add option to disable github version fetch

.env.example

@@ -55,4 +55,6 @@ SESSION_DB_URI="sqlite:./sessions.db?mode=rwc"
 # interval in seconds for gathering .health endpoint data
 STATS_INTERVAL_S=900
 # Folder to use for git operations during nitter version checks
-GIT_SCRATCH_DIR="."
+GIT_SCRATCH_DIR="."
+# Disable github fetching of latest commit
+DISABLE_GITHUB_FETCH=false

scanner/src/lib.rs

@@ -12,7 +12,8 @@ use miette::{Context, IntoDiagnostic};
 use profile_parser::ProfileParser;
 use regex::{Regex, RegexBuilder};
 use reqwest::{
-    Client, ClientBuilder, header::{HeaderMap, HeaderValue}, retry
+    header::{HeaderMap, HeaderValue},
+    Client, ClientBuilder,
 };
 use sea_orm::{ConnectionTrait, DatabaseConnection, DbBackend, FromQueryResult, Statement};
 use thiserror::Error;
@@ -129,8 +130,9 @@ pub async fn run_scanner(
     config: ScannerConfig,
     app_state: AppState,
     disable_health_checks: bool,
+    disable_github_fetch: bool,
 ) -> miette::Result<()> {
-    let scanner = Scanner::new(db, config, app_state)
+    let scanner = Scanner::new(db, config, app_state, disable_github_fetch)
         .await
         .wrap_err("Initializing scanner!")?;
     scanner.schedule_cleanup().unwrap();
@@ -192,11 +194,13 @@ impl Scanner {
         db: DatabaseConnection,
         config: ScannerConfig,
         app_state: AppState,
+        disable_github_fetch: bool,
     ) -> miette::Result<Self> {
         let mut builder_regex_rss = RegexBuilder::new(&config.rss_content);
         builder_regex_rss.case_insensitive(true);
 
-        let version_checker = Mutex::new(VersionCheck::new(config.clone()).into_diagnostic()?);
+        let version_checker =
+            Mutex::new(VersionCheck::new(config.clone(), disable_github_fetch).into_diagnostic()?);
 
         let http_client = Self::client_builder(&config);
         let client_ipv4 = Scanner::client_builder(&config)

scanner/src/version_check.rs

@@ -4,6 +4,7 @@ use std::{collections::HashMap, path::Path};
 
 use entities::state::{scanner::ScannerConfig, CommitInfo};
 use git2::Repository;
+use tracing::warn;
 
 use crate::Result;
 
@@ -11,7 +12,6 @@ pub struct VersionCheck {
     config: ScannerConfig,
     commit_cache: HashMap<String, CommitCacheValue>,
     current_epoch: u8,
-    // TODO: this is ugly and unnecessary
     repository: Repository,
 }
 
@@ -22,10 +22,10 @@ struct CommitCacheValue {
 }
 
 impl VersionCheck {
-    pub(crate) fn new(config: ScannerConfig) -> Result<Self> {
-        let temp_dir = Path::new(&config.git_scratch_folder).join(format!("nitter_version_clone"));
-        tracing::debug!(git_check_dir=?temp_dir);
-        let repository = Repository::init(temp_dir)?;
+    pub(crate) fn new(config: ScannerConfig, disable_health_checks: bool) -> Result<Self> {
+        let clone_dir = Path::new(&config.git_scratch_folder).join(format!("nitter_version_clone"));
+        tracing::debug!(git_check_dir=?clone_dir);
+        let repository = Repository::init(clone_dir)?;
         let mut checker = VersionCheck {
             config,
             commit_cache: Default::default(),
@@ -33,7 +33,11 @@ impl VersionCheck {
             repository,
         };
 
-        checker.update_remote()?;
+        if !disable_health_checks {
+            checker.update_remote()?;
+        } else {
+            warn!("Version fetching disabled, attempting to use existing dir.");
+        }
         Ok(checker)
     }
     /// Increases the epoch, invalidating unused cache entries and removes them.
@@ -148,7 +152,7 @@ mod test {
 
     #[test]
     fn test_git_commit_exists() {
-        let mut checker = VersionCheck::new(Config::test_defaults()).unwrap();
+        let mut checker = VersionCheck::new(Config::test_defaults(), false).unwrap();
         assert_eq!(
             CommitInfo::Outdated,
             checker

src/main.rs

@@ -84,12 +84,14 @@ async fn _main() -> miette::Result<()> {
     let cache = entities::state::new();
 
     let disable_health_checks = require_env_str("DISABLE_HEALTH_CHECKS")? == "true";
+    let disable_github_fetch = require_env_str("DISABLE_GITHUB_FETCH")? == "true";
 
     scanner::run_scanner(
         pool.clone(),
         scanner_config.clone(),
         cache.clone(),
         disable_health_checks,
+        disable_github_fetch,
     )
     .await
     .wrap_err("Crash starting background scanner")?;