refactor!: return list of files from archive extract function

Author: brianheineman

postgresql_archive/src/archive.rs

@@ -4,7 +4,7 @@
 use crate::error::Result;
 use crate::{extractor, repository};
 use semver::{Version, VersionReq};
-use std::path::Path;
+use std::path::{Path, PathBuf};
 use tracing::instrument;
 
 /// Gets the version for the specified [version requirement](VersionReq). If a version for the
@@ -41,7 +41,7 @@ pub async fn get_archive(url: &str, version_req: &VersionReq) -> Result<(Version
 /// Returns an error if the extraction fails.
 #[allow(clippy::cast_precision_loss)]
 #[instrument(skip(bytes))]
-pub async fn extract(url: &str, bytes: &Vec<u8>, out_dir: &Path) -> Result<()> {
+pub async fn extract(url: &str, bytes: &Vec<u8>, out_dir: &Path) -> Result<Vec<PathBuf>> {
     let extractor_fn = extractor::registry::get(url)?;
     extractor_fn(bytes, out_dir)
 }

postgresql_archive/src/blocking/archive.rs

@@ -1,5 +1,5 @@
 use crate::{Version, VersionReq};
-use std::path::Path;
+use std::path::{Path, PathBuf};
 use std::sync::LazyLock;
 use tokio::runtime::Runtime;
 
@@ -33,7 +33,7 @@ pub fn get_archive(url: &str, version_req: &VersionReq) -> crate::Result<(Versio
 ///
 /// # Errors
 /// Returns an error if the extraction fails.
-pub fn extract(url: &str, bytes: &Vec<u8>, out_dir: &Path) -> crate::Result<()> {
+pub fn extract(url: &str, bytes: &Vec<u8>, out_dir: &Path) -> crate::Result<Vec<PathBuf>> {
     RUNTIME
         .handle()
         .block_on(async move { crate::extract(url, bytes, out_dir).await })

postgresql_archive/src/configuration/theseus/extractor.rs

@@ -17,11 +17,11 @@ use tracing::{debug, instrument, warn};
 /// Returns an error if the extraction fails.
 #[allow(clippy::cast_precision_loss)]
 #[instrument(skip(bytes))]
-pub fn extract(bytes: &Vec<u8>, out_dir: &Path) -> Result<()> {
+pub fn extract(bytes: &Vec<u8>, out_dir: &Path) -> Result<Vec<PathBuf>> {
+    let mut files = Vec::new();
     let input = BufReader::new(Cursor::new(bytes));
     let decoder = GzDecoder::new(input);
     let mut archive = Archive::new(decoder);
-    let mut files = 0;
     let mut extracted_bytes = 0;
 
     let parent_dir = if let Some(parent) = out_dir.parent() {
@@ -42,7 +42,7 @@ pub fn extract(bytes: &Vec<u8>, out_dir: &Path) -> Result<()> {
             out_dir.to_string_lossy()
         );
         remove_file(&lock_file)?;
-        return Ok(());
+        return Ok(files);
     }
 
     let extract_dir = tempfile::tempdir_in(parent_dir)?.into_path();
@@ -74,20 +74,20 @@ pub fn extract(bytes: &Vec<u8>, out_dir: &Path) -> Result<()> {
         } else if entry_type.is_file() {
             let mut output_file = File::create(&entry_name)?;
             copy(&mut entry, &mut output_file)?;
-
-            files += 1;
             extracted_bytes += entry_size;
 
             #[cfg(unix)]
             {
                 use std::os::unix::fs::PermissionsExt;
                 output_file.set_permissions(std::fs::Permissions::from_mode(file_mode))?;
             }
+            files.push(entry_name);
         } else if entry_type.is_symlink() {
             #[cfg(unix)]
             if let Some(symlink_target) = entry.link_name()? {
-                let symlink_path = entry_name;
+                let symlink_path = entry_name.clone();
                 std::os::unix::fs::symlink(symlink_target.as_ref(), symlink_path)?;
+                files.push(entry_name);
             }
         }
     }
@@ -113,13 +113,14 @@ pub fn extract(bytes: &Vec<u8>, out_dir: &Path) -> Result<()> {
         remove_file(lock_file)?;
     }
 
+    let number_of_files = files.len();
     debug!(
         "Extracting {} files totalling {}",
-        files.to_formatted_string(&Locale::en),
+        number_of_files.to_formatted_string(&Locale::en),
         human_bytes(extracted_bytes as f64)
     );
 
-    Ok(())
+    Ok(files)
 }
 
 /// Acquires a lock file in the [out_dir](Path) to prevent multiple processes from extracting the

postgresql_archive/src/configuration/zonky/extractor.rs

@@ -19,7 +19,8 @@ use zip::ZipArchive;
 #[allow(clippy::case_sensitive_file_extension_comparisons)]
 #[allow(clippy::cast_precision_loss)]
 #[instrument(skip(bytes))]
-pub fn extract(bytes: &Vec<u8>, out_dir: &Path) -> Result<()> {
+pub fn extract(bytes: &Vec<u8>, out_dir: &Path) -> Result<Vec<PathBuf>> {
+    let mut files = Vec::new();
     let parent_dir = if let Some(parent) = out_dir.parent() {
         parent
     } else {
@@ -38,7 +39,7 @@ pub fn extract(bytes: &Vec<u8>, out_dir: &Path) -> Result<()> {
             out_dir.to_string_lossy()
         );
         remove_file(&lock_file)?;
-        return Ok(());
+        return Ok(files);
     }
 
     let extract_dir = tempfile::tempdir_in(parent_dir)?.into_path();
@@ -66,7 +67,6 @@ pub fn extract(bytes: &Vec<u8>, out_dir: &Path) -> Result<()> {
     let input = BufReader::new(Cursor::new(archive_bytes));
     let decoder = XzDecoder::new(input);
     let mut archive = Archive::new(decoder);
-    let mut files = 0;
     let mut extracted_bytes = 0;
 
     for archive_entry in archive.entries()? {
@@ -92,20 +92,20 @@ pub fn extract(bytes: &Vec<u8>, out_dir: &Path) -> Result<()> {
         } else if entry_type.is_file() {
             let mut output_file = File::create(&entry_name)?;
             copy(&mut entry, &mut output_file)?;
-
-            files += 1;
             extracted_bytes += entry_size;
 
             #[cfg(unix)]
             {
                 use std::os::unix::fs::PermissionsExt;
                 output_file.set_permissions(std::fs::Permissions::from_mode(file_mode))?;
             }
+            files.push(entry_name);
         } else if entry_type.is_symlink() {
             #[cfg(unix)]
             if let Some(symlink_target) = entry.link_name()? {
-                let symlink_path = entry_name;
+                let symlink_path = entry_name.clone();
                 std::os::unix::fs::symlink(symlink_target.as_ref(), symlink_path)?;
+                files.push(entry_name);
             }
         }
     }
@@ -131,13 +131,14 @@ pub fn extract(bytes: &Vec<u8>, out_dir: &Path) -> Result<()> {
         remove_file(lock_file)?;
     }
 
+    let number_of_files = files.len();
     debug!(
         "Extracting {} files totalling {}",
-        files.to_formatted_string(&Locale::en),
+        number_of_files.to_formatted_string(&Locale::en),
         human_bytes(extracted_bytes as f64)
     );
 
-    Ok(())
+    Ok(files)
 }
 
 /// Acquires a lock file in the [out_dir](Path) to prevent multiple processes from extracting the

postgresql_archive/src/extractor/registry.rs

@@ -4,14 +4,14 @@ use crate::configuration::theseus;
 use crate::configuration::zonky;
 use crate::Error::{PoisonedLock, UnsupportedExtractor};
 use crate::Result;
-use std::path::Path;
+use std::path::{Path, PathBuf};
 use std::sync::{Arc, LazyLock, Mutex, RwLock};
 
 static REGISTRY: LazyLock<Arc<Mutex<RepositoryRegistry>>> =
     LazyLock::new(|| Arc::new(Mutex::new(RepositoryRegistry::default())));
 
 type SupportsFn = fn(&str) -> Result<bool>;
-type ExtractFn = fn(&Vec<u8>, &Path) -> Result<()>;
+type ExtractFn = fn(&Vec<u8>, &Path) -> Result<Vec<PathBuf>>;
 
 /// Singleton struct to store extractors
 #[allow(clippy::type_complexity)]
@@ -102,7 +102,7 @@ mod tests {
 
     #[test]
     fn test_register() -> Result<()> {
-        register(|url| Ok(url == "https://foo.com"), |_, _| Ok(()))?;
+        register(|url| Ok(url == "https://foo.com"), |_, _| Ok(Vec::new()))?;
         let url = "https://foo.com";
         let extractor = get(url)?;
         assert!(extractor(&Vec::new(), Path::new("foo")).is_ok());

postgresql_archive/tests/archive.rs

@@ -32,7 +32,8 @@ async fn test_get_archive_and_extract() -> anyhow::Result<()> {
     assert!(version_req.matches(&archive_version));
 
     let out_dir = tempfile::tempdir()?.path().to_path_buf();
-    extract(url, &archive, &out_dir).await?;
+    let files = extract(url, &archive, &out_dir).await?;
+    assert_eq!(1_271, files.len());
     remove_dir_all(&out_dir)?;
     Ok(())
 }

postgresql_archive/tests/zonky.rs

@@ -41,7 +41,8 @@ async fn test_get_archive_and_extract() -> anyhow::Result<()> {
     assert!(version_req.matches(&archive_version));
 
     let out_dir = tempfile::tempdir()?.path().to_path_buf();
-    extract(url, &archive, &out_dir).await?;
+    let files = extract(url, &archive, &out_dir).await?;
+    assert_eq!(1_019, files.len());
     remove_dir_all(&out_dir)?;
     Ok(())
 }