file_database

file_database is a local filesystem database for files and folders.

It manages one root directory and keeps an in-memory index so you can target items by ID instead of raw paths.

What problem this solves

It reduces the code needed for managing a file system by keeping track of everything for you

Install

Cargo.toml:

[dependencies]
file_database = "1.1.1"

Quick start

use file_database::{DatabaseError, DatabaseManager, ItemId};

fn main() -> Result<(), DatabaseError> {
  let mut db = DatabaseManager::create_database(".", "database")?;

    db.write_new(ItemId::id("notes.txt"), ItemId::database_id())?;
    db.overwrite_existing(ItemId::id("notes.txt"), b"hello world")?;

    let bytes = db.read_existing(ItemId::id("notes.txt"))?;
    assert_eq!(bytes, b"hello world");

    Ok(())
}

ItemId

An ItemId has two parts:

• name: shared key (example: "test_file.txt")
• index: stable slot under that shared key

This means duplicate names are allowed. ItemId::id("name") always means index 0. Use ItemId::with_index("name", i) for a specific slot. When IDs are auto-generated (for example when opening an existing directory), file names keep the extension because indexing uses file_name().

ItemId::database_id() is the root ID that refers to the database root directory.

Create and organize

create_database works as create-or-open: if the directory does not exist, it creates it

if the directory already exists, it opens it and indexes current contents recursively

• write_new(id, parent)
• rename(id, new_name)
• migrate_item(id, to_parent)
• duplicate_item(id, to_parent, new_name)
• delete(id, force)

Locate and list

• locate_absolute(id)
• locate_relative(id)
• get_all(sorted)
• get_by_parent(parent, sorted)
• get_parent(id)
• get_ids_by_name(name)
• get_ids_by_index(index)

Read and write file data

• Raw bytes:
  • overwrite_existing(id, data)
  • read_existing(id)
• JSON:
  • overwrite_existing_json(id, &value, pretty)
  • read_existing_json::<T>(id)
• Binary (bincode):
  • overwrite_existing_binary(id, &value)
  • read_existing_binary::<T>(id)
• Streaming overwrite:
  • overwrite_existing_from_reader(id, &mut reader)

Move across database boundaries

• import_item(from_external_path, to_database_parent)
• export_item(id, to_external_directory, mode) where mode is ExportMode::Copy or ExportMode::Move
• migrate_database(new_parent_dir)

Metadata

• get_file_information(id) returns FileInformation with:
  • name and extension
  • normalized size (FileSize)
  • unix timestamps and time_since_* values when available

Scan for external changes

If files are changed outside this library (for example, another tool drops files into the database), use:

• scan_for_changes(scan_from, policy, recursive)

Policy options:

• ScanPolicy::DetectOnly: detect new files, do not index them
• ScanPolicy::AddNew: detect and index new files
• ScanPolicy::RemoveNew: delete new files from disk and do not keep them in the added list

Important behavior: missing tracked items are always removed from the in-memory index during scan.

The result is ScanReport with:

• scanned scope (scan_from)
• added
• removed
• unchanged_count
• total_changed_count

GenPath

GenPath helps build base paths for database setup:

• GenPath::from_working_dir(steps)
• GenPath::from_exe(steps)
• GenPath::from_closest_match("dir_name")

Example:

use file_database::{DatabaseError, DatabaseManager, GenPath};

fn main() -> Result<(), DatabaseError> {
    let root = GenPath::from_working_dir(0)?;
    let _db = DatabaseManager::create_database(root, "database")?;
    Ok(())
}

Errors

All fallible functions return Result<_, DatabaseError>.

Common variants include:

• NoMatchingID
• NotADirectory
• NotAFile
• IdAlreadyExists
• RootIdUnsupported
• Io
• SerdeJson
• Bincode

Notes on indexing behavior

• Internal storage is HashMap<String, StableVec<PathBuf>>.
• index is a stable slot in the per-name StableVec, not a shifting position in a plain Vec.
• Different ItemId values can share the same name and still point to different paths.
• If one item is removed, other occupied slots keep their index.
• If you need all IDs for one shared name, call get_ids_by_name.

License

This crate is licensed under the MIT license

Feel free to create an issue in the repo if anything isn't working correctly

Repo

file_database