moment/src/python/image_provider.py

182 lines
5.3 KiB
Python
Raw Normal View History

import asyncio
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
import logging as log
import random
import re
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
from dataclasses import dataclass, field
from io import BytesIO
from pathlib import Path
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
from typing import Optional, Tuple
from urllib.parse import urlparse
import aiofiles
from PIL import Image as PILImage
import nio
from nio.api import ResizingMethod
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
from . import pyotherside, utils
from .pyotherside import ImageData, Size
POSFormat = int
CONCURRENT_DOWNLOADS_LIMIT = asyncio.BoundedSemaphore(8)
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
with BytesIO() as img_out:
PILImage.new("RGBA", (1, 1), (0, 0, 0, 0)).save(img_out, "PNG")
TRANSPARENT_1X1_PNG = (img_out.getvalue(), pyotherside.format_data)
@dataclass
class Thumbnail:
provider: "ImageProvider" = field()
mxc: str = field()
width: int = field()
height: int = field()
def __post_init__(self) -> None:
self.mxc = re.sub(r"#auto$", "", self.mxc)
if not re.match(r"^mxc://.+/.+", self.mxc):
raise ValueError(f"Invalid mxc URI: {self.mxc}")
@property
def server_size(self) -> Tuple[int, int]:
# https://matrix.org/docs/spec/client_server/latest#thumbnails
if self.width > 640 or self.height > 480:
return (800, 600)
if self.width > 320 or self.height > 240:
return (640, 480)
if self.width > 96 or self.height > 96:
return (320, 240)
if self.width > 32 or self.height > 32:
return (96, 96)
return (32, 32)
@property
def resize_method(self) -> ResizingMethod:
return ResizingMethod.scale \
if self.width > 96 or self.height > 96 else ResizingMethod.crop
@property
def http(self) -> str:
return nio.Api.mxc_to_http(self.mxc)
@property
def local_path(self) -> Path:
parsed = urlparse(self.mxc)
name = "%s.%03d.%03d.%s" % (
parsed.path.lstrip("/"),
self.server_size[0],
self.server_size[1],
self.resize_method.value,
)
return self.provider.cache / parsed.netloc / name
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
async def read_data(self, data: bytes, mime: Optional[str],
) -> Tuple[bytes, POSFormat]:
if mime == "image/svg+xml":
return (data, pyotherside.format_svg_data)
if mime in ("image/jpeg", "image/png"):
return (data, pyotherside.format_data)
try:
with BytesIO(data) as img_in:
image = PILImage.open(img_in)
if image.mode == "RGB":
return (data, pyotherside.format_rgb888)
if image.mode == "RGBA":
return (data, pyotherside.format_argb32)
with BytesIO() as img_out:
image.save(img_out, "PNG")
return (img_out.getvalue(), pyotherside.format_data)
except OSError as err:
log.warning("Unable to process image: %s - %r", self.http, err)
return TRANSPARENT_1X1_PNG
async def download(self) -> Tuple[bytes, POSFormat]:
client = random.choice(
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
tuple(self.provider.app.backend.clients.values()),
)
parsed = urlparse(self.mxc)
async with CONCURRENT_DOWNLOADS_LIMIT:
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
resp = await client.thumbnail(
server_name = parsed.netloc,
media_id = parsed.path.lstrip("/"),
width = self.server_size[0],
height = self.server_size[1],
method = self.resize_method,
)
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
if isinstance(resp, nio.ThumbnailError):
log.warning("Downloading thumbnail failed - %s", resp)
return TRANSPARENT_1X1_PNG
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
body, pos_format = await self.read_data(resp.body, resp.content_type)
self.local_path.parent.mkdir(parents=True, exist_ok=True)
async with aiofiles.open(self.local_path, "wb") as file:
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
# body might have been converted, always save the original image.
await file.write(resp.body)
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
return (body, pos_format)
async def local_read(self) -> Tuple[bytes, POSFormat]:
data = self.local_path.read_bytes()
with BytesIO(data) as data_io:
return await self.read_data(data, utils.guess_mime(data_io))
async def get_data(self) -> ImageData:
try:
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
data, pos_format = await self.local_read()
except (OSError, IOError, FileNotFoundError):
data, pos_format = await self.download()
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
with BytesIO(data) as img_in:
real_size = PILImage.open(img_in).size
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
return (bytearray(data), real_size, pos_format)
class ImageProvider:
def __init__(self, app) -> None:
self.app = app
self.cache = Path(self.app.appdirs.user_cache_dir) / "thumbnails"
self.cache.mkdir(parents=True, exist_ok=True)
def get(self, image_id: str, requested_size: Size) -> ImageData:
if requested_size[0] < 1 or requested_size[1] < 1:
raise ValueError(f"width or height < 1: {requested_size!r}")
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
try:
thumb = Thumbnail(self, image_id, *requested_size)
except ValueError as err:
log.warning(err)
data, pos_format = TRANSPARENT_1X1_PNG
return (bytearray(data), (1, 1), pos_format)
return asyncio.run_coroutine_threadsafe(
Big performance refactoring & various improvements Instead of passing all sorts of events for the JS to handle and manually add to different data models, we now handle everything we can in Python. For any change, the python models send a sync event with their contents (no more than 4 times per second) to JS, and the QSyncable library's JsonListModel takes care of converting it to a QML ListModel and sending the appropriate signals. The SortFilterProxyModel library is not used anymore, the only case where we need to filter/sort something now is when the user interacts with the "Filter rooms" or "Filter members" fields. These cases are handled by a simple JS function. We now keep separated room and timeline models for different accounts, the previous approach of sharing all the data we could between accounts created a lot of complications (local echoes, decrypted messages replacing others, etc). The users's own account profile changes are now hidden in the timeline. On startup, if all events for a room were only own profile changes, more events will be loaded. Any kind of image format supported by Qt is now handled by the pyotherside image provider, instead of just PNG/JPG. SVGs which previously caused errors are supported as well. The typing members bar paddings/margins are fixed. The behavior of the avatar/"upload a profile picture" overlay is fixed. Config files read from disk are now cached (TODO: make them reloadable again). Pylint is not used anymore because of all its annoying false warnings and lack of understanding for dataclasses, it is replaced by flake8 with a custom config and various plugins. Debug mode is now considered on if the program was compiled with the right option, instead of taking an argument from CLI. When on, C++ will set a flag in the Window QML component. The loading screen is now unloaded after the UI is ready, where previously it just stayed in the background invisible and wasted CPU. The overall refactoring and improvements make us now able to handle rooms with thousand of members and no lazy-loading, where previously everything would freeze and simply scrolling up to load past events in any room would block the UI for a few seconds.
2019-08-11 22:01:22 +10:00
thumb.get_data(), self.app.loop,
).result()