feat(#113): 100% blob storage model — PostgreSQL metadata + DedupService blobs

BREAKING CHANGE: Storage model completely rewritten. All file/folder
metadata now lives in PostgreSQL (storage schema). File content stored
as content-addressable blobs via DedupService. Filesystem directories
are no longer used for user storage.

New components:
- storage.folders / storage.files / storage.trash_items (PG schema)
- FolderDbRepository: virtual folders backed by PG
- FileBlobReadRepository: file reads via PG metadata + dedup blobs
- FileBlobWriteRepository: file writes via PG metadata + dedup blobs
- TrashDbRepository: soft-delete trash using is_trashed flags

Removed legacy FS components (~5500 lines deleted):
- FolderFsRepository, FileFsReadRepository, FileFsWriteRepository
- CompositeFileRepository, ParallelFileProcessor
- IdMappingService, IdMappingOptimizer, FileMetadataCache
- BufferPool, FileSystemUtils, RepositoryErrors
- TrashFsRepository, FolderFsRepositoryTrash

DI rewired: build_app_state() now requires PgPool (no FS fallback).
FileUploadService.new_with_read() and FileRetrievalService.new_with_cache()
constructors added for blob model (no write-behind needed).

Closes #113
This commit is contained in:
Dionisio
2026-02-14 17:54:25 +01:00
parent f25987e553
commit 3c7c16f07e
27 changed files with 1883 additions and 7429 deletions
@@ -0,0 +1,271 @@
//! PostgreSQL + Blob-backed file read repository.
//!
//! Implements `FileReadPort` using:
//! - `storage.files` table for metadata lookups
//! - `DedupPort` for reading content-addressable blobs from the filesystem
use async_trait::async_trait;
use bytes::Bytes;
use futures::Stream;
use sqlx::PgPool;
use std::sync::Arc;
use crate::application::ports::dedup_ports::DedupPort;
use crate::application::ports::storage_ports::FileReadPort;
use crate::common::errors::DomainError;
use crate::domain::entities::file::File;
use crate::domain::repositories::folder_repository::FolderRepository;
use crate::domain::services::path_service::StoragePath;
use super::folder_db_repository::FolderDbRepository;
/// File read repository backed by PostgreSQL metadata + blob storage.
pub struct FileBlobReadRepository {
pool: Arc<PgPool>,
dedup: Arc<dyn DedupPort>,
folder_repo: Arc<FolderDbRepository>,
}
impl FileBlobReadRepository {
pub fn new(
pool: Arc<PgPool>,
dedup: Arc<dyn DedupPort>,
folder_repo: Arc<FolderDbRepository>,
) -> Self {
Self {
pool,
dedup,
folder_repo,
}
}
/// Build a virtual StoragePath for a file.
async fn build_file_path(
&self,
folder_id: Option<&str>,
file_name: &str,
) -> Result<StoragePath, DomainError> {
if let Some(fid) = folder_id {
let folder_path = self.folder_repo.get_folder_path(fid).await?;
Ok(folder_path.join(file_name))
} else {
Ok(StoragePath::from_string(file_name))
}
}
/// Convert a database row into a `File` domain entity.
async fn row_to_file(
&self,
id: String,
name: String,
folder_id: Option<String>,
size: i64,
mime_type: String,
created_at: i64,
modified_at: i64,
) -> Result<File, DomainError> {
let storage_path = self
.build_file_path(folder_id.as_deref(), &name)
.await?;
File::with_timestamps(
id,
name,
storage_path,
size as u64,
mime_type,
folder_id,
created_at as u64,
modified_at as u64,
)
.map_err(|e| DomainError::internal_error("FileBlobRead", format!("entity: {e}")))
}
/// Get the blob hash for a file.
async fn get_blob_hash(&self, file_id: &str) -> Result<String, DomainError> {
sqlx::query_scalar::<_, String>(
"SELECT blob_hash FROM storage.files WHERE id = $1::uuid AND NOT is_trashed",
)
.bind(file_id)
.fetch_optional(self.pool.as_ref())
.await
.map_err(|e| DomainError::internal_error("FileBlobRead", format!("hash lookup: {e}")))?
.ok_or_else(|| DomainError::not_found("File", file_id))
}
}
#[async_trait]
impl FileReadPort for FileBlobReadRepository {
async fn get_file(&self, id: &str) -> Result<File, DomainError> {
let row = sqlx::query_as::<_, (String, String, Option<String>, i64, String, i64, i64)>(
r#"
SELECT id::text, name, folder_id::text, size, mime_type,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
FROM storage.files
WHERE id = $1::uuid AND NOT is_trashed
"#,
)
.bind(id)
.fetch_optional(self.pool.as_ref())
.await
.map_err(|e| DomainError::internal_error("FileBlobRead", format!("get: {e}")))?
.ok_or_else(|| DomainError::not_found("File", id))?;
self.row_to_file(row.0, row.1, row.2, row.3, row.4, row.5, row.6)
.await
}
async fn list_files(
&self,
folder_id: Option<&str>,
) -> Result<Vec<File>, DomainError> {
let rows: Vec<(String, String, Option<String>, i64, String, i64, i64)> =
if let Some(fid) = folder_id {
sqlx::query_as(
r#"
SELECT id::text, name, folder_id::text, size, mime_type,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
FROM storage.files
WHERE folder_id = $1::uuid AND NOT is_trashed
ORDER BY name
"#,
)
.bind(fid)
.fetch_all(self.pool.as_ref())
.await
} else {
sqlx::query_as(
r#"
SELECT id::text, name, folder_id::text, size, mime_type,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
FROM storage.files
WHERE folder_id IS NULL AND NOT is_trashed
ORDER BY name
"#,
)
.fetch_all(self.pool.as_ref())
.await
}
.map_err(|e| DomainError::internal_error("FileBlobRead", format!("list: {e}")))?;
let mut files = Vec::with_capacity(rows.len());
for (id, name, fid, size, mime, ca, ma) in rows {
files.push(self.row_to_file(id, name, fid, size, mime, ca, ma).await?);
}
Ok(files)
}
async fn get_file_content(&self, id: &str) -> Result<Vec<u8>, DomainError> {
let blob_hash = self.get_blob_hash(id).await?;
self.dedup.read_blob(&blob_hash).await
}
async fn get_file_stream(
&self,
id: &str,
) -> Result<Box<dyn Stream<Item = Result<Bytes, std::io::Error>> + Send>, DomainError> {
// Read blob as bytes and wrap in a single-chunk stream.
// For very large files, a true streaming implementation from the
// blob file would be better, but DedupPort API currently returns bytes.
let blob_hash = self.get_blob_hash(id).await?;
let content = self.dedup.read_blob_bytes(&blob_hash).await?;
let stream = futures::stream::once(async move { Ok(content) });
Ok(Box::new(stream))
}
async fn get_file_range_stream(
&self,
id: &str,
start: u64,
end: Option<u64>,
) -> Result<Box<dyn Stream<Item = Result<Bytes, std::io::Error>> + Send>, DomainError> {
let blob_hash = self.get_blob_hash(id).await?;
let content = self.dedup.read_blob_bytes(&blob_hash).await?;
let start = start as usize;
let end = end.map_or(content.len(), |e| e as usize).min(content.len());
if start >= content.len() {
return Ok(Box::new(futures::stream::empty()));
}
let slice = content.slice(start..end);
let stream = futures::stream::once(async move { Ok(slice) });
Ok(Box::new(stream))
}
async fn get_file_mmap(&self, id: &str) -> Result<Bytes, DomainError> {
let blob_hash = self.get_blob_hash(id).await?;
self.dedup.read_blob_bytes(&blob_hash).await
}
async fn get_file_path(&self, id: &str) -> Result<StoragePath, DomainError> {
let row = sqlx::query_as::<_, (String, Option<String>)>(
r#"
SELECT name, folder_id::text
FROM storage.files
WHERE id = $1::uuid AND NOT is_trashed
"#,
)
.bind(id)
.fetch_optional(self.pool.as_ref())
.await
.map_err(|e| DomainError::internal_error("FileBlobRead", format!("path: {e}")))?
.ok_or_else(|| DomainError::not_found("File", id))?;
self.build_file_path(row.1.as_deref(), &row.0).await
}
async fn get_parent_folder_id(&self, path: &str) -> Result<String, DomainError> {
// Walk the path to find the parent folder, searching by folder names
let path = path.trim_start_matches('/').trim_end_matches('/');
let segments: Vec<&str> = path.split('/').filter(|s| !s.is_empty()).collect();
if segments.is_empty() {
return Err(DomainError::not_found("Folder", "empty path"));
}
// For path "a/b/c/file.txt", the parent folder path is "a/b/c"
// But we don't know which part is folders vs filename.
// Walk segments trying to find matching folders.
let mut current_parent: Option<String> = None;
for segment in &segments {
let row = if let Some(ref pid) = current_parent {
sqlx::query_as::<_, (String,)>(
r#"
SELECT id::text FROM storage.folders
WHERE name = $1 AND parent_id = $2::uuid AND NOT is_trashed
"#,
)
.bind(segment)
.bind(pid)
.fetch_optional(self.pool.as_ref())
.await
} else {
sqlx::query_as::<_, (String,)>(
r#"
SELECT id::text FROM storage.folders
WHERE name = $1 AND parent_id IS NULL AND NOT is_trashed
"#,
)
.bind(segment)
.fetch_optional(self.pool.as_ref())
.await
}
.map_err(|e| DomainError::internal_error("FileBlobRead", format!("path walk: {e}")))?;
match row {
Some(r) => current_parent = Some(r.0),
None => break, // This segment is not a folder → it's the filename
}
}
current_parent.ok_or_else(|| {
DomainError::not_found("Folder", format!("parent for path: {path}"))
})
}
}
@@ -0,0 +1,435 @@
//! PostgreSQL + Blob-backed file write repository.
//!
//! Implements `FileWritePort` using:
//! - `storage.files` table for metadata
//! - `DedupPort` for content-addressable blob storage on the filesystem
use async_trait::async_trait;
use bytes::Bytes;
use futures::Stream;
use sqlx::PgPool;
use std::path::PathBuf;
use std::pin::Pin;
use std::sync::Arc;
use crate::application::ports::dedup_ports::DedupPort;
use crate::application::ports::storage_ports::FileWritePort;
use crate::common::errors::DomainError;
use crate::domain::entities::file::File;
use crate::domain::repositories::folder_repository::FolderRepository;
use crate::domain::services::path_service::StoragePath;
use super::folder_db_repository::FolderDbRepository;
/// File write repository backed by PostgreSQL metadata + blob storage.
pub struct FileBlobWriteRepository {
pool: Arc<PgPool>,
dedup: Arc<dyn DedupPort>,
folder_repo: Arc<FolderDbRepository>,
}
impl FileBlobWriteRepository {
pub fn new(
pool: Arc<PgPool>,
dedup: Arc<dyn DedupPort>,
folder_repo: Arc<FolderDbRepository>,
) -> Self {
Self {
pool,
dedup,
folder_repo,
}
}
/// Build a virtual StoragePath for a file from its DB metadata.
async fn build_file_path(
&self,
folder_id: Option<&str>,
file_name: &str,
) -> Result<StoragePath, DomainError> {
if let Some(fid) = folder_id {
let folder_path = self.folder_repo.get_folder_path(fid).await?;
Ok(folder_path.join(file_name))
} else {
Ok(StoragePath::from_string(file_name))
}
}
/// Convert a database row into a `File` domain entity.
async fn row_to_file(
&self,
id: String,
name: String,
folder_id: Option<String>,
size: i64,
mime_type: String,
created_at: i64,
modified_at: i64,
) -> Result<File, DomainError> {
let storage_path = self
.build_file_path(folder_id.as_deref(), &name)
.await?;
File::with_timestamps(
id,
name,
storage_path,
size as u64,
mime_type,
folder_id,
created_at as u64,
modified_at as u64,
)
.map_err(|e| DomainError::internal_error("FileBlobWrite", format!("entity: {e}")))
}
/// Derive user_id from the parent folder, or error if folder_id is None.
async fn resolve_user_id(&self, folder_id: Option<&str>) -> Result<String, DomainError> {
match folder_id {
Some(fid) => self.folder_repo.get_folder_user_id(fid).await,
None => Err(DomainError::internal_error(
"FileBlobWrite",
"folder_id is required to determine file owner",
)),
}
}
}
#[async_trait]
impl FileWritePort for FileBlobWriteRepository {
async fn save_file(
&self,
name: String,
folder_id: Option<String>,
content_type: String,
content: Vec<u8>,
) -> Result<File, DomainError> {
let user_id = self.resolve_user_id(folder_id.as_deref()).await?;
let size = content.len() as i64;
// Store content in blob store
let dedup_result = self
.dedup
.store_bytes(&content, Some(content_type.clone()))
.await?;
let blob_hash = dedup_result.hash().to_string();
// Insert file metadata
let row = sqlx::query_as::<_, (String, i64, i64)>(
r#"
INSERT INTO storage.files (name, folder_id, user_id, blob_hash, size, mime_type)
VALUES ($1, $2::uuid, $3, $4, $5, $6)
RETURNING id::text,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
"#,
)
.bind(&name)
.bind(&folder_id)
.bind(&user_id)
.bind(&blob_hash)
.bind(size)
.bind(&content_type)
.fetch_one(self.pool.as_ref())
.await
.map_err(|e| {
if let sqlx::Error::Database(ref db_err) = e {
if db_err.code().as_deref() == Some("23505") {
return DomainError::already_exists(
"File",
format!("{name} already exists in folder"),
);
}
}
DomainError::internal_error("FileBlobWrite", format!("insert: {e}"))
})?;
tracing::info!(
"💾 BLOB WRITE: {} ({} bytes, hash: {})",
name,
size,
&blob_hash[..12]
);
self.row_to_file(
row.0,
name,
folder_id,
size,
content_type,
row.1,
row.2,
)
.await
}
async fn save_file_from_stream(
&self,
name: String,
folder_id: Option<String>,
content_type: String,
stream: Pin<Box<dyn Stream<Item = Result<Bytes, std::io::Error>> + Send>>,
) -> Result<File, DomainError> {
use futures::StreamExt;
// Collect stream into bytes (blobs are content-addressed, need full content for hash)
let mut content = Vec::new();
let mut stream = stream;
while let Some(chunk) = stream.next().await {
let chunk = chunk.map_err(|e| {
DomainError::internal_error("FileBlobWrite", format!("stream read: {e}"))
})?;
content.extend_from_slice(&chunk);
}
self.save_file(name, folder_id, content_type, content).await
}
async fn move_file(
&self,
file_id: &str,
target_folder_id: Option<String>,
) -> Result<File, DomainError> {
// If moving to a different folder, get the new user_id (must be same user)
let row = sqlx::query_as::<_, (String, String, Option<String>, i64, String, i64, i64)>(
r#"
UPDATE storage.files
SET folder_id = $1::uuid, updated_at = NOW()
WHERE id = $2::uuid AND NOT is_trashed
RETURNING id::text, name, folder_id::text, size, mime_type,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
"#,
)
.bind(&target_folder_id)
.bind(file_id)
.fetch_optional(self.pool.as_ref())
.await
.map_err(|e| DomainError::internal_error("FileBlobWrite", format!("move: {e}")))?
.ok_or_else(|| DomainError::not_found("File", file_id))?;
self.row_to_file(row.0, row.1, row.2, row.3, row.4, row.5, row.6)
.await
}
async fn rename_file(
&self,
file_id: &str,
new_name: &str,
) -> Result<File, DomainError> {
let row = sqlx::query_as::<_, (String, String, Option<String>, i64, String, i64, i64)>(
r#"
UPDATE storage.files
SET name = $1, updated_at = NOW()
WHERE id = $2::uuid AND NOT is_trashed
RETURNING id::text, name, folder_id::text, size, mime_type,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
"#,
)
.bind(new_name)
.bind(file_id)
.fetch_optional(self.pool.as_ref())
.await
.map_err(|e| {
if let sqlx::Error::Database(ref db_err) = e {
if db_err.code().as_deref() == Some("23505") {
return DomainError::already_exists(
"File",
format!("{new_name} already exists"),
);
}
}
DomainError::internal_error("FileBlobWrite", format!("rename: {e}"))
})?
.ok_or_else(|| DomainError::not_found("File", file_id))?;
self.row_to_file(row.0, row.1, row.2, row.3, row.4, row.5, row.6)
.await
}
async fn delete_file(&self, id: &str) -> Result<(), DomainError> {
// Get blob_hash before deleting so we can decrement ref
let hash = sqlx::query_scalar::<_, String>(
"SELECT blob_hash FROM storage.files WHERE id = $1::uuid",
)
.bind(id)
.fetch_optional(self.pool.as_ref())
.await
.map_err(|e| DomainError::internal_error("FileBlobWrite", format!("hash lookup: {e}")))?;
let result = sqlx::query("DELETE FROM storage.files WHERE id = $1::uuid")
.bind(id)
.execute(self.pool.as_ref())
.await
.map_err(|e| DomainError::internal_error("FileBlobWrite", format!("delete: {e}")))?;
if result.rows_affected() == 0 {
return Err(DomainError::not_found("File", id));
}
// Decrement blob reference
if let Some(h) = hash {
if let Err(e) = self.dedup.remove_reference(&h).await {
tracing::warn!("Failed to decrement blob ref for {}: {}", &h[..12], e);
}
}
Ok(())
}
async fn update_file_content(
&self,
file_id: &str,
content: Vec<u8>,
) -> Result<(), DomainError> {
// Get old blob hash to decrement ref
let old_hash = sqlx::query_scalar::<_, String>(
"SELECT blob_hash FROM storage.files WHERE id = $1::uuid",
)
.bind(file_id)
.fetch_optional(self.pool.as_ref())
.await
.map_err(|e| DomainError::internal_error("FileBlobWrite", format!("old hash: {e}")))?
.ok_or_else(|| DomainError::not_found("File", file_id))?;
// Store new content
let new_size = content.len() as i64;
let dedup_result = self.dedup.store_bytes(&content, None).await?;
let new_hash = dedup_result.hash().to_string();
// Update file metadata
sqlx::query(
r#"
UPDATE storage.files
SET blob_hash = $1, size = $2, updated_at = NOW()
WHERE id = $3::uuid
"#,
)
.bind(&new_hash)
.bind(new_size)
.bind(file_id)
.execute(self.pool.as_ref())
.await
.map_err(|e| DomainError::internal_error("FileBlobWrite", format!("update: {e}")))?;
// Decrement old blob ref (only if hash changed)
if old_hash != new_hash {
if let Err(e) = self.dedup.remove_reference(&old_hash).await {
tracing::warn!(
"Failed to decrement old blob ref {}: {}",
&old_hash[..12],
e
);
}
}
Ok(())
}
async fn register_file_deferred(
&self,
name: String,
folder_id: Option<String>,
content_type: String,
size: u64,
) -> Result<(File, PathBuf), DomainError> {
let user_id = self.resolve_user_id(folder_id.as_deref()).await?;
// For deferred registration we use a placeholder hash.
// The write-behind cache will call update_file_content later.
let placeholder_hash = "0000000000000000000000000000000000000000000000000000000000000000";
let row = sqlx::query_as::<_, (String, i64, i64)>(
r#"
INSERT INTO storage.files (name, folder_id, user_id, blob_hash, size, mime_type)
VALUES ($1, $2::uuid, $3, $4, $5, $6)
RETURNING id::text,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
"#,
)
.bind(&name)
.bind(&folder_id)
.bind(&user_id)
.bind(placeholder_hash)
.bind(size as i64)
.bind(&content_type)
.fetch_one(self.pool.as_ref())
.await
.map_err(|e| DomainError::internal_error("FileBlobWrite", format!("deferred: {e}")))?;
let file = self
.row_to_file(
row.0.clone(),
name,
folder_id,
size as i64,
content_type,
row.1,
row.2,
)
.await?;
// The target_path is not meaningful for blob storage (content goes to .blobs/)
// but the WriteBehindCache API requires it. We return a synthetic path.
let target_path = PathBuf::from(format!(".pending/{}", row.0));
Ok((file, target_path))
}
// ── Trash operations ──
async fn move_to_trash(&self, file_id: &str) -> Result<(), DomainError> {
let result = sqlx::query(
r#"
UPDATE storage.files
SET is_trashed = TRUE,
trashed_at = NOW(),
original_folder_id = folder_id,
updated_at = NOW()
WHERE id = $1::uuid AND NOT is_trashed
"#,
)
.bind(file_id)
.execute(self.pool.as_ref())
.await
.map_err(|e| DomainError::internal_error("FileBlobWrite", format!("trash: {e}")))?;
if result.rows_affected() == 0 {
return Err(DomainError::not_found("File", file_id));
}
Ok(())
}
async fn restore_from_trash(
&self,
file_id: &str,
_original_path: &str,
) -> Result<(), DomainError> {
let result = sqlx::query(
r#"
UPDATE storage.files
SET is_trashed = FALSE,
trashed_at = NULL,
folder_id = COALESCE(original_folder_id, folder_id),
original_folder_id = NULL,
updated_at = NOW()
WHERE id = $1::uuid AND is_trashed
"#,
)
.bind(file_id)
.execute(self.pool.as_ref())
.await
.map_err(|e| DomainError::internal_error("FileBlobWrite", format!("restore: {e}")))?;
if result.rows_affected() == 0 {
return Err(DomainError::not_found("File", file_id));
}
Ok(())
}
async fn delete_file_permanently(&self, file_id: &str) -> Result<(), DomainError> {
// Same as delete_file — removes from DB and decrements blob ref
self.delete_file(file_id).await
}
}
@@ -0,0 +1,603 @@
//! PostgreSQL-backed folder repository.
//!
//! Implements `FolderRepository` (and thus `FolderStoragePort`) using the
//! `storage.folders` table. Folders are purely virtual — no physical
//! directories are created on the filesystem.
use async_trait::async_trait;
use sqlx::PgPool;
use std::sync::Arc;
use crate::common::errors::DomainError;
use crate::domain::entities::folder::Folder;
use crate::domain::repositories::folder_repository::FolderRepository;
use crate::domain::services::path_service::StoragePath;
/// PostgreSQL-backed folder repository.
///
/// All folder metadata lives in the `storage.folders` table. The physical
/// filesystem is never touched for folder operations.
pub struct FolderDbRepository {
pool: Option<Arc<PgPool>>,
}
impl FolderDbRepository {
pub fn new(pool: Arc<PgPool>) -> Self {
Self { pool: Some(pool) }
}
/// Creates a stub instance for `AppState::default()`.
/// This is never called in production — only used for route scaffolding.
pub fn new_stub() -> Self {
Self { pool: None }
}
/// Get the pool, panicking if stub.
fn pool(&self) -> &PgPool {
self.pool.as_deref().expect("FolderDbRepository: pool not available (stub instance)")
}
// ── helpers ──────────────────────────────────────────────────
/// Build the full virtual path for a folder by walking up the `parent_id` chain.
async fn build_folder_path(&self, folder_id: &str) -> Result<StoragePath, DomainError> {
// CTE-based recursive query to build path segments
let _rows = sqlx::query_as::<_, (String,)>(
r#"
WITH RECURSIVE ancestors AS (
SELECT id, name, parent_id
FROM storage.folders
WHERE id = $1::uuid
UNION ALL
SELECT f.id, f.name, f.parent_id
FROM storage.folders f
JOIN ancestors a ON f.id = a.parent_id
)
SELECT name FROM ancestors ORDER BY name
"#,
)
.bind(folder_id)
.fetch_all(self.pool())
.await
.map_err(|e| DomainError::internal_error("FolderDb", format!("path query: {e}")))?;
// Actually we need a proper ordering. Let me rewrite with depth tracking.
// Re-query with depth.
let rows = sqlx::query_as::<_, (String, i32)>(
r#"
WITH RECURSIVE ancestors AS (
SELECT id, name, parent_id, 0 AS depth
FROM storage.folders
WHERE id = $1::uuid
UNION ALL
SELECT f.id, f.name, f.parent_id, a.depth + 1
FROM storage.folders f
JOIN ancestors a ON f.id = a.parent_id
)
SELECT name, depth FROM ancestors ORDER BY depth DESC
"#,
)
.bind(folder_id)
.fetch_all(self.pool())
.await
.map_err(|e| DomainError::internal_error("FolderDb", format!("path query: {e}")))?;
let path_parts: Vec<&str> = rows.iter().map(|(name, _)| name.as_str()).collect();
let path_str = path_parts.join("/");
Ok(StoragePath::from_string(&path_str))
}
/// Convert a database row into a `Folder` domain entity.
async fn row_to_folder(
&self,
id: String,
name: String,
parent_id: Option<String>,
created_at: i64,
modified_at: i64,
) -> Result<Folder, DomainError> {
let storage_path = self.build_folder_path(&id).await?;
Folder::with_timestamps(
id,
name,
storage_path,
parent_id,
created_at as u64,
modified_at as u64,
)
.map_err(|e| DomainError::internal_error("FolderDb", format!("entity: {e}")))
}
}
#[async_trait]
impl FolderRepository for FolderDbRepository {
async fn create_folder(
&self,
name: String,
parent_id: Option<String>,
) -> Result<Folder, DomainError> {
// Derive user_id from parent folder. Root-level folders require the
// caller to have set up the home folder beforehand (done during user
// registration).
let user_id: String = if let Some(ref pid) = parent_id {
sqlx::query_scalar::<_, String>(
"SELECT user_id FROM storage.folders WHERE id = $1::uuid",
)
.bind(pid)
.fetch_optional(self.pool())
.await
.map_err(|e| DomainError::internal_error("FolderDb", format!("parent lookup: {e}")))?
.ok_or_else(|| DomainError::not_found("Folder", pid))?
} else {
return Err(DomainError::internal_error(
"FolderDb",
"Cannot create root folder without user_id — use create_home_folder instead",
));
};
let row = sqlx::query_as::<_, (String, i64, i64)>(
r#"
INSERT INTO storage.folders (name, parent_id, user_id)
VALUES ($1, $2::uuid, $3)
RETURNING id::text,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
"#,
)
.bind(&name)
.bind(&parent_id)
.bind(&user_id)
.fetch_one(self.pool())
.await
.map_err(|e| {
if let sqlx::Error::Database(ref db_err) = e {
if db_err.code().as_deref() == Some("23505") {
return DomainError::already_exists(
"Folder",
format!("{name} already exists in parent"),
);
}
}
DomainError::internal_error("FolderDb", format!("insert: {e}"))
})?;
self.row_to_folder(row.0, name, parent_id, row.1, row.2)
.await
}
async fn get_folder(&self, id: &str) -> Result<Folder, DomainError> {
let row = sqlx::query_as::<_, (String, String, Option<String>, i64, i64)>(
r#"
SELECT id::text, name, parent_id::text,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
FROM storage.folders
WHERE id = $1::uuid AND NOT is_trashed
"#,
)
.bind(id)
.fetch_optional(self.pool())
.await
.map_err(|e| DomainError::internal_error("FolderDb", format!("get: {e}")))?
.ok_or_else(|| DomainError::not_found("Folder", id))?;
self.row_to_folder(row.0, row.1, row.2, row.3, row.4)
.await
}
async fn get_folder_by_path(
&self,
storage_path: &StoragePath,
) -> Result<Folder, DomainError> {
// Walk the path segments to find the folder.
let path_str = storage_path.to_string();
let segments: Vec<&str> = path_str.split('/').filter(|s| !s.is_empty()).collect();
if segments.is_empty() {
return Err(DomainError::not_found("Folder", "empty path"));
}
let mut current_parent: Option<String> = None;
let mut current_id = String::new();
for segment in &segments {
let row = if let Some(ref pid) = current_parent {
sqlx::query_as::<_, (String,)>(
r#"
SELECT id::text FROM storage.folders
WHERE name = $1 AND parent_id = $2::uuid AND NOT is_trashed
"#,
)
.bind(segment)
.bind(pid)
.fetch_optional(self.pool())
.await
} else {
sqlx::query_as::<_, (String,)>(
r#"
SELECT id::text FROM storage.folders
WHERE name = $1 AND parent_id IS NULL AND NOT is_trashed
"#,
)
.bind(segment)
.fetch_optional(self.pool())
.await
}
.map_err(|e| DomainError::internal_error("FolderDb", format!("path walk: {e}")))?
.ok_or_else(|| {
DomainError::not_found("Folder", format!("segment '{segment}' in path"))
})?;
current_id = row.0;
current_parent = Some(current_id.clone());
}
self.get_folder(&current_id).await
}
async fn list_folders(
&self,
parent_id: Option<&str>,
) -> Result<Vec<Folder>, DomainError> {
let rows: Vec<(String, String, Option<String>, i64, i64)> = if let Some(pid) = parent_id {
sqlx::query_as(
r#"
SELECT id::text, name, parent_id::text,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
FROM storage.folders
WHERE parent_id = $1::uuid AND NOT is_trashed
ORDER BY name
"#,
)
.bind(pid)
.fetch_all(self.pool())
.await
} else {
sqlx::query_as(
r#"
SELECT id::text, name, parent_id::text,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
FROM storage.folders
WHERE parent_id IS NULL AND NOT is_trashed
ORDER BY name
"#,
)
.fetch_all(self.pool())
.await
}
.map_err(|e| DomainError::internal_error("FolderDb", format!("list: {e}")))?;
let mut folders = Vec::with_capacity(rows.len());
for (id, name, pid, ca, ma) in rows {
folders.push(self.row_to_folder(id, name, pid, ca, ma).await?);
}
Ok(folders)
}
async fn list_folders_paginated(
&self,
parent_id: Option<&str>,
offset: usize,
limit: usize,
include_total: bool,
) -> Result<(Vec<Folder>, Option<usize>), DomainError> {
let total = if include_total {
let count: i64 = if let Some(pid) = parent_id {
sqlx::query_scalar(
"SELECT COUNT(*) FROM storage.folders WHERE parent_id = $1::uuid AND NOT is_trashed",
)
.bind(pid)
.fetch_one(self.pool())
.await
} else {
sqlx::query_scalar(
"SELECT COUNT(*) FROM storage.folders WHERE parent_id IS NULL AND NOT is_trashed",
)
.fetch_one(self.pool())
.await
}
.map_err(|e| DomainError::internal_error("FolderDb", format!("count: {e}")))?;
Some(count as usize)
} else {
None
};
let rows: Vec<(String, String, Option<String>, i64, i64)> = if let Some(pid) = parent_id {
sqlx::query_as(
r#"
SELECT id::text, name, parent_id::text,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
FROM storage.folders
WHERE parent_id = $1::uuid AND NOT is_trashed
ORDER BY name
LIMIT $2 OFFSET $3
"#,
)
.bind(pid)
.bind(limit as i64)
.bind(offset as i64)
.fetch_all(self.pool())
.await
} else {
sqlx::query_as(
r#"
SELECT id::text, name, parent_id::text,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
FROM storage.folders
WHERE parent_id IS NULL AND NOT is_trashed
ORDER BY name
LIMIT $1 OFFSET $2
"#,
)
.bind(limit as i64)
.bind(offset as i64)
.fetch_all(self.pool())
.await
}
.map_err(|e| DomainError::internal_error("FolderDb", format!("paginate: {e}")))?;
let mut folders = Vec::with_capacity(rows.len());
for (id, name, pid, ca, ma) in rows {
folders.push(self.row_to_folder(id, name, pid, ca, ma).await?);
}
Ok((folders, total))
}
async fn rename_folder(
&self,
id: &str,
new_name: String,
) -> Result<Folder, DomainError> {
sqlx::query(
r#"
UPDATE storage.folders
SET name = $1, updated_at = NOW()
WHERE id = $2::uuid AND NOT is_trashed
"#,
)
.bind(&new_name)
.bind(id)
.execute(self.pool())
.await
.map_err(|e| {
if let sqlx::Error::Database(ref db_err) = e {
if db_err.code().as_deref() == Some("23505") {
return DomainError::already_exists(
"Folder",
format!("{new_name} already exists"),
);
}
}
DomainError::internal_error("FolderDb", format!("rename: {e}"))
})?;
self.get_folder(id).await
}
async fn move_folder(
&self,
id: &str,
new_parent_id: Option<&str>,
) -> Result<Folder, DomainError> {
sqlx::query(
r#"
UPDATE storage.folders
SET parent_id = $1::uuid, updated_at = NOW()
WHERE id = $2::uuid AND NOT is_trashed
"#,
)
.bind(new_parent_id)
.bind(id)
.execute(self.pool())
.await
.map_err(|e| DomainError::internal_error("FolderDb", format!("move: {e}")))?;
self.get_folder(id).await
}
async fn delete_folder(&self, id: &str) -> Result<(), DomainError> {
// Hard delete folder and all descendants (CASCADE handles children)
let result = sqlx::query("DELETE FROM storage.folders WHERE id = $1::uuid")
.bind(id)
.execute(self.pool())
.await
.map_err(|e| DomainError::internal_error("FolderDb", format!("delete: {e}")))?;
if result.rows_affected() == 0 {
return Err(DomainError::not_found("Folder", id));
}
Ok(())
}
async fn folder_exists(&self, storage_path: &StoragePath) -> Result<bool, DomainError> {
// Try to find by walking the path
match self.get_folder_by_path(storage_path).await {
Ok(_) => Ok(true),
Err(e) if e.to_string().contains("not found") => Ok(false),
Err(e) => Err(e),
}
}
async fn get_folder_path(&self, id: &str) -> Result<StoragePath, DomainError> {
self.build_folder_path(id).await
}
// ── Trash operations ──
async fn move_to_trash(&self, folder_id: &str) -> Result<(), DomainError> {
// Soft-delete: set is_trashed = true and remember original parent
let result = sqlx::query(
r#"
UPDATE storage.folders
SET is_trashed = TRUE,
trashed_at = NOW(),
original_parent_id = parent_id,
updated_at = NOW()
WHERE id = $1::uuid AND NOT is_trashed
"#,
)
.bind(folder_id)
.execute(self.pool())
.await
.map_err(|e| DomainError::internal_error("FolderDb", format!("trash: {e}")))?;
if result.rows_affected() == 0 {
return Err(DomainError::not_found("Folder", folder_id));
}
// Also trash all files inside the folder (recursively)
sqlx::query(
r#"
WITH RECURSIVE descendants AS (
SELECT id FROM storage.folders WHERE id = $1::uuid
UNION ALL
SELECT f.id FROM storage.folders f JOIN descendants d ON f.parent_id = d.id
)
UPDATE storage.files
SET is_trashed = TRUE, trashed_at = NOW(), original_folder_id = folder_id
WHERE folder_id IN (SELECT id FROM descendants) AND NOT is_trashed
"#,
)
.bind(folder_id)
.execute(self.pool())
.await
.map_err(|e| DomainError::internal_error("FolderDb", format!("trash files: {e}")))?;
Ok(())
}
async fn restore_from_trash(
&self,
folder_id: &str,
_original_path: &str,
) -> Result<(), DomainError> {
// Restore: set is_trashed = false, restore parent_id from original_parent_id
let result = sqlx::query(
r#"
UPDATE storage.folders
SET is_trashed = FALSE,
trashed_at = NULL,
parent_id = COALESCE(original_parent_id, parent_id),
original_parent_id = NULL,
updated_at = NOW()
WHERE id = $1::uuid AND is_trashed
"#,
)
.bind(folder_id)
.execute(self.pool())
.await
.map_err(|e| DomainError::internal_error("FolderDb", format!("restore: {e}")))?;
if result.rows_affected() == 0 {
return Err(DomainError::not_found("Folder", folder_id));
}
// Also restore files that were trashed with this folder
sqlx::query(
r#"
WITH RECURSIVE descendants AS (
SELECT id FROM storage.folders WHERE id = $1::uuid
UNION ALL
SELECT f.id FROM storage.folders f JOIN descendants d ON f.parent_id = d.id
)
UPDATE storage.files
SET is_trashed = FALSE,
trashed_at = NULL,
folder_id = COALESCE(original_folder_id, folder_id),
original_folder_id = NULL
WHERE folder_id IN (SELECT id FROM descendants) AND is_trashed
"#,
)
.bind(folder_id)
.execute(self.pool())
.await
.map_err(|e| DomainError::internal_error("FolderDb", format!("restore files: {e}")))?;
Ok(())
}
async fn delete_folder_permanently(&self, folder_id: &str) -> Result<(), DomainError> {
// Permanently delete — CASCADE handles children
let result = sqlx::query("DELETE FROM storage.folders WHERE id = $1::uuid")
.bind(folder_id)
.execute(self.pool())
.await
.map_err(|e| DomainError::internal_error("FolderDb", format!("perm delete: {e}")))?;
if result.rows_affected() == 0 {
return Err(DomainError::not_found("Folder", folder_id));
}
Ok(())
}
}
// ── Extra helpers for blob-storage bootstrap ──
impl FolderDbRepository {
/// Creates a root-level home folder for a user.
/// This is called during user registration.
pub async fn create_home_folder(
&self,
user_id: &str,
name: &str,
) -> Result<Folder, DomainError> {
let row = sqlx::query_as::<_, (String, i64, i64)>(
r#"
INSERT INTO storage.folders (name, parent_id, user_id)
VALUES ($1, NULL, $2)
ON CONFLICT DO NOTHING
RETURNING id::text,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
"#,
)
.bind(name)
.bind(user_id)
.fetch_optional(self.pool())
.await
.map_err(|e| DomainError::internal_error("FolderDb", format!("home folder: {e}")))?;
match row {
Some((id, ca, ma)) => {
self.row_to_folder(id, name.to_string(), None, ca, ma).await
}
None => {
// Already exists — fetch it
let existing = sqlx::query_as::<_, (String, i64, i64)>(
r#"
SELECT id::text,
EXTRACT(EPOCH FROM created_at)::bigint,
EXTRACT(EPOCH FROM updated_at)::bigint
FROM storage.folders
WHERE name = $1 AND user_id = $2 AND parent_id IS NULL
"#,
)
.bind(name)
.bind(user_id)
.fetch_one(self.pool())
.await
.map_err(|e| {
DomainError::internal_error("FolderDb", format!("home fetch: {e}"))
})?;
self.row_to_folder(existing.0, name.to_string(), None, existing.1, existing.2)
.await
}
}
}
/// Returns user_id for a given folder. Used by file repositories.
pub async fn get_folder_user_id(&self, folder_id: &str) -> Result<String, DomainError> {
sqlx::query_scalar::<_, String>(
"SELECT user_id FROM storage.folders WHERE id = $1::uuid",
)
.bind(folder_id)
.fetch_optional(self.pool())
.await
.map_err(|e| DomainError::internal_error("FolderDb", format!("user_id lookup: {e}")))?
.ok_or_else(|| DomainError::not_found("Folder", folder_id))
}
}
+10
View File
@@ -11,6 +11,12 @@ mod settings_pg_repository;
mod transaction_utils;
mod user_pg_repository;
// ── Blob-storage repositories ──
pub mod file_blob_read_repository;
pub mod file_blob_write_repository;
pub mod folder_db_repository;
pub mod trash_db_repository;
pub use address_book_pg_repository::AddressBookPgRepository;
pub use calendar_event_pg_repository::CalendarEventPgRepository;
pub use calendar_pg_repository::CalendarPgRepository;
@@ -18,7 +24,11 @@ pub use contact_group_pg_repository::ContactGroupPgRepository;
pub use contact_persistence_dto::*;
pub use contact_pg_repository::ContactPgRepository;
pub use favorites_pg_repository::FavoritesPgRepository;
pub use file_blob_read_repository::FileBlobReadRepository;
pub use file_blob_write_repository::FileBlobWriteRepository;
pub use folder_db_repository::FolderDbRepository;
pub use recent_items_pg_repository::RecentItemsPgRepository;
pub use session_pg_repository::SessionPgRepository;
pub use settings_pg_repository::SettingsPgRepository;
pub use trash_db_repository::TrashDbRepository;
pub use user_pg_repository::UserPgRepository;
@@ -0,0 +1,191 @@
//! PostgreSQL-backed trash repository.
//!
//! Implements `TrashRepository` using soft-delete columns in `storage.files`
//! and `storage.folders`. There is no separate trash table — trashed items
//! are files/folders with `is_trashed = TRUE`.
use async_trait::async_trait;
use chrono::{DateTime, Utc};
use sqlx::PgPool;
use std::sync::Arc;
use uuid::Uuid;
use crate::common::errors::{DomainError, Result};
use crate::domain::entities::trashed_item::{TrashedItem, TrashedItemType};
use crate::domain::repositories::trash_repository::TrashRepository;
/// Default retention period (days) used when computing deletion_date.
const _DEFAULT_RETENTION_DAYS: i64 = 30;
/// PostgreSQL-backed trash repository using soft-delete flags.
pub struct TrashDbRepository {
pool: Arc<PgPool>,
retention_days: i64,
}
impl TrashDbRepository {
pub fn new(pool: Arc<PgPool>, retention_days: u32) -> Self {
Self {
pool,
retention_days: retention_days as i64,
}
}
/// Convert a trash_items view row into a TrashedItem entity.
fn row_to_trashed_item(
&self,
id: Uuid,
name: String,
item_type: String,
user_id: String,
trashed_at: Option<DateTime<Utc>>,
) -> TrashedItem {
let trashed_at = trashed_at.unwrap_or_else(Utc::now);
let deletion_date = trashed_at + chrono::Duration::days(self.retention_days);
let item_type_enum = match item_type.as_str() {
"folder" => TrashedItemType::Folder,
_ => TrashedItemType::File,
};
let user_uuid = Uuid::parse_str(&user_id).unwrap_or_else(|_| Uuid::nil());
// In the soft-delete model, the trash entry ID is the same as the
// original item ID since there is no separate trash table.
TrashedItem::from_raw(
id, // trash entry id (same as original)
id, // original item id
user_uuid, // owner
item_type_enum,
name.clone(),
String::new(), // original_path — not stored separately in soft-delete model
trashed_at,
deletion_date,
)
}
}
#[async_trait]
impl TrashRepository for TrashDbRepository {
async fn add_to_trash(&self, _item: &TrashedItem) -> Result<()> {
// No-op: the actual flagging is done by FileWritePort::move_to_trash
// or FolderRepository::move_to_trash. This method exists for interface
// compatibility with the TrashService.
Ok(())
}
async fn get_trash_items(&self, user_id: &Uuid) -> Result<Vec<TrashedItem>> {
let rows = sqlx::query_as::<_, (Uuid, String, String, String, Option<DateTime<Utc>>)>(
r#"
SELECT id, name, item_type, user_id, trashed_at
FROM storage.trash_items
WHERE user_id = $1
ORDER BY trashed_at DESC
"#,
)
.bind(user_id.to_string())
.fetch_all(self.pool.as_ref())
.await
.map_err(|e| {
DomainError::internal_error("TrashDb", format!("list: {e}"))
})?;
Ok(rows
.into_iter()
.map(|(id, name, item_type, uid, trashed_at)| {
self.row_to_trashed_item(id, name, item_type, uid, trashed_at)
})
.collect())
}
async fn get_trash_item(
&self,
id: &Uuid,
user_id: &Uuid,
) -> Result<Option<TrashedItem>> {
let row = sqlx::query_as::<_, (Uuid, String, String, String, Option<DateTime<Utc>>)>(
r#"
SELECT id, name, item_type, user_id, trashed_at
FROM storage.trash_items
WHERE id = $1 AND user_id = $2
"#,
)
.bind(id)
.bind(user_id.to_string())
.fetch_optional(self.pool.as_ref())
.await
.map_err(|e| {
DomainError::internal_error("TrashDb", format!("get: {e}"))
})?;
Ok(row.map(|(id, name, item_type, uid, trashed_at)| {
self.row_to_trashed_item(id, name, item_type, uid, trashed_at)
}))
}
async fn restore_from_trash(&self, _id: &Uuid, _user_id: &Uuid) -> Result<()> {
// No-op: the actual restore is done by FileWritePort::restore_from_trash
// or FolderRepository::restore_from_trash. The TrashService also removes
// the index entry — which in the soft-delete model means the flag is
// already cleared.
Ok(())
}
async fn delete_permanently(&self, _id: &Uuid, _user_id: &Uuid) -> Result<()> {
// No-op: the actual delete is done by FileWritePort::delete_file_permanently
// or FolderRepository::delete_folder_permanently.
Ok(())
}
async fn clear_trash(&self, user_id: &Uuid) -> Result<()> {
// Delete all trashed files for this user
sqlx::query(
"DELETE FROM storage.files WHERE user_id = $1 AND is_trashed = TRUE",
)
.bind(user_id.to_string())
.execute(self.pool.as_ref())
.await
.map_err(|e| {
DomainError::internal_error("TrashDb", format!("clear files: {e}"))
})?;
// Delete all trashed folders for this user
sqlx::query(
"DELETE FROM storage.folders WHERE user_id = $1 AND is_trashed = TRUE",
)
.bind(user_id.to_string())
.execute(self.pool.as_ref())
.await
.map_err(|e| {
DomainError::internal_error("TrashDb", format!("clear folders: {e}"))
})?;
Ok(())
}
async fn get_expired_items(&self) -> Result<Vec<TrashedItem>> {
let cutoff = Utc::now() - chrono::Duration::days(self.retention_days);
let rows = sqlx::query_as::<_, (Uuid, String, String, String, Option<DateTime<Utc>>)>(
r#"
SELECT id, name, item_type, user_id, trashed_at
FROM storage.trash_items
WHERE trashed_at < $1
ORDER BY trashed_at ASC
"#,
)
.bind(cutoff)
.fetch_all(self.pool.as_ref())
.await
.map_err(|e| {
DomainError::internal_error("TrashDb", format!("expired: {e}"))
})?;
Ok(rows
.into_iter()
.map(|(id, name, item_type, uid, trashed_at)| {
self.row_to_trashed_item(id, name, item_type, uid, trashed_at)
})
.collect())
}
}