Start of a huge refactoring

src/obikmer2/Cargo.toml

@@ -15,6 +15,7 @@ obisys        = { path = "../obisys" }
 obikindex     = { path = "../obikindex", default-features = false }
 obikindexer   = { path = "../obikindexer" }
 obikalgorithm = { path = "../obikalgorithm" }
+obikmerge     = { path = "../obikmerge" }
 obifastwrite  = { path = "../obifastwrite" }
 obiskbuilder  = { path = "../obiskbuilder" }
 clap          = { version = "4", features = ["derive"] }

src/obikmer2/src/cmd/merge/mod.rs

@@ -0,0 +1,104 @@
+use std::path::PathBuf;
+
+use clap::Args;
+use obikalgorithm::Algorithm;
+use obikindex::KmerIndex;
+use obikmerge::{Merge, MergeMode};
+use obisys::{Progress, Reporter, Stage, progress_bar};
+use tracing::info;
+
+#[derive(Args)]
+pub struct MergeArgs {
+    /// Source index directories to merge
+    #[arg(required = true)]
+    pub sources: Vec<PathBuf>,
+
+    /// Output index directory
+    #[arg(short, long)]
+    pub output: PathBuf,
+
+    /// Overwrite output directory if it already exists
+    #[arg(long, default_value_t = false)]
+    pub force: bool,
+
+    /// Force presence/absence mode even if all sources have count data
+    #[arg(long, default_value_t = false)]
+    pub force_presence: bool,
+
+    /// Disambiguate duplicate genome labels by appending .1, .2, … instead of erroring
+    #[arg(long, default_value_t = false)]
+    pub rename_duplicates: bool,
+}
+
+pub fn run(args: MergeArgs) {
+    let sources: Vec<KmerIndex> = args
+        .sources
+        .iter()
+        .map(|p| {
+            info!("opening source index: {}", p.display());
+            KmerIndex::open(p).unwrap_or_else(|e| {
+                eprintln!("error opening source index {}: {e}", p.display());
+                std::process::exit(1);
+            })
+        })
+        .collect();
+
+    // Auto-detect mode: count if all sources have count data, presence otherwise.
+    // --force-presence overrides to presence regardless.
+    let all_have_counts = sources.iter().all(|s| s.meta().config.with_counts);
+    let mode = if !args.force_presence && all_have_counts {
+        MergeMode::Count
+    } else {
+        MergeMode::Presence
+    };
+    info!(
+        "merge mode: {}",
+        if mode == MergeMode::Count {
+            "count"
+        } else {
+            "presence/absence"
+        }
+    );
+
+    let source_refs: Vec<&KmerIndex> = sources.iter().collect();
+
+    let n_genomes: usize = sources
+        .iter()
+        .map(|s| {
+            s.meta()
+                .genomes()
+                .unwrap_or_else(|e| {
+                    eprintln!("error reading index metadata: {e}");
+                    std::process::exit(1);
+                })
+                .len()
+        })
+        .sum();
+    info!(
+        "merging {} index(es), {} genome(s) total → {}",
+        sources.len(),
+        n_genomes,
+        args.output.display()
+    );
+
+    let mut rep = Reporter::new();
+    let t = Stage::start("merge");
+    let n_partitions = source_refs.first().map(|s| s.n_partitions()).unwrap_or(0);
+    let pb = progress_bar("merge", n_partitions as u64, "partitions");
+
+    let mut merge = Merge::new(&source_refs, &args.output, mode)
+        .force(args.force)
+        .rename_duplicates(args.rename_duplicates)
+        .on_progress(|_: Progress| pb.inc(1));
+
+    let dst = merge.run().unwrap_or_else(|e| {
+        eprintln!("error merging: {e}");
+        std::process::exit(1);
+    });
+    pb.finish_and_clear();
+    rep.push(t.stop());
+
+    info!("merge done — output at {}", dst.dir().display());
+    merge.reporter().print();
+    rep.print();
+}

src/obikmer2/src/cmd/mod.rs

@@ -1,2 +1,3 @@
 pub mod index;
+pub mod merge;
 pub mod superkmer;

src/obikmer2/src/main.rs

@@ -17,6 +17,8 @@ enum Commands {
     Index(cmd::index::IndexArgs),
     /// Extract super-k-mers from input sequences and scatter them by partition
     Superkmer(cmd::superkmer::SuperkmerArgs),
+    /// Merge multiple genome indexes into one
+    Merge(cmd::merge::MergeArgs),
 }
 
 fn main() {
@@ -31,5 +33,6 @@ fn main() {
     match cli.command {
         Commands::Index(args) => cmd::index::run(args),
         Commands::Superkmer(args) => cmd::superkmer::run(args),
+        Commands::Merge(args) => cmd::merge::run(args),
     }
 }