Uyanide/WallReel
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases 7 Wiki Activity

refactor: improve image loading logic in image_item for better memory performance

Browse Source
This commit is contained in:
Uyanide
2026-02-07 07:16:02 +01:00
parent cf81e01d85
commit f114999d6d
5 changed files with 49 additions and 31 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+29 -1
View File
@@ -35,7 +35,7 @@ It might not be that worthy to write a QtWidget application for such a small fea
> [!Warning]
>
> This script will ask for `sudo` permission if the prefix is set to a system directory like `/usr/local`. Please make sure you have read and trust the script before proceeding.
> This script will ask for root permissions if the prefix is set to a system directory like `/usr/local`. Please make sure you have read and trust the script before proceeding.
## How to use
@@ -65,3 +65,31 @@ By default, the path of the selected wallpaper will be output to stdout. If you
```
`action.confirm` should be a executable followed by a couple of arguments, where `%1` will be replaced by the path of the selected wallpaper.
## CLI
```
Usage: wallpaper-carousel [options]
Options:
-h, --help Displays help on commandline options.
-v, --version Displays version information.
-V, --verbose Set log level to DEBUG (default is INFO)
-q, --quiet Suppress all log output
-d, --append-dir <dir> Append an additional wallpaper search directory
-c, --config-file <file> Specify a custom configuration file
```
A few things to notice:
- It's generally not necessary to provide any CLI arguments, I would recommend using the config file to customize the behavior instead. However, it is still possible to control some essential options via CLI.
- All logs are directed to stderr by default. Only the full path of the selected wallpaper (if any) will be sent to stdout. This allows easy piping of the output to other programs.
- The `--append-dir` option can be used multiple times to add multiple directories.
- It is quite obvious that some options are conflicting with each other (e.g. `--verbose` and `--quiet`). If mutually exclusive options are provided together, the behavior is undefined and can be changed without notice in future versions.
- Paths passed via CLI options are tested before any further operation is performed. That is to say, if an invalid path is provided, the program will exit with an error before any further action, and you won't even have a chance to see a window.
On the contrary, paths provided in the config file are only tested when they are actually used (e.g. when searching for wallpapers). And most errors will be ignored silently (with a warning log).
src/image_item.cpp
+11 -8
View File
@@ -1,7 +1,7 @@
/*
* @Author: Uyanide pywang0608@foxmail.com
* @Date: 2025-11-30 20:32:27
* @LastEditTime: 2026-01-15 07:09:29
* @LastEditTime: 2026-02-07 07:13:49
* @Description: Image item widget for displaying an image.
*/
#include "image_item.h"
@@ -14,10 +14,10 @@ using namespace GeneralLogger;
ImageData* ImageData::create(const QString& p, const int initWidth, const int initHeight) {
ImageData* data = new ImageData(p);
data->image = new QImage();
// Use QImageReader for better performance
QImageReader reader(p);
reader.setAutoTransform(true);
if (!reader.canRead()) {
warn(QString("Failed to load image from path: %1").arg(p));
delete data;
@@ -28,7 +28,7 @@ ImageData* ImageData::create(const QString& p, const int initWidth, const int in
const QSize originalSize = reader.size();
// Scale the image to fit the target size while maintaining aspect ratio
if (originalSize.isValid()) {
if (originalSize.isValid() && reader.supportsOption(QImageIOHandler::ScaledSize)) {
double widthRatio = (double)targetSize.width() / originalSize.width();
double heightRatio = (double)targetSize.height() / originalSize.height();
double scaleFactor = std::max(widthRatio, heightRatio);
@@ -37,19 +37,22 @@ ImageData* ImageData::create(const QString& p, const int initWidth, const int in
reader.setScaledSize(scaledSize);
}
if (!reader.read(data->image)) {
QImage tempImage;
if (!reader.read(&tempImage)) {
warn(QString("Failed to load image from path: %1").arg(p));
delete data;
return nullptr;
}
// Crop to target size if necessary
if (data->image->size() != targetSize) {
int x = (data->image->width() - targetSize.width()) / 2;
int y = (data->image->height() - targetSize.height()) / 2;
*data->image = data->image->copy(x, y, targetSize.width(), targetSize.height());
if (tempImage.size() != targetSize) {
int x = (tempImage.width() - targetSize.width()) / 2;
int y = (tempImage.height() - targetSize.height()) / 2;
tempImage = tempImage.copy(x, y, targetSize.width(), targetSize.height());
}
data->image = new QImage(std::move(tempImage));
return data;
}
src/images_carousel.h
+6 -6
View File
@@ -1,7 +1,7 @@
/*
* @Author: Uyanide pywang0608@foxmail.com
* @Date: 2025-08-05 01:22:53
* @LastEditTime: 2026-01-15 07:43:44
* @LastEditTime: 2026-01-18 06:39:44
* @Description: Animated carousel widget for displaying and selecting images.
*/
#ifndef IMAGES_CAROUSEL_H
@@ -24,13 +24,13 @@
Two different image loading strategies:
- With loading screen: load all images directly
1. appendImages called -> increace m_addedImagesCount & spawn all ImageLoader
threads
1. appendImages called -> increace m_addedImagesCount & disable UI updates &
spawn and start all ImageLoader threads
2. Each ImageLoader calls _insertImageQueue with queued connection
3. _insertImageQueue calls _insertImage directly
- Without loading screen: queue loaded images and insert them in batches
1. appendImages called -> increace m_addedImagesCount & spawn all ImageLoader
threads and a timer m_imageInsertQueueTimer, disable UI updates
1. appendImages called -> increace m_addedImagesCount & spawn and start all
ImageLoader threads and a timer m_imageInsertQueueTimer & disable animations
2. Each ImageLoader calls _insertImageQueue with queued connection
3. _insertImageQueue enqueues the ImageData
4. m_imageInsertQueueTimer calls _processImageInsertQueue every
@@ -38,7 +38,7 @@ Two different image loading strategies:
5. _processImageInsertQueue processes up to s_processBatchSize items from the
queue and calls _insertImage for each
The stop logic is identical:
The stop logic is identical regardless of whether loading screen is used or not:
- Force stop
1. Set m_stopSign to true
2. ImageLoader::run checks m_stopSign and returns early if true
src/main.cpp
+2 -2
View File
@@ -1,7 +1,7 @@
/*
* @Author: Uyanide pywang0608@foxmail.com
* @Date: 2025-08-05 00:37:58
* @LastEditTime: 2026-01-15 07:51:54
* @LastEditTime: 2026-01-15 14:11:06
* @Description: Argument parser and entry point.
*/
#include <QApplication>
@@ -71,7 +71,7 @@ static class AppOptions {
const QCommandLineOption helpOption = parser.addHelpOption();
const QCommandLineOption versionOption = parser.addVersionOption();
QCommandLineOption verboseOption(QStringList() << "V" << "verbose", "Set log level to DEBUG");
QCommandLineOption verboseOption(QStringList() << "V" << "verbose", "Set log level to DEBUG (default is INFO)");
parser.addOption(verboseOption);
QCommandLineOption quietOption(QStringList() << "q" << "quiet", "Suppress all log output");
src/utils.h
+1 -14
View File
@@ -1,7 +1,7 @@
/*
* @Author: Uyanide pywang0608@foxmail.com
* @Date: 2025-11-30 20:59:57
* @LastEditTime: 2026-01-15 09:55:26
* @LastEditTime: 2026-01-18 06:36:13
* @Description: THE utils header that every project needs :)
*/
@@ -110,19 +110,6 @@ static QString splitNameFromPath(const QString& path) {
* @return false
*/
inline bool checkImageFile(const QString& filePath) {
static const QStringList validExtensions = {
".jpg",
".jpeg",
".jfif",
".png",
".bmp",
".gif",
".webp",
".tiff",
".avif",
".heic",
".heif"};
// check if exist
if (!QFile::exists(filePath)) {
return false;