path: root/cgi/img.py

# coding=utf-8
import struct
import math
#import random
import os
import subprocess
import logging
from io import BytesIO

from settings import Settings
from database import *
from framework import *

try:  # Windows needs stdio set for binary mode.
    import msvcrt
    msvcrt.setmode(0, os.O_BINARY)  # stdin  = 0
    msvcrt.setmode(1, os.O_BINARY)  # stdout = 1
except ImportError:
    pass


def processImage(post, data, t, originalname, spoiler=False):
    """
    Take all post data from <post>, process uploaded file in <data>, and calculate
    file names using datetime <t>
    Returns updated <post> with file and thumb values
    """
    board = Settings._.BOARD

    used_filetype = None

    # get image information
    content_type, width, height, size, extra = getImageInfo(data)
    #raise Exception(repr(getImageInfo(data)))
    logging.info("{} {} {}".format(content_type, width, height))

    # check the size is fine
    if size > int(board["maxsize"])*1024:
        raise UserError(_("File too big. The maximum file size is: %s") % board['maxsize'])

    # check if file is supported
    for filetype in board['filetypes']:
        if content_type == filetype['mime']:
            used_filetype = filetype
            break

    if not used_filetype:
        raise UserError(_("File type not supported."))

    # check if file is already posted
    is_duplicate = checkFileDuplicate(data)
    if checkFileDuplicate(data)[0]:
        raise UserError(_("This image has already been posted %s.") % ('<a href="' + Settings.BOARDS_URL + board['dir'] + '/res/' + str(is_duplicate[1]) + '.html#' + str(is_duplicate[2]) + '">' + _("here") + '</a>'))

    # prepare file names
    if used_filetype['preserve_name']:
        file_base = os.path.splitext(originalname)[0]  # use original filename
    else:
        file_base = '%d' % int(t * 1000)  # generate timestamp name
    file_name = file_base + "." + used_filetype['ext']
    file_thumb_name = file_base + "s.jpg"

    # prepare paths
    file_path = Settings.IMAGES_DIR + board["dir"] + "/src/" + file_name
    file_thumb_path = Settings.IMAGES_DIR + \
        board["dir"] + "/thumb/" + file_thumb_name
    file_mobile_path = Settings.IMAGES_DIR + \
        board["dir"] + "/mobile/" + file_thumb_name
    file_cat_path = Settings.IMAGES_DIR + \
        board["dir"] + "/cat/" + file_thumb_name

    # remove EXIF data if necessary for privacy
    if content_type == 'image/jpeg':
        data = removeExifData(data)

    # write file
    f = open(file_path, "wb")
    try:
        f.write(data)
    finally:
        f.close()

    # set maximum dimensions
    maxsize = int(board['thumb_px'])

    post["file"] = file_name
    post["image_width"] = width
    post["image_height"] = height

    # Do we need to thumbnail it?
    if not used_filetype['image']:
        # make thumbnail
        logging.debug("Generating thumbnail")
        file_thumb_width, file_thumb_height = getThumbDimensions(width, height, maxsize)
        
        try:
            if used_filetype['ffmpeg_thumb']:
                # use ffmpeg to make thumbnail
                if used_filetype['mime'][:5] == 'video':
                    # Create preview for video AND spoiler it if necessary
                    call_wrap([
                        Settings.FFMPEG_PATH, '-strict', '-2', '-ss', '0', '-i', file_path,
                        '-v', 'quiet', '-an', '-vframes', '1', '-f', 'mjpeg', '-vf', 'scale=%d:%d' % (
                            file_thumb_width, file_thumb_height),
                        '-threads', '1', file_thumb_path])
                    if spoiler:
                        call_wrap([Settings.CONVERT_PATH, file_thumb_path, "-limit", "thread",
                            "1", "-background", "white", "-flatten", "-resize",
                            "%dx%d" % (file_thumb_width, file_thumb_height), "-blur", "0x12",
                            "-gravity", "center", "-fill", "rgba(0,0,0, .6)",
                            "-draw", "rectangle 0,%d,%d,%d" % ((file_thumb_height/2)-10, file_thumb_width, (file_thumb_height/2)+7),
                            "-fill", "white", "-font", "Liberation-Sans", "-annotate", "0", "Alerta de spoiler",
                            "-quality", str(Settings.THUMB_QUALITY), file_thumb_path])
                elif used_filetype['mime'][:5] == 'audio':
                    # we do an exception and use png for audio waveform thumbnails since they
                    # 1. are smaller 2. allow for transparency
                    file_thumb_name = file_thumb_name[:-3] + "png"
                    file_thumb_path = file_thumb_path[:-3] + "png"
                    file_mobile_path = file_mobile_path[:-3] + "png"
                    file_cat_path = file_cat_path[:-3] + "png"

                    if int(board['thumb_px']) > 149:
                        file_thumb_width = board['thumb_px']
                        file_thumb_height = float(int(board['thumb_px'])/2)
                    else:
                        file_thumb_width = 150
                        file_thumb_height = 75

                    call_wrap([Settings.FFMPEG_PATH, '-t', '300', '-i', file_path,
                        '-filter_complex', 'showwavespic=s=%dx%d:split_channels=1' % (
                        int(file_thumb_width), int(file_thumb_height)),
                        '-frames:v', '1', '-threads', '1', file_thumb_path])
            else:
                # use imagemagick to make thumbnail
                args = [Settings.CONVERT_PATH, file_path, "-limit", "thread", "1", "-background",
                        "white", "-flatten", "-resize",  "%dx%d" % (file_thumb_width, file_thumb_height)]
                if spoiler:
                    args += ["-blur", "0x12", "-gravity", "center", "-fill", "rgba(0,0,0, .6)", "-draw", "rectangle 0,%d,%d,%d" % (
                        (file_thumb_height/2)-10, file_thumb_width, (file_thumb_height/2)+7), "-fill", "white", 
                        "-font", "Liberation-Sans", "-annotate", "0", "Alerta de spoiler"]
                args += ["-quality", str(Settings.THUMB_QUALITY), file_thumb_path]

                # generate thumbnails
                call_wrap(args)
        except subprocess.CalledProcessError as e:
            os.remove(file_path)
            logging.error("Thumbnail creation failure: " + str(e.output))
            raise UserError(_("Thumbnail creation failure.") + ' ('+str(e.returncode)+')')

        # check if thumbnail was truly created
        try:
            open(file_thumb_path)
        except:
            os.remove(file_path)
            raise UserError(_("Thumbnail creation failure."))

        # create extra thumbnails (catalog/mobile)
        subprocess.call([Settings.CONVERT_PATH, file_thumb_path, "-limit", "thread",
                         "1", "-resize",  "100x100", "-quality", "75", file_mobile_path])
        if not post["parentid"]:
            subprocess.call([Settings.CONVERT_PATH, file_thumb_path, "-limit",
                             "thread", "1", "-resize",  "150x150", "-quality", "60", file_cat_path])

        post["thumb"] = file_thumb_name
        post["thumb_width"] = file_thumb_width
        post["thumb_height"] = file_thumb_height
    else:
        # Don't thumbnail and use mime image
        if board["board_type"] == 0:
            post["thumb"] = used_filetype['image']
            post["thumb_width"] = 120
            post["thumb_height"] = 120
        else:
            post["thumb"] = used_filetype['image'].split(".")[0] + '_small.png'
            post["thumb_width"] = 90
            post["thumb_height"] = 90

    # calculate size (bytes)
    post["file_size"] = len(data)

    # add additional metadata, if any
    post["message"] += extraInfo(content_type, file_name, file_path)

    # file md5
    post["file_hex"] = getMD5b(data)

    return post


def extraInfo(mime, file_name, file_path):
    board = Settings._.BOARD

    if mime in ['audio/ogg', 'audio/opus', 'audio/mpeg', 'video/webm', 'video/mp4']:
        info = ffprobe_f(file_path)
        extra = {}
        credit_str = ""

        if mime == 'video/webm':
            for s in info['streams']:
                if 'width' in s:
                    stream = s
        else:
            stream = info['streams'][0]

        extra['codec'] = stream.get('codec_name', '')
        format = info['format']

        if 'bit_rate' in format:
            extra['codec'] += ' ~%d kbps' % int(int(format['bit_rate']) / 1000)
        if 'tags' in format:
            extra['title'] = format['tags'].get(
                'TITLE', format['tags'].get('title', ''))
            extra['artist'] = format['tags'].get(
                'ARTIST', format['tags'].get('artist', ''))
            if extra['title'] or extra['artist']:
                credit_str = ' - '.join((extra['artist'],
                                         extra['title'])) + ' '
        if 'tags' in stream:
            extra['title'] = stream['tags'].get('TITLE', '')
            extra['artist'] = stream['tags'].get('ARTIST', '')
            if extra['title'] or extra['artist']:
                credit_str = ' - '.join((extra['artist'],
                                         extra['title'])) + ' '

        return '<hr /><small>%s(%s)</small>' % (credit_str, extra['codec'])

    elif mime in ['audio/mod', 'audio/xm', 'audio/s3m']:
        ext = mime.split('/')[1].upper()
        url = '/cgi/play/%s/%s' % (board['dir'], file_name)
        return '<hr /><small>Módulo tracker (%s) [<a href="%s" target="_blank">Click para escuchar</a>]</small>' % (ext, url)

    return ''


def getImageInfo(data):
    size = len(data)
    height = 0
    width = 0
    extra = {}
    content_type = ""

    # handle GIFs
    if (size >= 10) and data[:6] in (b"GIF87a", b"GIF89a"):
        # Check to see if content_type is correct
        content_type = "image/gif"
        w, h = struct.unpack("<HH", data[6:10])
        width = int(w)
        height = int(h)

    # See PNG 2. Edition spec (http://www.w3.org/TR/PNG/)
    # Bytes 0-7 are below, 4-byte chunk length, then 'IHDR'
    # and finally the 4-byte width, height
    elif ((size >= 24) and data.startswith(b"\211PNG\r\n\032\n")
          and (data[12:16] == b"IHDR")):
        content_type = "image/png"
        w, h = struct.unpack(">LL", data[16:24])
        width = int(w)
        height = int(h)

    # Maybe this is for an older PNG version.
    elif (size >= 16) and data.startswith(b"\211PNG\r\n\032\n"):
        # Check to see if we have the right content type
        content_type = "image/png"
        w, h = struct.unpack(">LL", data[8:16])
        width = int(w)
        height = int(h)

    # handle JPEGs
    elif (size >= 2) and data.startswith(b"\377\330"):
        content_type = "image/jpeg"
        jpeg = BytesIO(data)
        jpeg.read(2)
        b = jpeg.read(1)
        try:
            while (b and ord(b) != 0xDA):
                while (ord(b) != 0xFF):
                    b = jpeg.read
                while (ord(b) == 0xFF):
                    b = jpeg.read(1)
                if (ord(b) >= 0xC0 and ord(b) <= 0xC3):
                    jpeg.read(3)
                    h, w = struct.unpack(">HH", jpeg.read(4))
                    break
                else:
                    jpeg.read(int(struct.unpack(">H", jpeg.read(2))[0])-2)
                b = jpeg.read(1)
            width = int(w)
            height = int(h)
        except struct.error:
            raise
        except ValueError:
            raise

    # handle WebP
    if data[:4] == b'RIFF' and data[8:12] == b'WEBP':
        chunk = data[12:]
        if chunk[:4] == b"VP8 " and chunk[11:14] == b"\x9d\x01\x2a":
            # Lossy VP8
            w, h = struct.unpack("HH", chunk[14:18])
            width  = w & 0x3fff
            height = h & 0x3fff
            content_type = "image/webp"
        elif chunk[:4] == b"VP8L":
            # Lossless VP8
            b0, b1, b2, b3 = struct.unpack("BBBB", chunk[9:13])
            width  = 1 + (((b1 & 0x3F) << 8) | b0)
            height = 1 + (((b3 & 0xF) << 10) | (b2 << 2) | ((b1 & 0xC0) >> 6))
            content_type = "image/webp"

    # handle WebM
    elif (size >= 4) and data.startswith(b"\x1A\x45\xDF\xA3"):
        content_type = "video/webm"
        info = ffprobe(data)

    # handle mp4
    elif (size >= 8) and data[4:12] in [b"ftypmp42", b"ftypisom"]:
        content_type = "video/mp4"

    # handle ogg formats (vorbis/opus)
    elif (size >= 64) and data[:4] == b"OggS":
        if data[28:35] == b"\x01vorbis":
            content_type = "audio/ogg"
        elif data[28:36] == b"OpusHead":
            content_type = "audio/opus"

    # handle MP3
    elif (size >= 64) and (data[:3] == b"ID3" or data[:3] == b"\xFF\xFB"):
        content_type = "audio/mpeg"

    # handle MOD
    elif (size >= 64) and data[1080:1084] == b"M.K.":
        content_type = "audio/mod"

    # handle XM
    elif (size >= 64) and data.startswith(b"Extended Module:"):
        content_type = "audio/xm"

    # handle S3M
    elif (size >= 64) and data[25:32] == b"\x00\x00\x00\x1A\x10\x00\x00":
        content_type = "audio/s3m"

    # handle PDF
    elif (size >= 4) and data[:7] == b"%PDF-1.":
        content_type = "application/pdf"

    # handle Shockwave Flash
    elif (size >= 3) and data[:3] in [b"CWS", b"FWS"]:
        content_type = "application/x-shockwave-flash"

    # handle torrent
    elif (size >= 11) and data[:11] == b"d8:announce":
        content_type = "application/x-bittorrent"

    # handle PDF
    elif (size >= 2) and data[:2] == b"PK":
        content_type = "application/epub+zip"

    if content_type.startswith("video"):
        info = ffprobe(data)
        for stream in info['streams']:
            if 'width' in stream:
                width = stream['width']
                height = stream['height']
                break

    return content_type, width, height, size, extra


def ffprobe(data):
    import json
    p = subprocess.Popen([Settings.FFPROBE_PATH, '-v', 'quiet', '-print_format', 'json', '-show_format', '-show_streams', '-'],
                         stdout=subprocess.PIPE, stdin=subprocess.PIPE, stderr=subprocess.PIPE)

    out = p.communicate(input=data)[0]
    return json.loads(out)


def ffprobe_f(filename):
    import json

    p = subprocess.Popen([Settings.FFPROBE_PATH, '-v', 'quiet', '-print_format', 'json', '-show_format', '-show_streams', filename],
                         stdout=subprocess.PIPE, stdin=subprocess.PIPE, stderr=subprocess.PIPE)

    out = p.communicate()[0]
    return json.loads(out)


def call_wrap(args):
    logging.info("Calling: " + repr(args))
    subprocess.check_output(args, stderr=subprocess.STDOUT)


def getThumbDimensions(width, height, maxsize):
    """
    Calculate dimensions to use for a thumbnail with maximum width/height of
    <maxsize>, keeping aspect ratio
    """
    if not width or not height:
        return 0, 0

    wratio = (float(maxsize) / float(width))
    hratio = (float(maxsize) / float(height))

    if (width <= maxsize) and (height <= maxsize):
        return width, height
    else:
        if (wratio * height) < maxsize:
            thumb_height = math.ceil(wratio * height)
            thumb_width = maxsize
        else:
            thumb_width = math.ceil(hratio * width)
            thumb_height = maxsize

    return int(thumb_width), int(thumb_height)


def checkFileDuplicate(data):
    """
    Check that the file <data> does not already exist in a live post on the
    current board by calculating its hex and checking it against the database
    """
    board = Settings._.BOARD

    file_hex = getMD5b(data)
    post = FetchOne("SELECT `id`, `parentid` FROM `posts` WHERE `file_hex` = %s AND `boardid` = %s AND IS_DELETED = 0 LIMIT 1", (file_hex, board['id']))
    if post:
        if post["parentid"]:
            return True, post["parentid"], post["id"]
        else:
            return True, post["id"], post["id"]
    else:
        return False, 0, 0


def getJpegSegments(data):
    if data[0:2] != b"\xff\xd8":
        raise UserError("Given data isn't JPEG.")

    head = 2
    segments = [b"\xff\xd8"]
    while 1:
        if data[head: head + 2] == b"\xff\xda":
            yield data[head:]
            break
        else:
            length = struct.unpack(">H", data[head + 2: head + 4])[0]
            endPoint = head + length + 2
            seg = data[head: endPoint]
            yield seg
            head = endPoint

        if (head >= len(data)):
            raise UserDataError("Wrong JPEG data.")


def removeExifData(src_data):
    exif = None

    for seg in getJpegSegments(src_data):
        if seg[0:2] == b"\xff\xe1" and seg[4:10] == b"Exif\x00\x00":
            exif = seg
            break

    if exif:
        return src_data.replace(exif, b"")
    else:
        return src_data