[package]
edition = "2024"
name = "content-extractor-rl-cli"
version = "0.1.3"
authors = ["Sandeep Singh Sandhu <sandeep.sandhu@gmx.com>"]
build = false
publish = true
autolib = false
autobins = false
autoexamples = false
autotests = false
autobenches = false
description = "RL-based article extraction from HTML using Deep Q-Networks and heuristic fallback"
homepage = "https://github.com/sandeepsandhu/content-extractor-rl"
documentation = "https://docs.rs/content-extractor-rl"
readme = "README.md"
keywords = [
"article",
"extraction",
"html",
"reinforcement",
"nlp",
]
categories = [
"text-processing",
"web-programming",
"science",
]
license = "MIT OR Apache-2.0"
repository = "https://github.com/sandeepsandhu/content-extractor-rl"
resolver = "2"
[features]
cuda = ["content-extractor-rl/cuda"]
default = []
mlflow = ["content-extractor-rl/mlflow-rs"]
[[bin]]
name = "content-extractor-rl"
path = "src/main.rs"
[dependencies.algorithm]
version = "0.1.18"
[dependencies.anyhow]
version = "1.0.102"
[dependencies.bzip2]
version = "0.6.1"
[dependencies.chrono]
version = "0.4"
features = ["serde"]
[dependencies.clap]
version = "4.6.1"
features = ["derive"]
[dependencies.content-extractor-rl]
version = "0.1.1"
features = ["mlflow-rs"]
[dependencies.indicatif]
version = "0.18.3"
[dependencies.rand]
version = "0.10"
[dependencies.rayon]
version = "1.11.0"
[dependencies.serde_json]
version = "1.0.150"
[dependencies.tokio]
version = "1.48"
features = ["full"]
[dependencies.tracing]
version = "0.1.44"
[dependencies.tracing-appender]
version = "0.2.4"
[dependencies.tracing-subscriber]
version = "0.3.23"
features = [
"env-filter",
"fmt",
"local-time",
]
[dependencies.walkdir]
version = "2.5"