CharBuf*
IxManager_make_snapshot_filename(IndexManager *self)
{
VArray *files = Folder_List(self->folder);
u32_t i, max;
i32_t max_gen = 0;
for (i = 0, max = VA_Get_Size(files); i < max; i++) {
CharBuf *file = (CharBuf*)VA_Fetch(files, i);
if ( CB_Starts_With_Str(file, "snapshot_", 9)
&& CB_Ends_With_Str(file, ".json", 5)
) {
i32_t gen = IxFileNames_extract_gen(file);
if (gen > max_gen) {
max_gen = gen;
}
}
}
DECREF(files);
{
i32_t new_gen = max_gen + 1;
CharBuf *base_36 = StrHelp_to_base36(new_gen);
CharBuf *snapfile = CB_newf("snapshot_%o.json", base_36);
DECREF(base_36);
return snapfile;
}
}
String*
IxFileNames_latest_snapshot(Folder *folder) {
DirHandle *dh = Folder_Open_Dir(folder, NULL);
String *retval = NULL;
uint64_t latest_gen = 0;
if (!dh) { RETHROW(INCREF(Err_get_error())); }
while (DH_Next(dh)) {
String *entry = DH_Get_Entry(dh);
if (Str_Starts_With_Utf8(entry, "snapshot_", 9)
&& Str_Ends_With_Utf8(entry, ".json", 5)
) {
uint64_t gen = IxFileNames_extract_gen(entry);
if (gen > latest_gen) {
latest_gen = gen;
DECREF(retval);
retval = Str_Clone(entry);
}
}
DECREF(entry);
}
DECREF(dh);
return retval;
}
String*
IxManager_Make_Snapshot_Filename_IMP(IndexManager *self) {
IndexManagerIVARS *const ivars = IxManager_IVARS(self);
Folder *folder = (Folder*)CERTIFY(ivars->folder, FOLDER);
DirHandle *dh = Folder_Open_Dir(folder, NULL);
uint64_t max_gen = 0;
if (!dh) { RETHROW(INCREF(Err_get_error())); }
while (DH_Next(dh)) {
String *entry = DH_Get_Entry(dh);
if (Str_Starts_With_Utf8(entry, "snapshot_", 9)
&& Str_Ends_With_Utf8(entry, ".json", 5)
) {
uint64_t gen = IxFileNames_extract_gen(entry);
if (gen > max_gen) { max_gen = gen; }
}
DECREF(entry);
}
DECREF(dh);
uint64_t new_gen = max_gen + 1;
char base36[StrHelp_MAX_BASE36_BYTES];
StrHelp_to_base36(new_gen, &base36);
return Str_newf("snapshot_%s.json", &base36);
}
Segment*
IxManager_make_new_segment(IndexManager *self, Snapshot *snapshot)
{
VArray *files = Snapshot_List(snapshot);
u32_t i, max;
i32_t highest_seg_num = 0;
CharBuf *seg_name = CB_new(20);
Segment *segment;
/* Find highest seg num. */
for (i = 0, max = VA_Get_Size(files); i < max; i++) {
CharBuf *file = (CharBuf*)VA_Fetch(files, i);
if (CB_Starts_With_Str(file, "seg_", 4)) {
i32_t seg_num = IxFileNames_extract_gen(file);
if (seg_num > highest_seg_num) {
highest_seg_num = seg_num;
}
}
}
/* Create segment with num one greater than current max. */
S_cat_seg_name(seg_name, highest_seg_num + 1);
segment = Seg_new(seg_name, self->folder);
DECREF(seg_name);
DECREF(files);
return segment;
}
CharBuf*
IxManager_make_snapshot_filename(IndexManager *self)
{
Folder *folder = (Folder*)CERTIFY(self->folder, FOLDER);
DirHandle *dh = Folder_Open_Dir(folder, NULL);
CharBuf *entry;
uint64_t max_gen = 0;
if (!dh) { RETHROW(INCREF(Err_get_error())); }
entry = DH_Get_Entry(dh);
while (DH_Next(dh)) {
if ( CB_Starts_With_Str(entry, "snapshot_", 9)
&& CB_Ends_With_Str(entry, ".json", 5)
) {
uint64_t gen = IxFileNames_extract_gen(entry);
if (gen > max_gen) { max_gen = gen; }
}
}
DECREF(dh);
{
uint64_t new_gen = max_gen + 1;
char base36[StrHelp_MAX_BASE36_BYTES];
StrHelp_to_base36(new_gen, &base36);
return CB_newf("snapshot_%s.json", &base36);
}
}
static void
S_test_extract_gen(TestBatchRunner *runner, const char *name, uint64_t gen,
const char *test_name) {
StackString *source = SSTR_WRAP_UTF8(name, strlen(name));
TEST_TRUE(runner, IxFileNames_extract_gen((String*)source) == gen,
test_name);
}
void
Indexer_prepare_commit(Indexer *self)
{
VArray *seg_readers = PolyReader_Get_Seg_Readers(self->polyreader);
uint32_t num_seg_readers = VA_Get_Size(seg_readers);
bool_t merge_happened = false;
if ( !self->write_lock || self->prepared ) {
THROW(ERR, "Can't call Prepare_Commit() more than once");
}
// Merge existing index data.
if (num_seg_readers) {
merge_happened = S_maybe_merge(self, seg_readers);
}
// Add a new segment and write a new snapshot file if...
if ( Seg_Get_Count(self->segment) // Docs/segs added.
|| merge_happened // Some segs merged.
|| !Snapshot_Num_Entries(self->snapshot) // Initializing index.
|| DelWriter_Updated(self->del_writer)
) {
Folder *folder = self->folder;
Schema *schema = self->schema;
Snapshot *snapshot = self->snapshot;
CharBuf *old_schema_name = S_find_schema_file(snapshot);
uint64_t schema_gen = old_schema_name
? IxFileNames_extract_gen(old_schema_name) + 1
: 1;
char base36[StrHelp_MAX_BASE36_BYTES];
CharBuf *new_schema_name;
StrHelp_to_base36(schema_gen, &base36);
new_schema_name = CB_newf("schema_%s.json", base36);
// Finish the segment, write schema file.
SegWriter_Finish(self->seg_writer);
Schema_Write(schema, folder, new_schema_name);
if (old_schema_name) {
Snapshot_Delete_Entry(snapshot, old_schema_name);
}
Snapshot_Add_Entry(snapshot, new_schema_name);
DECREF(new_schema_name);
// Write temporary snapshot file.
DECREF(self->snapfile);
self->snapfile = IxManager_Make_Snapshot_Filename(self->manager);
CB_Cat_Trusted_Str(self->snapfile, ".temp", 5);
Folder_Delete(folder, self->snapfile);
Snapshot_Write_File(snapshot, folder, self->snapfile);
self->needs_commit = true;
}
// Close reader, so that we can delete its files if appropriate.
PolyReader_Close(self->polyreader);
self->prepared = true;
}
void
Indexer_Prepare_Commit_IMP(Indexer *self) {
IndexerIVARS *const ivars = Indexer_IVARS(self);
Vector *seg_readers = PolyReader_Get_Seg_Readers(ivars->polyreader);
size_t num_seg_readers = Vec_Get_Size(seg_readers);
bool merge_happened = false;
if (!ivars->write_lock || ivars->prepared) {
THROW(ERR, "Can't call Prepare_Commit() more than once");
}
// Merge existing index data.
if (num_seg_readers) {
merge_happened = S_maybe_merge(self, seg_readers);
}
// Add a new segment and write a new snapshot file if...
if (Seg_Get_Count(ivars->segment) // Docs/segs added.
|| merge_happened // Some segs merged.
|| !Snapshot_Num_Entries(ivars->snapshot) // Initializing index.
|| DelWriter_Updated(ivars->del_writer)
) {
Folder *folder = ivars->folder;
Schema *schema = ivars->schema;
Snapshot *snapshot = ivars->snapshot;
// Derive snapshot and schema file names.
DECREF(ivars->snapfile);
String *snapfile = IxManager_Make_Snapshot_Filename(ivars->manager);
ivars->snapfile = Str_Cat_Trusted_Utf8(snapfile, ".temp", 5);
DECREF(snapfile);
uint64_t schema_gen = IxFileNames_extract_gen(ivars->snapfile);
char base36[StrHelp_MAX_BASE36_BYTES];
StrHelp_to_base36(schema_gen, &base36);
String *new_schema_name = Str_newf("schema_%s.json", base36);
// Finish the segment, write schema file.
SegWriter_Finish(ivars->seg_writer);
Schema_Write(schema, folder, new_schema_name);
String *old_schema_name = S_find_schema_file(snapshot);
if (old_schema_name) {
Snapshot_Delete_Entry(snapshot, old_schema_name);
}
Snapshot_Add_Entry(snapshot, new_schema_name);
DECREF(new_schema_name);
// Write temporary snapshot file.
Folder_Delete(folder, ivars->snapfile);
Snapshot_Write_File(snapshot, folder, ivars->snapfile);
ivars->needs_commit = true;
}
// Close reader, so that we can delete its files if appropriate.
PolyReader_Close(ivars->polyreader);
ivars->prepared = true;
}
int64_t
IxManager_Highest_Seg_Num_IMP(IndexManager *self, Snapshot *snapshot) {
Vector *files = Snapshot_List(snapshot);
uint64_t highest_seg_num = 0;
UNUSED_VAR(self);
for (size_t i = 0, max = Vec_Get_Size(files); i < max; i++) {
String *file = (String*)Vec_Fetch(files, i);
if (Seg_valid_seg_name(file)) {
uint64_t seg_num = IxFileNames_extract_gen(file);
if (seg_num > highest_seg_num) { highest_seg_num = seg_num; }
}
}
DECREF(files);
return (int64_t)highest_seg_num;
}
int64_t
IxManager_highest_seg_num(IndexManager *self, Snapshot *snapshot)
{
VArray *files = Snapshot_List(snapshot);
uint32_t i, max;
uint64_t highest_seg_num = 0;
UNUSED_VAR(self);
for (i = 0, max = VA_Get_Size(files); i < max; i++) {
CharBuf *file = (CharBuf*)VA_Fetch(files, i);
if (Seg_valid_seg_name(file)) {
uint64_t seg_num = IxFileNames_extract_gen(file);
if (seg_num > highest_seg_num) { highest_seg_num = seg_num; }
}
}
DECREF(files);
return (int64_t)highest_seg_num;
}
void
Snapshot_Write_File_IMP(Snapshot *self, Folder *folder, String *path) {
SnapshotIVARS *const ivars = Snapshot_IVARS(self);
Hash *all_data = Hash_new(0);
Vector *list = Snapshot_List(self);
// Update path.
DECREF(ivars->path);
if (path != NULL && Str_Get_Size(path) != 0) {
ivars->path = Str_Clone(path);
}
else {
String *latest = IxFileNames_latest_snapshot(folder);
uint64_t gen = latest ? IxFileNames_extract_gen(latest) + 1 : 1;
char base36[StrHelp_MAX_BASE36_BYTES];
StrHelp_to_base36(gen, &base36);
ivars->path = Str_newf("snapshot_%s.json", &base36);
DECREF(latest);
}
// Don't overwrite.
if (Folder_Exists(folder, ivars->path)) {
THROW(ERR, "Snapshot file '%o' already exists", ivars->path);
}
// Sort, then store file names.
Vec_Sort(list);
Hash_Store_Utf8(all_data, "entries", 7, (Obj*)list);
// Create a JSON-izable data structure.
Hash_Store_Utf8(all_data, "format", 6,
(Obj*)Str_newf("%i32", (int32_t)Snapshot_current_file_format));
Hash_Store_Utf8(all_data, "subformat", 9,
(Obj*)Str_newf("%i32", (int32_t)Snapshot_current_file_subformat));
// Write out JSON-ized data to the new file.
Json_spew_json((Obj*)all_data, folder, ivars->path);
DECREF(all_data);
}
void
Snapshot_write_file(Snapshot *self, Folder *folder, const CharBuf *path)
{
Hash *all_data = Hash_new(0);
VArray *list = Snapshot_List(self);
// Update path.
DECREF(self->path);
if (path) {
self->path = CB_Clone(path);
}
else {
CharBuf *latest = IxFileNames_latest_snapshot(folder);
uint64_t gen = latest ? IxFileNames_extract_gen(latest) + 1 : 1;
char base36[StrHelp_MAX_BASE36_BYTES];
StrHelp_to_base36(gen, &base36);
self->path = CB_newf("snapshot_%s.json", &base36);
DECREF(latest);
}
// Don't overwrite.
if (Folder_Exists(folder, self->path)) {
THROW(ERR, "Snapshot file '%o' already exists", self->path);
}
// Sort, then store file names.
VA_Sort(list, NULL, NULL);
Hash_Store_Str(all_data, "entries", 7, (Obj*)list);
// Create a JSON-izable data structure.
Hash_Store_Str(all_data, "format", 6,
(Obj*)CB_newf("%i32", (int32_t)Snapshot_current_file_format) );
Hash_Store_Str(all_data, "subformat", 9,
(Obj*)CB_newf("%i32", (int32_t)Snapshot_current_file_subformat) );
// Write out JSON-ized data to the new file.
Json_spew_json((Obj*)all_data, folder, self->path);
DECREF(all_data);
}
void
Snapshot_write_file(Snapshot *self, Folder *folder, const CharBuf *filename)
{
Hash *all_data = Hash_new(0);
VArray *list = Snapshot_List(self);
/* Update filename. */
DECREF(self->filename);
if (filename) {
self->filename = CB_Clone(filename);
}
else {
CharBuf *latest = IxFileNames_latest_snapshot(folder);
i32_t gen = latest ? IxFileNames_extract_gen(latest) + 1 : 1;
CharBuf *base_36 = StrHelp_to_base36(gen);
self->filename = CB_newf("snapshot_%o.json", base_36);
DECREF(latest);
DECREF(base_36);
}
/* Don't overwrite. */
if (Folder_Exists(folder, self->filename)) {
THROW("Snapshot file '%o' already exists", self->filename);
}
/* Sort, then store file names. */
VA_Sort(list, NULL);
Hash_Store_Str(all_data, "entries", 7, (Obj*)list);
/* Create a JSON-izable data structure. */
Hash_Store_Str(all_data, "format", 6,
(Obj*)CB_newf("%i32", (i32_t)Snapshot_current_file_format) );
/* Write out JSON-ized data to the new file. */
Json_spew_json((Obj*)all_data, folder, self->filename);
DECREF(all_data);
}
PolyReader*
PolyReader_do_open(PolyReader *self, Obj *index, Snapshot *snapshot,
IndexManager *manager) {
PolyReaderIVARS *const ivars = PolyReader_IVARS(self);
Folder *folder = S_derive_folder(index);
uint64_t last_gen = 0;
PolyReader_init(self, NULL, folder, snapshot, manager, NULL);
DECREF(folder);
if (manager) {
if (!S_obtain_deletion_lock(self)) {
DECREF(self);
THROW(LOCKERR, "Couldn't get deletion lock");
}
}
while (1) {
CharBuf *target_snap_file;
// If a Snapshot was supplied, use its file.
if (snapshot) {
target_snap_file = Snapshot_Get_Path(snapshot);
if (!target_snap_file) {
THROW(ERR, "Supplied snapshot objects must not be empty");
}
else {
CB_Inc_RefCount(target_snap_file);
}
}
else {
// Otherwise, pick the most recent snap file.
target_snap_file = IxFileNames_latest_snapshot(folder);
// No snap file? Looks like the index is empty. We can stop now
// and return NULL.
if (!target_snap_file) { break; }
}
// Derive "generation" of this snapshot file from its name.
uint64_t gen = IxFileNames_extract_gen(target_snap_file);
// Get a read lock on the most recent snapshot file if indicated.
if (manager) {
if (!S_obtain_read_lock(self, target_snap_file)) {
DECREF(self);
THROW(LOCKERR, "Couldn't get read lock for %o",
target_snap_file);
}
}
// Testing only.
if (PolyReader_race_condition_debug1) {
ZombieCharBuf *temp = ZCB_WRAP_STR("temp", 4);
if (Folder_Exists(folder, (CharBuf*)temp)) {
bool success = Folder_Rename(folder, (CharBuf*)temp,
PolyReader_race_condition_debug1);
if (!success) { RETHROW(INCREF(Err_get_error())); }
}
PolyReader_debug1_num_passes++;
}
// If a Snapshot object was passed in, the file has already been read.
// If that's not the case, we must read the file we just picked.
if (!snapshot) {
struct try_read_snapshot_context context;
context.snapshot = ivars->snapshot;
context.folder = folder;
context.path = target_snap_file;
Err *error = Err_trap(S_try_read_snapshot, &context);
if (error) {
S_release_read_lock(self);
DECREF(target_snap_file);
if (last_gen < gen) { // Index updated, so try again.
DECREF(error);
last_gen = gen;
continue;
}
else { // Real error.
if (manager) { S_release_deletion_lock(self); }
RETHROW(error);
}
}
}
/* It's possible, though unlikely, for an Indexer to delete files
* out from underneath us after the snapshot file is read but before
* we've got SegReaders holding open all the required files. If we
* failed to open something, see if we can find a newer snapshot file.
* If we can, then the exception was due to the race condition. If
* not, we have a real exception, so throw an error. */
struct try_open_elements_context context;
context.self = self;
context.seg_readers = NULL;
Err *error = Err_trap(S_try_open_elements, &context);
if (error) {
S_release_read_lock(self);
DECREF(target_snap_file);
if (last_gen < gen) { // Index updated, so try again.
DECREF(error);
last_gen = gen;
}
else { // Real error.
if (manager) { S_release_deletion_lock(self); }
RETHROW(error);
}
}
else { // Succeeded.
S_init_sub_readers(self, (VArray*)context.seg_readers);
DECREF(context.seg_readers);
DECREF(target_snap_file);
break;
}
}
if (manager) { S_release_deletion_lock(self); }
return self;
}
void
S_try_open_elements(void *context) {
struct try_open_elements_context *args
= (struct try_open_elements_context*)context;
PolyReader *self = args->self;
PolyReaderIVARS *const ivars = PolyReader_IVARS(self);
VArray *files = Snapshot_List(ivars->snapshot);
Folder *folder = PolyReader_Get_Folder(self);
uint32_t num_segs = 0;
uint64_t latest_schema_gen = 0;
CharBuf *schema_file = NULL;
// Find schema file, count segments.
for (uint32_t i = 0, max = VA_Get_Size(files); i < max; i++) {
CharBuf *entry = (CharBuf*)VA_Fetch(files, i);
if (Seg_valid_seg_name(entry)) {
num_segs++;
}
else if (CB_Starts_With_Str(entry, "schema_", 7)
&& CB_Ends_With_Str(entry, ".json", 5)
) {
uint64_t gen = IxFileNames_extract_gen(entry);
if (gen > latest_schema_gen) {
latest_schema_gen = gen;
if (!schema_file) { schema_file = CB_Clone(entry); }
else { CB_Mimic(schema_file, (Obj*)entry); }
}
}
}
// Read Schema.
if (!schema_file) {
DECREF(files);
THROW(ERR, "Can't find a schema file.");
}
else {
Hash *dump = (Hash*)Json_slurp_json(folder, schema_file);
if (dump) { // read file successfully
DECREF(ivars->schema);
ivars->schema = (Schema*)CERTIFY(
VTable_Load_Obj(SCHEMA, (Obj*)dump), SCHEMA);
DECREF(dump);
DECREF(schema_file);
schema_file = NULL;
}
else {
CharBuf *mess = MAKE_MESS("Failed to parse %o", schema_file);
DECREF(schema_file);
DECREF(files);
Err_throw_mess(ERR, mess);
}
}
VArray *segments = VA_new(num_segs);
for (uint32_t i = 0, max = VA_Get_Size(files); i < max; i++) {
CharBuf *entry = (CharBuf*)VA_Fetch(files, i);
// Create a Segment for each segmeta.
if (Seg_valid_seg_name(entry)) {
int64_t seg_num = IxFileNames_extract_gen(entry);
Segment *segment = Seg_new(seg_num);
// Bail if reading the file fails (probably because it's been
// deleted and a new snapshot file has been written so we need to
// retry).
if (Seg_Read_File(segment, folder)) {
VA_Push(segments, (Obj*)segment);
}
else {
CharBuf *mess = MAKE_MESS("Failed to read %o", entry);
DECREF(segment);
DECREF(segments);
DECREF(files);
Err_throw_mess(ERR, mess);
}
}
}
// Sort the segments by age.
VA_Sort(segments, NULL, NULL);
// Open individual SegReaders.
struct try_open_segreader_context seg_context;
seg_context.schema = PolyReader_Get_Schema(self);
seg_context.folder = folder;
seg_context.snapshot = PolyReader_Get_Snapshot(self);
seg_context.segments = segments;
seg_context.result = NULL;
args->seg_readers = VA_new(num_segs);
Err *error = NULL;
for (uint32_t seg_tick = 0; seg_tick < num_segs; seg_tick++) {
seg_context.seg_tick = seg_tick;
error = Err_trap(S_try_open_segreader, &seg_context);
if (error) {
break;
}
VA_Push(args->seg_readers, (Obj*)seg_context.result);
seg_context.result = NULL;
}
DECREF(segments);
DECREF(files);
if (error) {
DECREF(args->seg_readers);
args->seg_readers = NULL;
RETHROW(error);
}
}
void
BGMerger_Prepare_Commit_IMP(BackgroundMerger *self) {
BackgroundMergerIVARS *const ivars = BGMerger_IVARS(self);
Vector *seg_readers = PolyReader_Get_Seg_Readers(ivars->polyreader);
uint32_t num_seg_readers = Vec_Get_Size(seg_readers);
uint32_t segs_merged = 0;
if (ivars->prepared) {
THROW(ERR, "Can't call Prepare_Commit() more than once");
}
// Maybe merge existing index data.
if (num_seg_readers) {
segs_merged = S_maybe_merge(self);
}
if (!segs_merged) {
// Nothing merged. Leave `needs_commit` false and bail out.
ivars->prepared = true;
return;
}
// Finish the segment and write a new snapshot file.
else {
Folder *folder = ivars->folder;
Snapshot *snapshot = ivars->snapshot;
// Write out new deletions.
if (DelWriter_Updated(ivars->del_writer)) {
// Only write out if they haven't all been applied.
if (segs_merged != num_seg_readers) {
DelWriter_Finish(ivars->del_writer);
}
}
// Finish the segment.
SegWriter_Finish(ivars->seg_writer);
// Grab the write lock.
S_obtain_write_lock(self);
if (!ivars->write_lock) {
RETHROW(INCREF(Err_get_error()));
}
// Write temporary snapshot file.
DECREF(ivars->snapfile);
String *snapfile = IxManager_Make_Snapshot_Filename(ivars->manager);
ivars->snapfile = Str_Cat_Trusted_Utf8(snapfile, ".temp", 5);
DECREF(snapfile);
Folder_Delete(folder, ivars->snapfile);
Snapshot_Write_File(snapshot, folder, ivars->snapfile);
// Determine whether the index has been updated while this background
// merge process was running.
String *start_snapfile
= Snapshot_Get_Path(PolyReader_Get_Snapshot(ivars->polyreader));
Snapshot *latest_snapshot
= Snapshot_Read_File(Snapshot_new(), ivars->folder, NULL);
String *latest_snapfile = Snapshot_Get_Path(latest_snapshot);
bool index_updated
= !Str_Equals(start_snapfile, (Obj*)latest_snapfile);
if (index_updated) {
/* See if new deletions have been applied since this
* background merge process started against any of the
* segments we just merged away. If that's true, we need to
* write another segment which applies the deletions against
* the new composite segment.
*/
S_merge_updated_deletions(self);
// Add the fresh content to our snapshot. (It's important to
// run this AFTER S_merge_updated_deletions, because otherwise
// we couldn't tell whether the deletion counts changed.)
Vector *files = Snapshot_List(latest_snapshot);
for (uint32_t i = 0, max = Vec_Get_Size(files); i < max; i++) {
String *file = (String*)Vec_Fetch(files, i);
if (Str_Starts_With_Utf8(file, "seg_", 4)) {
int64_t gen = (int64_t)IxFileNames_extract_gen(file);
if (gen > ivars->cutoff) {
Snapshot_Add_Entry(ivars->snapshot, file);
}
}
}
DECREF(files);
// Since the snapshot content has changed, we need to rewrite it.
Folder_Delete(folder, ivars->snapfile);
Snapshot_Write_File(snapshot, folder, ivars->snapfile);
}
DECREF(latest_snapshot);
ivars->needs_commit = true;
}
// Close reader, so that we can delete its files if appropriate.
PolyReader_Close(ivars->polyreader);
ivars->prepared = true;
}
