def main():
# Transcription and Cleaning
url = input("Enter the URL = ")
sec = pafy.new(url).length
print(f"\nVideo duration in sec = {sec}\n")
# THRESHOLDS
DYNAMIC_INTERVAL = (sec / 60) * 100
if sec <= 900: # 0-15 min
NUM_KEYWORDS = 15
SUMMARY_PERCENT = 60
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 6
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 900 < sec <= 1800: # 15-30 min
NUM_KEYWORDS = 18
SUMMARY_PERCENT = 50
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 5
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 1800 < sec <= 2700: # 30-45 min
NUM_KEYWORDS = 20
SUMMARY_PERCENT = 40
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 2700 < sec <= 3600: # 45-60 min
NUM_KEYWORDS = 22
SUMMARY_PERCENT = 35
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 3600 < sec <= 7200: # 1-2 hr
NUM_KEYWORDS = 25
SUMMARY_PERCENT = 30
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
else: # More than 2 hr
NUM_KEYWORDS = 30
SUMMARY_PERCENT = 25
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
start = time.perf_counter()
yt = YoutubeTranscribe(url)
text = yt.youtube_transcribe()
# Keywords Extractor
num_keywords = NUM_KEYWORDS
words = KeywordsExtractor(text, num_keywords)
keywords = words.ExtractKeywords()
print(f'\nKeywords:\n {keywords}')
# Summarization
summ = Summarizer()
percentage = SUMMARY_PERCENT
summary_result = summ.summary(text, percentage)
print(f'\nSummary:\n {summary_result}')
# Keyframe Extraction (Output : 'out' folder)
print("\nExtracting Keyframes\n")
ip = ImageProcessing(url, keywords)
ip.img_processing(text_threshold=NON_TEXT_LEN,
dis_threshold=SIMILAR_DISTANCE,
jump=INTERVAL_KEYFRAMES)
# Paragraph and Headings (Output : paragraph_headings.txt)
print("\nGenerating Paragraphs and Headings\n")
pf = ParaFormation(summary_result)
list_para = pf.paragraph(similarity_threshold=SENTENCE_SIMILARITY,
word_threshold=WORDS_PER_PARA,
percent_reduce=PERCENT_REDUCE)
ph = ParaHeadings(list_para)
# title_para = ph.get_titles_paras(sentence_threshold = SENTENCES_PER_PARA, training = HEADING_TRAINING, heading_threshold = TOP_HEADINGS)
title_para = ph.get_titles_paras(sentence_threshold=SENTENCES_PER_PARA)
# Final Notes (Includes Web Scraping)
print("\nGenerating Final Notes\n")
scraped_results = Scrapper(keywords, 2, 2, 2)
s = scraped_results.web_scrape()
notes = Notes(url, s)
notes.generate_notes()
print("\nBrevis-Notes.docx and Brevis-Notes.pdf(on Windows) Generated\n")
if os.path.exists('res'):
shutil.rmtree('res')
finish = time.perf_counter()
print(f'Serial: Finished in {round(finish-start, 2)} second(s)')
def main():
url = input("Enter the URL = ")
#Thresholds
sec = pafy.new(url).length
print(f"\nVideo duration in sec = {sec}\n")
# THRESHOLDS
DYNAMIC_INTERVAL = (sec/60) * 100
if sec <= 900: # 0-15 min
NUM_KEYWORDS = 15
SUMMARY_PERCENT = 60
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 6
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 900 < sec <= 1800: # 15-30 min
NUM_KEYWORDS = 18
SUMMARY_PERCENT = 50
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 5
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 1800 < sec <= 2700: # 30-45 min
NUM_KEYWORDS = 20
SUMMARY_PERCENT = 40
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 2700 < sec <= 3600: # 45-60 min
NUM_KEYWORDS = 22
SUMMARY_PERCENT = 35
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 3600 < sec <= 7200: # 1-2 hr
NUM_KEYWORDS = 25
SUMMARY_PERCENT = 30
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
else: # More than 2 hr
NUM_KEYWORDS = 30
SUMMARY_PERCENT = 25
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
#Starting the timer
start = time.perf_counter()
# Transcription and Cleaning
yt = YoutubeTranscribe(url)
text = yt.youtube_transcribe()
#Declaring a multiprocessing queue to exchange data between various functions
Q=multiprocessing.Queue()
#Running keywords and summary Processes parallely
key_ext=multiprocessing.Process(target=Process_Extract_Keywords , args=(url,text,Q,NUM_KEYWORDS,NON_TEXT_LEN,SIMILAR_DISTANCE,INTERVAL_KEYFRAMES))
summ_ext=multiprocessing.Process(target=Process_Get_Summary , args=(text,SUMMARY_PERCENT,SENTENCE_SIMILARITY,WORDS_PER_PARA,PERCENT_REDUCE,SENTENCES_PER_PARA))
#Starting both process simultaneously
key_ext.start()
summ_ext.start()
#Checking if the process have finished execution
key_ext.join()
summ_ext.join()
#Fetching scraped links from the Queue
scraped_res = Q.get()
#Generating final notes
notes = Notes(url,scraped_res)
notes.generate_notes()
print("\nBrevis-Notes.docx and Brevis-Notes.pdf(on Windows) Generated\n")
#Removing the temporary res folder
if os.path.exists('res'):
shutil.rmtree('res')
#Stopping the timer
end=time.perf_counter()
#Printing the time taken by the program for execution
print(f"Finished in {round(end-start, 3)} second(s)")
def main():
# Parallel
url = input("Enter the URL = ")
start = time.perf_counter()
# Transcription and Cleaning
yt = YoutubeTranscribe(url)
text = yt.youtube_transcribe()
# Level1
with ThreadPoolExecutor() as executor:
'''
Type1:
level1_results1 = executor.submit(Test(10,20).RecArea)
print(type(level1_results1))
print(dir(level1_results1))
print(level1_results1.result())
Type2:
level1_results1 = list(executor.map(Test().RecArea,[10],[20]))
print(level1_results1[0])
'''
# Keywords Extractor
# num_keywords=int(input("Enter number of keywords to be extracted : "))
num_keywords = 10
level1_results1 = executor.submit(
KeywordsExtractor(text, num_keywords).ExtractKeywords)
# Summarization
percentage = 40
level1_results2 = list(
executor.map(Summarizer().summary, [text], [percentage]))
print(f"\nKeywords:\n {level1_results1.result()}")
print(f"\nSummary:\n {level1_results2[0]}")
# Level2
with ThreadPoolExecutor() as executor:
# Keyframe Extraction (Output : 'out' folder)
print("\nExtracting Keyframes\n")
level2_results1 = list(
executor.map(
ImageProcessing(url, level1_results1.result()).img_processing,
[50], [20], [1000]))
# Paragraph and Headings (Output : paragraph_headings.txt)
print("\nGenerating Paragraphs and Headings\n")
level2_results2 = executor.submit(
ParaFormation(level1_results2[0]).paragraph)
print("\nScraping Web\n")
level2_results3 = executor.submit(
Scrapper(level1_results1.result(), 2, 2, 2).web_scrape)
print(len(os.listdir(os.path.join('res', 'out'))),
"images extracted in 'out' folder")
ph = ParaHeadings(level2_results2.result())
title_para = ph.get_titles_paras(sentence_threshold=2)
# Final Notes
notes = Notes(url, level2_results3.result())
notes.generate_notes()
print("\nBrevis-Notes.docx and Brevis-Notes.pdf(on Windows) Generated\n")
if os.path.exists('res'):
shutil.rmtree('res')
finish = time.perf_counter()
print(f'Parallel: Finished in {round(finish-start, 2)} second(s)')
def gen():
global video_url
global keywords
global path
global json_result
global text
global summary_result
global scrape_json
global option
sec = pafy.new(video_url).length
print(f"\nVideo duration in sec = {sec}\n")
# THRESHOLDS
DYNAMIC_INTERVAL = (sec / 60) * 100
if sec <= 900: # 0-15 min
NUM_KEYWORDS = 15
SUMMARY_PERCENT = 60
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 6
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 900 < sec <= 1800: # 15-30 min
NUM_KEYWORDS = 18
SUMMARY_PERCENT = 50
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 5
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 1800 < sec <= 2700: # 30-45 min
NUM_KEYWORDS = 20
SUMMARY_PERCENT = 40
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 2700 < sec <= 3600: # 45-60 min
NUM_KEYWORDS = 22
SUMMARY_PERCENT = 35
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 3600 < sec <= 7200: # 1-2 hr
NUM_KEYWORDS = 25
SUMMARY_PERCENT = 30
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
else: # More than 2 hr
NUM_KEYWORDS = 30
SUMMARY_PERCENT = 25
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
start = time.perf_counter()
# if option == "Overview":
# percentage = 50
# elif option == "Notes":
# percentage = 60
# elif option == "Notes+Ref":
# percentage = 80
#Running keywords and summary Processes parallely
key_ext = multiprocessing.Process(target=Process_Extract_Keywords,
args=(video_url, text, NUM_KEYWORDS,
NON_TEXT_LEN, SIMILAR_DISTANCE,
INTERVAL_KEYFRAMES))
summ_ext = multiprocessing.Process(
target=Process_Get_Summary,
args=(text, SUMMARY_PERCENT, SENTENCE_SIMILARITY, WORDS_PER_PARA,
PERCENT_REDUCE, SENTENCES_PER_PARA))
#Starting both process simultaneously
key_ext.start()
summ_ext.start()
#Checking if the process have finished execution
key_ext.join()
summ_ext.join()
if option == "Overview" or option == "Notes":
scrape_json = {}
#Generating final notes
notes = Notes(video_url, scrape_json)
notes.generate_notes()
print("\nBrevis-Notes.docx and Brevis-Notes.pdf(on Windows) Generated\n")
with ZipFile('Brevis_Notes.zip', 'w') as zip:
print("Writing zip")
if os.path.exists(os.path.join('res', 'Brevis-Notes.pdf')):
zip.write(os.path.join('res', 'Brevis-Notes.pdf'),
arcname='Brevis-Notes.pdf')
zip.write(os.path.join('res', 'Brevis-Notes.docx'),
arcname='Brevis-Notes.docx')
path = os.path.abspath("Brevis_Notes.zip")
if os.path.exists('res'):
shutil.rmtree('res')
finish = time.perf_counter()
print(f'Gen Function: Finished in {round(finish-start, 2)} second(s)')
def gen():
global video_url
global keywords
global path
global json_result
global text
global summary_result
global scrape_json
global option
sec = pafy.new(video_url).length
print(f"\nVideo duration in sec = {sec}\n")
# THRESHOLDS
DYNAMIC_INTERVAL = (sec / 60) * 100
if sec <= 900: # 0-15 min
NUM_KEYWORDS = 15
SUMMARY_PERCENT = 60
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 6
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 900 < sec <= 1800: # 15-30 min
NUM_KEYWORDS = 18
SUMMARY_PERCENT = 50
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 5
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 1800 < sec <= 2700: # 30-45 min
NUM_KEYWORDS = 20
SUMMARY_PERCENT = 40
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 2700 < sec <= 3600: # 45-60 min
NUM_KEYWORDS = 22
SUMMARY_PERCENT = 35
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
elif 3600 < sec <= 7200: # 1-2 hr
NUM_KEYWORDS = 25
SUMMARY_PERCENT = 30
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
else: # More than 2 hr
NUM_KEYWORDS = 30
SUMMARY_PERCENT = 25
NON_TEXT_LEN = 50
SIMILAR_DISTANCE = 20
INTERVAL_KEYFRAMES = DYNAMIC_INTERVAL
SENTENCE_SIMILARITY = 0.35
WORDS_PER_PARA = 20
PERCENT_REDUCE = 0.6
SENTENCES_PER_PARA = 4
# HEADING_TRAINING = 500
# TOP_HEADINGS = 3
start = time.perf_counter()
if option == "Overview":
if not os.path.exists(os.path.join('res', 'out')):
os.mkdir(os.path.join('res', 'out'))
elif option == "Notes" or option == "Notes+Ref":
# Keyframe Extraction (Output : 'out' folder)
print("\nExtracting Keyframes\n")
ip = ImageProcessing(video_url, keywords)
ip.img_processing(text_threshold=NON_TEXT_LEN,
dis_threshold=SIMILAR_DISTANCE,
jump=INTERVAL_KEYFRAMES)
# Paragraph and Headings (Output : paragraph_headings.txt)
print("\nGenerating Paragraphs and Headings\n")
pf = ParaFormation(summary_result)
list_para = pf.paragraph(similarity_threshold=SENTENCE_SIMILARITY,
word_threshold=WORDS_PER_PARA,
percent_reduce=PERCENT_REDUCE)
ph = ParaHeadings(list_para)
title_para = ph.get_titles_paras(sentence_threshold=SENTENCES_PER_PARA)
# Final Notes (Includes Web Scraping)
print("\nGenerating Final Notes\n")
if option == "Overview" or option == "Notes":
scrape_json = {}
#scraped_results = Scrapper(scrape_keywords,2,2,2)
#s = scraped_results.web_scrape()
notes = Notes(video_url, scrape_json)
notes.generate_notes()
print("\nBrevis-Notes.docx Generated\n")
with ZipFile('Brevis_Notes.zip', 'w') as zip:
print("Writing zip")
if os.path.exists(os.path.join('res', 'Brevis-Notes.pdf')):
zip.write(os.path.join('res', 'Brevis-Notes.pdf'),
arcname='Brevis-Notes.pdf')
zip.write(os.path.join('res', 'Brevis-Notes.docx'),
arcname='Brevis-Notes.docx')
path = os.path.abspath("Brevis_Notes.zip")
if os.path.exists('res'):
shutil.rmtree('res')
finish = time.perf_counter()
print(f'Gen Function: Finished in {round(finish-start, 2)} second(s)')