feat: rewrite indexing to local storage, improve UX

README.md

@@ -12,17 +12,14 @@ Early alpha. Works on my machines.
 
 Python 3.14 is required. You can get it using [uv](https://github.com/astral-sh/uv).
 
-Also you will need one of backup backends.
-[Restic](https://github.com/restic/restic) and [Rustic](https://github.com/rustic-rs/rustic) are supported.
+Also you will need [Rustic](https://github.com/rustic-rs/rustic) as a backup backend. v0.11.4 is fully compatible.
 
 ```bash
-apt update && apt install -y restic
-# or download rustic binary from https://github.com/rustic-rs/rustic/releases/tag/v0.11.3
+pacman -S rustic
+# or get the binary binary from https://github.com/rustic-rs/rustic/releases/tag/v0.11.4
 
-# get using uv
+# and install lohup using uv
 uv tool install git+https://github.com/kam1sh/lohup
-# or by pip
-pip install 
 ```
 
 ## Usage
@@ -41,8 +38,8 @@ lohup snapshots --repo cloud
 ## Core features
 
 * Configuration in TOML
-* Restic/Rustic as a backup driver
-* Backup hooks, such as create/remove btrfs filesystem snapshot
+* Local database for snapshot information
+* Backup hooks, such as create/remove btrfs snapshot before backup
 * Environment variables
 
 ### Features in TODO
@@ -53,6 +50,7 @@ lohup snapshots --repo cloud
 * Metrics and OpenTelemetry support
 * Binary releases, win+linux amd64 along with Python wheel
 * Include files in profiles with `paths-from` 
+* Pluggable engines
 
 ## Development
 
@@ -66,4 +64,4 @@ uv run pytest --tb=short
 uv run pyinstaller -F lohup-cli.spec
 ```
 
-[1] in my native language loh (лох) means "dummy" or "loser". So, yeah, backups are for losers.
+[1] in Russian language loh (лох) means "dummy" or "loser". So, yeah, backups are for losers.

src/lohup/app.py

@@ -1,44 +1,41 @@
-from lohup.engine import Snapshot
-from lohup.storage.storage import Storage
-from lohup.config import LocalRepository, S3Repository
+import hashlib
 import subprocess as procs
+from collections import defaultdict
 
 from lohup import config as tomlconf
+from lohup.config import LocalRepository, S3Repository
+from lohup.engine import Snapshot, SnapshotInfo
 from lohup.logger import BasicLogger, LoggerProto, LogLevel
-from lohup.restic import Restic
 from lohup.rustic import Rustic
+from lohup.storage import refresher
+from lohup.storage.file import SnapshotFile
+from lohup.storage.snapshot import SnapshotRecord
+from lohup.storage.storage import Storage
 
 
 class Lohup:
-    subsystem: str | None
     _config: tomlconf.TomlConfig | None
+    _storage: Storage | None
 
-    def __init__(self, config_path: str | None, logger: LoggerProto | None = None, cache: bool = False):
+    def __init__(self, config_path: str | None, logger: LoggerProto | None = None):
         self._config_path = config_path or "lohup.toml"
         self._config = None
-        self.subsystem = None
-        self._cache = cache
+        self._storage = None
         self.log = logger or BasicLogger(level=LogLevel.INFO)
 
     def load(self):
         self._config = tomlconf.TomlConfig.from_file(self._config_path)
-        self.subsystem = self.config.settings.subsystem
-        if self.subsystem not in ("restic", "rustic"):
-            raise KeyError(f"Invalid subsystem: {self.subsystem}")
-        if self._cache:
-            pth = self.config.settings.cache_dir
-            if not pth.exists():
-                pth.mkdir(parents=True)
-                pth.joinpath("CACHEDIR.TAG").write_text(
-                    "Signature: 8a477f597d28d172789f06886806bc55"
-                )
-            self._cache = Storage.from_file(pth / "db.sqlite3")
+        pth = self.config.settings.data_dir
+        if not pth.exists():
+            pth.mkdir(parents=True)
+        name = f"{self._checksum()}.db"
+        self._storage = Storage.from_file(pth / name)
 
-    def invoke_direct(self, repo: str | None, args: tuple[str, ...]):
+    def invoke_direct(self, repo: str | None, args: list[str]):
         spec = self._repo_by_name(repo)
         engine = self._engine_for(spec)
         with engine:
-            engine.run(args)
+            engine.call(args)
 
     def _exechooks(self, hooks: list):
         for h in hooks:
@@ -77,19 +74,12 @@ class Lohup:
             raise KeyError(f"No repo attached to profile: {profile.name!r}")
         return default
 
-    def _engine_for(self, repo: tomlconf.Repository) -> Restic | Rustic:
-        match self.subsystem:
-            case "rustic":
-                return Rustic(
-                    repo,
-                    log=self.log,
-                    conf_dir=self.config.settings.build_dir,
-                    storage=self.storage,
-                )
-            case "restic":
-                return Restic(repo, log=self.log)
-            case x:
-                raise KeyError(f"Unknown subsystem: {x}")
+    def _engine_for(self, repo: tomlconf.Repository) -> Rustic:
+        return Rustic(
+            repo,
+            log=self.log,
+            conf_dir=self.config.settings.build_dir,
+        )
 
     def backup(self, profile: str):
         spec = self._profile_for(profile)
@@ -115,48 +105,116 @@ class Lohup:
             self._exechooks(self.config.hooks.after_all)
         self.log.info("Finished!")
 
-    def snapshots(self, repo: str | None, profile: str | None = None):
+    def snapshots(
+        self, repo: str | None, profile: str | None = None
+    ) -> dict[str, list[Snapshot]]:
         spec = self._repo_by_name(repo)
-        engine = self._engine_for(spec)
-        with engine:
-            result = engine.snapshots(profile=profile)
-        if cache := self.storage:
-            total = []
-            existing = {x.id for x in total}
-            for arr in result.values():
-                total.extend(arr)
-            with cache.session() as session:
-                print(len(Snapshot.findall(session)))
-                response = Snapshot.update_all(session, repo=spec.name, items=total)
-                session.commit()
-            if not response["untouched"]:
-                return result
-            to_delete = existing.difference(response["untouched"])
-            if not to_delete:
-                return result
-            print(response)
-            with cache.session() as session:
-                Snapshot.delete_all(session, repo=spec.name, ids=list(to_delete), profile=profile)
+        if self.storage.is_empty():
+            raise ValueError(
+                "Local database is empty. Update it using 'fetch' subcommand."
+            )
+        with self.storage.session() as session:
+            records = SnapshotRecord.all_by_repository(session, spec.name)
+            if profile is not None:
+                records = filter(lambda r: r.profile == profile, records)
+            items = [Snapshot.from_record(x) for x in records]
+        result = defaultdict(list)
+        for item in items:
+            result[item.profile].append(item)
+        for name in result:
+            result[name].sort(key=lambda x: x.time)
         return result
 
-    def profile_info(self, repo: str | None, profile: str, offset: int):
+    def profile_info(
+        self, repo: str | None, profile: str, offset: int, depth: int
+    ) -> Snapshot:
         spec = self._repo_by_name(repo)
-        engine = self._engine_for(spec)
-        with engine:
-            all_snapshots = engine.snapshots(profile=profile)
-            if not isinstance(all_snapshots, dict):
-                raise ValueError("internal error - unknown snapshots format")
-            snapshots = all_snapshots[profile]
+        with self.storage.session() as session:
+            snapshots = SnapshotRecord.all_by_profile(
+                session, repo=spec.name, profile=profile
+            )
+            if offset >= len(snapshots):
+                raise IndexError(
+                    f"Requested offset {offset} exceeds snapshots count ({len(snapshots)})"
+                )
             snapshot = snapshots[offset]
-            info = engine.info(snapshot.id)
-            return info
+            items = SnapshotFile.find_all(
+                session, repo=spec, snapshot=snapshot, depth=depth
+            )
+            info = SnapshotInfo.from_db(items)
+            result = Snapshot.from_record(snapshot)
+        if info is None:
+            raise KeyError("Internal error - snapshot has no files.")
+        result.info = info
+        return result
 
-    def snapshot_info(self, repo: str, ref: str):
+    def snapshot_info(self, repo: str, ref: str, depth: int) -> Snapshot:
         spec = self._repo_by_name(repo)
-        engine = self._engine_for(spec)
-        with engine:
-            info = engine.info(ref)
-            return info
+        with self.storage.session() as session:
+            snapshot = SnapshotRecord.find_by_key(session, repo, key=ref)
+            if snapshot is None:
+                raise KeyError("No such snapshot.")
+            files = SnapshotFile.find_all(
+                session, repo=spec, snapshot=snapshot, depth=depth
+            )
+            info = SnapshotInfo.from_db(files)
+            result = Snapshot.from_record(snapshot)
+        if info is None:
+            raise KeyError("Internal error - snapshot has no files.")
+        result.info = info
+        return result
+
+    def refresh(self, clean: bool, force: bool = False):
+        if clean:
+            self.storage.recreate()
+        first = list(self.config.repos.values())[0]
+        rustic = self._engine_for(first)
+        ref = refresher.Refresher(storage=self.storage, config=self.config, force=force)
+        repos = ref.update_repos(self.config.repos)
+        # repo name -> json from rustic
+        infos = {}
+        for name in repos.added:
+            rustic.repo = self.config.repos[name]
+            with rustic:
+                data = rustic.repoinfo()
+            if data is None:
+                repos.added.remove(name)
+                repos.message(f"Repository {name} does not exist, skipping")
+                continue
+            infos[name] = data
+        ref.add_repos(infos)
+        pulled = []
+        for repo in self.config.repos.values():
+            rustic.repo = repo
+            with rustic:
+                profiles = rustic.snapshots()
+            for snapshots in profiles.values():
+                pulled.extend(snapshots)
+        snapshots_result = ref.update_snapshots(pulled)
+        to_check = set(snapshots_result.added)
+        to_check.update(snapshots_result.updated)
+        by_repo = defaultdict(list)
+        for key in to_check:
+            snapshot = snapshots_result.items[key]
+            by_repo[snapshot.repo].append(snapshot)
+        # key id -> record
+        snapshots: dict[str, SnapshotRecord] = {}
+        with self.storage.session() as session:
+            for repo in by_repo:
+                for row in SnapshotRecord.all_by_repository(session, repo):
+                    snapshots[row.key] = row
+        # snapshot id -> snapshot info
+        for key in to_check:
+            item = snapshots_result.items[key]
+            snapshot = snapshots[key]
+            rustic.repo = self.config.repos[item.repo]
+            with rustic:
+                self.log.info(
+                    f"Fetching file tree of snapshot {key} (repo={item.repo})"
+                )
+                info = rustic.info(key)
+            ref.add_files(info=info, snapshot_id=snapshot.id)
+        return repos, snapshots_result
 
     def _invoke_profile(self, restic, profile: tomlconf.Profile):
         with restic as engine:
@@ -186,9 +244,12 @@ class Lohup:
         return self._config
 
     @property
-    def storage(self) -> Storage | None:
-        if self._cache is None:
+    def storage(self) -> Storage:
+        if not self._storage:
             raise ValueError("Storage is not available.")
-        if isinstance(self._cache, bool):
-            return None
-        return self._cache
+        return self._storage
+
+    def _checksum(self):
+        key = str(self.config.path.absolute())
+        hashsum = hashlib.sha3_256(key.encode(), usedforsecurity=False)
+        return hashsum.hexdigest()[:12]

src/lohup/cli.py

@@ -1,5 +1,5 @@
 import re
-from typing import no_type_check
+from typing import no_type_check, cast
 
 import click
 import humanize
@@ -11,11 +11,10 @@ from lohup.config import ConfigError
 
 @click.group()
 @click.option("--config", envvar="LOHUP_CONFIG", default="lohup.toml")
-@click.option("--cache", type=bool, default=True)
 @click.pass_context
-def cli(ctx, config, cache):
+def cli(ctx, config):
     log = logger.CliLogger(level=logger.LogLevel.DEBUG)
-    ctx.obj = Lohup(config_path=config, logger=log, cache=cache)
+    ctx.obj = Lohup(config_path=config, logger=log)
     try:
         ctx.obj.load()
     except ConfigError as e:
@@ -31,7 +30,7 @@ def restic(obj: Lohup, repo: str | None, args: tuple[str, ...]):
     """
     Pass command to restic
     """
-    obj.invoke_direct(repo, args)
+    obj.invoke_direct(repo, list(args))
 
 
 @cli.command()
@@ -66,24 +65,9 @@ def snapshots(obj: Lohup, repo: str | None, profile: str | None):
         items = raw_items.copy()
         items.sort(key=lambda x: x.time, reverse=True)
         for snap in items:
-            click.echo(f"\tSnapshot {snap.id}:")
-            started = humanize.naturaltime(snap.time)
-            text = click.style(started, fg="blue")
-            took = humanize.naturaldelta(snap.duration)
-            click.echo(f"\t\tStarted: {text} (took {took})")
-            ratio = snap.changes_ratio * 100
-            ratio = click.style(f"{ratio:.1f}%", fg="blue")
-            click.echo(
-                f"\t\tFiles changed (since previous): {snap.total_changes} ({ratio})"
-            )
-            uncompressed = humanize.naturalsize(snap.size, binary=True)
-            compressed = humanize.naturalsize(snap.size_compressed, binary=True)
-            total_bytes = snap.bytes_read
-            ratio = snap.size / total_bytes if total_bytes else 0
-            ratio = click.style(f"{ratio:.1f}%", fg="blue")
-            click.echo(f"\t\tDiff size: {compressed=!s}, {uncompressed=!s} ({ratio})")
-            click.echo(f"\t\tHost: {snap.hostname}")
-            click.echo()
+            lines = _write_snapshot(snap)
+            click.echo(f"\tSnapshot {snap.id}:\n\t\t", nl=False)
+            click.echo("\n\t\t".join(lines))
 
 
 @cli.command()
@@ -110,14 +94,50 @@ def info(obj: Lohup, repo: str | None, depth: int, ref: str):
         offset = 0
         if (value := match.group(2)) is not None:
             offset = int(value.removeprefix("^"))
-        info = obj.profile_info(repo, profile=profile, offset=offset)
+        snapshot = obj.profile_info(repo, profile=profile, offset=offset, depth=depth)
     elif repo:
-        info = obj.snapshot_info(repo=repo, ref=ref)
+        snapshot = obj.snapshot_info(repo=repo, ref=ref, depth=depth)
     else:
         raise click.ClickException("Requested info by ID without repository name.")
+    info = cast(engine.SnapshotInfo, snapshot.info)
     info.build_tree(depth=depth)
+    lines = _write_snapshot(snapshot)
+    click.echo(f"Snapshot {snapshot.id}")
+    click.echo("\n".join(lines))
     for line in info.show(depth=depth):
         click.echo(line)
 
+@cli.command()
+@click.option("--clean", is_flag=True, help="Refresh database from scratch")
+@click.option("--force", is_flag=True, help="Ignore items TTL")
+@click.pass_obj
+def update(obj: Lohup, clean: bool, force: bool):
+    repos, snapshots = obj.refresh(clean=clean, force=force)
+    click.echo("Repository updates:")
+    click.echo(repos.text())
+    click.echo("Snapshot updates:")
+    click.echo(snapshots.text())
+
+
+def _write_snapshot(snap: engine.Snapshot):
+    lines = []
+    started = click.style(humanize.naturaltime(snap.time), fg="blue")
+    took = humanize.naturaldelta(snap.duration)
+    lines.append(f"Started: {started} (took {took})")
+    ratio = snap.changes_ratio * 100
+    ratio = click.style(f"{ratio:.1f}%", fg="blue")
+    lines.append(
+        f"Files changed (since previous): {snap.total_changes} ({ratio})"
+    )
+    uncompressed = humanize.naturalsize(snap.size, binary=True)
+    compressed = humanize.naturalsize(snap.size_compressed, binary=True)
+    total = humanize.naturalsize(snap.bytes_read)
+    ratio = snap.size / snap.bytes_read if snap.bytes_read else 0
+    ratio = click.style(f"{ratio:.1f}%", fg="blue")
+    lines.append(f"Size: {total}")
+    lines.append(f"Diff size: {compressed=!s}, {uncompressed=!s} ({ratio})")
+    lines.append(f"Host: {snap.hostname}")
+    return lines
+
 if __name__ == "__main__":
     cli()

src/lohup/config.py

@@ -34,8 +34,7 @@ class Settings:
     backup_base_dir: Path
     globalvars: dict[str, str]
     build_dir: Path
-    subsystem: str
-    cache_dir: Path
+    data_dir: Path
 
     @staticmethod
     def load(conf: dict, validator: DeepValidator):
@@ -46,7 +45,7 @@ class Settings:
                     f"backup-base-dir: got ({pth!r}): variables are not allowed"
                 )
             pth = Path(pth)
-            if not pth.is_absolute:
+            if not pth.is_absolute():
                 validator.error(
                     f"backup-base-dir: got ({pth!r}): only absolute paths are supported"
                 )
@@ -57,14 +56,13 @@ class Settings:
                 validator.error(f"tmp-dir: got {tmpdir!r}: variables are not allowed")
             tmp_path = Path(tmpdir)
         elif platform.system() == "Windows":
-            tmp_path = Path.home().joinpath("AppData", "Local", "Temp", "lohup")
-        cache_path = Path.home().joinpath(".cache", "lohup")
+            tmp_path = Path.home().joinpath("AppData", "Local", "lohup")
+        data_dir = Path.home().joinpath(".local", "share", "lohup")
         settings = Settings(
             backup_base_dir=basepath,
             globalvars=conf.get("globals") or {},
             build_dir=tmp_path,
-            subsystem=conf.get("subsystem-name", "restic"),
-            cache_dir=cache_path
+            data_dir=data_dir
         )
         for key, value in settings.globalvars.items():
             with validator.inner(f"variable {key!r}") as inner:
@@ -260,6 +258,7 @@ class TomlConfig:
     hooks: HookSet
     templater: ValidatedTemplater
     profiles: dict[str, Profile]
+    path: Path
 
     @staticmethod
     def from_file(name: str):
@@ -349,5 +348,6 @@ class TomlConfig:
             hooks=hooks,
             templater=templater,
             profiles=profiles,
+            path=path
         )
         return toml

src/lohup/engine.py

@@ -1,37 +1,44 @@
-from lohup.storage.snapshot import SnapshotRecord
-from sqlalchemy.orm import Session
-import humanize.filesize
-from typing_extensions import cast
+import json
+import typing as ty
 from collections import defaultdict
-from lohup.storage.file import SnapshotFile
-from sqlalchemy import select, delete
-from lohup.storage.storage import Storage
-from pathlib import PosixPath
 from dataclasses import dataclass, field
 from datetime import datetime
-import typing as ty
+from pathlib import PosixPath, Path
+
+import humanize.filesize
+from sqlalchemy import delete, select
+from sqlalchemy.orm import Session
+from typing_extensions import cast
+
+from lohup.storage.file import SnapshotFile
+from lohup.storage.snapshot import SnapshotRecord
 
 
 @dataclass
 class Snapshot:
     id: str
+    repo: str
     profile: str
     tags: list[str]
     hostname: str
     time: datetime
     end: datetime
-    raw: dict[str, ty.Any]
+    raw: dict[str, ty.Any] = field(repr=False)
     parent: str | None
     _engine: str
+    info: SnapshotInfo | None = field(default=None)
 
     @staticmethod
-    def from_rustic(item: dict[str, ty.Any], group: dict[str, ty.Any]) -> Snapshot:
+    def from_rustic(
+        item: dict[str, ty.Any], group: dict[str, ty.Any], repo: str
+    ) -> Snapshot:
         profile = ""
         for tag in group["tags"]:
             if tag.startswith("lohup.profile="):
                 profile = tag.removeprefix("lohup.profile=")
         return Snapshot(
             id=item["id"][:8],
+            repo=repo,
             profile=profile,
             tags=group["tags"],
             hostname=group["hostname"],
@@ -42,44 +49,35 @@ class Snapshot:
             _engine="rustic",
         )
 
-    @staticmethod
-    def from_restic(item: dict[str, ty.Any]) -> Snapshot:
-        profile = ""
-        for tag in item["tags"]:
-            if tag.startswith("lohup.profile="):
-                profile = tag.removeprefix("lohup.profile=")
-        return Snapshot(
-            id=item["short_id"],
-            profile=profile,
-            tags=item["tags"],
-            hostname=item["hostname"],
-            time=datetime.fromisoformat(item["time"]),
-            end=datetime.fromisoformat(item["summary"]["backup_end"]),
-            raw=item,
-            parent=item.get("parent"),
-            _engine="restic",
-        )
-
     @staticmethod
     def findall(session: Session):
         out = []
         stmt = select(SnapshotRecord)
         for row in session.scalars(stmt):
-            out.append(Snapshot(
-                id=row.key,
-                profile=row.profile,
-                tags=[],
-                hostname="",
-                time=row.created,
-                end=row.created,
-                raw={},
-                parent=row.parent,
-                _engine=""
-            ))
+            out.append(Snapshot.from_record(row))
         return out
 
     @staticmethod
-    def update_all(session: Session, repo: str, items: list["Snapshot"]) -> dict[str, list[str]]:
+    def from_record(row: SnapshotRecord) -> Snapshot:
+        info = json.loads(row.raw_info)
+        return Snapshot(
+            id=row.key,
+            repo=row.repo.name,
+            profile=row.profile,
+            tags=[],
+            hostname=info["hostname"],
+            time=row.created,
+            end=row.created,
+            raw=info,
+            parent=row.parent,
+            _engine="",
+            # record=row,
+        )
+
+    @staticmethod
+    def update_all(
+        session: Session, repo: str, items: list["Snapshot"]
+    ) -> dict[str, list[str]]:
         stmt = select(SnapshotRecord.key).where(SnapshotRecord.repo == repo)
         stored = set(session.scalars(stmt))
         result = {
@@ -89,13 +87,15 @@ class Snapshot:
         for snapshot in items:
             if snapshot.id not in stored:
                 result["added"].append(snapshot.id)
-                session.add(SnapshotRecord(
-                    repo=repo,
-                    key=snapshot.id,
-                    profile=snapshot.profile,
-                    parent=snapshot.parent,
-                    created=snapshot.time,
-                ))
+                session.add(
+                    SnapshotRecord(
+                        repo=repo,
+                        key=snapshot.id,
+                        profile=snapshot.profile,
+                        parent=snapshot.parent,
+                        created=snapshot.time,
+                    )
+                )
             else:
                 result["untouched"].append(snapshot.id)
         return result
@@ -147,7 +147,6 @@ class Snapshot:
     def duration(self):
         return self.end - self.time
 
-
 @dataclass(slots=True)
 class FileItem:
     path: str
@@ -181,6 +180,7 @@ class Directory:
     def flags(self):
         return "d"
 
+
 T = ty.TypeVar("T")
 
 
@@ -202,15 +202,19 @@ class SnapshotInfo:
     root: Directory | None = field(default=None)
 
     @staticmethod
-    def load_db(
-        storage: Storage, repo: str, snapshot: str
-    ) -> "SnapshotInfo" | None:
+    def from_db(items: list[SnapshotFile]) -> "SnapshotInfo" | None:
         info = SnapshotInfo()
-        with storage.session() as session:
-            query = select(SnapshotFile).filter_by(repo=repo, snapshot=snapshot)
-            result = session.scalars(query)
-            for row in result:
-                parent = PosixPath(row.path).parent
+        for row in items:
+            parent = PosixPath(row.path).parent
+            if row.kind.startswith("d"):
+                item = Directory(
+                    path=row.path,
+                    parent=str(parent),
+                    name=row.name,
+                    size=row.size,
+                    date=row.mtime,
+                )
+            else:
                 item = FileItem(
                     path=row.path,
                     parent=str(parent),
@@ -219,59 +223,68 @@ class SnapshotInfo:
                     size=row.size,
                     date=row.mtime,
                 )
-                info.tree[item.path] = item
-            if not info.tree:
-                return None
+            info.tree[item.path] = item
+        if not info.tree:
+            return None
         return info
 
-    def save(self, storage: Storage, repo: str, snapshot: str):
-        with storage.session() as session:
-            exists = session.scalars(
-                select(SnapshotFile).filter_by(repo=repo, snapshot=snapshot).limit(1)
-            )
-            if exists.first():
-                return
-            for file in self.tree.values():
-                session.add(
-                    SnapshotFile(
-                        repo=repo,
-                        snapshot=snapshot,
-                        kind=file.flags,
-                        path=file.path,
-                        name=file.name,
-                        size=file.size,
-                        mtime=file.date,
-                    )
+    def save(self, session: Session, snapshot_id: int):
+        exists = session.scalars(
+            select(SnapshotFile).filter_by(snapshot_id=snapshot_id).limit(1)
+        )
+        if exists.first():
+            return
+        for file in self.tree.values():
+            session.add(
+                SnapshotFile(
+                    snapshot_id=snapshot_id,
+                    kind=file.flags,
+                    path=file.path,
+                    name=file.name,
+                    size=file.size,
+                    mtime=file.date,
+                    depth=file.depth,
                 )
-            session.commit()
+            )
 
-    def _itertree(self, depth: int) -> ty.Generator[FileItem | Directory]:
+    def _itertree(self, depth: int | None) -> ty.Generator[FileItem | Directory]:
         for pth in self.tree:
             value = self.tree[pth]
-            if value.depth > depth:
+            if depth and value.depth > depth:
                 continue
             yield value
 
-    def build_tree(self, depth: int):
+    def build_tree(self, depth: int | None):
         root: Directory | None = None
+        nearest: FileItem | None = None
         for file in self._itertree(depth):
-            file = cast(FileItem, file)
             pathobj = PosixPath(file.path)
-            if not file.flags.startswith("d"):
+            if isinstance(file, FileItem):
+                if nearest is None or not pathobj.is_relative_to(nearest.path):
+                    nearest = file
                 continue
-            item = Directory(
-                path=file.path, parent=file.parent, name=file.name, date=file.date
-            )
-            self.tree[file.path] = item
             if root is None or not pathobj.is_relative_to(root.path):
-                root = item
+                root = file
         if root is None:
-            raise ValueError("Failed to resolve root directory")
+            if nearest is None:
+                raise ValueError("No root or any file to fetch.")
+            parent = Path(nearest.parent)
+            root = Directory(
+                path=nearest.parent,
+                parent=str(parent.parent),
+                name=parent.name,
+                date=nearest.date,
+            )
+            self.tree.setdefault(root.path, root)
         for file in self._itertree(depth):
             if isinstance(file, FileItem):
                 self.tree[file.parent].children.append(file)  # ty: ignore
                 continue
-            file.size = sum(x.size for x in self.tree.values() if x.path.startswith(f"{file.path}/"))
+            if file.size != 0:
+                continue
+            file.size = sum(
+                x.size for x in self.tree.values() if x.path.startswith(f"{file.path}/")
+            )
         for file in self._itertree(depth):
             if parent := self.tree.get(file.parent):
                 parent.children.append(file)  # ty: ignore
@@ -308,7 +321,7 @@ class SnapshotInfo:
             file = current.value
             prefix = "  " * (file.depth - 1)
             line = f"{file.name}/" if isinstance(file, Directory) else file.name
-            size = "0"
+            size = "0 Bytes"
             if file.size:
                 size = humanize.filesize.naturalsize(file.size)
             lines.append(f"{size:>10} {prefix}{line}")

src/lohup/restic.py

@@ -1,98 +0,0 @@
-import json
-import os
-import subprocess as procs
-from collections import defaultdict
-from dataclasses import dataclass, field
-from pathlib import Path
-
-from lohup import config, engine
-from lohup.logger import BasicLogger, CliLogger
-
-
-@dataclass
-class Restic:
-    repo: config.Repository
-    log: CliLogger | BasicLogger
-    binary: str = field(default="restic")
-
-    def environ(self):
-        env = os.environ.copy()
-        env["RESTIC_PASSWORD_FILE"] = self.repo.repo_key_file.value
-        match self.repo:
-            case config.S3Repository():
-                if value := self.repo.region:
-                    env["AWS_DEFAULT_REGION"] = value
-                if masked := self.repo.access_key_file:
-                    env["AWS_ACCESS_KEY_ID"] = Path(masked.value).read_text().strip()
-                if masked := self.repo.secret_key_file:
-                    env["AWS_SECRET_ACCESS_KEY"] = (
-                        Path(masked.value).read_text().strip()
-                    )
-                path = self.repo.path.lstrip("/")
-                url = f"{self.repo.endpoint}/{self.repo.bucket}/{path}"
-                env["RESTIC_REPOSITORY"] = f"s3:{url}"
-            case config.LocalRepository():
-                env["RESTIC_REPOSITORY"] = self.repo.path
-        return env
-
-    def run(self, args):
-        cmd, env = self._prepare(args)
-        procs.check_call(cmd, env=env)
-
-    def backup(self, profile: config.Profile):
-        args = ["backup", "--tag", profile.name]
-        args.extend(profile.cli_args)
-        match profile:
-            case config.PathsProfile():
-                for pth in profile.exclude_paths:
-                    args.extend(["-e", pth])
-                args.extend(profile.paths)
-                self.run(args)
-            case config.CommandProfile():
-                args.append("--stdin")
-                match profile.command:
-                    case str(x):
-                        self.pipe_stdout(args, src_cmd=x.split())
-                    case list(x):
-                        self.pipe_stdout(args, src_cmd=x)
-                    case _:
-                        raise NotImplementedError(profile.command)
-
-    def pipe_stdout(self, args: list[str], src_cmd: list[str]):
-        cmd = [self.binary] + args
-        env = self.environ()
-        with procs.Popen(cmd, stdin=procs.PIPE, env=env) as restic:
-            procs.check_call(src_cmd, stdout=restic.stdin)
-
-    def snapshots(self, format="text", profile: str | None = None):
-        cmd, env = self._prepare(["snapshots", "--compact"])
-        if format == "json":
-            cmd.append("--json")
-        if profile:
-            cmd.extend(["--tag", f"lohup.profile={profile}"])
-        result = procs.check_output(cmd, env=env, encoding="utf-8")
-        if format == "json":
-            content = json.loads(result)
-            snapshots = defaultdict(list)
-            for x in content:
-                snap = engine.Snapshot.from_restic(x)
-                snapshots[snap.name].append(snap)
-            return snapshots
-        return result
-
-    def info(self, ref: str) -> engine.SnapshotInfo:
-        raise NotImplementedError
-
-    def _prepare(self, args: list[str]):
-        if self.repo is None:
-            raise ValueError("repo not set")
-        cmd = [self.binary]
-        cmd.extend(args)
-        env = self.environ()
-        return cmd, env
-
-    def __enter__(self):
-        return self
-
-    def __exit__(self, type, value, traceback):
-        return

src/lohup/rustic.py

@@ -1,13 +1,12 @@
-from lohup.storage.storage import Storage
-from typing import cast, IO
-import traceback
-from datetime import datetime
-import re
 import json
+import re
 import subprocess as procs
+import traceback
 from collections import defaultdict
 from dataclasses import dataclass, field
+from datetime import datetime
 from pathlib import Path, PosixPath
+from typing import IO, Any, cast
 
 import tomlkit
 
@@ -20,7 +19,6 @@ class Rustic:
     repo: config.Repository
     log: LoggerProto
     conf_dir: Path
-    storage: Storage | None
     binary: str = field(default="rustic")
 
     @property
@@ -56,7 +54,14 @@ class Rustic:
         out = [self.binary, "-P", str(self.conf_dir / "rustic")]
         return out
 
-    def run(self, args):
+    def run(self, args: list[str], combined_output=True) -> procs.CompletedProcess[str]:
+        cmd = self._cmdline()
+        cmd.extend(args)
+        stderr = procs.STDOUT if combined_output else procs.PIPE
+        proc = procs.run(cmd, stdout=procs.PIPE, stderr=stderr, text=True)
+        return proc
+
+    def call(self, args: list[str]):
         cmd = self._cmdline()
         cmd.extend(args)
         procs.check_call(cmd)
@@ -75,7 +80,7 @@ class Rustic:
                 for pth in profile.exclude_paths:
                     args.extend(["--glob", f"!{pth}"])
                 args.extend(profile.paths)
-                self.run(args)
+                self.call(args)
             case config.CommandProfile():
                 args.append("-")
                 match profile.command:
@@ -104,17 +109,12 @@ class Rustic:
         for group in content:
             key = group["group_key"]
             for x in group["snapshots"]:
-                snap = engine.Snapshot.from_rustic(x, group=key)
+                snap = engine.Snapshot.from_rustic(x, group=key, repo=self.repo.name)
                 grouped[snap.name].append(snap)
                 snapshots.append(snap)
         return grouped
 
-    def info(self, snapshot: str):
-        if self.storage:
-            if value := engine.SnapshotInfo.load_db(
-                self.storage, repo=self.repo.name, snapshot=snapshot
-            ):
-                return value
+    def info(self, snapshot: str) -> engine.SnapshotInfo:
         result = engine.SnapshotInfo()
 
         def read_(match):
@@ -126,7 +126,7 @@ class Rustic:
                 mtime = datetime.strptime(mtime_raw, "%d %b %Y %H:%M")
             path = match.group("path")
             pth = PosixPath(path)
-            result.tree[path] = engine.FileItem(
+            item = engine.FileItem(
                 path=path,
                 parent=str(pth.parent),
                 name=pth.name,
@@ -134,6 +134,14 @@ class Rustic:
                 size=int(size),
                 date=mtime,
             )
+            if flags.startswith("d"):
+                item = engine.Directory(
+                    path=path,
+                    parent=item.parent,
+                    name=pth.name,
+                    date=mtime,
+                )
+            result.tree[path] = item
 
         cmd = self._cmdline()
         cmd.extend(["ls", "-l", snapshot])
@@ -147,7 +155,7 @@ class Rustic:
             for line in stream:
                 match = pattern.match(line.strip())
                 if not match:
-                    print(f"unmatched line: {line!s}")
+                    self.log.warning(f"unmatched line: {line!s}")
                     continue
                 try:
                     read_(match)
@@ -155,10 +163,19 @@ class Rustic:
                     traceback.print_exc()
         if not result.tree:
             raise KeyError("No such snapshot")
-        if self.storage:
-            result.save(self.storage, repo=self.repo.name, snapshot=snapshot)
+        result.build_tree(depth=None)
         return result
 
+    def repoinfo(self) -> dict[str, Any] | None:
+        proc = self.run(["repoinfo", "--json"], combined_output=False)
+        if proc.returncode == 0:
+            return json.loads(proc.stdout)
+        if "No repository config file found" in proc.stderr:
+            return None
+        exc = ValueError("Process returned error")
+        exc.add_note(f"Output: {proc.stdout}{proc.stderr}")
+        raise exc
+
     def __enter__(self):
         self.write_config()
         return self

src/lohup/storage/file.py

@@ -1,18 +1,59 @@
 from datetime import datetime
-from sqlalchemy import String, Text, BigInteger, DateTime
-from sqlalchemy.orm import Mapped, mapped_column
-from typing import final
+from typing import TYPE_CHECKING, final
+
+from sqlalchemy import (
+    BigInteger,
+    DateTime,
+    ForeignKey,
+    Integer,
+    String,
+    Text,
+    select,
+)
+from sqlalchemy.orm import Mapped, Session, mapped_column, relationship
+
+from lohup.config import Repository
 from lohup.storage.base import Base
 
+if TYPE_CHECKING:
+    from lohup.storage.snapshot import SnapshotRecord
+
 
 @final
 class SnapshotFile(Base):
     __tablename__ = "files"
     id: Mapped[int] = mapped_column(primary_key=True)
-    repo: Mapped[str] = mapped_column(String(64))
-    snapshot: Mapped[str] = mapped_column(String(64))
+    snapshot_id: Mapped[int] = mapped_column(ForeignKey("snapshots.id"))
+    snapshot: Mapped["SnapshotRecord"] = relationship()
+    # file|dir
     kind: Mapped[str] = mapped_column(String(10))
     path: Mapped[str] = mapped_column(Text())
+    # how deep file is in the snapshot tree
+    depth: Mapped[int] = mapped_column(Integer())
     name: Mapped[str] = mapped_column(Text())
+    # sum of files for directories
     size: Mapped[int] = mapped_column(BigInteger())
     mtime: Mapped[datetime] = mapped_column(DateTime())
+
+    @staticmethod
+    def find_all(
+        session: Session,
+        repo: Repository,
+        snapshot: "SnapshotRecord",
+        depth: int | None = None,
+    ) -> list["SnapshotFile"]:
+        from lohup.storage.repository import RepositoryRecord
+        from lohup.storage.snapshot import SnapshotRecord
+
+        stmt = (
+            select(SnapshotFile)
+            .join(SnapshotFile.snapshot)
+            .join(SnapshotRecord.repo)
+            .where(
+                SnapshotRecord.id == snapshot.id,
+                RepositoryRecord.name == repo.name
+            )
+        )
+        if depth is not None:
+            stmt = stmt.where(SnapshotFile.depth <= depth)
+        return list(session.scalars(stmt))

src/lohup/storage/refresher.py

@@ -0,0 +1,148 @@
+import json
+from lohup.storage.snapshot import SnapshotRecord
+from lohup.engine import Snapshot, SnapshotInfo
+import typing
+import typing as ty
+from dataclasses import dataclass, field
+from datetime import datetime, timedelta
+
+from sqlalchemy import select, delete
+
+from lohup import config
+from lohup.storage.repository import RepositoryRecord
+from lohup.storage.storage import Storage
+
+T = typing.TypeVar("T")
+
+
+@dataclass
+class UpdateResult[T]:
+    items: dict[str, T] = field(default_factory=dict)
+    added: list[str] = field(default_factory=list)
+    unchanged: list[str] = field(default_factory=list)
+    updated: list[str] = field(default_factory=list)
+    removed: list[str] = field(default_factory=list)
+    _messages: list[str] = field(default_factory=list)
+
+    def message(self, line: str):
+        self._messages.append(line)
+
+    def text(self) -> str:
+        lines = []
+        if self.added:
+            lines.append(f"{len(self.added)} added")
+        if self.updated:
+            lines.append(f"{len(self.updated)} updated")
+        if self.removed:
+            lines.append(f"{len(self.removed)} removed")
+        if self.unchanged:
+            lines.append(f"{len(self.unchanged)} untouched")
+        lines.extend(self._messages)
+        return "\n".join(lines)
+
+
+@dataclass
+class Refresher:
+    storage: Storage
+    config: config.TomlConfig
+    force: bool
+
+    def update_repos(self, repos: dict[str, config.Repository]) -> UpdateResult[str]:
+        result = UpdateResult[str]()
+        existing = []
+        with self.storage.session() as session:
+            stmt = select(RepositoryRecord)
+            rows = session.scalars(stmt)
+            existing.extend(rows)
+        existing_names = {x.name for x in existing}
+        to_remove = []
+        for name in repos:
+            if name not in existing_names:
+                result.added.append(name)
+        for row in existing:
+            delta = datetime.now() - row.updated
+            if row.name not in repos:
+                to_remove.append(row.id)
+                result.removed.append(row.name)
+            elif self.force or delta > timedelta(hours=8):
+                result.updated.append(row.name)
+            else:
+                result.unchanged.append(row.name)
+        if to_remove:
+            stmt = delete(RepositoryRecord).where(RepositoryRecord.id.in_(to_remove))
+            with self.storage.session() as session:
+                session.execute(stmt)
+        return result
+
+    def add_repos(self, items: dict[str, ty.Any]):
+        pending = []
+        for name, info in items.items():
+            data = {x["tpe"]: x for x in info["files"]["repo"]}
+            size = sum(x["size"] for x in data.values())
+            pending.append(
+                RepositoryRecord(
+                    name=name, size=size, updated=datetime.now()
+                )
+            )
+        with self.storage.session() as session:
+            session.add_all(pending)
+            session.commit()
+
+    def update_snapshots(self, pulled: list[Snapshot]) -> UpdateResult[Snapshot]:
+        reponames = {r.repo for r in pulled}
+        result = UpdateResult[Snapshot]()
+        result.items = {x.id: x for x in pulled}
+        records = []
+        with self.storage.session() as session:
+            stmt = (
+                select(SnapshotRecord)
+                .join(SnapshotRecord.repo)
+                .where(RepositoryRecord.name.in_(reponames))
+            )
+            rows = session.scalars(stmt)
+            records.extend(rows)
+        for row in records:
+            if row.key not in result.items:
+                result.items[row.key] = Snapshot.from_record(row)
+        record_keys = {x.key for x in records}
+        existing_ids = {x.id for x in pulled}
+        for snapshot in pulled:
+            if snapshot.id not in record_keys:
+                result.added.append(snapshot.id)
+        rm_ids = []
+        for row in records:
+            if row.key not in existing_ids:
+                result.removed.append(row.key)
+                rm_ids.append(row.id)
+            else:
+                result.unchanged.append(row.key)
+        if result.removed:
+            stmt = delete(SnapshotRecord).where(SnapshotRecord.id.in_(rm_ids))
+            with self.storage.session() as session:
+                session.execute(stmt)
+        pending = [x for x in result.items.values() if x.id in result.added]
+        if pending:
+            self.add_snapshots(pending)
+        return result
+
+    def add_snapshots(self, items: list[Snapshot]):
+        stmt = select(RepositoryRecord)
+        with self.storage.session() as session:
+            rows = session.scalars(stmt)
+            repos = {x.name: x for x in rows}
+            for item in items:
+                record = SnapshotRecord(
+                    repo_id=repos[item.repo].id,
+                    key=item.id,
+                    profile=item.profile,
+                    parent=item.parent,
+                    created=item.time,
+                    raw_info=json.dumps(item.raw),
+                )
+                session.add(record)
+            session.commit()
+
+    def add_files(self, snapshot_id: int, info: SnapshotInfo):
+        with self.storage.session() as session:
+            info.save(session, snapshot_id)
+            session.commit()

src/lohup/storage/repository.py

@@ -0,0 +1,20 @@
+from datetime import datetime
+from sqlalchemy import String, DateTime, Integer
+from sqlalchemy.orm import Mapped, mapped_column, relationship
+from lohup.storage.base import Base
+from typing import final, TYPE_CHECKING
+
+if TYPE_CHECKING:
+    from lohup.storage.snapshot import SnapshotRecord
+
+
+@final
+class RepositoryRecord(Base):
+    __tablename__ = "repositories"
+    id: Mapped[int] = mapped_column(primary_key=True)
+    name: Mapped[str] = mapped_column(String(128))
+    # total repository size
+    size: Mapped[int] = mapped_column(Integer())
+    updated: Mapped[datetime] = mapped_column(DateTime())
+
+    snapshots: Mapped[list["SnapshotRecord"]] = relationship()

src/lohup/storage/snapshot.py

@@ -1,17 +1,50 @@
 from datetime import datetime
-from sqlalchemy import String, DateTime
-from sqlalchemy.orm import Mapped, mapped_column
 from typing import final
+
+from sqlalchemy import DateTime, ForeignKey, String, select
+from sqlalchemy.orm import Mapped, Session, mapped_column, relationship
+
 from lohup.storage.base import Base
+from lohup.storage.repository import RepositoryRecord
+
 
 @final
 class SnapshotRecord(Base):
     __tablename__ = "snapshots"
     id: Mapped[int] = mapped_column(primary_key=True)
-    repo: Mapped[str] = mapped_column(String(64))
+    repo_id: Mapped[int] = mapped_column(ForeignKey("repositories.id"))
+    repo: Mapped[RepositoryRecord] = relationship(back_populates="snapshots")
     key: Mapped[str] = mapped_column(String(64))
     profile: Mapped[str] = mapped_column(String(32))
     parent: Mapped[str] = mapped_column(String(64), nullable=True)
+    raw_info: Mapped[str] = mapped_column(String())
     created: Mapped[datetime] = mapped_column(DateTime())
 
+    @staticmethod
+    def all_by_repository(session: Session, repo: str) -> list["SnapshotRecord"]:
+        stmt = (
+            select(SnapshotRecord)
+            .join(SnapshotRecord.repo)
+            .where(RepositoryRecord.name == repo)
+            .order_by(SnapshotRecord.created)
+        )
+        return list(session.scalars(stmt))
 
+    @staticmethod
+    def all_by_profile(session: Session, repo: str, profile: str) -> list["SnapshotRecord"]:
+        stmt = (
+            select(SnapshotRecord)
+            .join(SnapshotRecord.repo)
+            .where(RepositoryRecord.name == repo, SnapshotRecord.profile == profile)
+            .order_by(SnapshotRecord.created)
+        )
+        return list(session.scalars(stmt))
+
+    @staticmethod
+    def find_by_key(session: Session, repo: str, key: str) -> "SnapshotRecord" | None:
+        stmt = (
+            select(SnapshotRecord)
+            .join(SnapshotRecord.repo)
+            .where(RepositoryRecord.name == repo, SnapshotRecord.key == key)
+        )
+        return session.scalar(stmt)

src/lohup/storage/storage.py

@@ -1,10 +1,11 @@
+from lohup.storage.repository import RepositoryRecord
 from lohup.storage.snapshot import SnapshotRecord
 import typing as ty
 from contextlib import contextmanager
 from dataclasses import dataclass
 from pathlib import Path
 
-from sqlalchemy import Engine, create_engine
+from sqlalchemy import Engine, create_engine, select, func
 from sqlalchemy.orm import Session
 
 from lohup.storage.base import Base
@@ -28,11 +29,22 @@ class Storage:
         with Session(self.engine) as sess:
             yield sess
 
+    def recreate(self):
+        Base.metadata.drop_all(self.engine)
+        self.create()
+
     def create(self):
         Base.metadata.create_all(
             self.engine,
             tables=[  # ty: ignore[invalid-argument-type]
-                SnapshotFile.__table__,
+                RepositoryRecord.__table__,
                 SnapshotRecord.__table__,
+                SnapshotFile.__table__,
             ],
         )
+
+    def is_empty(self):
+        with self.session() as session:
+            stmt = select(func.count()).select_from(RepositoryRecord)
+            result = session.scalar(stmt)
+        return not result