Use raw CLI inputs for random seed instead of derived inputs and use …

…proper hashing algorithm

Signed-off-by: Alex Saveau <[email protected]>

README.md

@@ -24,7 +24,7 @@ Generate a reproducibly random tree in the current directory with *approximately
 ```console
 $ ftzz ./simple -n 1M
 About 1,000,000 files will be generated in approximately 1,000 directories distributed across a tree of maximum depth 5 where each directory contains approximately 4 other directories.
-Created 1,002,953 files across 1,193 directories.
+Created 1,003,229 files across 1,259 directories.
 
 ```
 
@@ -33,7 +33,7 @@ Generate *exactly* 1 million files:
 ```console
 $ ftzz ./exact -en 1M
 Exactly 1,000,000 files will be generated in approximately 1,000 directories distributed across a tree of maximum depth 5 where each directory contains approximately 4 other directories.
-Created 1,000,000 files across 1,193 directories.
+Created 1,000,000 files across 1,259 directories.
 
 ```
 
@@ -42,7 +42,7 @@ Generate ~10_000 files with ~1 MB of random data spread across them:
 ```console
 $ ftzz ./with_data -n 10K -b 1M
 About 10,000 files will be generated in approximately 1,000 directories distributed across a tree of maximum depth 5 where each directory contains approximately 4 other directories. Each file will contain approximately 100 bytes of random data.
-Created 9,652 files (960.1 KB) across 930 directories.
+Created 9,312 files (924.6 KB) across 1,570 directories.
 
 ```
 
@@ -53,11 +53,11 @@ change the starting seed:
 ```console
 $ ftzz ./unseeded -n 100
 About 100 files will be generated in approximately 100 directories distributed across a tree of maximum depth 5 where each directory contains approximately 3 other directories.
-Created 32 files across 214 directories.
+Created 45 files across 198 directories.
 
 $ ftzz ./seeded -n 100 --seed 42 # Or $RANDOM
 About 100 files will be generated in approximately 100 directories distributed across a tree of maximum depth 5 where each directory contains approximately 3 other directories.
-Created 78 files across 59 directories.
+Created 83 files across 110 directories.
 
 ```
 

api.golden

@@ -166,8 +166,18 @@ pub struct ftzz::NumFilesWithRatio
 impl ftzz::NumFilesWithRatio
 pub fn ftzz::NumFilesWithRatio::from_num_files(num_files: core::num::nonzero::NonZeroU64) -> Self
 pub fn ftzz::NumFilesWithRatio::new(num_files: core::num::nonzero::NonZeroU64, file_to_dir_ratio: core::num::nonzero::NonZeroU64) -> core::result::Result<Self, ftzz::NumFilesWithRatioError>
+impl core::clone::Clone for ftzz::NumFilesWithRatio
+pub fn ftzz::NumFilesWithRatio::clone(&self) -> ftzz::NumFilesWithRatio
+impl core::cmp::Eq for ftzz::NumFilesWithRatio
+impl core::cmp::PartialEq for ftzz::NumFilesWithRatio
+pub fn ftzz::NumFilesWithRatio::eq(&self, other: &ftzz::NumFilesWithRatio) -> bool
 impl core::fmt::Debug for ftzz::NumFilesWithRatio
 pub fn ftzz::NumFilesWithRatio::fmt(&self, f: &mut core::fmt::Formatter<'_>) -> core::fmt::Result
+impl core::hash::Hash for ftzz::NumFilesWithRatio
+pub fn ftzz::NumFilesWithRatio::hash<__H: core::hash::Hasher>(&self, state: &mut __H)
+impl core::marker::Copy for ftzz::NumFilesWithRatio
+impl core::marker::StructuralEq for ftzz::NumFilesWithRatio
+impl core::marker::StructuralPartialEq for ftzz::NumFilesWithRatio
 impl core::marker::Send for ftzz::NumFilesWithRatio
 impl core::marker::Sync for ftzz::NumFilesWithRatio
 impl core::marker::Unpin for ftzz::NumFilesWithRatio
@@ -181,6 +191,10 @@ pub fn ftzz::NumFilesWithRatio::try_from(value: U) -> core::result::Result<T, <T
 impl<T, U> core::convert::TryInto<U> for ftzz::NumFilesWithRatio where U: core::convert::TryFrom<T>
 pub type ftzz::NumFilesWithRatio::Error = <U as core::convert::TryFrom<T>>::Error
 pub fn ftzz::NumFilesWithRatio::try_into(self) -> core::result::Result<U, <U as core::convert::TryFrom<T>>::Error>
+impl<T> alloc::borrow::ToOwned for ftzz::NumFilesWithRatio where T: core::clone::Clone
+pub type ftzz::NumFilesWithRatio::Owned = T
+pub fn ftzz::NumFilesWithRatio::clone_into(&self, target: &mut T)
+pub fn ftzz::NumFilesWithRatio::to_owned(&self) -> T
 impl<T> core::any::Any for ftzz::NumFilesWithRatio where T: 'static + core::marker::Sized
 pub fn ftzz::NumFilesWithRatio::type_id(&self) -> core::any::TypeId
 impl<T> core::borrow::Borrow<T> for ftzz::NumFilesWithRatio where T: core::marker::Sized

src/generator.rs

@@ -8,6 +8,7 @@ use std::{
     cmp::max,
     fmt::Write,
     fs::create_dir_all,
+    hash::{DefaultHasher, Hash, Hasher},
     num::{NonZeroU64, NonZeroUsize},
     path::PathBuf,
     process::ExitCode,
@@ -38,7 +39,7 @@ pub enum Error {
     RuntimeCreation,
 }
 
-#[derive(Debug)]
+#[derive(Copy, Clone, Eq, PartialEq, Hash, Debug)]
 pub struct NumFilesWithRatio {
     num_files: NonZeroU64,
     file_to_dir_ratio: NonZeroU64,
@@ -246,7 +247,11 @@ fn validated_options(
         bytes_per_file,
         dirs_per_dir,
         max_depth,
-        seed,
+        seed: {
+            let mut hasher = DefaultHasher::new();
+            (num_files_with_ratio, max_depth, seed).hash(&mut hasher);
+            hasher.finish()
+        },
         human_info: HumanInfo {
             dirs_per_dir: dirs_per_dir.round() as usize,
             total_dirs: num_dirs.round() as usize,
@@ -405,12 +410,7 @@ async fn run_generator_async(
     let bytes = NonZeroU64::new(bytes);
     let dynamic = DynamicGenerator {
         num_dirs_distr: truncatable_normal(dirs_per_dir),
-        random: {
-            let seed = ((files.get().wrapping_add(max_depth.into()) as f64 * dirs_per_dir) as u64)
-                .wrapping_add(seed);
-            log!(Level::Debug, "Starting seed: {seed}");
-            Xoshiro256PlusPlus::seed_from_u64(seed)
-        },
+        random: Xoshiro256PlusPlus::seed_from_u64(seed),
 
         bytes: bytes.map(|_| GeneratorBytes {
             num_bytes_distr: truncatable_normal(bytes_per_file),

testdata/cmds/examples.md

@@ -3,7 +3,7 @@
 ```console
 $ ftzz -n 1K -b 100 --files-exact --bytes-exact all-exact-explicit
 Exactly 1,000 files will be generated in approximately 1,000 directories distributed across a tree of maximum depth 5 where each directory contains approximately 4 other directories. Each file will contain approximately 0 bytes of random data totaling exactly 100 bytes.
-Created 1,000 files (100 B) across 506 directories.
+Created 1,000 files (100 B) across 983 directories.
 
 ```
 
@@ -21,7 +21,7 @@ Info output:
 ```console
 $ ftzz -vv -n 1K verbose
 About 1,000 files will be generated in approximately 1,000 directories distributed across a tree of maximum depth 5 where each directory contains approximately 4 other directories.
-INFO  [ftzz::generator] Starting config: Configuration { root_dir: "verbose", files: 1000, bytes: 0, files_exact: false, bytes_exact: false, fill_byte: None, dirs_per_dir: 3.9810717055349727, bytes_per_file: 0.0, max_depth: 5, seed: 0, human_info: HumanInfo { dirs_per_dir: 4, total_dirs: 1000, bytes_per_files: 0 } }
-Created 835 files across 476 directories.
+INFO  [ftzz::generator] Starting config: Configuration { root_dir: "verbose", files: 1000, bytes: 0, files_exact: false, bytes_exact: false, fill_byte: None, dirs_per_dir: 3.9810717055349727, bytes_per_file: 0.0, max_depth: 5, seed: 2268264718836998093, human_info: HumanInfo { dirs_per_dir: 4, total_dirs: 1000, bytes_per_files: 0 } }
+Created 578 files across 1,033 directories.
 
 ```

testdata/cmds/numbers.md

@@ -3,7 +3,7 @@ Plain nums:
 ```console
 $ ftzz --files 1000 --max-depth 10 --ftd-ratio 100 plain
 About 1,000 files will be generated in approximately 10 directories distributed across a tree of maximum depth 10 where each directory contains approximately 1 other directory.
-Created 994 files across 92 directories.
+Created 996 files across 42 directories.
 
 ```
 
@@ -12,7 +12,7 @@ SI numbers:
 ```console
 $ ftzz --files 1K --max-depth 0.01K --ftd-ratio 0.0001M si
 About 1,000 files will be generated in approximately 10 directories distributed across a tree of maximum depth 10 where each directory contains approximately 1 other directory.
-Created 994 files across 92 directories.
+Created 996 files across 42 directories.
 
 ```
 
@@ -21,7 +21,7 @@ Commas:
 ```console
 $ ftzz --files 1,000 --max-depth 0,010 --ftd-ratio 1,0,0 commas
 About 1,000 files will be generated in approximately 10 directories distributed across a tree of maximum depth 10 where each directory contains approximately 1 other directory.
-Created 994 files across 92 directories.
+Created 996 files across 42 directories.
 
 ```
 
@@ -30,6 +30,6 @@ Underscores:
 ```console
 $ ftzz --files 1_000 --max-depth 1_0 --ftd-ratio 1_0_0 underscores
 About 1,000 files will be generated in approximately 10 directories distributed across a tree of maximum depth 10 where each directory contains approximately 1 other directory.
-Created 994 files across 92 directories.
+Created 996 files across 42 directories.
 
 ```