Added bloom filter support.

In particular, we add a new FilterPolicy class.  An instance
of this class can be supplied in Options when opening a
database.  If supplied, the instance is used to generate
summaries of keys (e.g., a bloom filter) which are placed in
sstables.  These summaries are consulted by DB::Get() so we
can avoid reading sstable blocks that are guaranteed to not
contain the key we are looking for.

This change provides one implementation of FilterPolicy
based on bloom filters.

Other changes:
- Updated version number to 1.4.
- Some build tweaks.
- C binding for CompactRange.
- A few more benchmarks: deleteseq, deleterandom, readmissing, seekrandom.
- Minor .gitignore update.
This commit is contained in:
Sanjay Ghemawat
2012-04-17 08:36:46 -07:00
parent bc1ee4d25e
commit 85584d497e
38 changed files with 2070 additions and 570 deletions

110
db/c.cc
View File

@@ -10,6 +10,7 @@
#include "leveldb/comparator.h"
#include "leveldb/db.h"
#include "leveldb/env.h"
#include "leveldb/filter_policy.h"
#include "leveldb/iterator.h"
#include "leveldb/options.h"
#include "leveldb/status.h"
@@ -21,8 +22,10 @@ using leveldb::CompressionType;
using leveldb::DB;
using leveldb::Env;
using leveldb::FileLock;
using leveldb::FilterPolicy;
using leveldb::Iterator;
using leveldb::Logger;
using leveldb::NewBloomFilterPolicy;
using leveldb::NewLRUCache;
using leveldb::Options;
using leveldb::RandomAccessFile;
@@ -78,6 +81,47 @@ struct leveldb_comparator_t : public Comparator {
virtual void FindShortSuccessor(std::string* key) const { }
};
struct leveldb_filterpolicy_t : public FilterPolicy {
void* state_;
void (*destructor_)(void*);
const char* (*name_)(void*);
char* (*create_)(
void*,
const char* const* key_array, const size_t* key_length_array,
int num_keys,
size_t* filter_length);
unsigned char (*key_match_)(
void*,
const char* key, size_t length,
const char* filter, size_t filter_length);
virtual ~leveldb_filterpolicy_t() {
(*destructor_)(state_);
}
virtual const char* Name() const {
return (*name_)(state_);
}
virtual void CreateFilter(const Slice* keys, int n, std::string* dst) const {
std::vector<const char*> key_pointers(n);
std::vector<size_t> key_sizes(n);
for (int i = 0; i < n; i++) {
key_pointers[i] = keys[i].data();
key_sizes[i] = keys[i].size();
}
size_t len;
char* filter = (*create_)(state_, &key_pointers[0], &key_sizes[0], n, &len);
dst->append(filter, len);
free(filter);
}
virtual bool KeyMayMatch(const Slice& key, const Slice& filter) const {
return (*key_match_)(state_, key.data(), key.size(),
filter.data(), filter.size());
}
};
struct leveldb_env_t {
Env* rep;
bool is_default;
@@ -218,6 +262,17 @@ void leveldb_approximate_sizes(
delete[] ranges;
}
void leveldb_compact_range(
leveldb_t* db,
const char* start_key, size_t start_key_len,
const char* limit_key, size_t limit_key_len) {
Slice a, b;
db->rep->CompactRange(
// Pass NULL Slice if corresponding "const char*" is NULL
(start_key ? (a = Slice(start_key, start_key_len), &a) : NULL),
(limit_key ? (b = Slice(limit_key, limit_key_len), &b) : NULL));
}
void leveldb_destroy_db(
const leveldb_options_t* options,
const char* name,
@@ -340,6 +395,12 @@ void leveldb_options_set_comparator(
opt->rep.comparator = cmp;
}
void leveldb_options_set_filter_policy(
leveldb_options_t* opt,
leveldb_filterpolicy_t* policy) {
opt->rep.filter_policy = policy;
}
void leveldb_options_set_create_if_missing(
leveldb_options_t* opt, unsigned char v) {
opt->rep.create_if_missing = v;
@@ -407,6 +468,55 @@ void leveldb_comparator_destroy(leveldb_comparator_t* cmp) {
delete cmp;
}
leveldb_filterpolicy_t* leveldb_filterpolicy_create(
void* state,
void (*destructor)(void*),
char* (*create_filter)(
void*,
const char* const* key_array, const size_t* key_length_array,
int num_keys,
size_t* filter_length),
unsigned char (*key_may_match)(
void*,
const char* key, size_t length,
const char* filter, size_t filter_length),
const char* (*name)(void*)) {
leveldb_filterpolicy_t* result = new leveldb_filterpolicy_t;
result->state_ = state;
result->destructor_ = destructor;
result->create_ = create_filter;
result->key_match_ = key_may_match;
result->name_ = name;
return result;
}
void leveldb_filterpolicy_destroy(leveldb_filterpolicy_t* filter) {
delete filter;
}
leveldb_filterpolicy_t* leveldb_filterpolicy_create_bloom(int bits_per_key) {
// Make a leveldb_filterpolicy_t, but override all of its methods so
// they delegate to a NewBloomFilterPolicy() instead of user
// supplied C functions.
struct Wrapper : public leveldb_filterpolicy_t {
const FilterPolicy* rep_;
~Wrapper() { delete rep_; }
const char* Name() const { return rep_->Name(); }
void CreateFilter(const Slice* keys, int n, std::string* dst) const {
return rep_->CreateFilter(keys, n, dst);
}
bool KeyMayMatch(const Slice& key, const Slice& filter) const {
return rep_->KeyMayMatch(key, filter);
}
static void DoNothing(void*) { }
};
Wrapper* wrapper = new Wrapper;
wrapper->rep_ = NewBloomFilterPolicy(bits_per_key);
wrapper->state_ = NULL;
wrapper->destructor_ = &Wrapper::DoNothing;
return wrapper;
}
leveldb_readoptions_t* leveldb_readoptions_create() {
return new leveldb_readoptions_t;
}

View File

@@ -122,6 +122,31 @@ static const char* CmpName(void* arg) {
return "foo";
}
// Custom filter policy
static unsigned char fake_filter_result = 1;
static void FilterDestroy(void* arg) { }
static const char* FilterName(void* arg) {
return "TestFilter";
}
static char* FilterCreate(
void* arg,
const char* const* key_array, const size_t* key_length_array,
int num_keys,
size_t* filter_length) {
*filter_length = 4;
char* result = malloc(4);
memcpy(result, "fake", 4);
return result;
}
unsigned char FilterKeyMatch(
void* arg,
const char* key, size_t length,
const char* filter, size_t filter_length) {
CheckCondition(filter_length == 4);
CheckCondition(memcmp(filter, "fake", 4) == 0);
return fake_filter_result;
}
int main(int argc, char** argv) {
leveldb_t* db;
leveldb_comparator_t* cmp;
@@ -131,6 +156,7 @@ int main(int argc, char** argv) {
leveldb_readoptions_t* roptions;
leveldb_writeoptions_t* woptions;
char* err = NULL;
int run = -1;
snprintf(dbname, sizeof(dbname), "/tmp/leveldb_c_test-%d",
((int) geteuid()));
@@ -180,6 +206,14 @@ int main(int argc, char** argv) {
CheckNoError(err);
CheckGet(db, roptions, "foo", "hello");
StartPhase("compactall");
leveldb_compact_range(db, NULL, 0, NULL, 0);
CheckGet(db, roptions, "foo", "hello");
StartPhase("compactrange");
leveldb_compact_range(db, "a", 1, "z", 1);
CheckGet(db, roptions, "foo", "hello");
StartPhase("writebatch");
{
leveldb_writebatch_t* wb = leveldb_writebatch_create();
@@ -279,6 +313,49 @@ int main(int argc, char** argv) {
CheckGet(db, roptions, "foo", NULL);
CheckGet(db, roptions, "bar", NULL);
CheckGet(db, roptions, "box", "c");
leveldb_options_set_create_if_missing(options, 1);
leveldb_options_set_error_if_exists(options, 1);
}
StartPhase("filter");
for (run = 0; run < 2; run++) {
// First run uses custom filter, second run uses bloom filter
CheckNoError(err);
leveldb_filterpolicy_t* policy;
if (run == 0) {
policy = leveldb_filterpolicy_create(
NULL, FilterDestroy, FilterCreate, FilterKeyMatch, FilterName);
} else {
policy = leveldb_filterpolicy_create_bloom(10);
}
// Create new database
leveldb_close(db);
leveldb_destroy_db(options, dbname, &err);
leveldb_options_set_filter_policy(options, policy);
db = leveldb_open(options, dbname, &err);
CheckNoError(err);
leveldb_put(db, woptions, "foo", 3, "foovalue", 8, &err);
CheckNoError(err);
leveldb_put(db, woptions, "bar", 3, "barvalue", 8, &err);
CheckNoError(err);
leveldb_compact_range(db, NULL, 0, NULL, 0);
fake_filter_result = 1;
CheckGet(db, roptions, "foo", "foovalue");
CheckGet(db, roptions, "bar", "barvalue");
if (phase == 0) {
// Must not find value when custom filter returns false
fake_filter_result = 0;
CheckGet(db, roptions, "foo", NULL);
CheckGet(db, roptions, "bar", NULL);
fake_filter_result = 1;
CheckGet(db, roptions, "foo", "foovalue");
CheckGet(db, roptions, "bar", "barvalue");
}
leveldb_options_set_filter_policy(options, NULL);
leveldb_filterpolicy_destroy(policy);
}
StartPhase("cleanup");

View File

@@ -25,15 +25,20 @@
// overwrite -- overwrite N values in random key order in async mode
// fillsync -- write N/100 values in random key order in sync mode
// fill100K -- write N/1000 100K values in random order in async mode
// deleteseq -- delete N keys in sequential order
// deleterandom -- delete N keys in random order
// readseq -- read N times sequentially
// readreverse -- read N times in reverse order
// readrandom -- read N times in random order
// readmissing -- read N missing keys in random order
// readhot -- read N times in random order from 1% section of DB
// seekrandom -- N random seeks
// crc32c -- repeated crc32c of 4K of data
// acquireload -- load N*1000 times
// Meta operations:
// compact -- Compact the entire DB
// stats -- Print DB stats
// sstables -- Print sstable info
// heapprofile -- Dump a heap profile (if supported by this port)
static const char* FLAGS_benchmarks =
"fillseq,"
@@ -85,6 +90,10 @@ static int FLAGS_cache_size = -1;
// Maximum number of files to keep open at the same time (use default if == 0)
static int FLAGS_open_files = 0;
// Bloom filter bits per key.
// Negative means use default settings.
static int FLAGS_bloom_bits = -1;
// If true, do not destroy the existing database. If you set this
// flag and also specify a benchmark that wants a fresh database, that
// benchmark will fail.
@@ -293,6 +302,7 @@ struct ThreadState {
class Benchmark {
private:
Cache* cache_;
const FilterPolicy* filter_policy_;
DB* db_;
int num_;
int value_size_;
@@ -378,6 +388,9 @@ class Benchmark {
public:
Benchmark()
: cache_(FLAGS_cache_size >= 0 ? NewLRUCache(FLAGS_cache_size) : NULL),
filter_policy_(FLAGS_bloom_bits >= 0
? NewBloomFilterPolicy(FLAGS_bloom_bits)
: NULL),
db_(NULL),
num_(FLAGS_num),
value_size_(FLAGS_value_size),
@@ -399,6 +412,7 @@ class Benchmark {
~Benchmark() {
delete db_;
delete cache_;
delete filter_policy_;
}
void Run() {
@@ -457,11 +471,19 @@ class Benchmark {
method = &Benchmark::ReadReverse;
} else if (name == Slice("readrandom")) {
method = &Benchmark::ReadRandom;
} else if (name == Slice("readmissing")) {
method = &Benchmark::ReadMissing;
} else if (name == Slice("seekrandom")) {
method = &Benchmark::SeekRandom;
} else if (name == Slice("readhot")) {
method = &Benchmark::ReadHot;
} else if (name == Slice("readrandomsmall")) {
reads_ /= 1000;
method = &Benchmark::ReadRandom;
} else if (name == Slice("deleteseq")) {
method = &Benchmark::DeleteSeq;
} else if (name == Slice("deleterandom")) {
method = &Benchmark::DeleteRandom;
} else if (name == Slice("readwhilewriting")) {
num_threads++; // Add extra thread for writing
method = &Benchmark::ReadWhileWriting;
@@ -478,7 +500,9 @@ class Benchmark {
} else if (name == Slice("heapprofile")) {
HeapProfile();
} else if (name == Slice("stats")) {
PrintStats();
PrintStats("leveldb.stats");
} else if (name == Slice("sstables")) {
PrintStats("leveldb.sstables");
} else {
if (name != Slice()) { // No error message for empty name
fprintf(stderr, "unknown benchmark '%s'\n", name.ToString().c_str());
@@ -669,6 +693,7 @@ class Benchmark {
options.create_if_missing = !FLAGS_use_existing_db;
options.block_cache = cache_;
options.write_buffer_size = FLAGS_write_buffer_size;
options.filter_policy = filter_policy_;
Status s = DB::Open(options, FLAGS_db, &db_);
if (!s.ok()) {
fprintf(stderr, "open error: %s\n", s.ToString().c_str());
@@ -743,10 +768,28 @@ class Benchmark {
void ReadRandom(ThreadState* thread) {
ReadOptions options;
std::string value;
int found = 0;
for (int i = 0; i < reads_; i++) {
char key[100];
const int k = thread->rand.Next() % FLAGS_num;
snprintf(key, sizeof(key), "%016d", k);
if (db_->Get(options, key, &value).ok()) {
found++;
}
thread->stats.FinishedSingleOp();
}
char msg[100];
snprintf(msg, sizeof(msg), "(%d of %d found)", found, num_);
thread->stats.AddMessage(msg);
}
void ReadMissing(ThreadState* thread) {
ReadOptions options;
std::string value;
for (int i = 0; i < reads_; i++) {
char key[100];
const int k = thread->rand.Next() % FLAGS_num;
snprintf(key, sizeof(key), "%016d.", k);
db_->Get(options, key, &value);
thread->stats.FinishedSingleOp();
}
@@ -765,6 +808,54 @@ class Benchmark {
}
}
void SeekRandom(ThreadState* thread) {
ReadOptions options;
std::string value;
int found = 0;
for (int i = 0; i < reads_; i++) {
Iterator* iter = db_->NewIterator(options);
char key[100];
const int k = thread->rand.Next() % FLAGS_num;
snprintf(key, sizeof(key), "%016d", k);
iter->Seek(key);
if (iter->Valid() && iter->key() == key) found++;
delete iter;
thread->stats.FinishedSingleOp();
}
char msg[100];
snprintf(msg, sizeof(msg), "(%d of %d found)", found, num_);
thread->stats.AddMessage(msg);
}
void DoDelete(ThreadState* thread, bool seq) {
RandomGenerator gen;
WriteBatch batch;
Status s;
for (int i = 0; i < num_; i += entries_per_batch_) {
batch.Clear();
for (int j = 0; j < entries_per_batch_; j++) {
const int k = seq ? i+j : (thread->rand.Next() % FLAGS_num);
char key[100];
snprintf(key, sizeof(key), "%016d", k);
batch.Delete(key);
thread->stats.FinishedSingleOp();
}
s = db_->Write(write_options_, &batch);
if (!s.ok()) {
fprintf(stderr, "del error: %s\n", s.ToString().c_str());
exit(1);
}
}
}
void DeleteSeq(ThreadState* thread) {
DoDelete(thread, true);
}
void DeleteRandom(ThreadState* thread) {
DoDelete(thread, false);
}
void ReadWhileWriting(ThreadState* thread) {
if (thread->tid > 0) {
ReadRandom(thread);
@@ -799,9 +890,9 @@ class Benchmark {
db_->CompactRange(NULL, NULL);
}
void PrintStats() {
void PrintStats(const char* key) {
std::string stats;
if (!db_->GetProperty("leveldb.stats", &stats)) {
if (!db_->GetProperty(key, &stats)) {
stats = "(failed)";
}
fprintf(stdout, "\n%s\n", stats.c_str());
@@ -861,6 +952,8 @@ int main(int argc, char** argv) {
FLAGS_write_buffer_size = n;
} else if (sscanf(argv[i], "--cache_size=%d%c", &n, &junk) == 1) {
FLAGS_cache_size = n;
} else if (sscanf(argv[i], "--bloom_bits=%d%c", &n, &junk) == 1) {
FLAGS_bloom_bits = n;
} else if (sscanf(argv[i], "--open_files=%d%c", &n, &junk) == 1) {
FLAGS_open_files = n;
} else if (strncmp(argv[i], "--db=", 5) == 0) {

View File

@@ -87,12 +87,14 @@ static void ClipToRange(T* ptr, V minvalue, V maxvalue) {
}
Options SanitizeOptions(const std::string& dbname,
const InternalKeyComparator* icmp,
const InternalFilterPolicy* ipolicy,
const Options& src) {
Options result = src;
result.comparator = icmp;
ClipToRange(&result.max_open_files, 20, 50000);
ClipToRange(&result.write_buffer_size, 64<<10, 1<<30);
ClipToRange(&result.block_size, 1<<10, 4<<20);
result.filter_policy = (src.filter_policy != NULL) ? ipolicy : NULL;
ClipToRange(&result.max_open_files, 20, 50000);
ClipToRange(&result.write_buffer_size, 64<<10, 1<<30);
ClipToRange(&result.block_size, 1<<10, 4<<20);
if (result.info_log == NULL) {
// Open a log file in the same directory as the db
src.env->CreateDir(dbname); // In case it does not exist
@@ -112,7 +114,9 @@ Options SanitizeOptions(const std::string& dbname,
DBImpl::DBImpl(const Options& options, const std::string& dbname)
: env_(options.env),
internal_comparator_(options.comparator),
options_(SanitizeOptions(dbname, &internal_comparator_, options)),
internal_filter_policy_(options.filter_policy),
options_(SanitizeOptions(
dbname, &internal_comparator_, &internal_filter_policy_, options)),
owns_info_log_(options_.info_log != options.info_log),
owns_cache_(options_.block_cache != options.block_cache),
dbname_(dbname),

View File

@@ -105,6 +105,7 @@ class DBImpl : public DB {
// Constant after construction
Env* const env_;
const InternalKeyComparator internal_comparator_;
const InternalFilterPolicy internal_filter_policy_;
const Options options_; // options_.comparator == &internal_comparator_
bool owns_info_log_;
bool owns_cache_;
@@ -185,6 +186,7 @@ class DBImpl : public DB {
// it is not equal to src.info_log.
extern Options SanitizeOptions(const std::string& db,
const InternalKeyComparator* icmp,
const InternalFilterPolicy* ipolicy,
const Options& src);
} // namespace leveldb

File diff suppressed because it is too large Load Diff

View File

@@ -98,6 +98,26 @@ void InternalKeyComparator::FindShortSuccessor(std::string* key) const {
}
}
const char* InternalFilterPolicy::Name() const {
return user_policy_->Name();
}
void InternalFilterPolicy::CreateFilter(const Slice* keys, int n,
std::string* dst) const {
// We rely on the fact that the code in table.cc does not mind us
// adjusting keys[].
Slice* mkey = const_cast<Slice*>(keys);
for (int i = 0; i < n; i++) {
mkey[i] = ExtractUserKey(keys[i]);
// TODO(sanjay): Suppress dups?
}
user_policy_->CreateFilter(keys, n, dst);
}
bool InternalFilterPolicy::KeyMayMatch(const Slice& key, const Slice& f) const {
return user_policy_->KeyMayMatch(ExtractUserKey(key), f);
}
LookupKey::LookupKey(const Slice& user_key, SequenceNumber s) {
size_t usize = user_key.size();
size_t needed = usize + 13; // A conservative estimate

View File

@@ -8,6 +8,7 @@
#include <stdio.h>
#include "leveldb/comparator.h"
#include "leveldb/db.h"
#include "leveldb/filter_policy.h"
#include "leveldb/slice.h"
#include "leveldb/table_builder.h"
#include "util/coding.h"
@@ -123,6 +124,17 @@ class InternalKeyComparator : public Comparator {
int Compare(const InternalKey& a, const InternalKey& b) const;
};
// Filter policy wrapper that converts from internal keys to user keys
class InternalFilterPolicy : public FilterPolicy {
private:
const FilterPolicy* const user_policy_;
public:
explicit InternalFilterPolicy(const FilterPolicy* p) : user_policy_(p) { }
virtual const char* Name() const;
virtual void CreateFilter(const Slice* keys, int n, std::string* dst) const;
virtual bool KeyMayMatch(const Slice& key, const Slice& filter) const;
};
// Modules in this directory should keep internal keys wrapped inside
// the following class instead of plain strings so that we do not
// incorrectly use string comparisons instead of an InternalKeyComparator.

View File

@@ -48,7 +48,8 @@ class Repairer {
: dbname_(dbname),
env_(options.env),
icmp_(options.comparator),
options_(SanitizeOptions(dbname, &icmp_, options)),
ipolicy_(options.filter_policy),
options_(SanitizeOptions(dbname, &icmp_, &ipolicy_, options)),
owns_info_log_(options_.info_log != options.info_log),
owns_cache_(options_.block_cache != options.block_cache),
next_file_number_(1) {
@@ -99,6 +100,7 @@ class Repairer {
std::string const dbname_;
Env* const env_;
InternalKeyComparator const icmp_;
InternalFilterPolicy const ipolicy_;
Options const options_;
bool owns_info_log_;
bool owns_cache_;

View File

@@ -42,23 +42,18 @@ TableCache::~TableCache() {
delete cache_;
}
Iterator* TableCache::NewIterator(const ReadOptions& options,
uint64_t file_number,
uint64_t file_size,
Table** tableptr) {
if (tableptr != NULL) {
*tableptr = NULL;
}
Status TableCache::FindTable(uint64_t file_number, uint64_t file_size,
Cache::Handle** handle) {
Status s;
char buf[sizeof(file_number)];
EncodeFixed64(buf, file_number);
Slice key(buf, sizeof(buf));
Cache::Handle* handle = cache_->Lookup(key);
if (handle == NULL) {
*handle = cache_->Lookup(key);
if (*handle == NULL) {
std::string fname = TableFileName(dbname_, file_number);
RandomAccessFile* file = NULL;
Table* table = NULL;
Status s = env_->NewRandomAccessFile(fname, &file);
s = env_->NewRandomAccessFile(fname, &file);
if (s.ok()) {
s = Table::Open(*options_, file, file_size, &table);
}
@@ -68,13 +63,28 @@ Iterator* TableCache::NewIterator(const ReadOptions& options,
delete file;
// We do not cache error results so that if the error is transient,
// or somebody repairs the file, we recover automatically.
return NewErrorIterator(s);
} else {
TableAndFile* tf = new TableAndFile;
tf->file = file;
tf->table = table;
*handle = cache_->Insert(key, tf, 1, &DeleteEntry);
}
}
return s;
}
TableAndFile* tf = new TableAndFile;
tf->file = file;
tf->table = table;
handle = cache_->Insert(key, tf, 1, &DeleteEntry);
Iterator* TableCache::NewIterator(const ReadOptions& options,
uint64_t file_number,
uint64_t file_size,
Table** tableptr) {
if (tableptr != NULL) {
*tableptr = NULL;
}
Cache::Handle* handle = NULL;
Status s = FindTable(file_number, file_size, &handle);
if (!s.ok()) {
return NewErrorIterator(s);
}
Table* table = reinterpret_cast<TableAndFile*>(cache_->Value(handle))->table;
@@ -86,6 +96,22 @@ Iterator* TableCache::NewIterator(const ReadOptions& options,
return result;
}
Status TableCache::Get(const ReadOptions& options,
uint64_t file_number,
uint64_t file_size,
const Slice& k,
void* arg,
void (*saver)(void*, const Slice&, const Slice&)) {
Cache::Handle* handle = NULL;
Status s = FindTable(file_number, file_size, &handle);
if (s.ok()) {
Table* t = reinterpret_cast<TableAndFile*>(cache_->Value(handle))->table;
s = t->InternalGet(options, k, arg, saver);
cache_->Release(handle);
}
return s;
}
void TableCache::Evict(uint64_t file_number) {
char buf[sizeof(file_number)];
EncodeFixed64(buf, file_number);

View File

@@ -35,6 +35,15 @@ class TableCache {
uint64_t file_size,
Table** tableptr = NULL);
// If a seek to internal key "k" in specified file finds an entry,
// call (*handle_result)(arg, found_key, found_value).
Status Get(const ReadOptions& options,
uint64_t file_number,
uint64_t file_size,
const Slice& k,
void* arg,
void (*handle_result)(void*, const Slice&, const Slice&));
// Evict any entry for the specified file number
void Evict(uint64_t file_number);
@@ -43,6 +52,8 @@ class TableCache {
const std::string dbname_;
const Options* options_;
Cache* cache_;
Status FindTable(uint64_t file_number, uint64_t file_size, Cache::Handle**);
};
} // namespace leveldb

View File

@@ -255,35 +255,34 @@ void Version::AddIterators(const ReadOptions& options,
}
}
// If "*iter" points at a value or deletion for user_key, store
// either the value, or a NotFound error and return true.
// Else return false.
static bool GetValue(const Comparator* cmp,
Iterator* iter, const Slice& user_key,
std::string* value,
Status* s) {
if (!iter->Valid()) {
return false;
}
// Callback from TableCache::Get()
namespace {
enum SaverState {
kNotFound,
kFound,
kDeleted,
kCorrupt,
};
struct Saver {
SaverState state;
const Comparator* ucmp;
Slice user_key;
std::string* value;
};
}
static void SaveValue(void* arg, const Slice& ikey, const Slice& v) {
Saver* s = reinterpret_cast<Saver*>(arg);
ParsedInternalKey parsed_key;
if (!ParseInternalKey(iter->key(), &parsed_key)) {
*s = Status::Corruption("corrupted key for ", user_key);
return true;
}
if (cmp->Compare(parsed_key.user_key, user_key) != 0) {
return false;
}
switch (parsed_key.type) {
case kTypeDeletion:
*s = Status::NotFound(Slice()); // Use an empty error message for speed
break;
case kTypeValue: {
Slice v = iter->value();
value->assign(v.data(), v.size());
break;
if (!ParseInternalKey(ikey, &parsed_key)) {
s->state = kCorrupt;
} else {
if (s->ucmp->Compare(parsed_key.user_key, s->user_key) == 0) {
s->state = (parsed_key.type == kTypeValue) ? kFound : kDeleted;
if (s->state == kFound) {
s->value->assign(v.data(), v.size());
}
}
}
return true;
}
static bool NewestFirst(FileMetaData* a, FileMetaData* b) {
@@ -361,21 +360,27 @@ Status Version::Get(const ReadOptions& options,
last_file_read = f;
last_file_read_level = level;
Iterator* iter = vset_->table_cache_->NewIterator(
options,
f->number,
f->file_size);
iter->Seek(ikey);
const bool done = GetValue(ucmp, iter, user_key, value, &s);
if (!iter->status().ok()) {
s = iter->status();
delete iter;
Saver saver;
saver.state = kNotFound;
saver.ucmp = ucmp;
saver.user_key = user_key;
saver.value = value;
s = vset_->table_cache_->Get(options, f->number, f->file_size,
ikey, &saver, SaveValue);
if (!s.ok()) {
return s;
} else {
delete iter;
if (done) {
}
switch (saver.state) {
case kNotFound:
break; // Keep searching in other files
case kFound:
return s;
case kDeleted:
s = Status::NotFound(Slice()); // Use empty error message for speed
return s;
case kCorrupt:
s = Status::Corruption("corrupted key for ", user_key);
return s;
}
}
}
}