fix(package): ignore all files that start with a dot

server/src/db/query/package.rs

@@ -4,6 +4,9 @@ use sea_orm::{sea_query::IntoCondition, *};
 use sea_query::{Alias, Expr, Query, SelectStatement};
 use serde::Deserialize;
 
+/// How many fields may be inserted at once into the database.
+const PACKAGE_INSERT_LIMIT: usize = 1000;
+
 #[derive(Deserialize)]
 pub struct Filter {
     repo: Option<i32>,
@@ -160,23 +163,34 @@ pub async fn insert(
                 .iter()
                 .map(|s| (PackageRelatedEnum::Optdepend, s)),
         );
+    let related = crate::util::Chunked::new(related, PACKAGE_INSERT_LIMIT);
 
-    PackageRelated::insert_many(related.map(|(t, s)| package_related::ActiveModel {
-        package_id: Set(pkg_entry.id),
-        r#type: Set(t),
-        name: Set(s.to_string()),
-    }))
-    .on_empty_do_nothing()
-    .exec(&txn)
-    .await?;
+    for chunk in related {
+        PackageRelated::insert_many(
+            chunk
+                .into_iter()
+                .map(|(t, s)| package_related::ActiveModel {
+                    package_id: Set(pkg_entry.id),
+                    r#type: Set(t),
+                    name: Set(s.to_string()),
+                }),
+        )
+        .on_empty_do_nothing()
+        .exec(&txn)
+        .await?;
+    }
 
-    PackageFile::insert_many(pkg.files.iter().map(|s| package_file::ActiveModel {
-        package_id: Set(pkg_entry.id),
-        path: Set(s.display().to_string()),
-    }))
-    .on_empty_do_nothing()
-    .exec(&txn)
-    .await?;
+    let files = crate::util::Chunked::new(pkg.files, PACKAGE_INSERT_LIMIT);
+
+    for chunk in files {
+        PackageFile::insert_many(chunk.into_iter().map(|s| package_file::ActiveModel {
+            package_id: Set(pkg_entry.id),
+            path: Set(s.display().to_string()),
+        }))
+        .on_empty_do_nothing()
+        .exec(&txn)
+        .await?;
+    }
 
     txn.commit().await?;
 

server/src/main.rs

@@ -3,6 +3,7 @@ mod config;
 pub mod db;
 mod error;
 mod repo;
+mod util;
 mod web;
 
 pub use config::{Config, DbConfig, FsConfig};

server/src/repo/package.rs

@@ -13,8 +13,6 @@ use libarchive::{
 };
 use sea_orm::ActiveValue::Set;
 
-const IGNORED_FILES: [&str; 5] = [".BUILDINFO", ".INSTALL", ".MTREE", ".PKGINFO", ".CHANGELOG"];
-
 #[derive(Debug, Clone)]
 pub struct Package {
     pub path: PathBuf,
@@ -158,11 +156,9 @@ impl Package {
             let entry = entry?;
             let path_name = entry.pathname();
 
-            if !IGNORED_FILES.iter().any(|p| p == &path_name) {
+            if !path_name.starts_with('.') {
                 files.push(PathBuf::from(path_name));
-            }
-
-            if path_name == ".PKGINFO" {
+            } else if path_name == ".PKGINFO" {
                 info = Some(PkgInfo::parse(entry)?);
             }
         }

server/src/util.rs

@@ -0,0 +1,23 @@
+pub struct Chunked<I> {
+    iter: I,
+    chunk_size: usize,
+}
+
+impl<I: Iterator> Chunked<I> {
+    pub fn new<T: IntoIterator<IntoIter = I>>(into: T, chunk_size: usize) -> Self {
+        Self {
+            iter: into.into_iter(),
+            chunk_size,
+        }
+    }
+}
+
+// https://users.rust-lang.org/t/how-to-breakup-an-iterator-into-chunks/87915/5
+impl<I: Iterator> Iterator for Chunked<I> {
+    type Item = Vec<I::Item>;
+
+    fn next(&mut self) -> Option<Self::Item> {
+        Some(self.iter.by_ref().take(self.chunk_size).collect())
+            .filter(|chunk: &Vec<_>| !chunk.is_empty())
+    }
+}