def get_recipe(url, use_filter):
# target Binging with Babish video
transcript = YouTubeTranscriptApi.get_transcript(url)
# Get verbs from db
verbs = []
with open("../../datasets/cooking_verbs.txt", 'r') as verb_file:
verbs = verb_file.readlines()
# ingredient extraction ============================
db = Ingredients()
ingreds = set([])
actual_ingredients = []
res = {}
i = 0
print('>>> Getting ingredients')
for subtitle in transcript:
i += 1
text = subtitle['text']
cur_ingredients = db.parse_ingredients(text)
measurements = db.parse_measurements(text)
if len(cur_ingredients) > 0:
print('>>> Ingredients from line ' + str(i) + ': ')
print(cur_ingredients)
ingreds |= set(cur_ingredients)
actual_ingredients += get_actual_ingredients(cur_ingredients,
measurements)
print('>>> Ingredients detected: ')
print(actual_ingredients)
res['ingredients'] = actual_ingredients
# ==================================================
video_file = download_video(url)
# steps extraction =================================
times = ['seconds', 'minutes', 'hours', 'second', 'minute', 'hour']
instructions = []
pictures = []
if use_filter:
print('>>> Generating steps')
i = 0
for subtitle in transcript:
text = subtitle['text']
# Write all lines
# with open('before_filter.txt', 'a') as before_file:
# before_file.write(text + '\n')
# Remove lines without an ingredient, cooking verb, or time measurement
for target in (list(ingreds) + verbs):
if (target in text or len([t for t in times if (t in text)])):
instructions.append({'step': text})
if i % PICTURE_FREQUENCY == 0:
pictures.append(subtitle['start'])
print('>>> KEEPING LINE: ' + text)
# with open('after_filter.txt', 'a') as after_file:
# after_file.write(text + '\n')
# break
i += 1
break
res['instructions'] = instructions
print('>>> Steps Generation Complete')
# ==================================================
# frames extraction ================================
print('>>> Extracting Frames...')
file_names = extract_frames(video_file, pictures)
i = 0
for file_name in file_names:
res['instructions'][i]['image'] = file_name
i += PICTURE_FREQUENCY
if os.path.exists(video_file):
os.remove(video_file)
# ==================================================
print('>>> finished. result: ')
print(res)
return res
