Fix clippy lints, add 7z fallback for failed unrar

src/extractors/compress_tools.rs

@@ -63,9 +63,8 @@ impl<'a> Iterator for Extractor<'a> {
 
     fn next(&mut self) -> Option<Self::Item> {
         if self.plugin_file_paths.is_none() {
-            match self.list_plugins() {
+            if let Err(err) = self.list_plugins() {
-                Err(err) => return Some(Err(err).context(ExtractorError)),
+                return Some(Err(err).context(ExtractorError));
-                _ => {}
             }
         }
         if let Some(plugin_file_paths) = &mut self.plugin_file_paths {

src/main.rs

@@ -123,9 +123,9 @@ async fn extract_with_unrar(
 
     let mut plugin_file_paths = Vec::new();
     let list = Archive::new(&temp_file_path.to_string_lossy().to_string())?.list();
-    if let Ok(list) = list {
+    match list {
-        for entry in list {
+        Ok(list) => {
-            if let Ok(entry) = entry {
+            for entry in list.flatten() {
                 if let Some(extension) = entry.filename.extension() {
                     if entry.is_file()
                         && (extension == "esp" || extension == "esm" || extension == "esl")
@@ -135,7 +135,7 @@ async fn extract_with_unrar(
                 }
             }
         }
-    } else {
+        Err(_) => {
             if !checked_metadata {
                 warn!("failed to read archive and server has no metadata, skipping file");
                 file::update_unable_to_extract_plugins(&pool, db_file.id, true).await?;
@@ -145,12 +145,13 @@ async fn extract_with_unrar(
                 panic!("failed to read archive, but server had metadata");
             }
         }
+    }
     info!(
         num_plugin_files = plugin_file_paths.len(),
         "listed plugins in downloaded archive"
     );
 
-    if plugin_file_paths.len() > 0 {
+    if !plugin_file_paths.is_empty() {
         info!("uncompressing downloaded archive");
         let extract = Archive::new(&temp_file_path.to_string_lossy().to_string())?
             .extract_to(temp_dir.path().to_string_lossy().to_string());
@@ -163,14 +164,11 @@ async fn extract_with_unrar(
             }
             Ok(extract) => extract
         };
-        match extract.process() {
+        if let Err(err) = extract.process() {
-            Err(err) => {
             warn!(error = %err, "failed to extract with unrar");
             file::update_unable_to_extract_plugins(&pool, db_file.id, true).await?;
             return Ok(())
         }
-            _ => {}
-        }
 
         for file_path in plugin_file_paths.iter() {
             info!(
@@ -285,15 +283,17 @@ pub async fn main() -> Result<()> {
                 }
                 let db_file = file::insert(
                     &pool,
-                    api_file.name,
+                    &file::UnsavedFile {
-                    api_file.file_name,
+                        name: api_file.name,
-                    api_file.file_id as i32,
+                        file_name: api_file.file_name,
-                    db_mod.id,
+                        nexus_file_id: api_file.file_id as i32,
-                    api_file.category,
+                        mod_id: db_mod.id,
-                    api_file.version,
+                        category: api_file.category,
-                    api_file.mod_version,
+                        version: api_file.version,
-                    api_file.size,
+                        mod_version: api_file.mod_version,
-                    api_file.uploaded_at,
+                        size: api_file.size,
+                        uploaded_at: api_file.uploaded_at,
+                    },
                 )
                 .await?;
 
@@ -350,14 +350,11 @@ pub async fn main() -> Result<()> {
 
                 let mut initial_bytes = [0; 8];
                 tokio_file.seek(SeekFrom::Start(0)).await?;
-                match tokio_file.read_exact(&mut initial_bytes).await {
+                if let Err(err) = tokio_file.read_exact(&mut initial_bytes).await {
-                    Err(err) => {
                     warn!(error = %err, "failed to read initial bytes, skipping file");
                     file::update_unable_to_extract_plugins(&pool, db_file.id, true).await?;
                     continue;
                 }
-                    _ => {}
-                }
                 let kind = match infer::get(&initial_bytes) {
                     Some(kind) => kind,
                     None => {
@@ -378,8 +375,15 @@ pub async fn main() -> Result<()> {
                         // https://github.com/libarchive/libarchive/issues/373, https://github.com/libarchive/libarchive/issues/1426
                         tokio_file.seek(SeekFrom::Start(0)).await?;
                         let mut file = tokio_file.try_clone().await?.into_std().await;
-                        extract_with_unrar(&mut file, &pool, &db_file, &db_mod, checked_metadata)
+                        match extract_with_unrar(&mut file, &pool, &db_file, &db_mod, checked_metadata).await {
-                            .await?;
+                            Ok(_) => Ok(()),
+                            Err(err) => {
+                                // unrar failed to extract rar file (e.g. archive has unicode filenames)
+                                // Attempt to uncompress the archive using `7z` unix command instead
+                                warn!(error = %err, "failed to extract file with unrar, extracting whole archive with 7z instead");
+                                extract_with_7zip(&mut file, &pool, &db_file, &db_mod).await
+                            }
+                        }?;
                     }
                     _ => {
                         tokio_file.seek(SeekFrom::Start(0)).await?;

src/models/cell.rs

@@ -73,7 +73,7 @@ pub async fn batched_insert<'a>(
         let mut ys: Vec<Option<i32>> = vec![];
         let mut world_ids: Vec<Option<i32>> = vec![];
         let mut is_persistents: Vec<bool> = vec![];
-        batch.into_iter().for_each(|unsaved_cell| {
+        batch.iter().for_each(|unsaved_cell| {
             form_ids.push(unsaved_cell.form_id);
             masters.push(unsaved_cell.master);
             xs.push(unsaved_cell.x);

src/models/file.rs

@@ -23,6 +23,19 @@ pub struct File {
     pub unable_to_extract_plugins: bool,
 }
 
+#[derive(Debug)]
+pub struct UnsavedFile<'a> {
+    pub name: &'a str,
+    pub file_name: &'a str,
+    pub nexus_file_id: i32,
+    pub mod_id: i32,
+    pub category: Option<&'a str>,
+    pub version: Option<&'a str>,
+    pub mod_version: Option<&'a str>,
+    pub size: i64,
+    pub uploaded_at: NaiveDateTime,
+}
+
 #[instrument(level = "debug", skip(pool))]
 pub async fn get_by_nexus_file_id(
     pool: &sqlx::Pool<sqlx::Postgres>,
@@ -59,17 +72,9 @@ pub async fn get_processed_nexus_file_ids_by_mod_id(
 }
 
 #[instrument(level = "debug", skip(pool))]
-pub async fn insert(
+pub async fn insert<'a>(
     pool: &sqlx::Pool<sqlx::Postgres>,
-    name: &str,
+    unsaved_file: &UnsavedFile<'a>,
-    file_name: &str,
-    nexus_file_id: i32,
-    mod_id: i32,
-    category: Option<&str>,
-    version: Option<&str>,
-    mod_version: Option<&str>,
-    size: i64,
-    uploaded_at: NaiveDateTime,
 ) -> Result<File> {
     sqlx::query_as!(
         File,
@@ -80,15 +85,15 @@ pub async fn insert(
             SET (name, file_name, category, version, mod_version, uploaded_at, updated_at) =
             (EXCLUDED.name, EXCLUDED.file_name, EXCLUDED.category, EXCLUDED.version, EXCLUDED.mod_version, EXCLUDED.uploaded_at, now())
             RETURNING *",
-        name,
+        unsaved_file.name,
-        file_name,
+        unsaved_file.file_name,
-        nexus_file_id,
+        unsaved_file.nexus_file_id,
-        mod_id,
+        unsaved_file.mod_id,
-        category,
+        unsaved_file.category,
-        version,
+        unsaved_file.version,
-        mod_version,
+        unsaved_file.mod_version,
-        size,
+        unsaved_file.size,
-        uploaded_at
+        unsaved_file.uploaded_at
     )
     .fetch_one(pool)
     .await

src/models/game_mod.rs

@@ -106,7 +106,7 @@ pub async fn batched_insert<'a>(
         let mut categories: Vec<Option<&str>> = vec![];
         let mut descriptions: Vec<Option<&str>> = vec![];
         let mut game_ids: Vec<i32> = vec![];
-        batch.into_iter().for_each(|unsaved_mod| {
+        batch.iter().for_each(|unsaved_mod| {
             names.push(unsaved_mod.name);
             nexus_mod_ids.push(unsaved_mod.nexus_mod_id);
             authors.push(unsaved_mod.author);

src/models/plugin.rs

@@ -21,19 +21,24 @@ pub struct Plugin {
     pub created_at: NaiveDateTime,
 }
 
+#[derive(Debug)]
+pub struct UnsavedPlugin<'a> {
+    pub name: &'a str,
+    pub hash: i64,
+    pub file_id: i32,
+    pub version: f64,
+    pub size: i64,
+    pub author: Option<&'a str>,
+    pub description: Option<&'a str>,
+    pub masters: &'a [&'a str],
+    pub file_name: &'a str,
+    pub file_path: &'a str,
+}
+
 #[instrument(level = "debug", skip(pool))]
-pub async fn insert(
+pub async fn insert<'a>(
     pool: &sqlx::Pool<sqlx::Postgres>,
-    name: &str,
+    unsaved_plugin: &UnsavedPlugin<'a>,
-    hash: i64,
-    file_id: i32,
-    version: f64,
-    size: i64,
-    author: Option<&str>,
-    description: Option<&str>,
-    masters: &[&str],
-    file_name: &str,
-    file_path: &str,
 ) -> Result<Plugin> {
     // sqlx doesn't understand slices of &str with the query_as! macro: https://github.com/launchbadge/sqlx/issues/280
     sqlx::query_as(
@@ -45,16 +50,16 @@ pub async fn insert(
             (EXCLUDED.name, EXCLUDED.hash, EXCLUDED.version, EXCLUDED.author, EXCLUDED.description, EXCLUDED.masters, EXCLUDED.file_name, now())
             RETURNING *"#,
     )
-    .bind(name)
+    .bind(unsaved_plugin.name)
-    .bind(hash)
+    .bind(unsaved_plugin.hash)
-    .bind(file_id)
+    .bind(unsaved_plugin.file_id)
-    .bind(version)
+    .bind(unsaved_plugin.version)
-    .bind(size)
+    .bind(unsaved_plugin.size)
-    .bind(author)
+    .bind(unsaved_plugin.author)
-    .bind(description)
+    .bind(unsaved_plugin.description)
-    .bind(masters)
+    .bind(unsaved_plugin.masters)
-    .bind(file_name)
+    .bind(unsaved_plugin.file_name)
-    .bind(file_path)
+    .bind(unsaved_plugin.file_path)
     .fetch_one(pool)
     .await
     .context("Failed to insert plugin")

src/models/plugin_cell.rs

@@ -57,7 +57,7 @@ pub async fn batched_insert<'a>(
         let mut plugin_ids: Vec<i32> = vec![];
         let mut cell_ids: Vec<i32> = vec![];
         let mut editor_ids: Vec<Option<&str>> = vec![];
-        batch.into_iter().for_each(|unsaved_plugin_cell| {
+        batch.iter().for_each(|unsaved_plugin_cell| {
             plugin_ids.push(unsaved_plugin_cell.plugin_id);
             cell_ids.push(unsaved_plugin_cell.cell_id);
             editor_ids.push(unsaved_plugin_cell.editor_id);

src/models/plugin_world.rs

@@ -57,10 +57,10 @@ pub async fn batched_insert<'a>(
         let mut plugin_ids: Vec<i32> = vec![];
         let mut world_ids: Vec<i32> = vec![];
         let mut editor_ids: Vec<&str> = vec![];
-        batch.into_iter().for_each(|unsaved_plugin_world| {
+        batch.iter().for_each(|unsaved_plugin_world| {
             plugin_ids.push(unsaved_plugin_world.plugin_id);
             world_ids.push(unsaved_plugin_world.world_id);
-            editor_ids.push(unsaved_plugin_world.editor_id.clone());
+            editor_ids.push(unsaved_plugin_world.editor_id);
         });
         saved_plugin_worlds.append(
             &mut sqlx::query_as(

src/models/world.rs

@@ -52,7 +52,7 @@ pub async fn batched_insert<'a>(
     for batch in worlds.chunks(BATCH_SIZE) {
         let mut form_ids: Vec<i32> = vec![];
         let mut masters: Vec<&str> = vec![];
-        batch.into_iter().for_each(|unsaved_world| {
+        batch.iter().for_each(|unsaved_world| {
             form_ids.push(unsaved_world.form_id);
             masters.push(unsaved_world.master);
         });

src/nexus_api/files.rs

@@ -70,7 +70,7 @@ impl FilesResponse {
             .as_array()
             .ok_or_else(|| anyhow!("files value in API response is not an array"))?;
         let files: Vec<ApiFile> = files
-            .into_iter()
+            .iter()
             .map(|file| {
                 let file_id = file
                     .get("file_id")

src/nexus_scraper.rs

@@ -76,7 +76,6 @@ impl ModListResponse {
                     .attr("data-mod-id")
                     .expect("Missing mod id attribute")
                     .parse::<i32>()
-                    .ok()
                     .expect("Failed to parse mod id");
                 let name_elem = right
                     .select(&name_select)

src/plugin_processor.rs

@@ -9,7 +9,7 @@ use tracing::{info, warn};
 
 use crate::models::file::File;
 use crate::models::game_mod::Mod;
-use crate::models::plugin;
+use crate::models::{plugin, plugin::UnsavedPlugin};
 use crate::models::{cell, cell::UnsavedCell};
 use crate::models::{plugin_cell, plugin_cell::UnsavedPluginCell};
 use crate::models::{plugin_world, plugin_world::UnsavedPluginWorld};
@@ -36,7 +36,7 @@ pub async fn process_plugin(
     db_mod: &Mod,
     file_path: &str,
 ) -> Result<()> {
-    if plugin_buf.len() == 0 {
+    if plugin_buf.is_empty() {
         warn!("skipping processing of invalid empty plugin");
         return Ok(());
     }
@@ -58,16 +58,18 @@ pub async fn process_plugin(
             let masters: Vec<&str> = plugin.header.masters.iter().map(|s| s.borrow()).collect();
             let plugin_row = plugin::insert(
                 &pool,
-                &db_file.name,
+                &UnsavedPlugin {
-                hash as i64,
+                    name: &db_file.name,
-                db_file.id,
+                    hash: hash as i64,
-                plugin.header.version as f64,
+                    file_id: db_file.id,
-                plugin_buf.len() as i64,
+                    version: plugin.header.version as f64,
+                    size: plugin_buf.len() as i64,
                     author,
                     description,
-                &masters,
+                    masters: &masters,
-                &file_name,
+                    file_name: &file_name,
                     file_path,
+                },
             )
             .await?;