Jordan Terrell 7 years ago
commit f06812a00d
  1. 55
      CHANGELOG.txt
  2. 4
      Cargo.toml
  3. 1
      LICENSE
  4. 4
      librocksdb-sys/Cargo.toml
  5. 34
      librocksdb-sys/build.rs
  6. 2
      librocksdb-sys/rocksdb
  7. 2
      librocksdb-sys/rocksdb_lib_sources.txt
  8. 84
      src/db.rs
  9. 64
      src/db_options.rs
  10. 27
      src/lib.rs
  11. 520
      src/merge_operator.rs
  12. 149
      src/slice_transform.rs
  13. 73
      tests/test_column_family.rs
  14. 47
      tests/test_iterator.rs
  15. 48
      tests/test_slice_transform.rs

@ -1,7 +1,42 @@
Changelog
=========
0.8.2 (2017-12-28
0.10.0 (2018-03-17)
~~~~~~~~~~~~~~~~~~~~
* Bump rocksdb to 5.11.3 (spacejam)
**New features**
* Link with system rocksdb and snappy libs through envvars (ozkriff)
**Breaking Changes**
* Fix reverse iteration from a given key (ongardie)
0.9.1 (2018-02-10)
~~~~~~~~~~~~~~~~~~~~
**New features**
* SliceTransform support (spacejam)
0.9.0 (2018-02-10)
~~~~~~~~~~~~~~~~~~~~
**New features**
* Allow creating iterators over prefixes (glittershark)
**Breaking Changes**
* Open cfs with options (garyttierney, rrichardson)
* Non-Associative merge ops (rrichardson)
0.8.3 (2018-02-10)
~~~~~~~~~~~~~~~~~~~~
* Bump rocksdb to 5.10.2 (ongardie)
* Add Send marker to Options (iSynaptic)
* Expose advise_random_on_open option (ongardie)
0.8.2 (2017-12-28)
~~~~~~~~~~~~~~~~~~~~
* Bump rocksdb to 5.7.1 (jquesnelle)
@ -24,33 +59,39 @@ Changelog
0.7 (2017-07-26)
~~~~~~~~~~~~~~~~~~~~
**Breaking Changes**
**Breaking Changes**
* Bumped rocksdb to 5.4.6 (derekdreery)
* Remove `use_direct_writes` now that `use_direct_io_for_flush_and_compaction` exists (derekdreery)
**New features**
**New features**
* ReadOptions is now public (rschmukler)
* Implement Clone and AsRef<str> for Error (daboross)
* Support for `seek_for_prev` (kaedroho)
* Support for DirectIO (kaedroho)
**Internal cleanups**
**Internal cleanups**
* Fixed race condition in tests (debris)
* Move tests to the default `tests` directory (vmx)
0.6.1 (2017-03-13)
~~~~~~~~~~~~~~~~~~
**New features**
**New features**
* Support for raw iterator access (kaedroho)
0.6 (2016-12-18)
~~~~~~~~~~~~~~~~~~~~
**Breaking changes**
* Comparator function now returns an Ordering (alexreg)
**New features**
* Compaction filter (tmccombs)
* Support for backups (alexreg)
@ -64,9 +105,9 @@ Changelog
* All errors changed to use a new `rocksdb::Error` type (kaedroho, alexreg)
* Removed `Options.set_filter_deletes` as it was removed in RocksDB (kaedroho)
* Renamed `add_merge_operator` to `set_merge_operator` and `add_comparator` to `set_comparator` (kaedroho)
**New features**
* Windows support (development by jsgf and arkpar. ported by kaedroho)
* The RocksDB library is now built at crate compile-time and statically linked with the resulting binary (development by jsgf and arkpar. ported by kaedroho)
* Cleaned up and improved coverage and tests of the ffi module (alexreg)

@ -1,7 +1,7 @@
[package]
name = "rocksdb"
description = "Rust wrapper for Facebook's RocksDB embeddable database"
version = "0.8.2"
version = "0.10.0"
authors = ["Tyler Neely <t@jujit.su>", "David Greenberg <dsg123456789@gmail.com>"]
license = "Apache-2.0"
keywords = ["database", "embedded", "LSM-tree", "persistence"]
@ -19,4 +19,4 @@ valgrind = []
[dependencies]
libc = "0.2"
librocksdb-sys = { path = "librocksdb-sys", version = "5.7.1" }
librocksdb-sys = { path = "librocksdb-sys", version = "5.11.3" }

@ -187,6 +187,7 @@
same "printed page" as the copyright notice for easier
identification within third-party archives.
Copyright 2014 Tyler Neely
Copyright 2015 Tyler Neely
Copyright 2016 Tyler Neely
Copyright 2017 Tyler Neely

@ -1,6 +1,6 @@
[package]
name = "librocksdb-sys"
version = "5.7.1"
version = "5.11.3"
authors = ["Karl Hobley <karlhobley10@gmail.com>", "Arkadiy Paronyan <arkadiy@ethcore.io>"]
license = "MIT/Apache-2.0/BSD-3-Clause"
description = "Native bindings to librocksdb"
@ -22,6 +22,6 @@ libc = "0.2"
const-cstr = "0.2"
[build-dependencies]
cc = { version = "1.0", features = ["parallel"] }
cc = { version = "^1.0", features = ["parallel"] }
make-cmd = "0.1"
bindgen = "0.29"

@ -29,10 +29,7 @@ fn fail_on_empty_directory(name: &str) {
}
}
fn build_rocksdb() {
println!("cargo:rerun-if-changed=build.rs");
println!("cargo:rerun-if-changed=rocksdb/");
fn bindgen_rocksdb() {
let bindings = bindgen::Builder::default()
.header("rocksdb/include/rocksdb/c.h")
.hide_type("max_align_t") // https://github.com/rust-lang-nursery/rust-bindgen/issues/550
@ -44,7 +41,9 @@ fn build_rocksdb() {
bindings
.write_to_file(out_path.join("bindings.rs"))
.expect("unable to write rocksdb bindings");
}
fn build_rocksdb() {
let mut config = cc::Build::new();
config.include("rocksdb/include/");
config.include("rocksdb/");
@ -148,9 +147,32 @@ fn build_snappy() {
config.compile("libsnappy.a");
}
fn try_to_find_and_link_lib(lib_name: &str) -> bool {
if let Ok(lib_dir) = env::var(&format!("{}_LIB_DIR", lib_name)) {
println!("cargo:rustc-link-search=native={}", lib_dir);
let mode = match env::var_os(&format!("{}_STATIC", lib_name)) {
Some(_) => "static",
None => "dylib",
};
println!("cargo:rustc-link-lib={}={}", mode, lib_name.to_lowercase());
return true;
}
false
}
fn main() {
println!("cargo:rerun-if-changed=build.rs");
println!("cargo:rerun-if-changed=rocksdb/");
println!("cargo:rerun-if-changed=snappy/");
fail_on_empty_directory("rocksdb");
fail_on_empty_directory("snappy");
build_rocksdb();
build_snappy();
bindgen_rocksdb();
if !try_to_find_and_link_lib("ROCKSDB") {
build_rocksdb();
}
if !try_to_find_and_link_lib("SNAPPY") {
build_snappy();
}
}

@ -1 +1 @@
Subproject commit acf935e40f9d6f4c3d13c7d310def7064c1f1c95
Subproject commit dbd8fa09b823826dd2a30bc119dad7a6fa9a4c6d

File diff suppressed because one or more lines are too long

@ -14,7 +14,7 @@
//
use {DB, Error, Options, WriteOptions, ColumnFamily};
use {DB, Error, Options, WriteOptions, ColumnFamily, ColumnFamilyDescriptor};
use ffi;
use ffi_util::opt_bytes_to_ptr;
@ -466,10 +466,13 @@ impl DBIterator {
self.raw.seek_to_last();
self.direction = Direction::Reverse;
}
IteratorMode::From(key, dir) => {
// TODO: Should use seek_for_prev when reversing
IteratorMode::From(key, Direction::Forward) => {
self.raw.seek(key);
self.direction = dir;
self.direction = Direction::Forward;
}
IteratorMode::From(key, Direction::Reverse) => {
self.raw.seek_for_prev(key);
self.direction = Direction::Reverse;
}
};
@ -572,6 +575,16 @@ impl<'a> Drop for Snapshot<'a> {
}
}
impl ColumnFamilyDescriptor {
// Create a new column family descriptor with the specified name and options.
pub fn new<S>(name: S, options: Options) -> Self where S: Into<String> {
ColumnFamilyDescriptor {
name: name.into(),
options
}
}
}
impl DB {
/// Open a database with default options.
pub fn open_default<P: AsRef<Path>>(path: P) -> Result<DB, Error> {
@ -585,14 +598,17 @@ impl DB {
DB::open_cf(opts, path, &[])
}
/// Open a database with specified options and column family.
///
/// A column family must be created first by calling `DB::create_cf`.
/// Open a database with the given database options and column family names.
///
/// # Panics
///
/// * Panics if the column family doesn't exist.
/// Column families opened using this function will be created with default `Options`.
pub fn open_cf<P: AsRef<Path>>(opts: &Options, path: P, cfs: &[&str]) -> Result<DB, Error> {
let cfs_v = cfs.to_vec().iter().map(|name| ColumnFamilyDescriptor::new(*name, Options::default())).collect();
DB::open_cf_descriptors(opts, path, cfs_v)
}
/// Open a database with the given database options and column family names/options.
pub fn open_cf_descriptors<P: AsRef<Path>>(opts: &Options, path: P, cfs: Vec<ColumnFamilyDescriptor>) -> Result<DB, Error> {
let path = path.as_ref();
let cpath = match CString::new(path.to_string_lossy().as_bytes()) {
Ok(c) => c,
@ -621,17 +637,19 @@ impl DB {
db = ffi_try!(ffi::rocksdb_open(opts.inner, cpath.as_ptr() as *const _,));
}
} else {
let mut cfs_v = cfs.to_vec();
let mut cfs_v = cfs;
// Always open the default column family.
if !cfs_v.contains(&"default") {
cfs_v.push("default");
if !cfs_v.iter().any(|cf| cf.name == "default") {
cfs_v.push(ColumnFamilyDescriptor {
name: String::from("default"),
options: Options::default()
});
}
// We need to store our CStrings in an intermediate vector
// so that their pointers remain valid.
let c_cfs: Vec<CString> = cfs_v
.iter()
.map(|cf| CString::new(cf.as_bytes()).unwrap())
.map(|cf| CString::new(cf.name.as_bytes()).unwrap())
.collect();
let mut cfnames: Vec<_> = c_cfs.iter().map(|cf| cf.as_ptr()).collect();
@ -639,10 +657,8 @@ impl DB {
// These handles will be populated by DB.
let mut cfhandles: Vec<_> = cfs_v.iter().map(|_| ptr::null_mut()).collect();
// TODO(tyler) allow options to be passed in.
let mut cfopts: Vec<_> = cfs_v
.iter()
.map(|_| unsafe { ffi::rocksdb_options_create() as *const _ })
let mut cfopts: Vec<_> = cfs_v.iter()
.map(|cf| cf.options.inner as *const _)
.collect();
unsafe {
@ -666,7 +682,7 @@ impl DB {
}
for (n, h) in cfs_v.iter().zip(cfhandles) {
cf_map.insert(n.to_string(), ColumnFamily { inner: h });
cf_map.insert(n.name.clone(), ColumnFamily { inner: h });
}
}
@ -872,6 +888,12 @@ impl DB {
DBIterator::new(self, &opts, mode)
}
pub fn prefix_iterator<'a>(&self, prefix: &'a [u8]) -> DBIterator {
let mut opts = ReadOptions::default();
opts.set_prefix_same_as_start(true);
DBIterator::new(self, &opts, IteratorMode::From(prefix, Direction::Forward))
}
pub fn iterator_cf(
&self,
cf_handle: ColumnFamily,
@ -881,6 +903,16 @@ impl DB {
DBIterator::new_cf(self, cf_handle, &opts, mode)
}
pub fn prefix_iterator_cf<'a>(
&self,
cf_handle: ColumnFamily,
prefix: &'a [u8]
) -> Result<DBIterator, Error> {
let mut opts = ReadOptions::default();
opts.set_prefix_same_as_start(true);
DBIterator::new_cf(self, cf_handle, &opts, IteratorMode::From(prefix, Direction::Forward))
}
pub fn raw_iterator(&self) -> DBRawIterator {
let opts = ReadOptions::default();
DBRawIterator::new(self, &opts)
@ -1202,6 +1234,18 @@ impl ReadOptions {
);
}
}
pub fn set_prefix_same_as_start(&mut self, v: bool) {
unsafe {
ffi::rocksdb_readoptions_set_prefix_same_as_start(self.inner, v as c_uchar)
}
}
pub fn set_total_order_seek(&mut self, v:bool) {
unsafe {
ffi::rocksdb_readoptions_set_total_order_seek(self.inner, v as c_uchar)
}
}
}
impl Default for ReadOptions {

@ -11,20 +11,20 @@
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
//
use std::ffi::{CStr, CString};
use std::mem;
use libc::{self, c_int, c_uchar, c_uint, c_void, size_t, uint64_t};
use {BlockBasedOptions, DBCompactionStyle, DBCompressionType, DBRecoveryMode, Options,
WriteOptions};
use ffi;
use {BlockBasedOptions, DBCompactionStyle, DBCompressionType, DBRecoveryMode,
Options, WriteOptions};
use compaction_filter::{self, CompactionFilterCallback, CompactionFilterFn, filter_callback};
use comparator::{self, ComparatorCallback, CompareFn};
use ffi;
use libc::{self, c_int, c_uchar, c_uint, c_void, size_t, uint64_t};
use merge_operator::{self, MergeFn, MergeOperatorCallback, full_merge_callback,
partial_merge_callback};
use std::ffi::{CStr, CString};
use std::mem;
use slice_transform::SliceTransform;
pub fn new_cache(capacity: size_t) -> *mut ffi::rocksdb_cache_t {
unsafe { ffi::rocksdb_cache_create_lru(capacity) }
@ -149,6 +149,25 @@ impl Options {
}
}
/// If true, any column families that didn't exist when opening the database
/// will be created.
///
/// Default: `false`
///
/// # Example
///
/// ```
/// use rocksdb::Options;
///
/// let mut opts = Options::default();
/// opts.create_missing_column_families(true);
/// ```
pub fn create_missing_column_families(&mut self, create_missing_cfs: bool) {
unsafe {
ffi::rocksdb_options_set_create_missing_column_families(self.inner, create_missing_cfs as c_uchar);
}
}
/// Sets the compression algorithm that will be used for the bottommost level that
/// contain files. If level-compaction is used, this option will only affect
/// levels after base level.
@ -202,10 +221,14 @@ impl Options {
}
}
pub fn set_merge_operator(&mut self, name: &str, merge_fn: MergeFn) {
pub fn set_merge_operator(&mut self, name: &str,
full_merge_fn: MergeFn,
partial_merge_fn: Option<MergeFn>) {
let cb = Box::new(MergeOperatorCallback {
name: CString::new(name.as_bytes()).unwrap(),
merge_fn: merge_fn,
full_merge_fn: full_merge_fn,
partial_merge_fn: partial_merge_fn.unwrap_or(full_merge_fn),
});
unsafe {
@ -224,7 +247,7 @@ impl Options {
#[deprecated(since = "0.5.0",
note = "add_merge_operator has been renamed to set_merge_operator")]
pub fn add_merge_operator(&mut self, name: &str, merge_fn: MergeFn) {
self.set_merge_operator(name, merge_fn);
self.set_merge_operator(name, merge_fn, None);
}
/// Sets a compaction filter used to determine if entries should be kept, changed,
@ -280,6 +303,14 @@ impl Options {
}
}
pub fn set_prefix_extractor(&mut self, prefix_extractor: SliceTransform) {
unsafe {
ffi::rocksdb_options_set_prefix_extractor(
self.inner, prefix_extractor.inner
)
}
}
#[deprecated(since = "0.5.0", note = "add_comparator has been renamed to set_comparator")]
pub fn add_comparator(&mut self, name: &str, compare_fn: CompareFn) {
self.set_comparator(name, compare_fn);
@ -932,6 +963,17 @@ impl Options {
}
}
/// When set to true, reading SST files will opt out of the filesystem's
/// readahead. Setting this to false may improve sequential iteration
/// performance.
///
/// Default: `true`
pub fn set_advise_random_on_open(&mut self, advise: bool) {
unsafe {
ffi::rocksdb_options_set_advise_random_on_open(self.inner, advise as c_uchar)
}
}
/// Sets the number of levels for this database.
pub fn set_num_levels(&mut self, n: c_int) {
unsafe {

@ -30,6 +30,21 @@
//! db.delete(b"my key").unwrap();
//! ```
//!
//! Opening a database and a single column family with custom options:
//!
//! ```
//! use rocksdb::{DB, ColumnFamilyDescriptor, Options};
//! let mut cf_opts = Options::default();
//! cf_opts.set_max_write_buffer_number(16);
//! let cf = ColumnFamilyDescriptor::new("cf1", cf_opts);
//!
//! let mut db_opts = Options::default();
//! db_opts.create_missing_column_families(true);
//! db_opts.create_if_missing(true);
//!
//! let db = DB::open_cf_descriptors(&db_opts, "path/for/rocksdb/storage_with_cfs", vec![cf]).unwrap();
//! ```
//!
extern crate libc;
extern crate librocksdb_sys as ffi;
@ -43,12 +58,15 @@ pub mod merge_operator;
pub mod compaction_filter;
mod db;
mod db_options;
mod slice_transform;
pub use compaction_filter::Decision as CompactionDecision;
pub use db::{DBCompactionStyle, DBCompressionType, DBIterator, DBRawIterator, DBRecoveryMode,
DBVector, ReadOptions, Direction, IteratorMode, Snapshot, WriteBatch,
new_bloom_filter};
pub use slice_transform::SliceTransform;
pub use merge_operator::MergeOperands;
use std::collections::BTreeMap;
use std::error;
@ -64,6 +82,14 @@ pub struct DB {
path: PathBuf,
}
/// A descriptor for a RocksDB column family.
///
/// A description of the column family, containing the name and `Options`.
pub struct ColumnFamilyDescriptor {
name: String,
options: Options,
}
/// A simple wrapper round a string, used for errors reported from
/// ffi calls.
#[derive(Debug, Clone, PartialEq)]
@ -173,6 +199,7 @@ pub struct WriteOptions {
inner: *mut ffi::rocksdb_writeoptions_t,
}
/// An opaque type used to represent a column family. Returned from some functions, and used
/// in others
#[derive(Copy, Clone)]

@ -21,7 +21,7 @@
//! fn concat_merge(new_key: &[u8],
//! existing_val: Option<&[u8]>,
//! operands: &mut MergeOperands)
//! -> Vec<u8> {
//! -> Option<Vec<u8>> {
//!
//! let mut result: Vec<u8> = Vec::with_capacity(operands.size_hint().0);
//! existing_val.map(|v| {
@ -34,14 +34,14 @@
//! result.push(*e)
//! }
//! }
//! result
//! Some(result)
//! }
//!
//! fn main() {
//! let path = "path/to/rocksdb";
//! let mut opts = Options::default();
//! opts.create_if_missing(true);
//! opts.add_merge_operator("test operator", concat_merge);
//! opts.set_merge_operator("test operator", concat_merge, None);
//! let db = DB::open(&opts, path).unwrap();
//! let p = db.put(b"k1", b"a");
//! db.merge(b"k1", b"b");
@ -60,183 +60,397 @@ use std::mem;
use std::ptr;
use std::slice;
pub type MergeFn = fn(&[u8], Option<&[u8]>, &mut MergeOperands) -> Vec<u8>;
pub type MergeFn = fn(&[u8], Option<&[u8]>, &mut MergeOperands) -> Option<Vec<u8>>;
pub struct MergeOperatorCallback {
pub name: CString,
pub merge_fn: MergeFn,
pub name: CString,
pub full_merge_fn: MergeFn,
pub partial_merge_fn: MergeFn,
}
pub unsafe extern "C" fn destructor_callback(raw_cb: *mut c_void) {
let _: Box<MergeOperatorCallback> = mem::transmute(raw_cb);
let _: Box<MergeOperatorCallback> = mem::transmute(raw_cb);
}
pub unsafe extern "C" fn name_callback(raw_cb: *mut c_void) -> *const c_char {
let cb = &mut *(raw_cb as *mut MergeOperatorCallback);
cb.name.as_ptr()
let cb = &mut *(raw_cb as *mut MergeOperatorCallback);
cb.name.as_ptr()
}
pub unsafe extern "C" fn full_merge_callback(
raw_cb: *mut c_void,
raw_key: *const c_char,
key_len: size_t,
existing_value: *const c_char,
existing_value_len: size_t,
operands_list: *const *const c_char,
operands_list_len: *const size_t,
num_operands: c_int,
success: *mut u8,
new_value_length: *mut size_t,
) -> *mut c_char {
let cb = &mut *(raw_cb as *mut MergeOperatorCallback);
let operands = &mut MergeOperands::new(operands_list, operands_list_len, num_operands);
let key = slice::from_raw_parts(raw_key as *const u8, key_len as usize);
let oldval = slice::from_raw_parts(existing_value as *const u8, existing_value_len as usize);
let mut result = (cb.merge_fn)(key, Some(oldval), operands);
result.shrink_to_fit();
// TODO(tan) investigate zero-copy techniques to improve performance
let buf = libc::malloc(result.len() as size_t);
assert!(!buf.is_null());
*new_value_length = result.len() as size_t;
*success = 1 as u8;
ptr::copy(result.as_ptr() as *mut c_void, &mut *buf, result.len());
buf as *mut c_char
raw_cb: *mut c_void,
raw_key: *const c_char,
key_len: size_t,
existing_value: *const c_char,
existing_value_len: size_t,
operands_list: *const *const c_char,
operands_list_len: *const size_t,
num_operands: c_int,
success: *mut u8,
new_value_length: *mut size_t,
) -> *mut c_char {
let cb = &mut *(raw_cb as *mut MergeOperatorCallback);
let operands = &mut MergeOperands::new(operands_list, operands_list_len, num_operands);
let key = slice::from_raw_parts(raw_key as *const u8, key_len as usize);
let oldval =
if existing_value == ptr::null() {
None
} else {
Some(slice::from_raw_parts(existing_value as *const u8, existing_value_len as usize))
};
if let Some(mut result) = (cb.full_merge_fn)(key, oldval, operands) {
result.shrink_to_fit();
// TODO(tan) investigate zero-copy techniques to improve performance
let buf = libc::malloc(result.len() as size_t);
assert!(!buf.is_null());
*new_value_length = result.len() as size_t;
*success = 1 as u8;
ptr::copy(result.as_ptr() as *mut c_void, &mut *buf, result.len());
buf as *mut c_char
} else {
*success = 0 as u8;
ptr::null_mut() as *mut c_char
}
}
pub unsafe extern "C" fn partial_merge_callback(
raw_cb: *mut c_void,
raw_key: *const c_char,
key_len: size_t,
operands_list: *const *const c_char,
operands_list_len: *const size_t,
num_operands: c_int,
success: *mut u8,
new_value_length: *mut size_t,
) -> *mut c_char {
let cb = &mut *(raw_cb as *mut MergeOperatorCallback);
let operands = &mut MergeOperands::new(operands_list, operands_list_len, num_operands);
let key = slice::from_raw_parts(raw_key as *const u8, key_len as usize);
let mut result = (cb.merge_fn)(key, None, operands);
result.shrink_to_fit();
// TODO(tan) investigate zero-copy techniques to improve performance
let buf = libc::malloc(result.len() as size_t);
assert!(!buf.is_null());
*new_value_length = result.len() as size_t;
*success = 1 as u8;
ptr::copy(result.as_ptr() as *mut c_void, &mut *buf, result.len());
buf as *mut c_char
raw_cb: *mut c_void,
raw_key: *const c_char,
key_len: size_t,
operands_list: *const *const c_char,
operands_list_len: *const size_t,
num_operands: c_int,
success: *mut u8,
new_value_length: *mut size_t,
) -> *mut c_char {
let cb = &mut *(raw_cb as *mut MergeOperatorCallback);
let operands = &mut MergeOperands::new(operands_list, operands_list_len, num_operands);
let key = slice::from_raw_parts(raw_key as *const u8, key_len as usize);
if let Some(mut result) = (cb.partial_merge_fn)(key, None, operands) {
result.shrink_to_fit();
// TODO(tan) investigate zero-copy techniques to improve performance
let buf = libc::malloc(result.len() as size_t);
assert!(!buf.is_null());
*new_value_length = result.len() as size_t;
*success = 1 as u8;
ptr::copy(result.as_ptr() as *mut c_void, &mut *buf, result.len());
buf as *mut c_char
} else {
*success = 0 as u8;
ptr::null_mut::<c_char>()
}
}
pub struct MergeOperands {
operands_list: *const *const c_char,
operands_list_len: *const size_t,
num_operands: usize,
cursor: usize,
operands_list: *const *const c_char,
operands_list_len: *const size_t,
num_operands: usize,
cursor: usize,
}
impl MergeOperands {
fn new(
operands_list: *const *const c_char,
operands_list_len: *const size_t,
num_operands: c_int,
) -> MergeOperands {
assert!(num_operands >= 0);
MergeOperands {
operands_list: operands_list,
operands_list_len: operands_list_len,
num_operands: num_operands as usize,
cursor: 0,
}
}
fn new(
operands_list: *const *const c_char,
operands_list_len: *const size_t,
num_operands: c_int,
) -> MergeOperands {
assert!(num_operands >= 0);
MergeOperands {
operands_list: operands_list,
operands_list_len: operands_list_len,
num_operands: num_operands as usize,
cursor: 0,
}
}
}
impl<'a> Iterator for &'a mut MergeOperands {
type Item = &'a [u8];
fn next(&mut self) -> Option<&'a [u8]> {
if self.cursor == self.num_operands {
None
} else {
unsafe {
let base = self.operands_list as usize;
let base_len = self.operands_list_len as usize;
let spacing = mem::size_of::<*const *const u8>();
let spacing_len = mem::size_of::<*const size_t>();
let len_ptr = (base_len + (spacing_len * self.cursor)) as *const size_t;
let len = *len_ptr as usize;
let ptr = base + (spacing * self.cursor);
self.cursor += 1;
Some(mem::transmute(slice::from_raw_parts(
*(ptr as *const *const u8) as *const u8,
len,
)))
}
}
}
fn size_hint(&self) -> (usize, Option<usize>) {
let remaining = self.num_operands - self.cursor;
(remaining, Some(remaining))
}
type Item = &'a [u8];
fn next(&mut self) -> Option<&'a [u8]> {
if self.cursor == self.num_operands {
None
} else {
unsafe {
let base = self.operands_list as usize;
let base_len = self.operands_list_len as usize;
let spacing = mem::size_of::<*const *const u8>();
let spacing_len = mem::size_of::<*const size_t>();
let len_ptr = (base_len + (spacing_len * self.cursor)) as *const size_t;
let len = *len_ptr as usize;
let ptr = base + (spacing * self.cursor);
self.cursor += 1;
Some(mem::transmute(slice::from_raw_parts(
*(ptr as *const *const u8) as *const u8,
len,
)))
}
}
}
fn size_hint(&self) -> (usize, Option<usize>) {
let remaining = self.num_operands - self.cursor;
(remaining, Some(remaining))
}
}
#[cfg(test)]
#[allow(unused_variables)]
fn test_provided_merge(
new_key: &[u8],
existing_val: Option<&[u8]>,
operands: &mut MergeOperands,
) -> Vec<u8> {
let nops = operands.size_hint().0;
let mut result: Vec<u8> = Vec::with_capacity(nops);
if let Some(v) = existing_val {
for e in v {
result.push(*e);
}
}
for op in operands {
for e in op {
result.push(*e);
}
}
result
}
mod test {
use super::*;
fn test_provided_merge(
_new_key: &[u8],
existing_val: Option<&[u8]>,
operands: &mut MergeOperands,
) -> Option<Vec<u8>> {
let nops = operands.size_hint().0;
let mut result: Vec<u8> = Vec::with_capacity(nops);
if let Some(v) = existing_val {
for e in v {
result.push(*e);
}
}
for op in operands {
for e in op {
result.push(*e);
}
}
Some(result)
}
#[test]
fn mergetest() {
use {DB, Options};
let path = "_rust_rocksdb_mergetest";
let mut opts = Options::default();
opts.create_if_missing(true);
opts.set_merge_operator("test operator", test_provided_merge);
{
let db = DB::open(&opts, path).unwrap();
let p = db.put(b"k1", b"a");
assert!(p.is_ok());
let _ = db.merge(b"k1", b"b");
let _ = db.merge(b"k1", b"c");
let _ = db.merge(b"k1", b"d");
let _ = db.merge(b"k1", b"efg");
let m = db.merge(b"k1", b"h");
assert!(m.is_ok());
match db.get(b"k1") {
Ok(Some(value)) => {
match value.to_utf8() {
Some(v) => println!("retrieved utf8 value: {}", v),
None => println!("did not read valid utf-8 out of the db"),
}
}
Err(_) => println!("error reading value"),
_ => panic!("value not present"),
}
assert!(m.is_ok());
let r = db.get(b"k1");
assert!(r.unwrap().unwrap().to_utf8().unwrap() == "abcdefgh");
assert!(db.delete(b"k1").is_ok());
assert!(db.get(b"k1").unwrap().is_none());
}
assert!(DB::destroy(&opts, path).is_ok());
fn mergetest() {
use {DB, Options};
let path = "_rust_rocksdb_mergetest";
let mut opts = Options::default();
opts.create_if_missing(true);
opts.set_merge_operator("test operator", test_provided_merge, None);
{
let db = DB::open(&opts, path).unwrap();
let p = db.put(b"k1", b"a");
assert!(p.is_ok());
let _ = db.merge(b"k1", b"b");
let _ = db.merge(b"k1", b"c");
let _ = db.merge(b"k1", b"d");
let _ = db.merge(b"k1", b"efg");
let m = db.merge(b"k1", b"h");
assert!(m.is_ok());
match db.get(b"k1") {
Ok(Some(value)) => {
match value.to_utf8() {
Some(v) => println!("retrieved utf8 value: {}", v),
None => println!("did not read valid utf-8 out of the db"),
}
}
Err(_) => println!("error reading value"),
_ => panic!("value not present"),
}
assert!(m.is_ok());
let r = db.get(b"k1");
assert!(r.unwrap().unwrap().to_utf8().unwrap() == "abcdefgh");
assert!(db.delete(b"k1").is_ok());
assert!(db.get(b"k1").unwrap().is_none());
}
assert!(DB::destroy(&opts, path).is_ok());
}
unsafe fn to_slice<T: Sized>(p: &T) -> &[u8] {
::std::slice::from_raw_parts(
(p as *const T) as *const u8,
::std::mem::size_of::<T>(),
)
}
fn from_slice<T: Sized>(s: &[u8]) -> Option<&T> {
if ::std::mem::size_of::<T>() != s.len() {
println!("slice {:?} is len {}, but T is size {}", s, s.len(), ::std::mem::size_of::<T>());
None
} else {
unsafe {
Some(::std::mem::transmute(s.as_ptr()))
}
}
}
#[repr(packed)]
#[derive(Clone, Debug)]
struct ValueCounts {
num_a: u32,
num_b: u32,
num_c: u32,
num_d: u32,
}
fn test_counting_partial_merge(
_new_key: &[u8],
_existing_val: Option<&[u8]>,
operands: &mut MergeOperands,
) -> Option<Vec<u8>> {
let nops = operands.size_hint().0;
let mut result: Vec<u8> = Vec::with_capacity(nops);
for op in operands {
for e in op {
result.push(*e);
}
}
Some(result)
}
fn test_counting_full_merge(
_new_key: &[u8],
existing_val: Option<&[u8]>,
operands: &mut MergeOperands,
) -> Option<Vec<u8>> {
let mut counts : ValueCounts =
if let Some(v) = existing_val {
from_slice::<ValueCounts>(v).unwrap().clone()
} else {
ValueCounts {
num_a: 0,
num_b: 0,
num_c: 0,
num_d: 0 }
};
for op in operands {
for e in op {
match *e {
b'a' => counts.num_a += 1,
b'b' => counts.num_b += 1,
b'c' => counts.num_c += 1,
b'd' => counts.num_d += 1,
_ => {}
}
}
}
let slc = unsafe { to_slice(&counts) };
Some(slc.to_vec())
}
#[test]
fn counting_mergetest() {
use std::thread;
use std::sync::Arc;
use {DB, Options, DBCompactionStyle};
let path = "_rust_rocksdb_partial_mergetest";
let mut opts = Options::default();
opts.create_if_missing(true);
opts.set_compaction_style(DBCompactionStyle::Universal);
opts.set_min_write_buffer_number_to_merge(10);
opts.set_merge_operator("sort operator", test_counting_full_merge, Some(test_counting_partial_merge));
{
let db = Arc::new(DB::open(&opts, path).unwrap());
let _ = db.delete(b"k1");
let _ = db.delete(b"k2");
let _ = db.merge(b"k1", b"a");
let _ = db.merge(b"k1", b"b");
let _ = db.merge(b"k1", b"d");
let _ = db.merge(b"k1", b"a");
let _ = db.merge(b"k1", b"a");
let _ = db.merge(b"k1", b"efg");
for i in 0..500 {
let _ = db.merge(b"k2", b"c");
if i % 20 == 0 {
let _ = db.get(b"k2");
}
}
for i in 0..500 {
let _ = db.merge(b"k2", b"c");
if i % 20 == 0 {
let _ = db.get(b"k2");
}
}
db.compact_range(None, None);
let d1 = db.clone();
let d2 = db.clone();
let d3 = db.clone();
let h1 = thread::spawn(move || {
for i in 0..500 {
let _ = d1.merge(b"k2", b"c");
if i % 20 == 0 {
let _ = d1.get(b"k2");
}
}
for i in 0..500 {
let _ = d1.merge(b"k2", b"a");
if i % 20 == 0 {
let _ = d1.get(b"k2");
}
}
});
let h2 = thread::spawn(move || {
for i in 0..500 {
let _ = d2.merge(b"k2", b"b");
if i % 20 == 0 {
let _ = d2.get(b"k2");
}
}
for i in 0..500 {
let _ = d2.merge(b"k2", b"d");
if i % 20 == 0 {
let _ = d2.get(b"k2");
}
}
d2.compact_range(None, None);
});
h2.join().unwrap();
let h3 = thread::spawn(move || {
for i in 0..500 {
let _ = d3.merge(b"k2", b"a");
if i % 20 == 0 {
let _ = d3.get(b"k2");
}
}
for i in 0..500 {
let _ = d3.merge(b"k2", b"c");
if i % 20 == 0 {
let _ = d3.get(b"k2");
}
}
});
let m = db.merge(b"k1", b"b");
assert!(m.is_ok());
h3.join().unwrap();
h1.join().unwrap();
match db.get(b"k2") {
Ok(Some(value)) => {
match from_slice::<ValueCounts>(&*value) {
Some(v) => {
assert_eq!(v.num_a, 1000);
assert_eq!(v.num_b, 500);
assert_eq!(v.num_c, 2000);
assert_eq!(v.num_d, 500);
},
None => panic!("Failed to get ValueCounts from db"),
}
}
Err(e) => panic!("error reading value {:?}", e),
_ => panic!("value not present"),
}
match db.get(b"k1") {
Ok(Some(value)) => {
match from_slice::<ValueCounts>(&*value) {
Some(v) => {
assert_eq!(v.num_a, 3);
assert_eq!(v.num_b, 2);
assert_eq!(v.num_c, 0);
assert_eq!(v.num_d, 1);
},
None => panic!("Failed to get ValueCounts from db"),
}
}
Err(e) => panic!("error reading value {:?}", e),
_ => panic!("value not present"),
}
}
assert!(DB::destroy(&opts, path).is_ok());
}
}

@ -0,0 +1,149 @@
// Copyright 2018 Tyler Neely
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
use std::ffi::CString;
use std::mem;
use std::ptr;
use std::slice;
use libc::{self, c_char, c_void, size_t};
use ffi;
/// A SliceTranform is a generic pluggable way of transforming one string
/// to another. Its primary use-case is in configuring rocksdb
/// to store prefix blooms by setting prefix_extractor in
/// ColumnFamilyOptions.
pub struct SliceTransform {
pub inner: *mut ffi::rocksdb_slicetransform_t,
}
// NB we intentionally don't implement a Drop that passes
// through to rocksdb_slicetransform_destroy because
// this is currently only used (to my knowledge)
// by people passing it as a prefix extractor when
// opening a DB.
impl SliceTransform {
pub fn create(
name: &str,
transform_fn: TransformFn,
in_domain_fn: Option<InDomainFn>,
) -> SliceTransform{
let cb = Box::new(TransformCallback {
name: CString::new(name.as_bytes()).unwrap(),
transform_fn: transform_fn,
in_domain_fn: in_domain_fn,
});
let st = unsafe {
ffi::rocksdb_slicetransform_create(
mem::transmute(cb),
Some(slice_transform_destructor_callback),
Some(transform_callback),
// this is ugly, but I can't get the compiler
// not to barf with "expected fn pointer, found fn item"
// without this. sorry.
if let Some(_) = in_domain_fn {
Some(in_domain_callback)
} else {
None
},
// this None points to the deprecated InRange callback
None,
Some(slice_transform_name_callback),
)
};
SliceTransform {
inner: st
}
}
pub fn create_fixed_prefix(len: size_t) -> SliceTransform {
SliceTransform {
inner: unsafe {
ffi::rocksdb_slicetransform_create_fixed_prefix(len)
},
}
}
pub fn create_noop() -> SliceTransform {
SliceTransform {
inner: unsafe {
ffi::rocksdb_slicetransform_create_noop()
},
}
}
}
pub type TransformFn = fn(&[u8]) -> Vec<u8>;
pub type InDomainFn = fn(&[u8]) -> bool;
pub struct TransformCallback {
pub name: CString,
pub transform_fn: TransformFn,
pub in_domain_fn: Option<InDomainFn>,
}
pub unsafe extern "C" fn slice_transform_destructor_callback(
raw_cb: *mut c_void
) {
let transform: Box<TransformCallback> = mem::transmute(raw_cb);
drop(transform);
}
pub unsafe extern "C" fn slice_transform_name_callback(
raw_cb: *mut c_void
) -> *const c_char {
let cb = &mut *(raw_cb as *mut TransformCallback);
cb.name.as_ptr()
}
pub unsafe extern "C" fn transform_callback(
raw_cb: *mut c_void,
raw_key: *const c_char,
key_len: size_t,
dst_length: *mut size_t,
) -> *mut c_char {
let cb = &mut *(raw_cb as *mut TransformCallback);
let key = slice::from_raw_parts(raw_key as *const u8, key_len as usize);
let mut result = (cb.transform_fn)(key);
result.shrink_to_fit();
// copy the result into a C++ destroyable buffer
let buf = libc::malloc(result.len() as size_t);
assert!(!buf.is_null());
ptr::copy(result.as_ptr() as *mut c_void, &mut *buf, result.len());
*dst_length = result.len() as size_t;
buf as *mut c_char
}
pub unsafe extern "C" fn in_domain_callback(
raw_cb: *mut c_void,
raw_key: *const c_char,
key_len: size_t,
) -> u8 {
let cb = &mut *(raw_cb as *mut TransformCallback);
let key = slice::from_raw_parts(raw_key as *const u8, key_len as usize);
if (cb.in_domain_fn.unwrap())(key) {
1
} else {
0
}
}

@ -14,7 +14,7 @@
//
extern crate rocksdb;
use rocksdb::{DB, MergeOperands, Options};
use rocksdb::{DB, MergeOperands, Options, ColumnFamilyDescriptor};
#[test]
pub fn test_column_family() {
@ -24,7 +24,7 @@ pub fn test_column_family() {
{
let mut opts = Options::default();
opts.create_if_missing(true);
opts.set_merge_operator("test operator", test_provided_merge);
opts.set_merge_operator("test operator", test_provided_merge, None);
let mut db = DB::open(&opts, path).unwrap();
let opts = Options::default();
match db.create_cf("cf1", &opts) {
@ -38,7 +38,7 @@ pub fn test_column_family() {
// should fail to open db without specifying same column families
{
let mut opts = Options::default();
opts.set_merge_operator("test operator", test_provided_merge);
opts.set_merge_operator("test operator", test_provided_merge, None);
match DB::open(&opts, path) {
Ok(_) => {
panic!("should not have opened DB successfully without \
@ -56,7 +56,7 @@ pub fn test_column_family() {
// should properly open db when specyfing all column families
{
let mut opts = Options::default();
opts.set_merge_operator("test operator", test_provided_merge);
opts.set_merge_operator("test operator", test_provided_merge, None);
match DB::open_cf(&opts, path, &["cf1"]) {
Ok(_) => println!("successfully opened db with column family"),
Err(e) => panic!("failed to open db with column family: {}", e),
@ -91,6 +91,25 @@ pub fn test_column_family() {
assert!(DB::destroy(&Options::default(), path).is_ok());
}
#[test]
fn test_create_missing_column_family() {
let path = "_rust_rocksdb_missing_cftest";
// should be able to create new column families when opening a new database
{
let mut opts = Options::default();
opts.create_if_missing(true);
opts.create_missing_column_families(true);
match DB::open_cf(&opts, path, &["cf1"]) {
Ok(_) => println!("successfully created new column family"),
Err(e) => panic!("failed to create new column family: {}", e),
}
}
assert!(DB::destroy(&Options::default(), path).is_ok());
}
#[test]
#[ignore]
fn test_merge_operator() {
@ -98,7 +117,7 @@ fn test_merge_operator() {
// TODO should be able to write, read, merge, batch, and iterate over a cf
{
let mut opts = Options::default();
opts.set_merge_operator("test operator", test_provided_merge);
opts.set_merge_operator("test operator", test_provided_merge, None);
let db = match DB::open_cf(&opts, path, &["cf1"]) {
Ok(db) => {
println!("successfully opened db with column family");
@ -140,7 +159,7 @@ fn test_merge_operator() {
fn test_provided_merge(_: &[u8],
existing_val: Option<&[u8]>,
operands: &mut MergeOperands)
-> Vec<u8> {
-> Option<Vec<u8>> {
let nops = operands.size_hint().0;
let mut result: Vec<u8> = Vec::with_capacity(nops);
match existing_val {
@ -156,5 +175,45 @@ fn test_provided_merge(_: &[u8],
result.push(*e);
}
}
result
Some(result)
}
#[test]
pub fn test_column_family_with_options() {
let path = "_rust_rocksdb_cf_with_optionstest";
{
let mut cfopts = Options::default();
cfopts.set_max_write_buffer_number(16);
let cf_descriptor = ColumnFamilyDescriptor::new("cf1", cfopts);
let mut opts = Options::default();
opts.create_if_missing(true);
opts.create_missing_column_families(true);
let cfs = vec![cf_descriptor];
match DB::open_cf_descriptors(&opts, path, cfs) {
Ok(_) => println!("created db with column family descriptors succesfully"),
Err(e) => {
panic!("could not create new database with column family descriptors: {}", e);
}
}
}
{
let mut cfopts = Options::default();
cfopts.set_max_write_buffer_number(16);
let cf_descriptor = ColumnFamilyDescriptor::new("cf1", cfopts);
let opts = Options::default();
let cfs = vec![cf_descriptor];
match DB::open_cf_descriptors(&opts, path, cfs) {
Ok(_) => println!("succesfully re-opened database with column family descriptorrs"),
Err(e) => {
panic!("unable to re-open database with column family descriptors: {}", e);
}
}
}
assert!(DB::destroy(&Options::default(), path).is_ok());
}

@ -122,6 +122,11 @@ pub fn test_iterator() {
let expected = vec![(cba(&k2), cba(&v2)), (cba(&k1), cba(&v1))];
assert_eq!(iterator1.collect::<Vec<_>>(), expected);
}
{
let iterator1 = db.iterator(IteratorMode::From(b"zz", Direction::Reverse));
let expected = vec![(cba(&k4), cba(&v4)), (cba(&k3), cba(&v3))];
assert_eq!(iterator1.take(2).collect::<Vec<_>>(), expected);
}
{
let iterator1 = db.iterator(IteratorMode::From(b"k0", Direction::Forward));
assert!(iterator1.valid());
@ -132,13 +137,13 @@ pub fn test_iterator() {
let iterator4 = db.iterator(IteratorMode::From(b"k5", Direction::Forward));
assert!(!iterator4.valid());
let iterator5 = db.iterator(IteratorMode::From(b"k0", Direction::Reverse));
assert!(iterator5.valid());
assert!(!iterator5.valid());
let iterator6 = db.iterator(IteratorMode::From(b"k1", Direction::Reverse));
assert!(iterator6.valid());
let iterator7 = db.iterator(IteratorMode::From(b"k11", Direction::Reverse));
assert!(iterator7.valid());
let iterator8 = db.iterator(IteratorMode::From(b"k5", Direction::Reverse));
assert!(!iterator8.valid());
assert!(iterator8.valid());
}
{
let mut iterator1 = db.iterator(IteratorMode::From(b"k4", Direction::Forward));
@ -151,3 +156,41 @@ pub fn test_iterator() {
let opts = Options::default();
assert!(DB::destroy(&opts, path).is_ok());
}
fn key(k: &[u8]) -> Box<[u8]> { k.to_vec().into_boxed_slice() }
#[test]
pub fn test_prefix_iterator() {
let path = "_rust_rocksdb_prefixiteratortest";
{
let a1: Box<[u8]> = key(b"aaa1");
let a2: Box<[u8]> = key(b"aaa2");
let b1: Box<[u8]> = key(b"bbb1");
let b2: Box<[u8]> = key(b"bbb2");
let prefix_extractor = rocksdb::SliceTransform::create_fixed_prefix(3);
let mut opts = Options::default();
opts.create_if_missing(true);
opts.set_prefix_extractor(prefix_extractor);
let db = DB::open(&opts, path).unwrap();
assert!(db.put(&*a1, &*a1).is_ok());
assert!(db.put(&*a2, &*a2).is_ok());
assert!(db.put(&*b1, &*b1).is_ok());
assert!(db.put(&*b2, &*b2).is_ok());
{
let expected = vec![(cba(&a1), cba(&a1)), (cba(&a2), cba(&a2))];
let a_iterator = db.prefix_iterator(b"aaa");
assert_eq!(a_iterator.collect::<Vec<_>>(), expected)
}
{
let expected = vec![(cba(&b1), cba(&b1)), (cba(&b2), cba(&b2))];
let b_iterator = db.prefix_iterator(b"bbb");
assert_eq!(b_iterator.collect::<Vec<_>>(), expected)
}
}
}

@ -0,0 +1,48 @@
extern crate rocksdb;
use rocksdb::{DB, Options, SliceTransform};
#[test]
pub fn test_slice_transform() {
let path = "_rust_rocksdb_slicetransform_test";
let a1: Box<[u8]> = key(b"aaa1");
let a2: Box<[u8]> = key(b"aaa2");
let b1: Box<[u8]> = key(b"bbb1");
let b2: Box<[u8]> = key(b"bbb2");
fn first_three(k: &[u8]) -> Vec<u8> {
k.iter().take(3).cloned().collect()
}
let prefix_extractor = SliceTransform::create("first_three", first_three, None);
let mut opts = Options::default();
opts.create_if_missing(true);
opts.set_prefix_extractor(prefix_extractor);
let db = DB::open(&opts, path).unwrap();
assert!(db.put(&*a1, &*a1).is_ok());
assert!(db.put(&*a2, &*a2).is_ok());
assert!(db.put(&*b1, &*b1).is_ok());
assert!(db.put(&*b2, &*b2).is_ok());
fn cba(input: &Box<[u8]>) -> Box<[u8]> {
input.iter().cloned().collect::<Vec<_>>().into_boxed_slice()
}
fn key(k: &[u8]) -> Box<[u8]> { k.to_vec().into_boxed_slice() }
{
let expected = vec![(cba(&a1), cba(&a1)), (cba(&a2), cba(&a2))];
let a_iterator = db.prefix_iterator(b"aaa");
assert_eq!(a_iterator.collect::<Vec<_>>(), expected)
}
{
let expected = vec![(cba(&b1), cba(&b1)), (cba(&b2), cba(&b2))];
let b_iterator = db.prefix_iterator(b"bbb");
assert_eq!(b_iterator.collect::<Vec<_>>(), expected)
}
}
Loading…
Cancel
Save