defrag/app/migration/hardlink.py

import os
from pathlib import Path
from typing import Optional
from ..shared.logger import ProgressLogger

class HardlinkMigrationStrategy:

    def __init__(self, logger: Optional[ProgressLogger]=None):
        self.logger = logger

    def migrate(self, source: Path, destination: Path, verify: bool=True) -> bool:
        if not source.exists():
            if self.logger:
                self.logger.error(f'Source file does not exist: {source}')
            return False
        if not self._same_filesystem(source, destination.parent):
            if self.logger:
                self.logger.warning(f'Cannot hardlink across filesystems: {source} -> {destination}')
            return False
        destination.parent.mkdir(parents=True, exist_ok=True)
        try:
            os.link(source, destination)
            if verify:
                if not self._verify_hardlink(source, destination):
                    if self.logger:
                        self.logger.error(f'Verification failed: {source} -> {destination}')
                    destination.unlink()
                    return False
            return True
        except FileExistsError:
            if self.logger:
                self.logger.warning(f'Destination already exists: {destination}')
            return False
        except Exception as e:
            if self.logger:
                self.logger.error(f'Hardlink failed: {source} -> {destination}: {e}')
            return False

    def _same_filesystem(self, path1: Path, path2: Path) -> bool:
        try:
            stat1 = path1.stat()
            stat2 = path2.stat()
            return stat1.st_dev == stat2.st_dev
        except Exception:
            return False

    def _verify_hardlink(self, source: Path, destination: Path) -> bool:
        try:
            source_stat = source.stat()
            dest_stat = destination.stat()
            return source_stat.st_ino == dest_stat.st_ino
        except Exception:
            return False

    def can_migrate(self, source: Path, destination: Path) -> bool:
        if not source.exists():
            return False
        dest_dir = destination.parent
        if dest_dir.exists():
            return self._same_filesystem(source, dest_dir)
        parent = dest_dir.parent
        while not parent.exists() and parent != parent.parent:
            parent = parent.parent
        return parent.exists() and self._same_filesystem(source, parent)

    def estimate_time(self, source: Path) -> float:
        return 0.01

    def cleanup(self, source: Path) -> bool:
        return True

class SymlinkMigrationStrategy:

    def __init__(self, logger: Optional[ProgressLogger]=None, absolute_links: bool=True):
        self.logger = logger
        self.absolute_links = absolute_links

    def migrate(self, source: Path, destination: Path, verify: bool=True) -> bool:
        if not source.exists():
            if self.logger:
                self.logger.error(f'Source file does not exist: {source}')
            return False
        destination.parent.mkdir(parents=True, exist_ok=True)
        try:
            if self.absolute_links:
                target = source.resolve()
            else:
                target = os.path.relpath(source, destination.parent)
            destination.symlink_to(target)
            if verify:
                if not self._verify_symlink(destination, source):
                    if self.logger:
                        self.logger.error(f'Verification failed: {source} -> {destination}')
                    destination.unlink()
                    return False
            return True
        except FileExistsError:
            if self.logger:
                self.logger.warning(f'Destination already exists: {destination}')
            return False
        except Exception as e:
            if self.logger:
                self.logger.error(f'Symlink failed: {source} -> {destination}: {e}')
            return False

    def _verify_symlink(self, symlink: Path, expected_target: Path) -> bool:
        try:
            if not symlink.is_symlink():
                return False
            resolved = symlink.resolve()
            expected = expected_target.resolve()
            return resolved == expected
        except Exception:
            return False

    def can_migrate(self, source: Path, destination: Path) -> bool:
        if not source.exists():
            return False
        dest_dir = destination.parent
        if dest_dir.exists():
            return os.access(dest_dir, os.W_OK)
        return True

    def estimate_time(self, source: Path) -> float:
        return 0.01

    def cleanup(self, source: Path) -> bool:
        return True

class DedupHardlinkStrategy(HardlinkMigrationStrategy):

    def __init__(self, logger: Optional[ProgressLogger]=None):
        super().__init__(logger=logger)

    def deduplicate(self, canonical: Path, duplicate: Path) -> bool:
        if not canonical.exists():
            if self.logger:
                self.logger.error(f'Canonical file does not exist: {canonical}')
            return False
        if not duplicate.exists():
            if self.logger:
                self.logger.error(f'Duplicate file does not exist: {duplicate}')
            return False
        if self._verify_hardlink(canonical, duplicate):
            return True
        if not self._same_filesystem(canonical, duplicate):
            if self.logger:
                self.logger.warning(f'Cannot hardlink across filesystems: {canonical} -> {duplicate}')
            return False
        try:
            backup = duplicate.with_suffix(duplicate.suffix + '.bak')
            duplicate.rename(backup)
            os.link(canonical, duplicate)
            backup.unlink()
            return True
        except Exception as e:
            if self.logger:
                self.logger.error(f'Deduplication failed: {duplicate}: {e}')
            if backup.exists():
                backup.rename(duplicate)
            return False