feat: possible incremental backup implementation using new abstraction

src/server/backups.rs

@@ -5,6 +5,7 @@ use std::collections::{HashMap, HashSet};
 use std::fs::File;
 use std::io;
 use std::path::{Path, PathBuf};
+use std::sync::Arc;
 
 #[link(name = "c")]
 extern "C" {
@@ -14,27 +15,6 @@ extern "C" {
 
 const FILENAME_FORMAT: &str = "%Y-%m-%d_%H-%M-%S.tar.gz";
 
-pub enum BackupType {
-    Full,
-    Incremental,
-}
-
-/// Represents a successful backup
-pub struct Backup {
-    previous: Option<Box<Backup>>,
-    /// When the backup was started (also corresponds to the name)
-    start_time: chrono::DateTime<Utc>,
-    /// Type of the backup
-    type_: BackupType,
-    /// What files were added/modified in each part of the tarball.
-    pub added: HashMap<PathBuf, HashSet<PathBuf>>,
-    /// What files were removed in this backup, in comparison to the previous backup. For full
-    /// backups, this will always be empty, as they do not consider previous backups.
-    /// The map stores a separate list for each top-level directory, as the contents of these
-    /// directories can come for different source directories.
-    pub removed: HashMap<PathBuf, HashSet<PathBuf>>,
-}
-
 fn files(src_dir: PathBuf) -> io::Result<HashSet<PathBuf>> {
     let mut dirs = vec![src_dir.clone()];
     let mut files: HashSet<PathBuf> = HashSet::new();
@@ -58,8 +38,148 @@ fn files(src_dir: PathBuf) -> io::Result<HashSet<PathBuf>> {
     Ok(files)
 }
 
+/// Return false only if we can say with certainty that the file wasn't modified since the given
+/// timestamp, true otherwise.
+fn modified_since<T: AsRef<Path>>(time: chrono::DateTime<Utc>, path: T) -> bool {
+    let path = path.as_ref();
+
+    if let Ok(metadata) = path.metadata() {
+        let last_modified = metadata.modified();
+
+        if let Ok(last_modified) = last_modified {
+            let t: chrono::DateTime<Utc> = last_modified.into();
+            let t = t.with_timezone(&Local);
+
+            return t >= time;
+        }
+    }
+
+    false
+}
+
+#[derive(PartialEq)]
+pub enum BackupType {
+    Full,
+    Incremental,
+}
+
+#[derive(Debug)]
+pub enum BackupError {
+    NoFullAncestor,
+}
+
+type BackupResult<T> = Result<T, BackupError>;
+
+/// Represents the changes relative to the previous backup
+pub struct BackupDelta {
+    /// What files were added/modified in each part of the tarball.
+    pub added: HashMap<PathBuf, HashSet<PathBuf>>,
+    /// What files were removed in this backup, in comparison to the previous backup. For full
+    /// backups, this will always be empty, as they do not consider previous backups.
+    /// The map stores a separate list for each top-level directory, as the contents of these
+    /// directories can come for different source directories.
+    pub removed: HashMap<PathBuf, HashSet<PathBuf>>,
+}
+
+impl BackupDelta {
+    pub fn new() -> Self {
+        BackupDelta {
+            added: HashMap::new(),
+            removed: HashMap::new(),
+        }
+    }
+
+    /// Update the current state so that its result becomes the merge of itself and the other
+    /// state.
+    pub fn merge(&mut self, delta: &BackupDelta) {
+        for (dir, added) in delta.added.iter() {
+            // Files that were removed in the current state, but added in the new state, are no
+            // longer removed
+            if let Some(orig_removed) = self.removed.get_mut(dir) {
+                orig_removed.retain(|k| !added.contains(k));
+            }
+
+            // Newly added files are added to the state as well
+            if let Some(orig_added) = self.added.get_mut(dir) {
+                orig_added.extend(added.iter().cloned());
+            } else {
+                self.added.insert(dir.clone(), added.clone());
+            }
+        }
+
+        for (dir, removed) in delta.removed.iter() {
+            // Files that were originally added, but now deleted are removed from the added list
+            if let Some(orig_added) = self.added.get_mut(dir) {
+                orig_added.retain(|k| !removed.contains(k));
+            }
+
+            // Newly removed files are added to the state as well
+            if let Some(orig_removed) = self.removed.get_mut(dir) {
+                orig_removed.extend(removed.iter().cloned());
+            } else {
+                self.removed.insert(dir.clone(), removed.clone());
+            }
+        }
+    }
+
+    /// Modify the given state by applying this delta's changes to it
+    pub fn apply(&self, state: &mut HashMap<PathBuf, HashSet<PathBuf>>) {
+        // First we add new files, then we remove the old ones
+        for (dir, added) in self.added.iter() {
+            if let Some(current) = state.get_mut(dir) {
+                current.extend(added.iter().cloned());
+            } else {
+                state.insert(dir.clone(), added.clone());
+            }
+        }
+
+        for (dir, removed) in self.removed.iter() {
+            if let Some(current) = state.get_mut(dir) {
+                current.retain(|k| !removed.contains(k));
+            }
+        }
+    }
+}
+
+/// Represents a successful backup
+pub struct Backup {
+    previous: Option<Arc<Backup>>,
+    /// When the backup was started (also corresponds to the name)
+    start_time: chrono::DateTime<Utc>,
+    /// Type of the backup
+    type_: BackupType,
+    delta: BackupDelta,
+}
+
 impl Backup {
-    /// Create a new full backup
+    /// Calculate the full state of the backup by applying all its ancestors delta's in order,
+    /// starting from the last full ancestor.
+    pub fn state(&self) -> BackupResult<HashMap<PathBuf, HashSet<PathBuf>>> {
+        if self.type_ == BackupType::Full {
+            let mut state = HashMap::new();
+            self.delta.apply(&mut state);
+
+            Ok(state)
+        } else if let Some(previous) = &self.previous {
+            let mut state = previous.state()?;
+            self.delta.apply(&mut state);
+
+            Ok(state)
+        } else {
+            return Err(BackupError::NoFullAncestor);
+        }
+    }
+    /// Create a new Full backup, populated with the given directories.
+    ///
+    /// # Arguments
+    ///
+    /// * `backup_dir` - Directory to store archive in
+    /// * `dirs` - list of tuples `(path_in_tar, src_dir)` with `path_in_tar` the directory name
+    ///     under which `src_dir`'s contents should be stored in the archive
+    ///
+    /// # Returns
+    ///
+    /// The `Backup` instance describing this new backup.
     pub fn create<P: AsRef<Path>>(
         backup_dir: P,
         dirs: Vec<(PathBuf, PathBuf)>,
@@ -79,7 +199,7 @@ impl Backup {
             let files = files(src_dir.clone())?;
 
             for path in &files {
-                ar.append_path_with_name(dir_in_tar.join(&path), src_dir.join(&path))?;
+                ar.append_path_with_name(dir_in_tar.join(path), src_dir.join(path))?;
             }
 
             added.insert(dir_in_tar, files);
@@ -89,8 +209,58 @@ impl Backup {
             previous: None,
             type_: BackupType::Full,
             start_time,
+            delta: BackupDelta {
                 added,
                 removed: HashMap::new(),
+            },
+        })
+    }
+
+    /// Create a new incremental backup from a given previous backup
+    pub fn create_from<P: AsRef<Path>>(
+        previous: Arc<Backup>,
+        backup_dir: P,
+        dirs: Vec<(PathBuf, PathBuf)>,
+    ) -> io::Result<Self> {
+        let backup_dir = backup_dir.as_ref();
+        let start_time = chrono::offset::Utc::now();
+
+        let filename = format!("{}", start_time.format(FILENAME_FORMAT));
+        let path = backup_dir.join(filename);
+        let tar_gz = File::create(path)?;
+        let enc = GzEncoder::new(tar_gz, Compression::default());
+        let mut ar = tar::Builder::new(enc);
+
+        let previous_state = previous.state().unwrap();
+        let mut delta = BackupDelta::new();
+
+        for (dir_in_tar, src_dir) in dirs {
+            let files = files(src_dir.clone())?;
+            let added_files = files
+                .iter()
+                .filter(|p| modified_since(previous.start_time, p))
+                .cloned()
+                .collect::<HashSet<PathBuf>>();
+
+            for path in added_files.iter() {
+                ar.append_path_with_name(dir_in_tar.join(path), src_dir.join(path))?;
+            }
+
+            delta.added.insert(dir_in_tar.clone(), added_files);
+
+            if let Some(previous_files) = previous_state.get(&dir_in_tar) {
+                delta.removed.insert(
+                    dir_in_tar,
+                    previous_files.difference(&files).cloned().collect(),
+                );
+            }
+        }
+
+        Ok(Backup {
+            previous: Some(previous),
+            type_: BackupType::Incremental,
+            start_time,
+            delta,
         })
     }
 }
@@ -100,7 +270,7 @@ pub struct BackupManager {
     config_dir: PathBuf,
     world_dir: PathBuf,
     max_backups: u64,
-    last_backup: Option<Backup>,
+    last_backup: Option<Arc<Backup>>,
 }
 
 impl BackupManager {
@@ -124,10 +294,16 @@ impl BackupManager {
             (PathBuf::from("config"), self.config_dir.clone()),
             (PathBuf::from("worlds"), self.world_dir.clone()),
         ];
+
         if let Some(last_backup) = &self.last_backup {
-            todo!();
+            let clone = last_backup.clone();
+            self.last_backup = Some(Arc::new(Backup::create_from(
+                clone,
+                &self.backup_dir,
+                dirs,
+            )?));
         } else {
-            self.last_backup = Some(Backup::create(&self.backup_dir, dirs)?);
+            self.last_backup = Some(Arc::new(Backup::create(&self.backup_dir, dirs)?));
         }
 
         Ok(())