async def post_upload(request):
if request.method == 'GET':
return abort(404)
if 'file' not in request.files:
return response.redirect('/')
file = request.files['file']
file_body = file[0].body
file_name = file[0].name
file_type = file[0].type
if "" in [file_body, file_name, file_type]:
return response.redirect('/')
# check if is valid filetype
if 'video' in file_type:
index = uuid.uuid4().hex
create_file(index, file_body)
hashcode = compute_md5(file_body)
log.info('hash generated = {}'.format(hashcode))
b64hash = base64.b64encode(hashcode).decode("ascii")
es_search = es.search(index='videos',
body={"query": {
"match": {
"hash": b64hash
}
}})
if es_search['hits']['total'] > 0:
# check if truly have hash
es_record_found = es_search['hits']['hits']
for record in es_record_found:
if record['_source']['hash'] == b64hash:
# found 1 is enough
return response.redirect('/hash/{}'.format(b64hash))
# if cannot find, continue to create new job
table = dynamodb.Table('Videos')
table.put_item(
Item={
'id': index,
'title': file_name,
'hash': hashcode,
'job_status': 'Queue for Audio Extraction',
'transcript': None,
'subs': None,
'link': index,
'vote_count': 0,
'author': 'anonymous',
'upload_date': int(time.time()) # time in seconds
})
q.enqueue(aws_stuff, index, timeout='2h', job_id=index)
return response.redirect('/job/{}'.format(index))
else:
create_file(file_name, file_body)
trans = Transcribe()
trans.parseOutput(file_name)
return response.text(file_type)
async def retrieve_job(request, id):
jinja_response = {}
table = dynamodb.Table('Videos')
db_query = table.get_item(Key={'id': id}, ConsistentRead=True)
db_item = db_query.get('Item')
if db_item is None:
abort(404)
job_status = db_item['job_status']
title = {'title': db_item['title']}
timestamp = {'date': db_item['upload_date']}
author = {'author': db_item['author']}
count = db_item.get('vote_count')
if count is None:
count = 0
jinja_response.update({'status': job_status})
jinja_response.update(title)
jinja_response.update(timestamp)
jinja_response.update(author)
try:
duration = {'duration': db_item['video_length']}
jinja_response.update(duration)
except:
pass
if job_status == 'Sent Audio For Transcription':
transcribe = boto3.client('transcribe')
result = transcribe.get_transcription_job(TranscriptionJobName=id)
status = result['TranscriptionJob']['TranscriptionJobStatus']
jinja_response.update({'status': status})
if status == 'COMPLETED':
status = 'Transcription done'
trans_uri = result['TranscriptionJob']['Transcript'][
'TranscriptFileUri']
trans_file = './resources/trans{}'.format(id)
path_file = 'trans{}'.format(id)
urllib.request.urlretrieve(trans_uri, trans_file)
trans = Transcribe()
trans.parseOutput(trans_file)
srt_mem = trans.srt_mem(trans_file + '.srt')
table = dynamodb.Table('Videos')
trans_data = {}
with open(trans_file, "r") as f:
trans_data = json.load(f)
# Delete transcription file
try:
os.remove(trans_file)
except OSError:
pass
# Delete srt file in disk
try:
os.remove('./resources/{}'.format(path_file))
except OSError:
pass
# Delete audio file in S3 bucket
s3 = boto3.client('s3')
bucket = 'orbitalphase1'
audio_file = './resources/{}.flac'.format(id)
s3.delete_object(Bucket=bucket, Key=audio_file)
table.update_item(
Key={'id': id},
UpdateExpression=
"SET job_status=:job_status, subs=:subtitles, transcript=:transcript",
ExpressionAttributeValues={
':job_status': 'Transcription done',
':transcript': trans_data,
':subtitles': srt_mem
})
jinja_response.update({'status': status})
jinja_response.update({
'srt': id,
'flac': id,
'ready': True,
'count': count
})
return jinja_response
elif status == 'IN_PROGRESS':
status = 'Sent to AWS to Transcribe'
jinja_response.update({'status': status})
return jinja_response
else:
return jinja_response
elif job_status == 'Transcription done' or "Edited" in job_status:
jinja_response.update({
'flac': id,
'srt': id,
'ready': True,
'count': count
})
return jinja_response
else:
return jinja_response
async def post_transcribe(request):
if request.method == 'GET':
return abort(404)
Transcribe.parseOutput()
return response.text("TODO")
async def post_transcribe(request):
Transcribe.parseOutput()
return response.text("TODO")