Add size hints for Datman Backup on dir trees

datman/src/backup.rs

@@ -7,6 +7,7 @@ use dashmap::DashSet;
 use eyre::{bail, ensure, eyre, Context, ContextCompat};
 use indicatif::ProgressStyle;
 use patricia_tree::PatriciaMap;
+use std::cmp::max;
 use std::collections::{BTreeMap, BTreeSet, HashMap};
 use std::io::Write;
 use std::path::PathBuf;
@@ -250,6 +251,11 @@ async fn scan_dir_sources(
             let options = options.clone();
             joinset.spawn_blocking(move || -> eyre::Result<DirSourcePrep> {
                 let scan_entry_map = scan::scan(&path, &ignore).context("Failed to scan")?;
+                info!(
+                    "size estimate for {:?} (full scan): {}",
+                    path,
+                    summarise_scan_entry_map_size(&scan_entry_map)
+                );
 
                 // TODO This whole section is messy.
                 //      Maybe we should consider combining prepopulate_unmodified and limit_scan_entry_map_to_size
@@ -275,6 +281,12 @@ async fn scan_dir_sources(
                             bail!("bug: intersecting prepop and prune keys: {ix_keys:?}");
                         }
 
+                        info!(
+                            "size estimate for {:?} (differential): {}",
+                            path,
+                            summarise_scan_entry_map_size(&pruned)
+                        );
+
                         (cfm, pruned, prepopulated)
                     } else {
                         (
@@ -285,10 +297,18 @@ async fn scan_dir_sources(
                     };
 
                 let pruned_scan_entry_map = match options.gradual {
-                    Some(gradual_size_limit) => limit_scan_entry_map_to_size(
-                        pruned_scan_entry_map,
-                        gradual_size_limit.as_u64(),
-                    ),
+                    Some(gradual_size_limit) => {
+                        let limited = limit_scan_entry_map_to_size(
+                            pruned_scan_entry_map,
+                            gradual_size_limit.as_u64(),
+                        );
+                        info!(
+                            "size estimate for {:?} (gradual/limited): {}",
+                            path,
+                            summarise_scan_entry_map_size(&limited)
+                        );
+                        limited
+                    }
                     None => pruned_scan_entry_map,
                 };
 
@@ -313,6 +333,24 @@ async fn scan_dir_sources(
     Ok(result)
 }
 
+fn summarise_scan_entry_map_size(scan_entry_map: &PatriciaMap<ScanEntry>) -> String {
+    let mut num_bytes = 0u64;
+
+    for (_, entry) in scan_entry_map.iter() {
+        num_bytes += match entry {
+            ScanEntry::NormalFile { size, .. } => max(*size, 4096),
+            _ => 4096,
+        };
+    }
+
+    let num_files = scan_entry_map.len();
+
+    format!(
+        "{num_files} files ({})",
+        ByteSize(num_bytes).to_string_as(true)
+    )
+}
+
 struct BackupDirSourcesReturn {
     pub chunkmaps: BTreeMap<BloblogId, IndexBloblogEntry>,
     pub dir_source_returns: Vec<(DirSourcePrep, PatriciaMap<Option<(RecursiveChunkRef, u64)>>)>,