[package]
edition = "2024"
rust-version = "1.88"
name = "ruwex"
version = "0.1.0"
authors = ["Magnus Manske <magnusmanske@proton.me>"]
build = false
exclude = ["benchmarks/"]
autolib = false
autobins = false
autoexamples = false
autotests = false
autobenches = false
description = "Fast Rust rewrite of wikiextractor: extract and clean text from Wikimedia XML dumps"
homepage = "https://codeberg.org/magnusmanske/ruwex"
readme = "README.md"
keywords = [
"wikipedia",
"wikitext",
"dump",
"extraction",
"mediawiki",
]
categories = [
"text-processing",
"command-line-utilities",
"parser-implementations",
]
license = "AGPL-3.0-only"
repository = "https://codeberg.org/magnusmanske/ruwex"
[lib]
name = "ruwex"
path = "src/lib.rs"
[[bin]]
name = "cirrus-extract"
path = "src/bin/cirrus_extract.rs"
[[bin]]
name = "extractPage"
path = "src/bin/extract_page.rs"
[[bin]]
name = "wikiextractor"
path = "src/bin/wikiextractor.rs"
[[test]]
name = "e2e"
path = "tests/e2e.rs"
[[test]]
name = "golden"
path = "tests/golden.rs"
[dependencies.anyhow]
version = "1"
[dependencies.bzip2]
version = "0.6"
[dependencies.clap]
version = "4"
features = ["derive"]
[dependencies.crossbeam-channel]
version = "0.5"
[dependencies.env_logger]
version = "0.11"
[dependencies.flate2]
version = "1"
[dependencies.fst]
version = "0.4"
[dependencies.jiff]
version = "0.2"
[dependencies.log]
version = "0.4"
[dependencies.memmap2]
version = "0.9"
[dependencies.quick-xml]
version = "0.37"
[dependencies.regex]
version = "1"
[dependencies.serde_json]
version = "1"
[dependencies.thiserror]
version = "2"
[dev-dependencies.tempfile]
version = "3"
[profile.release]
lto = "thin"