more stats

src/bundledb/db.rs

@@ -525,10 +525,25 @@ impl BundleDb {
 
     pub fn statistics(&self) -> BundleStatistics {
         let bundles = self.list_bundles();
+        let bundles_meta: Vec<_> = bundles.iter().filter(|b| b.mode == BundleMode::Meta).collect();
+        let bundles_data: Vec<_> = bundles.iter().filter(|b| b.mode == BundleMode::Data).collect();
+        let mut hash_methods = HashMap::new();
+        let mut compressions = HashMap::new();
+        for bundle in &bundles {
+            *hash_methods.entry(bundle.hash_method).or_insert(0) += 1;
+            *compressions.entry(bundle.compression.clone()).or_insert(0) += 1;
+        }
         BundleStatistics {
+            hash_methods, compressions,
             raw_size: ValueStats::from_iter(|| bundles.iter().map(|b| b.raw_size as f32)),
             encoded_size: ValueStats::from_iter(|| bundles.iter().map(|b| b.encoded_size as f32)),
-            chunk_count: ValueStats::from_iter(|| bundles.iter().map(|b| b.chunk_count as f32))
+            chunk_count: ValueStats::from_iter(|| bundles.iter().map(|b| b.chunk_count as f32)),
+            raw_size_meta: ValueStats::from_iter(|| bundles_meta.iter().map(|b| b.raw_size as f32)),
+            encoded_size_meta: ValueStats::from_iter(|| bundles_meta.iter().map(|b| b.encoded_size as f32)),
+            chunk_count_meta: ValueStats::from_iter(|| bundles_meta.iter().map(|b| b.chunk_count as f32)),
+            raw_size_data: ValueStats::from_iter(|| bundles_data.iter().map(|b| b.raw_size as f32)),
+            encoded_size_data: ValueStats::from_iter(|| bundles_data.iter().map(|b| b.encoded_size as f32)),
+            chunk_count_data: ValueStats::from_iter(|| bundles_data.iter().map(|b| b.chunk_count as f32))
         }
     }
 }

src/bundledb/mod.rs

@@ -13,6 +13,7 @@ pub use self::uploader::BundleUploader;
 use prelude::*;
 
 use std::fmt;
+use std::collections::HashMap;
 use serde;
 use rand;
 
@@ -137,7 +138,15 @@ impl Default for BundleInfo {
 
 #[derive(Debug)]
 pub struct BundleStatistics {
-    raw_size: ValueStats,
+    pub raw_size: ValueStats,
-    encoded_size: ValueStats,
+    pub encoded_size: ValueStats,
-    chunk_count: ValueStats
+    pub chunk_count: ValueStats,
+    pub raw_size_meta: ValueStats,
+    pub encoded_size_meta: ValueStats,
+    pub chunk_count_meta: ValueStats,
+    pub raw_size_data: ValueStats,
+    pub encoded_size_data: ValueStats,
+    pub chunk_count_data: ValueStats,
+    pub hash_methods: HashMap<HashMethod, usize>,
+    pub compressions: HashMap<Option<Compression>, usize>
 }

src/cli/args.rs

@@ -75,6 +75,9 @@ pub enum Arguments {
         backup_name: Option<String>,
         inode: Option<String>
     },
+    Stats {
+        repo_path: PathBuf
+    },
     Copy {
         repo_path_src: PathBuf,
         backup_name_src: String,
@@ -464,6 +467,11 @@ pub fn parse() -> Result<(log::Level, Arguments), ErrorCode> {
             .arg(Arg::from_usage("<REPO>")
                 .help(tr!("Path of the repository"))
                 .validator(|val| validate_repo_path(val, true, Some(false), Some(false)))))
+        .subcommand(SubCommand::with_name("stats")
+            .about(tr!("Display statistics on a repository"))
+            .arg(Arg::from_usage("<REPO>")
+                .help(tr!("Path of the repository"))
+                .validator(|val| validate_repo_path(val, true, Some(false), Some(false)))))
         .subcommand(SubCommand::with_name("bundleinfo")
             .about(tr!("Display information on a bundle"))
             .arg(Arg::from_usage("<REPO>")
@@ -739,6 +747,15 @@ pub fn parse() -> Result<(log::Level, Arguments), ErrorCode> {
                 inode: inode.map(|v| v.to_string())
             }
         }
+        ("stats", Some(args)) => {
+            let (repository, _backup, _inode) = parse_repo_path(
+                args.value_of("REPO").unwrap(),
+                true,
+                Some(false),
+                Some(false)
+            ).unwrap();
+            Arguments::Stats { repo_path: repository }
+        }
         ("copy", Some(args)) => {
             let (repository_src, backup_src, _inode) =
                 parse_repo_path(args.value_of("SRC").unwrap(), true, Some(true), Some(false))

src/cli/mod.rs

@@ -313,6 +313,61 @@ fn print_repoinfo(info: &RepositoryInfo) {
     );
 }
 
+fn print_repostats(stats: &RepositoryStatistics) {
+    tr_println!("Index\n=====");
+    let index_usage = stats.index.count as f32 / stats.index.capacity as f32;
+    tr_println!("Size: {}", to_file_size(stats.index.size as u64));
+    tr_println!("Entries: {} / {}, {:.0}%", stats.index.count, stats.index.capacity, index_usage*100.0);
+    let disp = &stats.index.displacement;
+    tr_println!("Displacement:\n  - average: {:.1}\n  - stddev: {:.1}\n  - over {:.1}: {:.0}, {:.1}%\n  - maximum: {:.0}",
+        disp.avg, disp.stddev, disp.avg + 2.0 * disp.stddev, disp.count_xl, disp.count_xl as f32 / disp.count as f32 * 100.0, disp.max);
+    println!("");
+    tr_println!("Bundles (all)\n=============");
+    let rsize = &stats.bundles.raw_size;
+    tr_println!("Raw size:\n  - average: {}\n  - stddev: {}\n  - maximum: {}",
+        to_file_size(rsize.avg as u64), to_file_size(rsize.stddev as u64), to_file_size(rsize.max as u64));
+    let esize = &stats.bundles.encoded_size;
+    tr_println!("Encoded size:\n  - average: {}\n  - stddev: {}\n  - maximum: {}",
+        to_file_size(esize.avg as u64), to_file_size(esize.stddev as u64), to_file_size(esize.max as u64));
+    let ccount = &stats.bundles.chunk_count;
+    tr_println!("Chunk count:\n  - average: {:.1}\n  - stddev: {:.1}\n  - minimum: {:.0}\n  - maximum: {:.0}", ccount.avg, ccount.stddev, ccount.min, ccount.max);
+    println!("");
+    tr_println!("Meta bundles\n============");
+    let rsize = &stats.bundles.raw_size_meta;
+    tr_println!("Raw size:\n  - average: {}\n  - stddev: {}\n  - maximum: {}",
+        to_file_size(rsize.avg as u64), to_file_size(rsize.stddev as u64), to_file_size(rsize.max as u64));
+    let esize = &stats.bundles.encoded_size_meta;
+    tr_println!("Encoded size:\n  - average: {}\n  - stddev: {}\n  - maximum: {}",
+        to_file_size(esize.avg as u64), to_file_size(esize.stddev as u64), to_file_size(esize.max as u64));
+    let ccount = &stats.bundles.chunk_count_meta;
+    tr_println!("Chunk count:\n  - average: {:.1}\n  - stddev: {:.1}\n  - minimum: {:.0}\n  - maximum: {:.0}", ccount.avg, ccount.stddev, ccount.min, ccount.max);
+    println!("");
+    tr_println!("Data bundles\n============");
+    let rsize = &stats.bundles.raw_size_data;
+    tr_println!("Raw size:\n  - average: {}\n  - stddev: {}\n  - maximum: {}",
+        to_file_size(rsize.avg as u64), to_file_size(rsize.stddev as u64), to_file_size(rsize.max as u64));
+    let esize = &stats.bundles.encoded_size_data;
+    tr_println!("Encoded size:\n  - average: {}\n  - stddev: {}\n  - maximum: {}",
+        to_file_size(esize.avg as u64), to_file_size(esize.stddev as u64), to_file_size(esize.max as u64));
+    let ccount = &stats.bundles.chunk_count_data;
+    tr_println!("Chunk count:\n  - average: {:.1}\n  - stddev: {:.1}\n  - minimum: {:.0}\n  - maximum: {:.0}", ccount.avg, ccount.stddev, ccount.min, ccount.max);
+    println!("");
+    tr_println!("Bundle methods\n==============");
+    tr_println!("Compression:");
+    for (compr, &count) in &stats.bundles.compressions {
+        let compr_name = if let &Some(ref compr) = compr {
+            compr.to_string()
+        } else {
+            tr!("none").to_string()
+        };
+        tr_println!("  - {}: {}, {:.1}%", compr_name, count, count as f32 / stats.bundles.raw_size.count as f32 * 100.0);
+    }
+    tr_println!("Hash:");
+    for (hash, &count) in &stats.bundles.hash_methods {
+        tr_println!("  - {}: {}, {:.1}%", hash.name(), count, count as f32 / stats.bundles.raw_size.count as f32 * 100.0);
+    }
+}
+
 fn print_bundle(bundle: &StoredBundle) {
     tr_println!("Bundle {}", bundle.info.id);
     tr_println!("  - Mode: {:?}", bundle.info.mode);
@@ -859,10 +914,15 @@ pub fn run() -> Result<(), ErrorCode> {
                     print_backup(&backup);
                 }
             } else {
-                println!("{:?}", repo.statistics());
                 print_repoinfo(&repo.info());
             }
         }
+        Arguments::Stats {
+            repo_path
+        } => {
+            let mut repo = try!(open_repository(&repo_path, false));
+            print_repostats(&repo.statistics());
+        }
         Arguments::Mount {
             repo_path,
             backup_name,

src/util/compression.rs

@@ -34,7 +34,7 @@ quick_error!{
     }
 }
 
-#[derive(Clone, Debug, Copy, Eq, PartialEq)]
+#[derive(Clone, Debug, Copy, Eq, PartialEq, Hash)]
 pub enum CompressionMethod {
     Deflate, // Standardized
     Brotli, // Good speed and ratio
@@ -49,7 +49,7 @@ serde_impl!(CompressionMethod(u8) {
 });
 
 
-#[derive(Clone, Debug, Eq, PartialEq)]
+#[derive(Clone, Debug, Eq, PartialEq, Hash)]
 pub struct Compression {
     method: CompressionMethod,
     level: u8

src/util/hash.rs

@@ -105,7 +105,7 @@ impl<'a> Deserialize<'a> for Hash {
 }
 
 
-#[derive(Debug, Clone, Copy, Eq, PartialEq)]
+#[derive(Debug, Clone, Copy, Eq, PartialEq, Hash)]
 pub enum HashMethod {
     Blake2,
     Murmur3