From 9a0abd5faeacb3968db3da716d5df99d587c3395 Mon Sep 17 00:00:00 2001
From: Manuel Holtgrewe <manuel.holtgrewe@bih-charite.de>
Date: Thu, 1 Jun 2023 08:59:59 +0200
Subject: [PATCH] feat: adding "db-utils dump-meta" command (#56)

---
 src/db_utils/cli/dump_meta.rs | 78 +++++++++++++++++++++++++++++++++++
 src/db_utils/cli/mod.rs       |  1 +
 src/main.rs                   |  5 +++
 3 files changed, 84 insertions(+)
 create mode 100644 src/db_utils/cli/dump_meta.rs

diff --git a/src/db_utils/cli/dump_meta.rs b/src/db_utils/cli/dump_meta.rs
new file mode 100644
index 00000000..7c5a4568
--- /dev/null
+++ b/src/db_utils/cli/dump_meta.rs
@@ -0,0 +1,78 @@
+//! Implementation of `db-utils dump-meta` sub command.
+
+use clap::Parser;
+
+use crate::common;
+
+/// Command line arguments for `db-utils dump-meta` sub command.
+#[derive(Parser, Debug, Clone)]
+#[command(about = "Dump the metadata columns", long_about = None)]
+pub struct Args {
+    /// Path to input directory.
+    #[arg(long)]
+    pub path_in: String,
+}
+
+/// Main entry point for `db-utils dump-meta` sub command.
+pub fn run(common: &common::cli::Args, args: &Args) -> Result<(), anyhow::Error> {
+    tracing::info!("Starting 'db-utils dump-meta' command");
+    tracing::info!("common = {:#?}", &common);
+    tracing::info!("args = {:#?}", &args);
+
+    tracing::info!("Opening input database");
+    // List all column families in database and check that the meta column exists.
+    let cf_names = rocksdb::DB::list_cf(&rocksdb::Options::default(), &args.path_in)?;
+    if !cf_names.iter().any(|s| s == "meta") {
+        anyhow::bail!("input database does not contain a column family named 'meta'");
+    }
+    // Open database for reading.
+    let db_read = rocksdb::DB::open_cf_for_read_only(
+        &rocksdb::Options::default(),
+        &args.path_in,
+        ["meta"],
+        false,
+    )?;
+
+    // Iterate over all values in the "meta" column family.
+    println!("#key\tvalue");
+    let mut count = 0;
+    let cf_read = db_read.cf_handle("meta").unwrap();
+    let mut iter = db_read.raw_iterator_cf(&cf_read);
+    iter.seek(b"");
+    while iter.valid() {
+        if let Some(iter_value) = iter.value() {
+            let iter_key = iter.key().unwrap();
+            println!(
+                "{}\t{}",
+                String::from_utf8(iter_key.to_vec())?,
+                String::from_utf8(iter_value.to_vec())?
+            );
+            iter.next();
+            count += 1;
+        } else {
+            break;
+        }
+    }
+    println!("#rows\t{}", count);
+
+    Ok(())
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+
+    use clap_verbosity_flag::Verbosity;
+
+    #[test]
+    fn smoke_test_dump() -> Result<(), anyhow::Error> {
+        let common = common::cli::Args {
+            verbose: Verbosity::new(1, 0),
+        };
+        let args = Args {
+            path_in: String::from("tests/dbsnp/example/dbsnp.brca1.vcf.bgz.db"),
+        };
+
+        run(&common, &args)
+    }
+}
diff --git a/src/db_utils/cli/mod.rs b/src/db_utils/cli/mod.rs
index 9f3dc652..8247c304 100644
--- a/src/db_utils/cli/mod.rs
+++ b/src/db_utils/cli/mod.rs
@@ -1,3 +1,4 @@
 //! CLI for the database utilities.
 
 pub mod copy;
+pub mod dump_meta;
diff --git a/src/main.rs b/src/main.rs
index fbbfe30e..91e9f7f3 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -154,6 +154,8 @@ struct DbUtils {
 enum DbUtilsCommands {
     /// "copy" sub command
     Copy(db_utils::cli::copy::Args),
+    /// "dump-meta" sub command
+    DumpMeta(db_utils::cli::dump_meta::Args),
 }
 
 pub fn main() -> Result<(), anyhow::Error> {
@@ -211,6 +213,9 @@ pub fn main() -> Result<(), anyhow::Error> {
             },
             Commands::DbUtils(args) => match &args.command {
                 DbUtilsCommands::Copy(args) => db_utils::cli::copy::run(&cli.common, args)?,
+                DbUtilsCommands::DumpMeta(args) => {
+                    db_utils::cli::dump_meta::run(&cli.common, args)?
+                }
             },
         }