[package]
edition = "2021"
name = "bgustcrawler"
version = "0.2.0"
authors = ["B-GUST <augustbenitogroup@gmail.com>"]
build = false
autolib = false
autobins = false
autoexamples = false
autotests = false
autobenches = false
description = "Distributed web crawler and scheduler integrated with Kafka, SurrealDB, and S3 storage"
homepage = "https://bgustops.com"
documentation = "https://docs.bgustops.com"
readme = "README.md"
keywords = [
"crawler",
"spider",
"distributed-ingest",
"kafka",
"rust",
]
categories = [
"web-programming",
"database",
]
license-file = "LICENSE"
repository = "https://github.com/B-GUST/bgustecosystem"
[lib]
name = "bgustcrawler"
path = "src/lib.rs"
[[bin]]
name = "bgustcrawler-cli"
path = "src/main.rs"
[dependencies.anyhow]
version = "1.0"
[dependencies.aws-config]
version = "1.8.17"
[dependencies.aws-sdk-s3]
version = "1.133.0"
[dependencies.bgustdown]
version = "0.2.0"
[dependencies.bgustscraper]
version = "0.2.0"
[dependencies.burn]
version = "0.14"
features = [
"train",
"wgpu",
"vision",
]
[dependencies.futures-util]
version = "0.3"
[dependencies.hf-hub]
version = "0.3"
[dependencies.rdkafka]
version = "0.39.0"
features = ["cmake-build"]
[dependencies.reqwest]
version = "0.11"
features = [
"json",
"stream",
]
[dependencies.scraper]
version = "0.19"
[dependencies.serde]
version = "1.0"
features = ["derive"]
[dependencies.serde_json]
version = "1.0"
[dependencies.sha2]
version = "0.11.0"
[dependencies.surrealdb]
version = "3.1.0"
features = ["protocol-http"]
[dependencies.tokenizers]
version = "0.19"
[dependencies.tokio]
version = "1.36"
features = ["full"]
[dependencies.tokio-cron-scheduler]
version = "0.15.1"
[dependencies.tokio-util]
version = "0.7.18"
[dependencies.url]
version = "2.5"
[dependencies.urlencoding]
version = "2.1"
[dependencies.uuid]
version = "1.23.1"
features = ["v4"]