feat: add k-mer index rebuild and compaction feature

This commit introduces a new `rebuild` CLI subcommand that reconstructs an existing multi-layer k-mer index into a compact, single-layer index. It implements a configurable filtering pipeline supporting min/max genome fraction/count and total count thresholds, parallel partition processing via `rayon`, and CLI progress tracking. The change also restructures module declarations across `obikindex` and `obikpartitionner` to integrate the new rebuild and layer-handling logic.

src/obikindex/src/lib.rs

@@ -5,6 +5,7 @@ mod distance;
 mod dump;
 mod index;
 mod merge;
+mod rebuild;
 
 pub use error::{OKIError, OKIResult};
 pub use distance::{DistanceMetric, DistanceOutput};

src/obikindex/src/rebuild.rs

@@ -0,0 +1,116 @@
+use std::fs;
+use std::io;
+use std::path::Path;
+use std::time::Duration;
+
+use indicatif::{ProgressBar, ProgressStyle};
+use obikpartitionner::{KmerFilter, KmerPartition, MergeMode};
+use obisys::{Reporter, Stage};
+use rayon::prelude::*;
+use tracing::info;
+
+use crate::error::{OKIError, OKIResult};
+use crate::index::KmerIndex;
+use crate::meta::IndexMeta;
+use crate::state::{IndexState, SENTINEL_INDEXED};
+
+impl KmerIndex {
+    /// Rebuild `src` into a new compact single-layer index at `output`.
+    ///
+    /// Only k-mers whose per-genome row passes every filter in `filters` are
+    /// written. If `filters` is empty every k-mer is kept (pure compaction).
+    ///
+    /// `mode` controls whether the output stores counts or presence/absence.
+    /// A count source may be rebuilt in presence mode; a presence source
+    /// cannot be rebuilt in count mode.
+    pub fn rebuild<P: AsRef<Path>>(
+        output: P,
+        src: &KmerIndex,
+        filters: &[Box<dyn KmerFilter>],
+        mode: MergeMode,
+        force: bool,
+        rep: &mut Reporter,
+    ) -> OKIResult<Self> {
+        let output = output.as_ref();
+
+        if src.state() != IndexState::Indexed {
+            return Err(OKIError::NotIndexed(src.root_path.clone()));
+        }
+
+        if mode == MergeMode::Count && !src.meta.config.with_counts {
+            return Err(OKIError::InvalidInput(
+                "cannot rebuild in count mode from a presence-only source index".into(),
+            ));
+        }
+
+        if output.exists() {
+            if force {
+                fs::remove_dir_all(output)?;
+            } else {
+                return Err(OKIError::Io(io::Error::new(
+                    io::ErrorKind::AlreadyExists,
+                    format!("{}: output directory already exists", output.display()),
+                )));
+            }
+        }
+
+        // ── Create output directory + metadata ────────────────────────────────
+        fs::create_dir_all(output)?;
+        let mut meta = IndexMeta::new(src.meta.config.clone());
+        meta.config.with_counts = mode == MergeMode::Count;
+        meta.genomes = src.meta.genomes.clone();
+        meta.write(output)?;
+
+        let n_genomes = src.meta.genomes.len();
+        let n_partitions = src.partition.n_partitions();
+
+        // ── Create an empty destination KmerPartition ─────────────────────────
+        // Create the partitions/ subdirectory so KmerPartition::open_with_config works.
+        fs::create_dir_all(output.join(obikpartitionner::PARTITIONS_SUBDIR))?;
+        let dst_partition = KmerPartition::open_with_config(
+            output,
+            meta.config.kmer_size,
+            meta.config.minimizer_size,
+            meta.config.n_bits,
+        )?;
+
+        info!(
+            "rebuild: {} partition(s), {} genome(s), mode={:?}",
+            n_partitions, n_genomes, mode,
+        );
+
+        let t = Stage::start("rebuild");
+        let pb = ProgressBar::new(n_partitions as u64).with_style(
+            ProgressStyle::with_template("rebuild — [{bar:20}] {pos}/{len} | {msg}")
+                .unwrap()
+                .progress_chars("=> "),
+        );
+        pb.enable_steady_tick(Duration::from_millis(100));
+
+        let src_partition = &src.partition;
+
+        let errors: Vec<obiskio::SKError> = (0..n_partitions)
+            .into_par_iter()
+            .filter_map(|i| {
+                let result = dst_partition
+                    .rebuild_partition(src_partition, i, filters, mode, n_genomes)
+                    .err();
+                pb.inc(1);
+                result
+            })
+            .collect();
+
+        pb.finish_and_clear();
+
+        if let Some(e) = errors.into_iter().next() {
+            return Err(OKIError::Partition(e));
+        }
+
+        rep.push(t.stop());
+
+        // Write SENTINEL_INDEXED — output is ready to use.
+        fs::File::create(output.join(SENTINEL_INDEXED))?;
+
+        KmerIndex::open(output)
+    }
+}