pandora_cdosea/render.py

#!/usr/bin/python3
import os
import json
import subprocess
import string

from pi import random
from keywords import KEYWORDS
import ox
import ox.web.auth


base_url = 'https://cdosea.0x2620.org'

api = None

def get_api():
    global api
    if not api:
        api = ox.API(base_url + '/api/')
        api.signin(**ox.web.auth.get('cdosea'))

if os.path.exists('PATHS.json'):
    PATHS = json.load(open('PATHS.json'))
else:
    PATHS = {}

if os.path.exists('CLIPS.json'):
    CLIPS = json.load(open('CLIPS.json'))
else:
    CLIPS = {}

def get_path(id):
    global PATHS
    if id not in PATHS:
        get_api()
        info = api.findMedia({
            'query': {
                'conditions': [
                    {'key': 'id', 'operator': '==', 'value': id}
                ]
            },
            'keys': ['id', 'extension'],
            'range': [0, 1]
        })['data']['items'][0]
        path = os.path.join('cache', '%s.%s' % (info['id'], info['extension']))
        if not os.path.exists(path):
            url = '%s/%s/download/source/' % (base_url, id)
            print('get video', url)
        PATHS[id] = path
        with open('PATHS.json', 'w') as fd:
            json.dump(PATHS, fd, indent=4, sort_keys=True)
    return PATHS[id]

def get_clips(tag):
    global CLIPS
    if tag not in CLIPS:
        get_api()
        clips = api.findAnnotations({
            'query': {
                'conditions': [
                    {'key': 'layer', 'operator': '==', 'value': 'keywords'},
                    {'key': 'value', 'operator': '==', 'value': tag}
                ],
                'operator': '&'
            },
            'keys': ['id', 'in', 'out'],
            'range': [0, 10000]})['data']['items']
        for clip in clips:
            clip['path'] = get_path(clip['id'].split('/')[0])
            clip['duration'] = clip['out'] - clip['in']
            clip['tag'] = tag
        CLIPS[tag] = list(sorted(clips, key=lambda c: c['id']))
        with open('CLIPS.json', 'w') as fd:
            json.dump(CLIPS, fd, indent=4, sort_keys=True)
    return CLIPS[tag]

def random_choice(seq, items):
    n = n_ = len(items) - 1
    print('len', n)
    if n == 0:
        return items[0]
    r = seq()
    base = 10
    while n > 10:
        n /= 10
        print(r)
        r += seq()
        base += 10
    r = int(n_ * r / base)
    print('result', r, items)
    return items[r]

def splitint(number, by):
    div = int(number/by)
    mod = number % by
    return [div + 1 if i > (by - 1 - mod) else div for i in range(by)]

def filter_clips(clips, duration, max_duration=0):
    # 1 minute
    blur = 0.5
    low = 1
    high = 10

    # 2 minute
    blur = 1
    low = 2
    high = 20

    buckets = {}
    clips_ = []
    for tag in clips:
        for clip in clips[tag]:
            clip['tag'] = tag
            clips_.append(clip)
    clips_.sort(key=lambda c: c['duration'])
    print(clips_)
    size = splitint(len(clips_), 10)
    p = 0
    for i in range(10):
        buckets[i+1] = clips_[p:+p+size[i]]
        p += size[i]
    clips_ = {}
    print(buckets[duration])
    for clip in buckets[duration]:
        if clip['tag'] not in clips_:
            clips_[clip['tag']] = []
        clips_[clip['tag']].append(clip)
    return clips_

def sequence(seq, letter):
    tags = KEYWORDS[letter]
    clips = {tag: get_clips(tag) for tag in tags}
    result = {
        'clips': [],
        'text': [],
    }
    duration = 0
    MAX_DURATION = 65 * 2
    MIN_DURATION = 56 * 2
    while duration < MAX_DURATION and not duration >= MIN_DURATION:
        # clip duration: 1-10
        n = seq()
        if n == 0:
            n = 10

        max_duration = MAX_DURATION - duration
        clips_n = filter_clips(clips, n, max_duration)
        tags_n = list(sorted(clips_n.keys()))
        if not tags_n:
            print('NO tags for', letter, n)
        tag = random_choice(seq, tags_n)
        #if 'tiger' in tags_n:
        #    tag = 'tiger'
        clip = random_choice(seq, clips_n[tag])
        duration += clip['duration']
        result['clips'].append(clip.copy())
    for clip in result['clips']:
        if seq() == 0:
            clip['black'] = True
    position = last_text = 0
    while position < duration:
        n = seq()
        if n == 0:
            blank = {'blank': True, 'duration': position - last_text}
            result['text'].append(blank)
            n = seq()
            if n == 0:
                n = 10
            n = min(n, duration-position)
            text = {
                'text': random_choice(seq, tags_n),
                'duration': n
            }
            result['text'].append(text)
            position += n
            last_text = position
        else:
            position += n
    if last_text < duration:
        blank = {'blank': True, 'duration': duration - last_text}
        result['text'].append(blank)

    return result

if __name__ == '__main__':
    for n in range(10):
        seq = random(n)
        #for letter in string.ascii_uppercase:
        for letter in ('T', 'W'):
            r = sequence(seq, letter)
            tjson = 'output/%s%d.json' % (letter, n)
            with open(tjson, 'w') as fd:
                json.dump(r, fd, indent=4, sort_keys=True)
            print(json.dumps(r, indent=4, sort_keys=True))
            #print(sum([c['duration'] for c in r['clips']]))
            subprocess.call(['./render_mlt.py', tjson])