Two new config variables for allowing short scrapes

chorus-bin/src/bin/dump.rs

@@ -37,7 +37,7 @@ fn main() -> Result<(), Error> {
     config.allow_scraping = true;
 
     // Setup store
-    let store = Store::new(&config.data_directory, config.allow_scraping)?;
+    let store = Store::new(&config)?;
 
     let mut buffer: [u8; 128] = [0; 128];
     let (_incount, _outcount, filter) = Filter::from_json(b"{}", &mut buffer)?;

chorus-bin/src/main.rs

@@ -63,7 +63,7 @@ async fn main() -> Result<(), Error> {
     log::debug!(target: "Server", "Loaded config file.");
 
     // Setup store
-    let store = Store::new(&config.data_directory, config.allow_scraping)?;
+    let store = Store::new(&config)?;
     let _ = GLOBALS.store.set(store);
 
     // TLS setup

chorus-lib/src/config.rs

@@ -20,6 +20,8 @@ pub struct FriendlyConfig {
     pub user_hex_keys: Vec<String>,
     pub verify_events: bool,
     pub allow_scraping: bool,
+    pub allow_scrape_if_limited_to: u32,
+    pub allow_scrape_if_max_seconds: u64,
     pub max_subscriptions: usize,
     pub serve_ephemeral: bool,
     pub serve_relay_lists: bool,
@@ -45,6 +47,8 @@ impl Default for FriendlyConfig {
             user_hex_keys: vec![],
             verify_events: true,
             allow_scraping: false,
+            allow_scrape_if_limited_to: 100,
+            allow_scrape_if_max_seconds: 3600,
             max_subscriptions: 32,
             serve_ephemeral: true,
             serve_relay_lists: true,
@@ -72,6 +76,8 @@ impl FriendlyConfig {
             user_hex_keys,
             verify_events,
             allow_scraping,
+            allow_scrape_if_limited_to,
+            allow_scrape_if_max_seconds,
             max_subscriptions,
             serve_ephemeral,
             serve_relay_lists,
@@ -115,6 +121,8 @@ impl FriendlyConfig {
             user_hex_keys,
             verify_events,
             allow_scraping,
+            allow_scrape_if_limited_to,
+            allow_scrape_if_max_seconds,
             max_subscriptions,
             serve_ephemeral,
             serve_relay_lists,
@@ -142,6 +150,8 @@ pub struct Config {
     pub user_hex_keys: Vec<String>,
     pub verify_events: bool,
     pub allow_scraping: bool,
+    pub allow_scrape_if_limited_to: u32,
+    pub allow_scrape_if_max_seconds: u64,
     pub max_subscriptions: usize,
     pub serve_ephemeral: bool,
     pub serve_relay_lists: bool,

chorus-lib/src/store/mod.rs

@@ -3,6 +3,7 @@ pub use event_store::EventStore;
 
 mod migrations;
 
+use crate::config::Config;
 use crate::error::{ChorusError, Error};
 use crate::ip::IpData;
 use crate::types::{Event, Filter, Id, Kind, Pubkey, Time};
@@ -29,11 +30,13 @@ pub struct Store {
     deleted_events: Database<UnalignedSlice<u8>, Unit>,
     ip_data: Database<UnalignedSlice<u8>, UnalignedSlice<u8>>,
     allow_scraping: bool,
+    allow_scrape_if_limited_to: u32,
+    allow_scrape_if_max_seconds: u64,
 }
 
 impl Store {
     /// Setup persistent storage
-    pub fn new(data_directory: &str, allow_scraping: bool) -> Result<Store, Error> {
+    pub fn new(config: &Config) -> Result<Store, Error> {
         let mut builder = EnvOpenOptions::new();
         unsafe {
             builder.flags(EnvFlags::NO_TLS);
@@ -41,7 +44,7 @@ impl Store {
         builder.max_dbs(32);
         builder.map_size(1048576 * 1024 * 24); // 24 GB
 
-        let dir = format!("{}/lmdb", data_directory);
+        let dir = format!("{}/lmdb", &config.data_directory);
         fs::create_dir_all(&dir)?;
 
         let env = match builder.open(&dir) {
@@ -111,7 +114,7 @@ impl Store {
 
         txn.commit()?;
 
-        let event_map_file = format!("{}/event.map", data_directory);
+        let event_map_file = format!("{}/event.map", &config.data_directory);
         let events = EventStore::new(event_map_file)?;
 
         let store = Store {
@@ -126,7 +129,9 @@ impl Store {
             deleted_offsets,
             deleted_events,
             ip_data,
-            allow_scraping,
+            allow_scraping: config.allow_scraping,
+            allow_scrape_if_limited_to: config.allow_scrape_if_limited_to,
+            allow_scrape_if_max_seconds: config.allow_scrape_if_max_seconds,
         };
 
         // This is in migrations.rs
@@ -465,8 +470,9 @@ impl Store {
         } else {
             // SCRAPE:
             let maxtime = filter.until().0.min(Time::now().0);
-            let allow =
+            let allow = self.allow_scraping ||
-                self.allow_scraping || filter.limit() <= 100 || (maxtime - filter.since().0) < 3600;
+                filter.limit() <= self.allow_scrape_if_limited_to ||
+                (maxtime - filter.since().0) < self.allow_scrape_if_max_seconds;
             if !allow {
                 return Err(ChorusError::Scraper.into());
             }

contrib/chorus.toml

@@ -136,6 +136,28 @@ verify_events = true
 allow_scraping = false
 
 
+# This is a u32 count of events indicating a filter `limit` value under which a scrape is
+# allowed, irrespective of the `allow_scraping` setting. Such scrapes are not expensive due
+# to the limit.
+#
+# See `allow_scraping` to learn the definition of a scrape.
+#
+# The default is 100.
+#
+allow_scrape_if_limited_to = 100
+
+
+# This is a u64 number of seconds indicating a filter time range under which a scrape is
+# allowed, irrespective of the `allow_scraping` setting. Such scrapes are rarely expensive
+# due to the short time period.
+#
+# See `allow_scraping` to learn the definition of a scrape.
+#
+# The default is 3600.
+#
+allow_scrape_if_max_seconds = 100
+
+
 # This is an integer indicating the maximum number of subscriptions a connection can have open
 # at a given time.
 #

docs/CONFIG.md

@@ -121,6 +121,22 @@ The purpose of this setting is as a temporary setting that allows you to dump ev
 
 Default is false.
 
+### allow_scrape_if_limited_to
+
+This is a u32 count of events indicating a filter `limit` value under which a scrape is allowed, irrespective of the `allow_scraping` setting. Such scrapes are not expensive due to the limit.
+
+See `allow_scraping` to learn the definition of a scrape.
+
+The default is 100.
+
+### allow_scrape_if_max_seconds
+
+This is a u64 number of seconds indicating a filter time range under which a scrape is allowed, irrespective of the `allow_scraping` setting. Such scrapes are rarely expensive due to the short time period.
+
+See `allow_scraping` to learn the definition of a scrape.
+
+The default is 3600.
+
 ### max_subscriptions
 
 This is an integer indicating the maximum number of subscriptions a connection can have open at a given time.

sample/sample.config.toml

@@ -14,6 +14,8 @@ user_hex_keys = [
 ]
 verify_events = true
 allow_scraping = true
+allow_scrape_if_limited_to = 100
+allow_scrape_if_max_seconds = 3600
 max_subscriptions = 32
 serve_ephemeral = true
 serve_relay_lists = true