feat: add env_stat/1 NIF to expose map_size and used bytes

native/elmdb_nif/Cargo.toml

@@ -10,4 +10,5 @@ crate-type = ["cdylib"]
 [dependencies]
 rustler = { version = "0.36.2", features = ["nif_version_2_16"] }
 lmdb = "0.8"
+lmdb-sys = "0.8"
 lazy_static = "1.4"

native/elmdb_nif/src/lib.rs

@@ -28,6 +28,7 @@ use rustler::{Env, Term, NifResult, Error, Encoder, ResourceArc};
 use rustler::types::binary::Binary;
 use rustler::types::binary::OwnedBinary;
 use std::collections::{HashMap, HashSet, VecDeque};
+use std::mem;
 use std::sync::{Arc, Mutex};
 use std::path::Path;
 use lmdb::{Environment, EnvironmentFlags, Database, DatabaseFlags, Transaction, WriteFlags, Cursor};
@@ -1598,15 +1599,45 @@ struct DbOptions {
 #[rustler::nif]
 fn env_status<'a>(env: Env<'a>, env_handle: ResourceArc<LmdbEnv>) -> NifResult<Term<'a>> {
     let closed = env_handle.is_closed().map_err(|_| Error::BadArg)?;
-    
+
     let ref_count = {
         let ref_count = env_handle.ref_count.lock().map_err(|_| Error::BadArg)?;
         *ref_count
     };
-    
+
     Ok((atoms::ok(), closed, ref_count, env_handle.path.clone()).encode(env))
 }
 
+/// Returns the configured map_size and the approximate number of bytes currently
+/// used by the environment.
+///
+/// Used bytes are estimated as `(last_pgno + 1) * page_size`, which counts every
+/// page that has ever been allocated (including free-list pages).  The value is
+/// therefore an upper bound on live data size, but it is the right metric for
+/// deciding how close the environment is to hitting the map_size limit.
+///
+/// `lmdb 0.8` does not expose `mdb_env_info` in its safe API, so we call it
+/// through the `lmdb-sys` FFI layer directly.
+#[rustler::nif]
+fn env_stat<'a>(env: Env<'a>, env_handle: ResourceArc<LmdbEnv>) -> NifResult<Term<'a>> {
+    let (lmdb_env, _gen) = env_handle.ensure_open().map_err(|_| Error::BadArg)?;
+    let stat = lmdb_env.stat().map_err(|_| Error::BadArg)?;
+    let page_size = stat.page_size() as usize;
+
+    // mdb_env_info is not wrapped by the `lmdb` crate, so call it directly.
+    let (map_size, last_pgno) = unsafe {
+        let mut info: lmdb_sys::MDB_envinfo = mem::zeroed();
+        let rc = lmdb_sys::mdb_env_info(lmdb_env.env(), &mut info);
+        if rc != 0 {
+            return Err(Error::BadArg);
+        }
+        (info.me_mapsize, info.me_last_pgno)
+    };
+
+    let used_bytes = (last_pgno + 1) * page_size;
+    Ok((atoms::ok(), map_size, used_bytes).encode(env))
+}
+
 // Initialize the NIF module
 // explicit fuctions are deprecated but here is a list
 // [env_open, env_close, env_close_by_name, db_open, db_close, put, put_batch, get, list, flush, env_status]

src/elmdb.erl

@@ -9,7 +9,7 @@
 -module(elmdb).
 
 %% Environment management
--export([env_open/2, env_sync/1, env_close/1, env_close_by_name/1, env_status/1]).
+-export([env_open/2, env_sync/1, env_close/1, env_close_by_name/1, env_status/1, env_stat/1]).
 
 %% Database operations
 -export([db_open/2, db_close/1]).
@@ -119,6 +119,27 @@ env_close_by_name(_Path) ->
 env_status(_Env) ->
     erlang:nif_error(nif_not_loaded).
 
+%% @doc Get size statistics for an LMDB environment.
+%%
+%% Returns the configured map_size limit and an estimate of bytes currently in
+%% use.  Used bytes are calculated as `(last_pgno + 1) * page_size`, which
+%% counts every page ever allocated (including free-list pages) and therefore
+%% represents the high-water mark rather than live data only.  This is the
+%% correct value for gauging proximity to the map_size limit.
+%%
+%% Only meaningful for writable environments; calling this on a read-only
+%% environment (opened with no_lock) is safe but may return stale values if
+%% another process is the primary writer.
+%%
+%% @param Env Environment handle from env_open
+%% @returns {ok, MapSize, UsedBytes} where both values are non-negative integers
+%%          in bytes, or {error, badarg} if the environment cannot be opened.
+-spec env_stat(Env :: term()) ->
+    {ok, MapSize :: non_neg_integer(), UsedBytes :: non_neg_integer()} |
+    {error, badarg}.
+env_stat(_Env) ->
+    erlang:nif_error(nif_not_loaded).
+
 
 %%%===================================================================
 %%% Database Operations