- The ProgressWriter class now resets it's cursor position to the beginning of the current line so that any log inbetween increments overwrites the current progress bar.
- Bot generation will now display a progress bar.
This is a big one. 🔥
**Changes:**
- Targets next major release (v4.0.0)
- Switch runtimes from NodeJS to BunJS (bun.sh)
- Typescript library support moved to `ESNext`
- Typescript compile option `verbatimModuleSyntax` enabled
- Updated all interfaces to be imported explicitly as types
- Strict mode enabled
- Reduces the number of tsconfig files
- Pins all dep packages to specific patch versions
- Includes Bun lock file in repo (doesn't cause issues like the
package-lock did)
- Replaces Gulp with a new Typescript based build system
- Adds `core-js` as a workaround for Bun not playing nice with
`reflect-metadata`
- Removes `pkg` and `swc` (Yay Bun!)
- Updated package scripts and entry point system to be more intuitive
- Updated VSCode workspace configurations
- Updated `.gitignore` to align with updated project structure
- Updated Biome configuration to align with updated project structure
- `Program.ts` - Removes call to set encoding on the process
- `global.d.ts` - Added underscores to build globals to match other
global names
- `JsonUtil.ts` - Replaced old `fixJson` package with newer `jsonrepair`
package
- `HashUtil.ts` - Replaced old `buffer-crc32` package with built-in
`node:zlib` package
- `DatabaseImporter.ts` - Updates database validation object to be flat,
where the keys are the relative path to the file
- `BunTimer.ts` - Adds an easy to use timer class that's compatible with
nanoseconds.
**TODO:**
- Look into mod loading. I think we use a TS transpiler for mods and I
believe that can be removed now.
- Bun includes a number of APIs that can be used in place of Node's
packages (built-in or otherwise); HTTP server, WebSocket server, File
IO, Hashing, File Globing, Testing... Each of these should be utilized
where ever possible.
- Update in-repo documentation to reference BunJS instead of NodeJS.
Waffle wrote a PR (#986) that shows a progress bar when the database is
populating. I pulled it into a local branch and made some adjustments.
I've tested this on my end and it's working okay. In the compiled
project, and directly in terminals (that *I have installed*) it looks
great, but in VSCode's built in terminal you can see escape characters
at the beginning of the line:
I can't figure out why, or how to get rid of it, but it doesn't really
bother me much.
---------
Co-authored-by: waffle-lord <76401815+waffle-lord@users.noreply.github.com>
Adds a public `randNum` method to the `RandomUtil` class which can be seen as a "superseeded" version of the `randInt` method. It can handle generating random numbers, including both integers or floating-point numbers with support for providing a precision.
Adds a private `getNumberPrecision` method to the `RandomUtil` class which simply counts the number of fractional digits in a number.
Changes `randInt` to log a debug message saying that the ability to pass it a floating-point number is deprecated.
This started as a small update.
This pull request includes updates to the `RandomUtil` class to improve
the security and accuracy of random number generation by utilizing the
`node:crypto` module. Additionally, it enhances the documentation and
refactors several methods for clarity and performance.
Includes tests, both before and after the changes.
The only *possible* breaking change is that now the `getKey`,
`getKeyValue`, and `drawRandomFromDict` methods are typed with generics
instead of `any`. I tried to keep them as *generic* as possible, but we
should probably include this in a BE to ensure it doesn't break
anything.
No further changes are planned--Saved as a draft because it's late and I
still want to review.
---------
Co-authored-by: Chomp <27521899+chompDev@users.noreply.github.com>
We've created our own Git Large File Storage (LFS) server due to the
excessive cost of GitHub's bandwidth. Did you know all of their Ethernet
cables are actually gold coated diamonds?
This PR reverses the work done in #954 to compress large location JSON
files into a 7zip archive and handle the (de)compression of the archive.
Only JSON files within the `project/assets/database/` directory that are
larger than 5MB have been included in LFS. This translates to all of the
`looseLoot.json` files. The rest are small enough to be included in the
base repo.
A `.lfsconfig` file has been added to the root of the project to alert
git to the presence of the custom LFS server. This public server is
read-only. Write access is only available to developers within the
Single Player Tarkov GitHub organization.
<img
src="https://github.com/user-attachments/assets/7ddfec9b-5a9a-42e6-806d-fd419e4eaa4f"
width="250">
Due to LFS storage issues... This PR removes all current LFS files (the
location loot files) and replaces them with a single 7-zip archive. The
archive is stored in LFS, but has decreased in size by roughly 95%.
The location `.json` files are now git-ignored.
There are two new npm commands to aid in working with the archive:
- `npm run database:compress`
Compresses the JSON files into an archive *which can be committed into
the project*.
- `npm run database:decompress`
Decompresses the archive into the original JSON files located in the
working directory.
The gulp file that handles builds has been updated to ensure that the
archive files are *always* used when a build is processed, regardless of
if the JSON files are already present in the working directory.
In combination with my other PR handle all the serializers asynchronously too, this should almost result in a fully asynchronous sending of responses.
Reviewed-on: SPT/Server#428
Co-authored-by: Archangel <jesse@archangel.wtf>
Co-committed-by: Archangel <jesse@archangel.wtf>
When compiled natively to linux, logfiles are saved in main SPT folder instead of default location. It's caused by file paths being hardcoded as text.
This patch fixes logs location on linux build.
Reviewed-on: SPT/Server#390
Co-authored-by: witek <falls-toluene.0q@icloud.com>
Co-committed-by: witek <falls-toluene.0q@icloud.com>
(cherry picked from commit 3edf0f2ff9d6793d7441591f15dd309fef9871f5)
Boogidy, boogidy, boogidy. Let's go racing! 🏎️
Removes the over-complicated and super-slow setup we had with ESLint & Prettier in favour of Biome. The largest change with the formatting is moving from Allman braces to 1TBS braces. Other than that, it's *pretty much* the same. Ah, and that Biome runs formatting and linting on the entire project about x10 faster than the old system ran formatting on one file. Seriously, the guy who came up with that last solution should be fired. :runs:
I've kept all of the formatting and linting commands the same as before, with the main mamma-jamma being: `npm run format`, which applies formatting and linting changes to the entire project.
Formatting-on-save works (quickly!) by (1) ensuring that you're working within the VSC workspace (as you should be), and (2) have the recommended Biome VSC extension installed. The link to the Biome extension is in the README.
This limits our options on code formatting going forward; Biome, like prettier, is very opinionated with very few formatting options available. But I see this as a good thing. I'd rather spend my time arguing about which gun in Tarkov is the best, rather than coding brace styles...
...It's the TOZ, and it always will be. Don't DM me.
Co-authored-by: chomp <chomp@noreply.dev.sp-tarkov.com>
Reviewed-on: SPT/Server#383
Co-authored-by: Refringe <me@refringe.com>
Co-committed-by: Refringe <me@refringe.com>
Rebranded src code and scripts to SPT
Co-authored-by: clodan <clodan@clodan.com>
Reviewed-on: SPT-AKI/Server#345
Co-authored-by: Alex <clodan@noreply.dev.sp-tarkov.com>
Co-committed-by: Alex <clodan@noreply.dev.sp-tarkov.com>
This PR fixes an issue in RecursiveCloner where null value properties of any type (i.e. arrays, nullable primitives) were being converted to empty objects.
An example of where this is a problem can occur when starting a raid in 3.9. The client would attempt to convert the empty object (created by the RecursiveCloner) to an Array and it would result in a JSON error.
Reviewed-on: SPT-AKI/Server#338
Co-authored-by: paulov <v2k4gameplay@outlook.com>
Co-committed-by: paulov <v2k4gameplay@outlook.com>
This is the first pass of ESLint on the codebase.
ESLint formatting is less strict when it comes to line-length and line-breaks then dprint/biome, so if you see formatting that you don't like... fix it! It shouldn't require a configuration change.
- This should merge clean into master (when the time comes).
- This will not merge clean into `3.9.0-DEV`, but the conflicts aren't that bad.
Fixed ragfair memory leak where NPC player ids would continuously get added into the cache map, when those offers expire the entry on the dictionary would be left empty but the string allocation on the key would remain in memory, overtime it would stack up to massive amounts of memory usage
Co-authored-by: clodan <clodan@clodan.com>
Reviewed-on: SPT-AKI/Server#319
Co-authored-by: Alex <clodan@noreply.dev.sp-tarkov.com>
Co-committed-by: Alex <clodan@noreply.dev.sp-tarkov.com>
Fixes a number of Biome linting issues:
- Variable implicitly has the any type.
- Use a function type instead of a call signature.
- Reassigning a function parameter is confusing.
- Suppression comment is not being used
This is primarily to stop confusion when a user overwrites their `aki_data` folder with an old version, the data shown in logs/server console is now based on compile time data instead of runtime data.
- New build.json file added to the `obj/ide/` folder that gets populated with the build data on build
- Moved asset copying prior to packaging, so that `obj/ide/build.json` is available at package time
- Updated all references of core.commit, and core.buildTime to use globalThis
- Updated all references of core.akiVersion to use globalThis with a fallback if not found (When running in VSCode for example)
Co-authored-by: DrakiaXYZ <565558+TheDgtl@users.noreply.github.com>
Reviewed-on: SPT-AKI/Server#297
Co-authored-by: DrakiaXYZ <drakiaxyz@noreply.dev.sp-tarkov.com>
Co-committed-by: DrakiaXYZ <drakiaxyz@noreply.dev.sp-tarkov.com>
