale = min(scale_w, scale_h)
new_w = int((endX - startX) * scale)
new_h = int((endY - startY) * scale)
resized_frame = cv2.resize(cropped_frame, (new_w, new_h))
else:
# If no person detected, resize the entire frame while maintaining the aspect ratio
scale_w = width / w
scale_h = height / h
scale = min(scale_w, scale_h)
new_w = int(w * scale)
new_h = int(h * scale)
resized_frame = cv2.resize(frame, (new_w, new_h))
# Create a blank frame with the desired dimensions and center the resized frame
blank_frame = np.zeros((height, width, 3), dtype=np.uint8)
x_offset = (width - resized_frame.shape[1]) // 2
y_offset = (height - resized_frame.shape[0]) // 2
blank_frame[y_offset:y_offset + resized_frame.shape[0], x_offset:x_offset + resized_frame.shape[1]] = resized_frame
final_frame = blank_frame
out.write(final_frame)
cap.release()
out.release()
# Load the original video and extract the audio
original_clip = VideoFileClip(input_video_path)
audio_clip = original_clip.audio
# Load the processed video without audio
video_clip = VideoFileClip(temp_video_path)
# Combine the video with the extracted audio
final_clip = video_clip.set_audio(audio_clip)
final_clip.write_videofile(output_video_path, codec='libx264', audio_codec='aac', bitrate='5000k', fps=original_frame_rate)
one_parsed_content['reel'] = output_video_path
# Remove the temporary video file
os.remove(temp_video_path)
except Exception as e:
print(f"Exception: {e}")
def distance(box1, box2):
# Calculate the Euclidean distance between two boxes (x, y, width, height)
center1 = ((box1[0] + box1[2]) / 2, (box1[1] + box1[3]) / 2)
center2 = ((box2[0] + box2[2]) / 2, (box2[1] + box2[3]) / 2)
return ((center1[0] - center2[0]) ** 2 + (center1[1] - center2[1]) ** 2) ** 0.5
if __name__ == "__main__":
if len(sys.argv) != 2:
print("Usage: python download_video.py <youtube_video_id>")
sys.exit(1)
video_id = sys.argv[1]
transcript = get_transcript(video_id)
interesting_segment = analyze_transcript(transcript, video_id)
parsed_content = []
for seg in interesting_segment:
seg = seg.replace('```','').replace('json','')
try:
parsed_content += json.loads(seg)
except json.JSONDecodeError as e:
print(f"Exception: {e}")
segment_video(parsed_content, f'initialVideos/{video_id}.mp4', video_id)
all_subtitles = create_subtitles_for_segments(transcript, parsed_content)
face_change_threshold = 400
zoom_factor = 5
get_shorts(video_id, parsed_content, width=720, height=1080)