display.py

'''Google image slide show viewer.
Handles downloading, converting, and displaying images.
Images are chosen based on fixed search terms'''

import pygame, os, tempfile, random
import urllib, urllib.request, urllib.error, urllib.parse
import threading, time, datetime, re
import hashlib
import signal
import sys
import requests
import logging
import pickle

import config
from GIFImage import GIFImage
from pygame import display, image, Rect
from PIL import Image
from ImageDownloader import ImageDownloader
from SearchTermServer import SearchTermServer
from config import vars

formatter = logging.Formatter(fmt='%(asctime)s %(message)s', datefmt='%m/%d/%Y %I:%M:%S %p')

def make_logger(name, filename, level):
    handler = logging.FileHandler(filename)
    handler.setFormatter(formatter)

    logger = logging.getLogger(name)
    logger.setLevel(level)
    logger.addHandler(handler)

    return logger

main_logger = make_logger('main_logger', 'log.log', logging.DEBUG)
logger_store = {}

pygame.font.init()
CODE_DIR = os.path.dirname(__file__)
IMAGE_DIR = os.path.join(CODE_DIR, "images")
if not os.path.exists(IMAGE_DIR): os.mkdir(IMAGE_DIR)

#This path should exist already with loading images that come with the program
LOAD_IMAGE_DIR = os.path.join(CODE_DIR, "loading_imgs")

#currently the conversion cache never expires while this program is running.
CONVERT_CACHE = {}
display.init()

highest_res = display.list_modes()[0]
SCREEN_WIDTH = highest_res[0]
SCREEN_HEIGHT = highest_res[1]
# SCREEN_WIDTH = 1024
# SCREEN_HEIGHT = 768
IMAGE_SIZE = (SCREEN_WIDTH, SCREEN_HEIGHT)
FONT_SIZE = 48
FONT_COLOR = (255,255,0)
LOADING_FONT = pygame.font.Font(None, FONT_SIZE)
LOADING_FONT_DETAILED = pygame.font.Font(None, int(FONT_SIZE / 2))
DETAILED_PROGRESS = True
TEXT_PADDING = 5 #px of padding for text
RGB_BLACK = (0,0,0)
MOUSE_LEFT = 1
MOUSE_RIGHT = 3

#API only returns 100 pages of results. To get the maximum return, specify
#the max number of results per page which is 10
RESULTS_PER_PAGE = vars['results_per_page'] # must be between 1-10
LOADING_PAGE_THRESHOLD = RESULTS_PER_PAGE
CHUNK_SIZE = 8192
GOOGLE_API_URL = "https://www.googleapis.com/customsearch/v1?{}"
SEARCH_ENGINE_ID = vars['search_engine_id']
USER_AGENT = vars['user_agent']
API_KEY = vars['api_key']

IMAGES_LOCK = threading.Lock()
IMAGE_SIZES = [
    'xlarge',
    'xxlarge',
    'huge'
]
IMAGE_BLACKLIST_FILENAME = os.path.join(CODE_DIR,"urlblacklist")
MAX_FILE_AGE = 60 * 60 * 24 * 90 # 90 days
SHOW_IMAGE_POSITION = False

try:
    with open("qc.pickle", "rb") as qc_pickle_file:
        QUERY_CACHE = pickle.load(qc_pickle_file)
except (pickle.UnpicklingError, FileNotFoundError, TypeError):
    QUERY_CACHE = {} #key = query, value = search result page index

SCREEN_LOCK = threading.Lock()

try:
    with open(IMAGE_BLACKLIST_FILENAME) as blacklist:
        IMAGE_BLACKLIST = {url.strip() for url in blacklist.readlines()}
except IOError:
    IMAGE_BLACKLIST = {}


def assemble_query(query, img_size, index=1):
    parameters = {
        'key': API_KEY,
        'cx': SEARCH_ENGINE_ID,
        'filter': 1,
        'prettyPrint': 'true',
        'searchType': 'image',
        'imgSize': img_size,
        'fields': 'queries(nextPage/totalResults,nextPage/startIndex),items(link)',
        'num': RESULTS_PER_PAGE,
        'q': query,
        'start': index
    }
    return GOOGLE_API_URL.format(urllib.parse.urlencode(parameters))

def assemble_images():
    images = set()
    for term in vars['search_terms'] + vars['extra_images']:
        img_dir = os.path.join(IMAGE_DIR, term)

        if not os.path.exists(img_dir):
            continue

        for img in os.listdir(img_dir):
            if img.endswith('.log'):
                continue
            img_full_path = os.path.join(img_dir, img)
            images.add(img_full_path)

    return images

IMAGES = assemble_images()
REFRESH_EVENT = threading.Event()
server = SearchTermServer(IMAGE_DIR, IMAGES, IMAGES_LOCK, MAX_FILE_AGE, REFRESH_EVENT)

def get_center_width_offset(pil_image):
    iwidth = pil_image.size[0]
    return (SCREEN_WIDTH - iwidth) / 2

def get_center_height_offset(pil_image):
    iheight = pil_image.size[1]
    return (SCREEN_HEIGHT - iheight) / 2

def pil_image_convert(image_path):
    try:
        #will need to close this one
        return Image.open(CONVERT_CACHE[image_path]),True
    except KeyError:
        pil_image = Image.open(image_path)
        #no need to close this, as pil does it automatically during resize_image()

        new_pil_image = resize_image(pil_image)
        conv_filepath = tempfile.NamedTemporaryFile(delete=False)
        new_pil_image.save(conv_filepath, format=pil_image.format)
        conv_filepath.close()
        CONVERT_CACHE[image_path] = conv_filepath.name

        #don't need to close this image, PIL has not allocated a FP
        return new_pil_image,False

def resize_image(pil_image):
    #pil wil close pil_image automatically when thumbnail or resize() is called
    #the returned image does not have a file pointer so close() wont work on it either
    width = pil_image.size[0]
    height = pil_image.size[1]

    if width > SCREEN_WIDTH or height > SCREEN_HEIGHT:
        #Shrink image to screen size
        pil_image.thumbnail(IMAGE_SIZE)
        return pil_image

    width_to_screen_ratio = float(SCREEN_WIDTH) / width
    height_to_screen_ratio = float(SCREEN_HEIGHT) / height

    ratio_increase = min(width_to_screen_ratio, height_to_screen_ratio)

    new_width = int(float(width) * ratio_increase)
    new_height = int(float(height) * ratio_increase)

    #Or we return an aspect ratio preserved enlarged image to screen size
    return pil_image.resize((new_width, new_height))


def display_image(image_path, image_index):
    pil_image,cache_hit = pil_image_convert(image_path)

    coordinate_x = get_center_width_offset(pil_image)
    coordinate_y = get_center_height_offset(pil_image)

    if cache_hit:
        pil_image.close()

    surface_img = image.load(CONVERT_CACHE[image_path])

    SCREEN_LOCK.acquire()
    screen.fill(RGB_BLACK)
    screen.blit(surface_img, (coordinate_x, coordinate_y))
    if SHOW_IMAGE_POSITION:
        display_image_position(image_index)
    display.flip()
    SCREEN_LOCK.release()

def display_image_position(image_index):
    position = "{}/{}".format(image_index+1, len(IMAGES))
    text = LOADING_FONT_DETAILED.render(position, 1, FONT_COLOR)
    screen.blit(text, (0,0))


def search_for_images(search_term, img_sizes, num_urls_desired=RESULTS_PER_PAGE):

    term_logger = setup_term_logger(search_term)
    urls = set()

    #return if there are no more image sizes to try the search term against
    if not img_sizes:
        term_logger.info("Ran out of images to search for {}".format(search_term))
        return urls

    urls_found = 0
    img_size = img_sizes[0]
    next_start_index = QUERY_CACHE.get(search_term+img_size, 1)

    while urls_found < num_urls_desired:
        # API only returns a maximum of 100 results
        if next_start_index + RESULTS_PER_PAGE > 100:
            # if we have exhausted searching every image size, clear the cache and start again
            if len(img_sizes) == 1:
                for img_s in IMAGE_SIZES:
                    QUERY_CACHE[search_term + img_s] = 1
            else:
                # otherwise, try another image size
                return urls.union(search_for_images(search_term, img_sizes[1:], num_urls_desired - len(urls)))

        query_url = assemble_query(search_term, img_size, next_start_index)

        try:
            term_logger.info('Requesting url with term:{} size:{} start_index:{}'.format(search_term, img_size, next_start_index))
            r = requests.get(query_url)
            if r.status_code != requests.codes.ok:
                r.raise_for_status()
            json_data = r.json()

            if not json_data:
                #json data is empty if there are no more search results so try with another image size
                term_logger.info("No search results for {} {}".format(search_term, img_size))
                return urls.union(search_for_images(search_term, img_sizes[1:], num_urls_desired - len(urls)))

            for item in json_data['items']:
                link = item['link']

                #filter out x-raw-image:// urls
                if re.match('^https?://', link) and link not in urls and link not in IMAGE_BLACKLIST:
                    filename_hash = hashlib.sha1(link.encode('utf-8')).hexdigest()
                    filepath = os.path.join(IMAGE_DIR, search_term, filename_hash)
                    if filepath not in IMAGES:
                        urls.add(link)
                        urls_found += 1
                        term_logger.info("Found {} {} {}".format(filepath, img_size, filename_hash))
                        if urls_found == num_urls_desired:
                            break

            next_start_index = json_data['queries']['nextPage'][0]['startIndex']
            QUERY_CACHE[search_term+img_size] = next_start_index

        except (requests.exceptions.RequestException, KeyError) as e:
            error_str = 'Query Error: {}'.format(query_url)
            main_logger.info(error_str)
            term_logger.info(error_str)
            main_logger.info(e)
            term_logger.info(e)
            return urls

    return urls

def display_loading_progress(search_term, term_url_count, total_urls, urls_processed, term_count):

    percent_complete = (urls_processed/float(total_urls)) * 100
    percent_complete = int(percent_complete)
    msg = "{}%".format(percent_complete)

    loading_font_size = LOADING_FONT.get_ascent()
    text = LOADING_FONT.render(msg, 1, (random.randint(0,255), random.randint(0,255), random.randint(0,255)))
    x_coord = SCREEN_WIDTH - len(msg) * loading_font_size
    y_coord = 0

    SCREEN_LOCK.acquire()
    screen.fill(RGB_BLACK, Rect(x_coord, y_coord, len(msg) * loading_font_size, loading_font_size + TEXT_PADDING))
    screen.blit(text, (x_coord, y_coord))

    if not DETAILED_PROGRESS:
        display.flip()
        SCREEN_LOCK.release()
        return

    progress_font_size = LOADING_FONT_DETAILED.get_ascent()
    y_coord = SCREEN_HEIGHT - progress_font_size - TEXT_PADDING
    screen.fill(RGB_BLACK, Rect(0, y_coord, SCREEN_WIDTH, progress_font_size + TEXT_PADDING))

    msg = 'Total: {}/{} urls Search Term:"{}":{}/{} urls'.format(urls_processed, total_urls, search_term, term_count, term_url_count)
    text = LOADING_FONT_DETAILED.render(msg, 1, FONT_COLOR)
    screen.blit(text, (0, y_coord))
    display.flip()
    SCREEN_LOCK.release()

def display_file_download_progress(content_length, bytes_read, url, percent_complete):
    if not DETAILED_PROGRESS:
        return

    SCREEN_LOCK.acquire()
    progress_font_size = LOADING_FONT_DETAILED.get_ascent()
    y_coord_url = SCREEN_HEIGHT - progress_font_size*3 - TEXT_PADDING*3
    y_coord_dl = SCREEN_HEIGHT - progress_font_size*2 - TEXT_PADDING*2
    screen.fill(RGB_BLACK, Rect(0, y_coord_url, SCREEN_WIDTH, progress_font_size + TEXT_PADDING))
    screen.fill(RGB_BLACK, Rect(0, y_coord_dl, SCREEN_WIDTH, progress_font_size + TEXT_PADDING))

    text = LOADING_FONT_DETAILED.render(url, 1, FONT_COLOR)
    screen.blit(text, (0, y_coord_url))

    msg = "Download progress {}B/{}B".format(bytes_read, content_length)
    msg = msg + " {0:.2f}%".format(percent_complete)
    text = LOADING_FONT_DETAILED.render(msg, 1, FONT_COLOR)
    screen.blit(text, (0, y_coord_dl))

    display.flip()
    SCREEN_LOCK.release()

def clear_progress():
    font_size = LOADING_FONT_DETAILED.get_ascent()
    y_coord = SCREEN_HEIGHT - font_size*3 - TEXT_PADDING*3
    height = SCREEN_HEIGHT - y_coord
    screen.fill(RGB_BLACK, Rect(0, y_coord, SCREEN_WIDTH, height))
    display.flip()

def download_file(response, img, url):

    content_length = int(response.info().get('Content-Length').strip())
    bytes_read = 0
    percent_complete = 0

    while True:
        chunk = response.read(CHUNK_SIZE)

        if not chunk:
            return

        bytes_read += len(chunk)
        last_percent_complete = percent_complete
        percent_complete = (bytes_read/float(content_length)) * 100
        img.write(chunk)

        if percent_complete - last_percent_complete > 1:
            display_file_download_progress(content_length, bytes_read, url, percent_complete)

def download_images(term_dict, total_urls):

    urls_processed = 0
    successful_downloads = 0
    for search_term in term_dict:
        term_logger = logger_store[search_term]
        search_images_dir = os.path.join(IMAGE_DIR, search_term)
        url_count = len(term_dict[search_term])

        for i,url in enumerate(term_dict[search_term]):
            filename_hash = hashlib.sha1(url.encode('utf-8')).hexdigest()
            filename = os.path.join(search_images_dir, filename_hash)
            error = False

            with open(filename, 'wb') as img:
                try:
                    response = urllib.request.urlopen(urllib.request.Request(url, headers={'User-Agent': USER_AGENT}))
                    download_file(response, img, url)
                    IMAGES_LOCK.acquire()
                    IMAGES.add(filename)
                    IMAGES_LOCK.release()
                    term_logger.info("Downloaded {} {}".format(url, filename_hash))
                    successful_downloads = successful_downloads + 1
                except (urllib.error.HTTPError, urllib.error.URLError, AttributeError) as e:
                    error = True
                    main_logger.info(e)
                    term_logger.info(e)

                    IMAGE_BLACKLIST.add(url)
                    term_logger.info('Blacklisted url:{}'.format(url))

            if error and os.path.exists(filename):
                os.unlink(filename)
                term_logger.info("Failed to download {}".format(url))

            display_loading_progress(search_term, url_count, total_urls, urls_processed, i+1)
            urls_processed += 1
    return successful_downloads

def setup_term_logger(term):
    search_images_dir = os.path.join(IMAGE_DIR, term)
    if not os.path.exists(search_images_dir):
        os.mkdir(search_images_dir)
    try:
        logger = logger_store[term]
    except KeyError:
        logger = logger_store[term] = make_logger(term, os.path.join(search_images_dir, '{}.log'.format(term)), logging.DEBUG)
    return logger

def search_term_download():
    term_dict = {}
    total_urls = 0

    for term in vars['search_terms']:
        term_dict[term] = search_for_images(term, IMAGE_SIZES)
        total_urls += len(term_dict[term])

    items_downloaded = download_images(term_dict, total_urls)
    server.new_term_event.clear()

    if items_downloaded:
        REFRESH_EVENT.set()

def check_for_exit():
    for e in pygame.event.get():
        if e.type == pygame.KEYDOWN:
            if e.key == pygame.K_q:
                end()
            elif e.key == pygame.K_d:
                server.new_term_event.set()

def display_loading():
    '''Loading screen that displays if less than 10 images are available'''

    gifs = os.listdir(LOAD_IMAGE_DIR)
    loading_gif = GIFImage(os.path.join(LOAD_IMAGE_DIR, random.choice(gifs)))

    x_coord = get_center_width_offset(loading_gif.image)
    y_coord = get_center_height_offset(loading_gif.image)

    screen.fill(RGB_BLACK)

    while True:

        SCREEN_LOCK.acquire()
        loading_gif.render(screen, (x_coord,y_coord))
        display.flip()
        SCREEN_LOCK.release()

        check_for_exit()

        IMAGES_LOCK.acquire()
        if len(IMAGES) >= LOADING_PAGE_THRESHOLD:
            IMAGES_LOCK.release()
            return
        IMAGES_LOCK.release()


def scan_input():
    '''
    q - quits
    d - initiate download event
    arrow keys - advance the image
    i - display image position i.e 3/10
    delete - stop displaying the image and delete it. advance to the next image
    :return:
    '''
    global SHOW_IMAGE_POSITION

    e = pygame.event.poll()
    while e.type != pygame.NOEVENT:
        if e.type == pygame.QUIT:
            end()

        if e.type == pygame.KEYDOWN:
            if e.key == pygame.K_q:
                end()
            elif e.key == pygame.K_d:
                server.new_term_event.set()
            elif e.key == pygame.K_LEFT:
                return pygame.K_LEFT
            elif e.key == pygame.K_RIGHT:
                return pygame.K_RIGHT
            elif e.key == pygame.K_i:
                SHOW_IMAGE_POSITION = not SHOW_IMAGE_POSITION
            elif e.key == pygame.K_DELETE:
                return pygame.K_DELETE
        e = pygame.event.poll()

def idle_and_scan_input():
    start = datetime.datetime.now()
    next_tick = datetime.datetime.now()
    input_scan_rate = .1 # sec

    while (next_tick - start).seconds < vars['flip_frequency']:
        input = scan_input()
        if input:
            return input
        time.sleep(input_scan_rate)
        next_tick = datetime.datetime.now()

#*args for linux compatibility
def end(*args):
    main_logger.info('Exiting....')

    for conv_file in CONVERT_CACHE.values():
        os.unlink(conv_file)

    with open(IMAGE_BLACKLIST_FILENAME, 'w') as blacklist:
        blacklist.writelines('\n'.join(IMAGE_BLACKLIST))

    try:
        with open("qc.pickle",'wb') as qc_pickle_file:
            pickle.dump(QUERY_CACHE, qc_pickle_file, pickle.DEFAULT_PROTOCOL)
    except (pickle.PickleError, pickle.PicklingError) as e:
        main_logger.info('Failed to pickle query cache: {}'.format(e))

    config.save_config()
    server.shutdown()
    pygame.display.quit()
    pygame.quit()
    sys.exit(0)

def run():

    if not os.path.exists(IMAGE_DIR):
        os.mkdir(IMAGE_DIR)

    threading.Thread(target=server.serve_forever, daemon=True).start()
    ImageDownloader(IMAGES, search_term_download, server).start()

    IMAGES_LOCK.acquire()
    images = list(IMAGES)
    IMAGES_LOCK.release()

    i = 0

    while True:
        if len(images) < LOADING_PAGE_THRESHOLD:
            display_loading()

            #refresh images with those newly downloaded
            IMAGES_LOCK.acquire()
            images = list(IMAGES)
            IMAGES_LOCK.release()

        image = images[i]
        try:
            display_image(image, i)
        except IOError:
            i = (i + 1) % len(images)
            continue

        input = idle_and_scan_input()
        if input == pygame.K_LEFT:
            i = (i - 1) % len(images)
        elif input == pygame.K_DELETE:
            IMAGES_LOCK.acquire()
            IMAGES.remove(image)
            images = list(IMAGES)
            i = (i - 1) % len(images)
            IMAGES_LOCK.release()

            try:
                os.unlink(image)
            except OSError as e:
                main_logger.info(e)
        elif input == pygame.K_RIGHT or input is None:
            i = (i + 1) % len(images)

        # reload the images to display because 1) ImageCleaner cleaned out images
        # or 2) The server added/removed extra images to display ("extra_images")
        # or 3) ImageDownloader finished downloading new images
        if REFRESH_EVENT.is_set():
            IMAGES_LOCK.acquire()
            images = list(IMAGES)
            i = 0
            IMAGES_LOCK.release()
            REFRESH_EVENT.clear()


screen = display.set_mode(IMAGE_SIZE, pygame.FULLSCREEN)
# screen = display.set_mode(IMAGE_SIZE)
signal.signal(signal.SIGINT,end)
run()