oxdata/oxdata/poster/models.py

# -*- coding: utf-8 -*-
# vi:si:et:sw=4:sts=4:ts=4
from __future__ import division, print_function, absolute_import

import hashlib
import os.path
import socket
import urllib

from django.db import models

import ox
import ox.web.criterion
import ox.web.movieposterdb
import ox.web.imdb
import ox.web.impawards
import ox.web.apple
import ox.web.piratecinema

from lookup.models import MovieId

def getPosters(movie_id, url_prefix='', limit=lambda x, y: 0.3 < x/y < 1):
    if not movie_id:
        return {}
    get_poster_urls(movie_id)
    posters = {}
    if url_prefix.endswith('/'):
        url_prefix = url_prefix[:-1]
    for p in PosterCache.objects.all().filter(movie_id=movie_id, failed=False).order_by('-id'):
        if p.site not in posters:
            posters[p.site] = []
        poster = p.get()
        if poster and (not poster.width or not poster.height):
            p.failed = True
            p.status = "invalid image"
            p.image.delete()
            p.save()
            poster = None
        if poster:
            pjson = {}
            pjson['url'] = url_prefix + poster._get_url()
            pjson['width'] = poster.width
            pjson['height'] = poster.height
            if p.site not in ['other', 'wikipedia.org'] or limit(poster.width, poster.height):
                posters[p.site].append(pjson)
    for p in list(posters):
        if not posters[p]:
            del posters[p]
    return posters

def poster_path(url, filename):
    if not isinstance(url, bytes):
        url = url.encode('utf-8')
    h = hashlib.sha1(url).hexdigest()
    ext = 'jpg'
    if filename.endswith('.png'):
        ext = 'png'
    name = "%s.%s" % (h, ext)
    return os.path.join('posters', h[:2], h[2:4], h[4:6], name)

def postercache_path(i, f):
    return poster_path(i.url.encode('utf-8'), f)

class PosterCache(models.Model):
    class Meta:
        unique_together = ("movie_id", "url")

    created = models.DateTimeField(auto_now_add=True)
    modified = models.DateTimeField(auto_now=True)

    movie_id = models.ForeignKey(MovieId, related_name='postercache')
    url = models.CharField(max_length=1024)
    site = models.CharField(max_length=255)
    site_id = models.CharField(max_length=1024)
    image = models.ImageField(max_length=255, upload_to=postercache_path)
    status = models.CharField(max_length=1024, default='200')
    failed = models.BooleanField(default=False)

    def __str__(self):
        return '%s' % self.url

    def get(self):
        if not self.image and not self.failed:
            import ox.net
            url = self.url.encode('utf-8')
            name = hashlib.sha1(url).hexdigest()
            try:
                data = ox.net.read_url(url)
                self.image.name = poster_path(self.url, os.path.basename(self.url))
                ox.makedirs(os.path.dirname(self.image.path))
                with open(self.image.path, 'wb') as f:
                    f.write(data)
                self.save()
            except urllib.error.HTTPError as e:
                #import traceback
                #print traceback.print_exc()
                self.status = e.code
                self.failed = True
                self.save()
            except urllib.error.URLError as e:
                #import traceback
                #print traceback.print_exc()
                self.status = e.reason
                self.failed = True
                self.save()
            except socket.error as e:
                self.status = str(e)
                self.failed = True
                self.save()
            except:
                self.status = 'some error'
                self.failed = True
                self.save()
        if self.image:
            failed = False
            try:
                self.image.width
                if not self.image.width or not self.image.width:
                    failed = True
            except:
                failed = True
            if failed:
                self.failed = True
                self.status = "invalid image"
                self.image.delete()
                self.save()
        if not self.failed:
            return self.image

def get_poster_urls(m):
    def addPoster(url, site, site_id):
        if PosterCache.objects.all().filter(url=url, movie_id=m).count() == 0:
            p = PosterCache(url=url, site=site, site_id=site_id, movie_id=m)
            p.save()

    if m.imdb_id:
        #if settings.DEBUG:
        #    print 'imdb'
        poster = ox.web.imdb.get_movie_poster(m.imdb_id)
        if poster:
            addPoster(poster, 'imdb.com', m.imdb_id)
        #site is sometimes down
        #for poster in ox.web.movieposterdb.get_data(m.imdb_id)['posters']:
        #    addPoster(poster, 'movieposterdb.com', m.imdb_id)

        poster = ox.web.piratecinema.get_poster_url(m.imdb_id)
        if poster:
            addPoster(poster, 'piratecinema.org', m.imdb_id)

    if m.criterion_id:
        #if settings.DEBUG:
        #    print 'criterion', m.criterion_id
        for poster in ox.web.criterion.get_data(m.criterion_id)['posters']:
            addPoster(poster, 'criterion.com', m.criterion_id)
    if m.wikipedia_id:
        #if settings.DEBUG:
        #    print 'wikipedia'
        poster = ox.web.wikipedia.get_poster_url(m.wikipedia_id)
        if poster:
            if PosterCache.objects.all().filter(url=poster).count() == 0:
                addPoster(poster, 'wikipedia.org', m.wikipedia_id)
    if m.impawards_id:
        #if settings.DEBUG:
        #    print 'impawards'
        data = ox.web.impawards.get_data(m.impawards_id)
        if data and 'imdbId' in data:
            for poster in data['posters']:
                addPoster(poster, 'impawards.com', m.imdb_id)

    '''
    if m.title and m.director:
        data = ox.web.apple.get_movie_data(m.title, m.director)
        if data and 'poster' in data:
            addPoster(data['poster'], 'apple.com', m.imdb_id)
    '''
    #fixme: get 0xdb still, possibly imdb still as fallback?