Recursive, parallel crawl #11
6 changed files with 204 additions and 119 deletions
148
Cargo.lock
generated
148
Cargo.lock
generated
|
@ -205,6 +205,15 @@ dependencies = [
|
|||
"memchr",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "ansi_term"
|
||||
version = "0.12.1"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "d52a9bb7ec0cf484c551830a7ce27bd20d67eac647e1befb56b0be4ee39a55d2"
|
||||
dependencies = [
|
||||
"winapi",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "anyhow"
|
||||
version = "1.0.57"
|
||||
|
@ -444,26 +453,17 @@ dependencies = [
|
|||
|
||||
[[package]]
|
||||
name = "clap"
|
||||
version = "3.1.15"
|
||||
version = "2.34.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "85a35a599b11c089a7f49105658d089b8f2cf0882993c17daf6de15285c2c35d"
|
||||
checksum = "a0610544180c38b88101fecf2dd634b174a62eef6946f84dfc6a7127512b381c"
|
||||
dependencies = [
|
||||
"ansi_term",
|
||||
"atty",
|
||||
"bitflags",
|
||||
"clap_lex",
|
||||
"indexmap",
|
||||
"strsim",
|
||||
"termcolor",
|
||||
"strsim 0.8.0",
|
||||
"textwrap",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "clap_lex"
|
||||
version = "0.2.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "a37c35f1112dad5e6e0b1adaff798507497a18fceeb30cceb3bae7d1427b9213"
|
||||
dependencies = [
|
||||
"os_str_bytes",
|
||||
"unicode-width",
|
||||
"vec_map",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
|
@ -556,7 +556,7 @@ dependencies = [
|
|||
"ident_case",
|
||||
"proc-macro2 1.0.37",
|
||||
"quote 1.0.18",
|
||||
"strsim",
|
||||
"strsim 0.10.0",
|
||||
"syn 1.0.92",
|
||||
]
|
||||
|
||||
|
@ -1030,6 +1030,15 @@ version = "0.11.2"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "ab5ef0d4909ef3724cc8cce6ccc8572c5c817592e9285f5464f8e86f8bd3726e"
|
||||
|
||||
[[package]]
|
||||
name = "heck"
|
||||
version = "0.3.3"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "6d621efb26863f0e9924c6ac577e8275e5e6b77455db64ffa6c65c904e9e132c"
|
||||
dependencies = [
|
||||
"unicode-segmentation",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "heck"
|
||||
version = "0.4.0"
|
||||
|
@ -1318,7 +1327,6 @@ version = "0.1.0"
|
|||
dependencies = [
|
||||
"anyhow",
|
||||
"async-recursion",
|
||||
"clap",
|
||||
"derive-new",
|
||||
"futures",
|
||||
"lemmy_api_common",
|
||||
|
@ -1328,6 +1336,8 @@ dependencies = [
|
|||
"semver",
|
||||
"serde",
|
||||
"serde_json",
|
||||
"stderrlog",
|
||||
"structopt",
|
||||
"tokio",
|
||||
]
|
||||
|
||||
|
@ -1831,12 +1841,6 @@ dependencies = [
|
|||
"vcpkg",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "os_str_bytes"
|
||||
version = "6.0.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "8e22443d1643a904602595ba1cd8f7d896afe56d26712531c5ff73a15b2fbf64"
|
||||
|
||||
[[package]]
|
||||
name = "parking_lot"
|
||||
version = "0.11.2"
|
||||
|
@ -2065,6 +2069,30 @@ version = "0.1.1"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "925383efa346730478fb4838dbe9137d2a47675ad789c546d150a6e1dd4ab31c"
|
||||
|
||||
[[package]]
|
||||
name = "proc-macro-error"
|
||||
version = "1.0.4"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "da25490ff9892aab3fcf7c36f08cfb902dd3e71ca0f9f9517bea02a73a5ce38c"
|
||||
dependencies = [
|
||||
"proc-macro-error-attr",
|
||||
"proc-macro2 1.0.37",
|
||||
"quote 1.0.18",
|
||||
"syn 1.0.92",
|
||||
"version_check",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "proc-macro-error-attr"
|
||||
version = "1.0.4"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "a1be40180e52ecc98ad80b184934baf3d0d29f979574e439af5a55274b35f869"
|
||||
dependencies = [
|
||||
"proc-macro2 1.0.37",
|
||||
"quote 1.0.18",
|
||||
"version_check",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "proc-macro2"
|
||||
version = "0.4.30"
|
||||
|
@ -2612,6 +2640,19 @@ version = "0.5.2"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "6e63cff320ae2c57904679ba7cb63280a3dc4613885beafb148ee7bf9aa9042d"
|
||||
|
||||
[[package]]
|
||||
name = "stderrlog"
|
||||
version = "0.5.1"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "45a53e2eff3e94a019afa6265e8ee04cb05b9d33fe9f5078b14e4e391d155a38"
|
||||
dependencies = [
|
||||
"atty",
|
||||
"chrono",
|
||||
"log",
|
||||
"termcolor",
|
||||
"thread_local",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "string_cache"
|
||||
version = "0.8.4"
|
||||
|
@ -2638,12 +2679,42 @@ dependencies = [
|
|||
"quote 1.0.18",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "strsim"
|
||||
version = "0.8.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "8ea5119cdb4c55b55d432abb513a0429384878c15dde60cc77b1c99de1a95a6a"
|
||||
|
||||
[[package]]
|
||||
name = "strsim"
|
||||
version = "0.10.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "73473c0e59e6d5812c5dfe2a064a6444949f089e20eec9a2e5506596494e4623"
|
||||
|
||||
[[package]]
|
||||
name = "structopt"
|
||||
version = "0.3.26"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "0c6b5c64445ba8094a6ab0c3cd2ad323e07171012d9c98b0b15651daf1787a10"
|
||||
dependencies = [
|
||||
"clap",
|
||||
"lazy_static",
|
||||
"structopt-derive",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "structopt-derive"
|
||||
version = "0.4.18"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "dcb5ae327f9cc13b68763b5749770cb9e048a99bd9dfdfa58d0cf05d5f64afe0"
|
||||
dependencies = [
|
||||
"heck 0.3.3",
|
||||
"proc-macro-error",
|
||||
"proc-macro2 1.0.37",
|
||||
"quote 1.0.18",
|
||||
"syn 1.0.92",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "strum"
|
||||
version = "0.24.0"
|
||||
|
@ -2656,7 +2727,7 @@ version = "0.24.0"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "6878079b17446e4d3eba6192bb0a2950d5b14f0ed8424b852310e5a94345d0ef"
|
||||
dependencies = [
|
||||
"heck",
|
||||
"heck 0.4.0",
|
||||
"proc-macro2 1.0.37",
|
||||
"quote 1.0.18",
|
||||
"rustversion",
|
||||
|
@ -2730,9 +2801,12 @@ dependencies = [
|
|||
|
||||
[[package]]
|
||||
name = "textwrap"
|
||||
version = "0.15.0"
|
||||
version = "0.11.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "b1141d4d61095b28419e22cb0bbf02755f5e54e0526f97f1e3d1d160e60885fb"
|
||||
checksum = "d326610f408c7a4eb6f51c37c330e496b08506c9457c9d34287ecc38809fb060"
|
||||
dependencies = [
|
||||
"unicode-width",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "thiserror"
|
||||
|
@ -2756,11 +2830,11 @@ dependencies = [
|
|||
|
||||
[[package]]
|
||||
name = "thread_local"
|
||||
version = "1.1.4"
|
||||
version = "1.0.1"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "5516c27b78311c50bf42c071425c560ac799b11c30b31f87e3081965fe5e0180"
|
||||
checksum = "d40c6d1b69745a6ec6fb1ca717914848da4b44ae29d9b3080cbee91d72a69b14"
|
||||
dependencies = [
|
||||
"once_cell",
|
||||
"lazy_static",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
|
@ -2943,9 +3017,9 @@ dependencies = [
|
|||
|
||||
[[package]]
|
||||
name = "tracing-subscriber"
|
||||
version = "0.3.11"
|
||||
version = "0.3.6"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "4bc28f93baff38037f64e6f43d34cfa1605f27a49c34e8a04c5e78b0babf2596"
|
||||
checksum = "77be66445c4eeebb934a7340f227bfe7b338173d3f8c00a60a5a58005c9faecf"
|
||||
dependencies = [
|
||||
"sharded-slab",
|
||||
"thread_local",
|
||||
|
@ -3016,6 +3090,12 @@ dependencies = [
|
|||
"tinyvec",
|
||||
]
|
||||
|
||||
[[package]]
|
||||
name = "unicode-segmentation"
|
||||
version = "1.9.0"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "7e8820f5d777f6224dc4be3632222971ac30164d4a258d595640799554ebfd99"
|
||||
|
||||
[[package]]
|
||||
name = "unicode-width"
|
||||
version = "0.1.9"
|
||||
|
@ -3087,6 +3167,12 @@ version = "0.2.15"
|
|||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "accd4ea62f7bb7a82fe23066fb0957d48ef677f6eeb8215f372f52e48bb32426"
|
||||
|
||||
[[package]]
|
||||
name = "vec_map"
|
||||
version = "0.8.2"
|
||||
source = "registry+https://github.com/rust-lang/crates.io-index"
|
||||
checksum = "f1bddf1187be692e79c5ffeab891132dfb0f236ed36a43c7ed39f1165ee20191"
|
||||
|
||||
[[package]]
|
||||
name = "version_check"
|
||||
version = "0.9.4"
|
||||
|
|
|
@ -11,10 +11,11 @@ anyhow = "1.0.57"
|
|||
tokio = { version = "1.18.1", features = ["macros", "rt-multi-thread"] }
|
||||
futures = "0.3.21"
|
||||
serde_json = "1.0.81"
|
||||
clap = "3.1.15"
|
||||
semver = "1.0.9"
|
||||
once_cell = "1.10.0"
|
||||
lemmy_api_common = "0.16.0"
|
||||
async-recursion = "1.0.0"
|
||||
log = "0.4.17"
|
||||
derive-new = "0.5.9"
|
||||
stderrlog = "0.5.1"
|
||||
structopt = "0.3.26"
|
||||
|
|
90
src/crawl.rs
90
src/crawl.rs
|
@ -4,7 +4,7 @@ use anyhow::Error;
|
|||
use async_recursion::async_recursion;
|
||||
use futures::future::join_all;
|
||||
use lemmy_api_common::site::GetSiteResponse;
|
||||
use log::info;
|
||||
use log::debug;
|
||||
use semver::Version;
|
||||
use serde::Serialize;
|
||||
use std::collections::HashSet;
|
||||
|
@ -12,10 +12,11 @@ use std::ops::Deref;
|
|||
use std::sync::Arc;
|
||||
use tokio::sync::Mutex;
|
||||
|
||||
#[derive(Serialize, Debug)]
|
||||
pub struct InstanceDetails {
|
||||
pub domain: String,
|
||||
pub site_info: GetSiteResponse,
|
||||
#[derive(new)]
|
||||
pub struct CrawlJob {
|
||||
domain: String,
|
||||
current_distance: i32,
|
||||
params: Arc<CrawlParams>,
|
||||
}
|
||||
|
||||
#[derive(new)]
|
||||
|
@ -26,33 +27,62 @@ pub struct CrawlParams {
|
|||
crawled_instances: Arc<Mutex<HashSet<String>>>,
|
||||
}
|
||||
|
||||
#[derive(new)]
|
||||
pub struct CrawlJob {
|
||||
domain: String,
|
||||
current_depth: i32,
|
||||
params: Arc<CrawlParams>,
|
||||
#[derive(Serialize, Debug)]
|
||||
pub struct InstanceDetails {
|
||||
pub domain: String,
|
||||
pub site_info: GetSiteResponse,
|
||||
}
|
||||
|
||||
impl CrawlJob {
|
||||
#[async_recursion]
|
||||
pub async fn crawl(self) -> Result<Vec<Result<InstanceDetails, Error>>, Error> {
|
||||
// need to acquire and release mutix before recursing, otherwise it will deadlock
|
||||
pub async fn crawl(self) -> Vec<Result<InstanceDetails, Error>> {
|
||||
// need to acquire and release mutex before recursing, otherwise it will deadlock
|
||||
{
|
||||
let mut crawled_instances = self.params.crawled_instances.deref().lock().await;
|
||||
if crawled_instances.contains(&self.domain) {
|
||||
return Ok(vec![]);
|
||||
return vec![];
|
||||
} else {
|
||||
crawled_instances.insert(self.domain.clone());
|
||||
}
|
||||
}
|
||||
|
||||
if self.current_depth > self.params.max_depth
|
||||
if self.current_distance > self.params.max_depth
|
||||
|| self.params.exclude_domains.contains(&self.domain)
|
||||
{
|
||||
return Ok(vec![]);
|
||||
return vec![];
|
||||
}
|
||||
info!("Starting crawl for {}", &self.domain);
|
||||
|
||||
debug!("Starting crawl for {}, distance {}", &self.domain, &self.current_distance);
|
||||
let site_info = match self.fetch_instance_details().await {
|
||||
Ok(o) => o,
|
||||
Err(e) => return vec![Err(e)],
|
||||
};
|
||||
|
||||
if site_info.1 < self.params.min_lemmy_version {
|
||||
return vec![];
|
||||
}
|
||||
|
||||
let mut result = vec![];
|
||||
if let Some(federated) = &site_info.0.federated_instances {
|
||||
for domain in federated.linked.iter() {
|
||||
let crawl_job =
|
||||
CrawlJob::new(domain.clone(), self.current_distance + 1, self.params.clone());
|
||||
result.push(crawl_job.crawl());
|
||||
}
|
||||
}
|
||||
|
||||
let mut result2: Vec<Result<InstanceDetails, Error>> =
|
||||
join_all(result).await.into_iter().flatten().collect();
|
||||
debug!("Successfully finished crawl for {}", &self.domain);
|
||||
result2.push(Ok(InstanceDetails {
|
||||
domain: self.domain,
|
||||
site_info: site_info.0,
|
||||
}));
|
||||
|
||||
result2
|
||||
}
|
||||
|
||||
async fn fetch_instance_details(&self) -> Result<(GetSiteResponse, Version), Error> {
|
||||
let site_info_url = format!("https://{}/api/v3/site", &self.domain);
|
||||
let site_info = CLIENT
|
||||
.get(&site_info_url)
|
||||
|
@ -61,33 +91,7 @@ impl CrawlJob {
|
|||
.await?
|
||||
.json::<GetSiteResponse>()
|
||||
.await?;
|
||||
|
||||
let version = Version::parse(&site_info.version)?;
|
||||
if version < self.params.min_lemmy_version {
|
||||
return Ok(vec![]);
|
||||
}
|
||||
|
||||
let mut result = vec![];
|
||||
if let Some(federated) = &site_info.federated_instances {
|
||||
for domain in federated.linked.iter() {
|
||||
let crawl_job =
|
||||
CrawlJob::new(domain.clone(), self.current_depth + 1, self.params.clone());
|
||||
result.push(crawl_job.crawl());
|
||||
}
|
||||
}
|
||||
|
||||
let mut result2: Vec<Result<InstanceDetails, Error>> = join_all(result)
|
||||
.await
|
||||
.into_iter()
|
||||
.filter_map(|r| r.ok())
|
||||
.flat_map(|r| r.into_iter())
|
||||
.collect();
|
||||
info!("Successfully finished crawl for {}", &self.domain);
|
||||
result2.push(Ok(InstanceDetails {
|
||||
domain: self.domain,
|
||||
site_info,
|
||||
}));
|
||||
|
||||
Ok(result2)
|
||||
Ok((site_info, version))
|
||||
}
|
||||
}
|
||||
|
|
|
@ -1,4 +0,0 @@
|
|||
pub const DEFAULT_START_INSTANCES: &str = "lemmy.ml";
|
||||
pub const DEFAULT_MAX_CRAWL_DEPTH: &str = "20";
|
||||
pub const EXCLUDE_INSTANCES: &str =
|
||||
"ds9.lemmy.ml, enterprise.lemmy.ml, voyager.lemmy.ml, test.lemmy.ml";
|
10
src/lib.rs
10
src/lib.rs
|
@ -13,7 +13,6 @@ use std::time::Duration;
|
|||
use tokio::sync::Mutex;
|
||||
|
||||
pub mod crawl;
|
||||
pub mod defaults;
|
||||
|
||||
pub const REQUEST_TIMEOUT: Duration = Duration::from_secs(10);
|
||||
|
||||
|
@ -22,12 +21,12 @@ static CLIENT: Lazy<Client> = Lazy::new(Client::default);
|
|||
pub async fn start_crawl(
|
||||
start_instances: Vec<String>,
|
||||
exclude_domains: Vec<String>,
|
||||
max_depth: i32,
|
||||
max_distance: i32,
|
||||
) -> Result<Vec<InstanceDetails>, Error> {
|
||||
let params = Arc::new(CrawlParams::new(
|
||||
min_lemmy_version().await?,
|
||||
exclude_domains,
|
||||
max_depth,
|
||||
max_distance,
|
||||
Arc::new(Mutex::new(HashSet::new())),
|
||||
));
|
||||
let mut jobs = vec![];
|
||||
|
@ -36,12 +35,11 @@ pub async fn start_crawl(
|
|||
jobs.push(job.crawl());
|
||||
}
|
||||
|
||||
// TODO: optionally log the errors
|
||||
// TODO: log the errors
|
||||
let mut instance_details: Vec<InstanceDetails> = join_all(jobs)
|
||||
.await
|
||||
.into_iter()
|
||||
.filter_map(|r| r.ok())
|
||||
.flat_map(|r| r.into_iter())
|
||||
.flatten()
|
||||
.filter_map(|r| r.ok())
|
||||
.collect();
|
||||
|
||||
|
|
68
src/main.rs
68
src/main.rs
|
@ -1,52 +1,52 @@
|
|||
use anyhow::Error;
|
||||
use clap::{Arg, Command};
|
||||
use lemmy_stats_crawler::crawl::InstanceDetails;
|
||||
use lemmy_stats_crawler::defaults::{
|
||||
DEFAULT_MAX_CRAWL_DEPTH, DEFAULT_START_INSTANCES, EXCLUDE_INSTANCES,
|
||||
};
|
||||
use lemmy_stats_crawler::start_crawl;
|
||||
use serde::Serialize;
|
||||
use structopt::StructOpt;
|
||||
|
||||
#[derive(StructOpt, Debug)]
|
||||
#[structopt()]
|
||||
struct Parameters {
|
||||
#[structopt(short, long, default_value = "lemmy.ml")]
|
||||
start_instances: Vec<String>,
|
||||
#[structopt(short, long, default_value = "ds9.lemmy.ml, enterprise.lemmy.ml, voyager.lemmy.ml, test.lemmy.ml")]
|
||||
exclude_instances: Vec<String>,
|
||||
#[structopt(short, long, default_value = "20")]
|
||||
max_crawl_distance: i32,
|
||||
|
||||
/// Silence all output
|
||||
#[structopt(short, long)]
|
||||
quiet: bool,
|
||||
/// Verbose mode (-v, -vv, -vvv, etc)
|
||||
#[structopt(short = "v", long = "verbose", parse(from_occurrences))]
|
||||
verbose: usize,
|
||||
}
|
||||
|
||||
#[tokio::main]
|
||||
pub async fn main() -> Result<(), Error> {
|
||||
let matches = Command::new("Lemmy Stats Crawler")
|
||||
.arg(
|
||||
Arg::new("start-instances")
|
||||
.long("start-instances")
|
||||
.takes_value(true),
|
||||
)
|
||||
.arg(Arg::new("exclude").long("exclude").takes_value(true))
|
||||
.arg(
|
||||
Arg::new("max-crawl-depth")
|
||||
.long("max-crawl-depth")
|
||||
.takes_value(true),
|
||||
)
|
||||
.get_matches();
|
||||
let start_instances: Vec<String> = matches
|
||||
.value_of("start-instances")
|
||||
.unwrap_or(DEFAULT_START_INSTANCES)
|
||||
.split(',')
|
||||
.map(|s| s.trim().to_string())
|
||||
.collect();
|
||||
let exclude: Vec<String> = matches
|
||||
.value_of("exclude")
|
||||
.unwrap_or(EXCLUDE_INSTANCES)
|
||||
.split(',')
|
||||
.map(|s| s.trim().to_string())
|
||||
.collect();
|
||||
let max_crawl_depth: i32 = matches
|
||||
.value_of("max-crawl-depth")
|
||||
.unwrap_or(DEFAULT_MAX_CRAWL_DEPTH)
|
||||
.parse()?;
|
||||
let params = Parameters::from_args();
|
||||
|
||||
stderrlog::new()
|
||||
.module(module_path!())
|
||||
.quiet(params.quiet)
|
||||
.verbosity(params.verbose)
|
||||
.init()?;
|
||||
|
||||
eprintln!("Crawling...");
|
||||
let instance_details = start_crawl(start_instances, exclude, max_crawl_depth).await?;
|
||||
let instance_details = start_crawl(
|
||||
params.start_instances,
|
||||
params.exclude_instances,
|
||||
params.max_crawl_distance,
|
||||
)
|
||||
.await?;
|
||||
let total_stats = aggregate(instance_details);
|
||||
|
||||
println!("{}", serde_json::to_string_pretty(&total_stats)?);
|
||||
Ok(())
|
||||
}
|
||||
|
||||
// TODO: lemmy stores these numbers in SiteAggregates, would be good to simply use that as a member
|
||||
// (to avoid many members). but SiteAggregates also has id, site_id fields
|
||||
#[derive(Serialize)]
|
||||
struct TotalStats {
|
||||
crawled_instances: i32,
|
||||
|
|
Loading…
Reference in a new issue