fix views analytics (#885)

* fix views analytics

* update ip stripping

* update clickhouse tables

* fix broken queries

* Fix panics

* fix download undercounting

* fix packerator failing sometimes

* run prep
This commit is contained in:
Geometrically
2024-03-02 14:04:46 -07:00
committed by GitHub
parent 04d834187b
commit e2ffeab8fa
12 changed files with 207 additions and 140 deletions

View File

@@ -5,12 +5,15 @@ use crate::routes::ApiError;
use dashmap::{DashMap, DashSet};
use redis::cmd;
use sqlx::PgPool;
use std::collections::HashMap;
use std::net::Ipv6Addr;
const DOWNLOADS_NAMESPACE: &str = "downloads";
const VIEWS_NAMESPACE: &str = "views";
pub struct AnalyticsQueue {
views_queue: DashSet<PageView>,
downloads_queue: DashMap<String, Download>,
views_queue: DashMap<(u64, u64), Vec<PageView>>,
downloads_queue: DashMap<(u64, u64), Download>,
playtime_queue: DashSet<Playtime>,
}
@@ -24,26 +27,37 @@ impl Default for AnalyticsQueue {
impl AnalyticsQueue {
pub fn new() -> Self {
AnalyticsQueue {
views_queue: DashSet::with_capacity(1000),
views_queue: DashMap::with_capacity(1000),
downloads_queue: DashMap::with_capacity(1000),
playtime_queue: DashSet::with_capacity(1000),
}
}
pub fn add_view(&self, page_view: PageView) {
self.views_queue.insert(page_view);
fn strip_ip(ip: Ipv6Addr) -> u64 {
if let Some(ip) = ip.to_ipv4_mapped() {
let octets = ip.octets();
u64::from_be_bytes([octets[0], octets[1], octets[2], octets[3], 0, 0, 0, 0])
} else {
let octets = ip.octets();
u64::from_be_bytes([
octets[0], octets[1], octets[2], octets[3], octets[4], octets[5], octets[6],
octets[7],
])
}
}
pub fn add_view(&self, page_view: PageView) {
let ip_stripped = Self::strip_ip(page_view.ip);
self.views_queue
.entry((ip_stripped, page_view.project_id))
.or_default()
.push(page_view);
}
pub fn add_download(&self, download: Download) {
let ip_stripped = if let Some(ip) = download.ip.to_ipv4_mapped() {
let octets = ip.octets();
u64::from_be_bytes([0, 0, 0, 0, octets[0], octets[1], octets[2], octets[3]])
} else {
let octets = download.ip.octets();
u64::from_be_bytes([0, 0, 0, 0, octets[0], octets[1], octets[2], octets[3]])
};
let ip_stripped = Self::strip_ip(download.ip);
self.downloads_queue
.insert(format!("{}-{}", ip_stripped, download.project_id), download);
.insert((ip_stripped, download.project_id), download);
}
pub fn add_playtime(&self, playtime: Playtime) {
@@ -65,16 +79,6 @@ impl AnalyticsQueue {
let playtime_queue = self.playtime_queue.clone();
self.playtime_queue.clear();
if !views_queue.is_empty() {
let mut views = client.insert("views")?;
for view in views_queue {
views.write(&view).await?;
}
views.end().await?;
}
if !playtime_queue.is_empty() {
let mut playtimes = client.insert("playtime")?;
@@ -85,6 +89,78 @@ impl AnalyticsQueue {
playtimes.end().await?;
}
if !views_queue.is_empty() {
let mut views_keys = Vec::new();
let mut raw_views = Vec::new();
for (key, views) in views_queue {
views_keys.push(key);
raw_views.push((views, true));
}
let mut redis = redis.pool.get().await.map_err(DatabaseError::RedisPool)?;
let results = cmd("MGET")
.arg(
views_keys
.iter()
.map(|x| format!("{}:{}-{}", VIEWS_NAMESPACE, x.0, x.1))
.collect::<Vec<_>>(),
)
.query_async::<_, Vec<Option<u32>>>(&mut redis)
.await
.map_err(DatabaseError::CacheError)?;
let mut pipe = redis::pipe();
for (idx, count) in results.into_iter().enumerate() {
let key = &views_keys[idx];
let new_count = if let Some((views, monetized)) = raw_views.get_mut(idx) {
if let Some(count) = count {
println!("len: {} count: {}", views.len(), count);
if count > 3 {
*monetized = false;
continue;
}
if (count + views.len() as u32) > 3 {
*monetized = false;
}
count + (views.len() as u32)
} else {
views.len() as u32
}
} else {
1
};
pipe.atomic().set_ex(
format!("{}:{}-{}", VIEWS_NAMESPACE, key.0, key.1),
new_count,
6 * 60 * 60,
);
}
pipe.query_async(&mut *redis)
.await
.map_err(DatabaseError::CacheError)?;
let mut views = client.insert("views")?;
for (all_views, monetized) in raw_views {
for (idx, mut view) in all_views.into_iter().enumerate() {
if idx != 0 || !monetized {
view.monetized = false;
}
views.write(&view).await?;
}
}
views.end().await?;
}
if !downloads_queue.is_empty() {
let mut downloads_keys = Vec::new();
let raw_downloads = DashMap::new();
@@ -100,7 +176,7 @@ impl AnalyticsQueue {
.arg(
downloads_keys
.iter()
.map(|x| format!("{}:{}", DOWNLOADS_NAMESPACE, x))
.map(|x| format!("{}:{}-{}", DOWNLOADS_NAMESPACE, x.0, x.1))
.collect::<Vec<_>>(),
)
.query_async::<_, Vec<Option<u32>>>(&mut redis)
@@ -123,7 +199,7 @@ impl AnalyticsQueue {
};
pipe.atomic().set_ex(
format!("{}:{}", DOWNLOADS_NAMESPACE, key),
format!("{}:{}-{}", DOWNLOADS_NAMESPACE, key.0, key.1),
new_count,
6 * 60 * 60,
);
@@ -132,37 +208,46 @@ impl AnalyticsQueue {
.await
.map_err(DatabaseError::CacheError)?;
let version_ids = raw_downloads
.iter()
.map(|x| x.version_id as i64)
.collect::<Vec<_>>();
let project_ids = raw_downloads
.iter()
.map(|x| x.project_id as i64)
.collect::<Vec<_>>();
let mut transaction = pool.begin().await?;
let mut downloads = client.insert("downloads")?;
let mut version_downloads: HashMap<i64, i32> = HashMap::new();
let mut project_downloads: HashMap<i64, i32> = HashMap::new();
for (_, download) in raw_downloads {
*version_downloads
.entry(download.version_id as i64)
.or_default() += 1;
*project_downloads
.entry(download.project_id as i64)
.or_default() += 1;
downloads.write(&download).await?;
}
sqlx::query!(
"UPDATE versions
SET downloads = downloads + 1
WHERE id = ANY($1)",
&version_ids
sqlx::query(
"
UPDATE versions v
SET downloads = v.downloads + x.amount
FROM unnest($1::BIGINT[], $2::int[]) AS x(id, amount)
WHERE v.id = x.id
",
)
.bind(version_downloads.keys().copied().collect::<Vec<_>>())
.bind(version_downloads.values().copied().collect::<Vec<_>>())
.execute(&mut *transaction)
.await?;
sqlx::query!(
"UPDATE mods
SET downloads = downloads + 1
WHERE id = ANY($1)",
&project_ids
sqlx::query(
"
UPDATE mods m
SET downloads = m.downloads + x.amount
FROM unnest($1::BIGINT[], $2::int[]) AS x(id, amount)
WHERE m.id = x.id
",
)
.bind(project_downloads.keys().copied().collect::<Vec<_>>())
.bind(project_downloads.values().copied().collect::<Vec<_>>())
.execute(&mut *transaction)
.await?;