zulip/tools/setup/emoji/build_emoji

#!/usr/bin/env python3
#
# See docs/subsystems/emoji.md for a high-level explanation of how this system
# works.
import os
import shutil
import sys
import ujson

from typing import Any, Dict, Iterator, List, Optional

from emoji_setup_utils import generate_emoji_catalog, generate_codepoint_to_name_map, \
    get_emoji_code, generate_name_to_codepoint_map, emoji_names_for_picker, \
    EMOTICON_CONVERSIONS, REMAPPED_EMOJIS
from emoji_names import EMOJI_NAME_MAPS

ZULIP_PATH = os.path.join(os.path.dirname(os.path.abspath(__file__)), '../../../')
sys.path.append(ZULIP_PATH)

from scripts.lib.zulip_tools import generate_sha1sum_emoji, run_as_root

TARGET_EMOJI_DUMP = os.path.join(ZULIP_PATH, 'static', 'generated', 'emoji')
TARGET_EMOJI_STYLES = os.path.join(ZULIP_PATH, 'static', 'generated', 'emoji-styles')
EMOJI_CACHE_PATH = "/srv/zulip-emoji-cache"
EMOJI_SCRIPT_DIR_PATH = os.path.join(ZULIP_PATH, 'tools', 'setup', 'emoji')
NODE_MODULES_PATH = os.path.join(ZULIP_PATH, 'node_modules')


# The CSS for emoji spritesheet has somewhat tricky requirements.  One
# is that we want to be able to use the same emoji CSS classes for
# different display sizes of our emoji (e.g. reactions are smaller
# than inline message emoji, which are smaller than those in the emoji
# picker) while only downloading 1 copy of the spritesheet, having
# good browser rendering performance, and reusing as much common CSS
# as is possible.

# Our solution to those problem is to use the `background-size` (Which
# is e.g. 5700%) and background-position attributes to select the
# region of the spritesheet corresponding to the target sprite and
# display it properly scaled in an emoji span.
SPRITE_CSS_FILE_TEMPLATE = """\
div.emoji,
span.emoji
{
    display: inline-block;
    background-image: url(~emoji-datasource-%(emojiset)s/img/%(alt_name)s/sheets-256/64.png);
    background-size: %(background_size)s;
    background-repeat: no-repeat;

    /* Hide the text. */
    text-indent: 100%%;
    white-space: nowrap;
    overflow: hidden;
}

.emoji-1f419
{
    background-image: url(../emoji/images-google-64/1f419.png) !important;
    background-position: 0%% 0%% !important;
    background-size: contain !important;
}

%(emoji_positions)s
"""

EMOJI_POS_INFO_TEMPLATE = """\
.emoji-%(codepoint)s {
    background-position: %(pos_x)s %(pos_y)s;
}
"""

# change directory
os.chdir(EMOJI_SCRIPT_DIR_PATH)

if 'TRAVIS' in os.environ:
    # In Travis CI, we don't have root access
    EMOJI_CACHE_PATH = "/home/travis/zulip-emoji-cache"

def main() -> None:
    if not os.access(EMOJI_CACHE_PATH, os.W_OK):
        # Note: In production, this block will fail, since we don't
        # assume sudo access; but it should never run in production
        # anyway, because EMOJI_CACHE_PATH is created by puppet before
        # build_emoji would be run.
        run_as_root(["mkdir", "-p", EMOJI_CACHE_PATH])
        run_as_root(["chown", "%s:%s" % (os.getuid(), os.getgid()), EMOJI_CACHE_PATH])

    sha1_hexdigest = generate_sha1sum_emoji(ZULIP_PATH)
    source_emoji_dump = os.path.join(EMOJI_CACHE_PATH, sha1_hexdigest, 'emoji')
    success_stamp = os.path.join(source_emoji_dump, '.success-stamp')

    if not os.path.exists(success_stamp):
        print("Dumping emojis ...")
        dump_emojis(source_emoji_dump)
        with open(success_stamp, 'w') as f:
            f.close()

    print("build_emoji: Using cached emojis from {}".format(source_emoji_dump))
    if os.path.lexists(TARGET_EMOJI_DUMP):
        os.remove(TARGET_EMOJI_DUMP)
    os.symlink(source_emoji_dump, TARGET_EMOJI_DUMP)

    # These must not be symlinked so webpack can resolve module references.
    os.makedirs(TARGET_EMOJI_STYLES, exist_ok=True)
    to_remove = set(os.listdir(TARGET_EMOJI_STYLES))
    for filename in os.listdir(source_emoji_dump):
        if filename.endswith(".css"):
            shutil.copy2(os.path.join(source_emoji_dump, filename), TARGET_EMOJI_STYLES)
            to_remove.discard(filename)
    for filename in to_remove:
        os.remove(os.path.join(TARGET_EMOJI_STYLES, filename))

def percent(f: float) -> str:
    return '%0.3f%%' % (f * 100,)

def get_square_size(emoji_data: List[Dict[str, Any]]) -> int:
    """
    Spritesheets are usually NxN squares, and we have to
    infer N from the sheet_x/sheet_y values of emojis.
    """
    def get_offsets(emoji_data: List[Dict[str, Any]]) -> Iterator[int]:
        for emoji_dict in emoji_data:
            yield emoji_dict['sheet_x']
            yield emoji_dict['sheet_y']
            if 'skin_variations' in emoji_dict:
                for img_info in emoji_dict['skin_variations'].values():
                    yield img_info['sheet_x']
                    yield img_info['sheet_y']

    n = max(get_offsets(emoji_data)) + 1
    return n

def generate_sprite_css_files(cache_path: str,
                              emoji_data: List[Dict[str, Any]],
                              emojiset: str,
                              alt_name: str) -> None:
    """
    Spritesheets are usually NxN squares.
    """
    n = get_square_size(emoji_data)

    """
    Each single emoji is 64x64, with 1px gutters on every border.
    We just consider the gutters to be part of the image for
    simplicity reasons, so you can think of the spritesheet as
    an NxN square of 66x66 pre-padded emojis.  The CSS
    background-size parameter below says to size the background
    element as N times the size of the element that you're drawing.

    Note that we use percentages here, instead of absolute
    pixel values, because when we render emojis as actual elements,
    their size will vary depending on which part of the UI we're
    in (message emojis, emoji reactions, emoji popup, emoji
    popup showcase, etc.).

    (The next step is to offset the image; that will be in the
    upcoming loop.)
    """
    background_size = percent(n)

    emoji_positions = ""
    for emoji in emoji_data:
        if emoji["has_img_google"]:
            # Why is the test here has_img_google and not
            # emoji_is_universal?  Because we briefly supported all
            # Google emoji (not just the universal ones), we need to
            # ensure the spritesheet is setup to correctly display
            # those google emoji (in case anyone used them).

            """
            For background-position we need to use percentages.
            Absolute pixel values won't work, because the size
            of the background sprite image is proportional to
            the size of the element we're rendering, and we render
            elements in multiple sizes.

            The way that CSS background-position works is linear
            interpolation.  When you tell CSS background-position
            is "42% 37%", then in the `x` dimension it will align
            the image such that 42% of the background image is to
            the left of the 42% mark in the element itself.

            For simplicity assume we render the emoji as 66px
            (and everything will scale appropriately for other
            size images as long as we use percentages).

            The image size will be 66n.
            The left offset of the x-th emoji (including its
            padding) will be 66x.  And the element's width
            will be 66.

            So, solve this equation for `p`, where p is
            the ratio that we'll later express as a
            percentage:

                <image offset> = <offset of p% mark of element>
                (p * 66n) = 66x + p66
                p * n = x + p
                p * n - p = x
                p * (n - 1) = x
                p = x / (n - 1)

            If you ever want to change the code so that the
            gutters don't show up in the element, the algebra
            will get more complicated.
            """

            emoji_positions += EMOJI_POS_INFO_TEMPLATE % {
                'codepoint': get_emoji_code(emoji),
                'pos_x': percent(emoji["sheet_x"] / (n - 1)),
                'pos_y': percent(emoji["sheet_y"] / (n - 1)),
            }

    SPRITE_CSS_PATH = os.path.join(cache_path, '%s-sprite.css' % (emojiset,))
    with open(SPRITE_CSS_PATH, 'w') as f:
        f.write(SPRITE_CSS_FILE_TEMPLATE % {'emojiset': emojiset,
                                            'alt_name': alt_name,
                                            'emoji_positions': emoji_positions,
                                            'background_size': background_size,
                                            })

def setup_emoji_farms(cache_path: str, emoji_data: List[Dict[str, Any]]) -> None:
    def ensure_emoji_image(emoji_dict: Dict[str, Any],
                           src_emoji_farm: str,
                           target_emoji_farm: str) -> None:
        # We use individual images from emoji farm for rendering emojis
        # in notification messages. We have a custom emoji formatter in
        # notifications processing code that converts `span` tags to
        # `img` and that logic requires us to have non-qualified
        # `emoji_code` as file name for emoji.
        emoji_code = get_emoji_code(emoji_dict)
        img_file_name = emoji_code + '.png'
        src_file = os.path.join(src_emoji_farm, emoji_dict['image'])
        dst_file = os.path.join(target_emoji_farm, img_file_name)
        shutil.copy2(src_file, dst_file)

    def setup_emoji_farm(emojiset: str,
                         emoji_data: List[Dict[str, Any]],
                         alt_name: Optional[str]=None) -> None:
        # `alt_name` is an optional parameter that we use to avoid duplicating below
        # code. It is only used while setting up google-blob emojiset as it is just
        # a wrapper for an older version of emoji-datasource package due to which we
        # need to use 'google' at some places in this code. It has no meaning for other
        # emojisets and is just equivalent to `emojiset`.
        alt_name = alt_name or emojiset

        # Copy individual emoji images from npm packages.
        src_emoji_farm = os.path.join(
            NODE_MODULES_PATH, 'emoji-datasource-' + emojiset, 'img', alt_name, '64')
        target_emoji_farm = os.path.join(cache_path, 'images-' + emojiset + '-64')
        os.makedirs(target_emoji_farm, exist_ok=True)
        print("Copying individual image files...")
        for emoji_dict in emoji_data:
            if emoji_dict['has_img_' + alt_name]:
                ensure_emoji_image(emoji_dict, src_emoji_farm, target_emoji_farm)
                skin_variations = emoji_dict.get('skin_variations', {})
                for skin_tone, img_info in skin_variations.items():
                    if img_info['has_img_' + alt_name]:
                        ensure_emoji_image(img_info, src_emoji_farm, target_emoji_farm)

        # Copy zulip.png to the emoji farm.
        zulip_image = os.path.join(ZULIP_PATH, 'static', 'assets', 'zulip-emoji')
        for f in os.listdir(zulip_image):
            shutil.copy2(os.path.join(zulip_image, f), target_emoji_farm, follow_symlinks=False)

        # We hardcode octopus emoji image to Google emojiset's old
        # "cute octopus" image. Copy it to the emoji farms.
        input_img_file = os.path.join(EMOJI_SCRIPT_DIR_PATH, '1f419.png')
        output_img_file = os.path.join(cache_path, 'images-' + emojiset + '-64', '1f419.png')
        shutil.copyfile(input_img_file, output_img_file)

        generate_sprite_css_files(cache_path, emoji_data, emojiset, alt_name)

    # Setup standard emojisets.
    for emojiset in ['google', 'twitter']:
        setup_emoji_farm(emojiset, emoji_data)

    # Setup old google "blobs" emojiset.
    GOOGLE_BLOB_EMOJI_DATA_PATH = os.path.join(NODE_MODULES_PATH,
                                               'emoji-datasource-google-blob',
                                               'emoji.json')
    with open(GOOGLE_BLOB_EMOJI_DATA_PATH) as fp:
        blob_emoji_data = ujson.load(fp)
    setup_emoji_farm('google-blob', blob_emoji_data, 'google')

def setup_old_emoji_farm(cache_path: str,
                         emoji_map: Dict[str, str],
                         emoji_data: List[Dict[str, Any]]) -> None:
    # Code for setting up old emoji farm.
    os.chdir(cache_path)
    emoji_cache_path = os.path.join(cache_path, 'images', 'emoji')
    unicode_emoji_cache_path = os.path.join(cache_path, 'images', 'emoji', 'unicode')
    google_emoji_cache_path = os.path.join(cache_path, 'images-google-64')
    os.makedirs(emoji_cache_path, exist_ok=True)
    os.makedirs(unicode_emoji_cache_path, exist_ok=True)

    # Symlink zulip.png image file.
    image_file_path = os.path.join(google_emoji_cache_path, 'zulip.png')
    symlink_path = os.path.join(emoji_cache_path, 'zulip.png')
    os.symlink(image_file_path, symlink_path)

    unicode_symlink_path = os.path.join(unicode_emoji_cache_path, 'zulip.png')
    os.symlink(image_file_path, unicode_symlink_path)

    for name, codepoint in emoji_map.items():
        mapped_codepoint = REMAPPED_EMOJIS.get(codepoint, codepoint)
        image_file_path = os.path.join(google_emoji_cache_path, '{}.png'.format(mapped_codepoint))
        symlink_path = os.path.join(emoji_cache_path, '{}.png'.format(name))
        os.symlink(image_file_path, symlink_path)
        try:
            # `emoji_map` contains duplicate entries for the same codepoint with different
            # names. So creation of symlink for <codepoint>.png may throw `FileExistsError`.
            unicode_symlink_path = os.path.join(unicode_emoji_cache_path, '{}.png'.format(codepoint))
            os.symlink(image_file_path, unicode_symlink_path)
        except FileExistsError:
            pass

def generate_map_files(cache_path: str, emoji_catalog: Dict[str, List[str]]) -> None:
    # This function generates the data file consumed by webapp, mobile apps, bugdown etc.
    names = emoji_names_for_picker(EMOJI_NAME_MAPS)
    codepoint_to_name = generate_codepoint_to_name_map(EMOJI_NAME_MAPS)
    name_to_codepoint = generate_name_to_codepoint_map(EMOJI_NAME_MAPS)

    EMOJI_CODES_FILE_PATH = os.path.join(cache_path, 'emoji_codes.json')
    with open(EMOJI_CODES_FILE_PATH, 'w') as emoji_codes_file:
        ujson.dump({
            'names': names,
            'name_to_codepoint': name_to_codepoint,
            'codepoint_to_name': codepoint_to_name,
            'emoji_catalog': emoji_catalog,
            'emoticon_conversions': EMOTICON_CONVERSIONS,
        }, emoji_codes_file)

def dump_emojis(cache_path: str) -> None:
    with open('emoji_map.json') as emoji_map_file:
        emoji_map = ujson.load(emoji_map_file)

    # `emoji.json` or any other data file can be sourced from any of the supported
    # emojiset packages, they all contain the same data files.
    EMOJI_DATA_FILE_PATH = os.path.join(NODE_MODULES_PATH, 'emoji-datasource-google', 'emoji.json')
    with open(EMOJI_DATA_FILE_PATH) as emoji_data_file:
        emoji_data = ujson.load(emoji_data_file)
    emoji_catalog = generate_emoji_catalog(emoji_data, EMOJI_NAME_MAPS)

    # Setup emoji farms.
    if os.path.exists(cache_path):
        shutil.rmtree(cache_path)
    setup_emoji_farms(cache_path, emoji_data)
    setup_old_emoji_farm(cache_path, emoji_map, emoji_data)

    # Generate various map files.
    generate_map_files(cache_path, emoji_catalog)

if __name__ == "__main__":
    main()