Olib: the repository ships the shell and withholds the data source
O-lib is a free and open-source software application for PC
At a glance
- What is it?
- Olib is a cross-platform ebook desktop client built on Tauri, with a React frontend, a Rust backend and a bundled downloader. The repository deliberately contains no data-source implementation and states that cloning it will not build or run, and the three features that concern fetching books are described without any statement about the licensing of the books themselves.
- Who is it for?
- Olib fits someone who wants an offline reading shell with a local library, LAN sharing and an open catalogue feed, and it fits nobody who expects a clone to produce a working application. Verify four things first.
- Can I use it commercially?
- Yes, with conditions. GPL-3.0 is a copyleft licence: if you distribute software that includes it, you must release that software's source code under the same licence. Running it internally without distributing it does not trigger that obligation.
- Is it still maintained?
- Yes. The repository last received commits 3 days ago.
- What is it written in?
- Mainly TypeScript, according to GitHub's language statistics.
Answers come from the project's GitHub data, last synced on November 15, 2026, and from our analysis. They are not legal advice.
Star history
- shiyi-0x7f/o-lib3,713 ★ · +371
| Repository | 2026-09-24 | 2026-11-15 |
|---|---|---|
| shiyi-0x7f/o-lib | 3.3k | 3.7k |
371 Stars in the last 52 days, measured from the daily GitHub snapshot. Compare with other projects
Editorial analysis
The interface notice says a clone will not run
The most important paragraph in the file is a notice about interfaces. It says that to prevent the project from being abused, this repository hides the implementation of certain data interfaces, that the public repository contains only the application framework and the front-end interface, and that the part connecting to external data sources is a separate private module. It then says directly that cloning the repository cannot build and run out of the box, and that if you want to build it yourself you should find the relevant interfaces and implement the corresponding modules. The stated reason is so the project can exist healthily and for a long time. A second note repeats this after the development commands and says the module must be implemented before building and wired in as a local dependency, with references in a directory inside the repository. So the contract is a shell plus a written-down integration point.
Three of eight features are about obtaining books rather than reading them
The feature table has eight rows and three of them concern getting content in rather than reading it. One is download management with a built-in multi-threaded downloader supporting resume. One is multiple server routes, selectable, with subscription or manual import and a one-click speed test. One is multiple account switching. The remaining five cover search, a local shelf with favourites and download history, an interface style, sharing the local library over the local network with catalogue support other readers can subscribe to directly, and the absence of advertising or fees. The route row is the one that carries the weight: a subscribed, speed-tested list of alternate servers is the shape of a download manager, and the catalogue row names the open standard through which other reader applications can subscribe to a feed. The disclaimer beside all of it covers data provenance and local law and says nothing about the rights in the books.
A downloader binary ships inside the package, and building needs two toolchains
The technology list is short and specific. The application framework is a desktop webview wrapper, with a Rust backend and the system's own web view rather than a bundled browser. The Rust side is listed as four crates by name: an HTTP client, a SQLite binding, an async runtime and a web server, the last presumably serving the local-network sharing feature. The frontend is React with TypeScript and a bundler, four UI libraries by name, and a downloader described as bundled with the package. Requirements are Node eighteen or newer with a later version recommended, a stable Rust toolchain, and the per-platform prerequisites the framework documents. So this is not a JavaScript-only project: getting a development build going needs both toolchains, and the published artefact carries a third-party downloader binary with it. The full check sequence the file gives is worth reading as a whole, because it names two working directories and four commands:
# frontend type check
npm run build # tsc + vite build
# Rust compile and lint
cd src-tauri
cargo build
cargo clippy
# Rust integration tests (LAN/OPDS etc.)
cargo testThe manifest version is one patch behind the only tag
There is one release on record, tagged at the same moment as the last commit. Its title names the product and the version. The manifest in the repository declares the version one patch lower than that tag. So the tree you clone is not the tree the release was cut from, or the version field was missed, and the repository ships a script whose entire job is bumping the version, which means the mechanism exists and was not run. Two further naming details compound it. The manifest's package name is the product name with the framework name appended to it, which is harmless only because the package is marked private. And the front page says multi-platform installers are built automatically and published to the releases page by continuous integration, against a releases page carrying exactly one entry.
No test script, no test dependency, and the tests cover the part with no data source
The manifest defines five scripts: a development server, a build, a preview, a wrapper for the desktop tool and the version bump. There is no test script, and no test dependency of any kind in either the runtime or the development list. What testing the file does describe is Rust-side: compile the backend, run the static checker, and run integration tests which a comment labels as covering the local-network and catalogue features. That is the coherent part of the picture. The features covered by the only tests are exactly the two that need no external data source, which is what you would expect from a repository where that layer is absent. One label is also loose: the frontend type check is a script that compiles the type check and then produces a production bundle, so a step described as a check does real work on every run.
Two lockfiles, one documented package manager
The repository root carries both an npm lockfile and a pnpm lockfile. The documented path uses npm: the install command in the quick start is the npm one, the development command is npm, and the type check is invoked through npm. Nothing in the file mentions the second package manager, and nothing at the root selects one. A contributor who reaches for the tool the second lockfile implies will produce a different install, and the mismatch is invisible until two people report different behaviour from the same commit. The rest of the root is coherent for a shell of this kind: an HTML entry point, a configuration file for the bundler, two TypeScript configurations split the conventional way, and four directories for the desktop side, the front end, static assets and helper scripts.
Three links climb out of the repository, and none of them should
Three links in the front page are written with a parent path that leaves the repository. The report-a-problem link and the feature-suggestion link are the same path with different labels, so both break identically. The third one, in the build section, points at the releases page and also breaks. So on the rendered page none of the three report, suggest or download links resolves, and the ones that do work are the two external ones, a video channel profile and a social profile. That is worth knowing before you conclude a project has no issue tracker or no downloads, because from the repository alone it looks like it does. Two smaller things from the same file: the contribution steps use the older branch-creation syntax, and the file ends by asking for a star rather than for anything else.
No commercial cooperation, four names, and a hotkey nobody mentions
The project commits publicly to accepting no commercial cooperation of any kind and describes itself as non-profit, which is a stronger constraint than the licence alone and rules out the sponsorship and donation routes most permissively licensed projects rely on. That commitment sits next to four different names for one thing: the repository, the product as the file titles it, the package name in the manifest, and the release tag's title. And the manifest declares a global keyboard shortcut plugin that the feature table never mentions, in an application whose documented capabilities are reading, searching, downloading and sharing. A dependency that installs a system-wide hotkey is the kind of thing that belongs in a feature list, whatever the reason for it.
Editorial conclusion
Olib fits someone who wants an offline reading shell with a local library, LAN sharing and an open catalogue feed, and it fits nobody who expects a clone to produce a working application. Verify four things first. Read the interface notice before anything else, because the project states outright that the fetching layer is absent and that you are expected to find and implement it yourself, which makes this a front end plus a downloader rather than a usable client. Then read the feature table with that in mind, since three of its eight rows are about obtaining books rather than reading them. Decide how you will source content within your own jurisdiction, because the disclaimer covers data provenance and local law and says nothing about the rights in the books. And expect no automated frontend checks, since the manifest has no test script and no test dependency of any kind.
Frequently asked questions
What is the Olib ebook desktop client?
A cross-platform desktop reader built on a desktop webview wrapper, with a React and TypeScript front end, a Rust backend and a bundled downloader. It offers search, a local shelf with favourites and download history, resumable multi-threaded downloads, selectable server routes with subscription and speed testing, multiple accounts, and sharing the local library over the local network with support for an open ebook catalogue standard other readers can subscribe to.
Can I build Olib from the repository?
Not as it stands. The file states that this repository hides the implementation of its data interfaces, that only the application framework and the front end are included, and that cloning it cannot build and run out of the box. To build it you are told to find the relevant interfaces and implement the modules yourself, wiring the result in as a local dependency, with references inside the repository.
Is the Olib repository version the same as its release?
No. The manifest declares a version one patch below the single release tag on record, and that tag carries the same timestamp as the last commit. The repository also ships a script whose only job is bumping the version, so the mechanism for keeping them aligned exists in the tree.
Does Olib have automated tests?
Only on the Rust side. The manifest defines no test script and declares no test dependency, and the documentation's testing section covers compiling the backend, running the static checker and running integration tests that a comment labels as covering the local-network sharing and catalogue features. No front-end test runner appears anywhere in the manifest or at the repository root.
What package manager does Olib use?
The documentation uses npm throughout, including the install command and the development and type-check commands. The repository nevertheless carries both an npm lockfile and a pnpm lockfile at its root, and nothing in the file mentions the second one or selects between them.
Official sources
Add this badge to your README
If you maintain this project, the badge below links readers to this analysis and shows its maintenance status from the daily GitHub snapshot. Paste the markdown into your README; add ?metric=license or ?metric=stars to the image URL for a different field.
[](https://hysenlabs.com/projects/shiyi-0x7f-o-lib)Star history chart for your README
[](https://hysenlabs.com/projects/shiyi-0x7f-o-lib)Was this useful?