nostr/strfry
1
0
Fork 0
mirror of https://github.com/hoytech/strfry.git synced 2025-06-19 09:36:43 +00:00
Code Issues Packages Projects Releases Wiki Activity

don't throw exceptions if modifications are made during dict operations

Browse Source
This commit is contained in:
Doug Hoyte
2023-02-15 17:00:11 -05:00
parent decc3aea26
commit 30b8c387a2
1 changed files with 46 additions and 39 deletions
Download Patch File Download Diff File Expand all files Collapse all files

85
src/cmd_dict.cpp
View File

@ -44,32 +44,29 @@ void cmd_dict(const std::vector<std::string> &subArgs) {
Decompressor decomp; Decompressor decomp;
std::vector<uint64_t> levIds; std::vector<uint64_t> levIds;
{
auto txn = env.txn_ro();
auto filterGroup = NostrFilterGroup::unwrapped(tao::json::from_string(filterStr), MAX_U64); auto txn = env.txn_ro();
Subscription sub(1, "junkSub", filterGroup);
DBScanQuery query(sub);
while (1) { auto filterGroup = NostrFilterGroup::unwrapped(tao::json::from_string(filterStr), MAX_U64);
bool complete = query.process(txn, MAX_U64, false, [&](const auto &sub, uint64_t levId){ Subscription sub(1, "junkSub", filterGroup);
levIds.push_back(levId); DBScanQuery query(sub);
});
if (complete) break; while (1) {
} bool complete = query.process(txn, MAX_U64, false, [&](const auto &sub, uint64_t levId){
levIds.push_back(levId);
});
LI << "Filter matched " << levIds.size() << " records"; if (complete) break;
} }
LI << "Filter matched " << levIds.size() << " records";
if (args["stats"].asBool()) { if (args["stats"].asBool()) {
uint64_t totalSize = 0; uint64_t totalSize = 0;
uint64_t totalCompressedSize = 0; uint64_t totalCompressedSize = 0;
uint64_t numCompressed = 0; uint64_t numCompressed = 0;
auto txn = env.txn_ro();
btree_map<uint32_t, uint64_t> dicts; btree_map<uint32_t, uint64_t> dicts;
env.foreach_CompressionDictionary(txn, [&](auto &view){ env.foreach_CompressionDictionary(txn, [&](auto &view){
@ -112,22 +109,18 @@ void cmd_dict(const std::vector<std::string> &subArgs) {
std::string trainingBuf; std::string trainingBuf;
std::vector<size_t> trainingSizes; std::vector<size_t> trainingSizes;
{ if (levIds.size() > limit) {
auto txn = env.txn_ro(); LI << "Randomly selecting " << limit << " records";
std::random_device rd;
std::mt19937 g(rd());
std::shuffle(levIds.begin(), levIds.end(), g);
levIds.resize(limit);
}
if (levIds.size() > limit) { for (auto levId : levIds) {
LI << "Randomly selecting " << limit << " records"; std::string json = std::string(getEventJson(txn, decomp, levId));
std::random_device rd; trainingBuf += json;
std::mt19937 g(rd()); trainingSizes.emplace_back(json.size());
std::shuffle(levIds.begin(), levIds.end(), g);
levIds.resize(limit);
}
for (auto levId : levIds) {
std::string json = std::string(getEventJson(txn, decomp, levId));
trainingBuf += json;
trainingSizes.emplace_back(json.size());
}
} }
std::string dict(dictSize, '\0'); std::string dict(dictSize, '\0');
@ -137,19 +130,19 @@ void cmd_dict(const std::vector<std::string> &subArgs) {
auto ret = ZDICT_trainFromBuffer(dict.data(), dict.size(), trainingBuf.data(), trainingSizes.data(), trainingSizes.size()); auto ret = ZDICT_trainFromBuffer(dict.data(), dict.size(), trainingBuf.data(), trainingSizes.data(), trainingSizes.size());
if (ZDICT_isError(ret)) throw herr("zstd training failed: ", ZSTD_getErrorName(ret)); if (ZDICT_isError(ret)) throw herr("zstd training failed: ", ZSTD_getErrorName(ret));
{ txn.abort();
auto txn = env.txn_rw(); txn = env.txn_rw();
uint64_t newDictId = env.insert_CompressionDictionary(txn, dict); uint64_t newDictId = env.insert_CompressionDictionary(txn, dict);
std::cout << "Saved new dictionary, dictId = " << newDictId << std::endl; std::cout << "Saved new dictionary, dictId = " << newDictId << std::endl;
txn.commit(); txn.commit();
}
} else if (args["compress"].asBool()) { } else if (args["compress"].asBool()) {
if (dictId == 0) throw herr("specify --dictId or --decompress"); if (dictId == 0) throw herr("specify --dictId or --decompress");
auto txn = env.txn_rw(); txn.abort();
txn = env.txn_rw();
auto view = env.lookup_CompressionDictionary(txn, dictId); auto view = env.lookup_CompressionDictionary(txn, dictId);
if (!view) throw herr("couldn't find dictId ", dictId); if (!view) throw herr("couldn't find dictId ", dictId);
@ -166,7 +159,14 @@ void cmd_dict(const std::vector<std::string> &subArgs) {
std::string compressedData(500'000, '\0'); std::string compressedData(500'000, '\0');
for (auto levId : levIds) { for (auto levId : levIds) {
auto orig = getEventJson(txn, decomp, levId); std::string_view orig;
try {
orig = getEventJson(txn, decomp, levId);
} catch (std::exception &e) {
continue;
}
auto ret = ZSTD_compress_usingCDict(cctx, compressedData.data(), compressedData.size(), orig.data(), orig.size(), cdict); auto ret = ZSTD_compress_usingCDict(cctx, compressedData.data(), compressedData.size(), orig.data(), orig.size(), cdict);
if (ZDICT_isError(ret)) throw herr("zstd compression failed: ", ZSTD_getErrorName(ret)); if (ZDICT_isError(ret)) throw herr("zstd compression failed: ", ZSTD_getErrorName(ret));
@ -203,13 +203,20 @@ void cmd_dict(const std::vector<std::string> &subArgs) {
LI << "Original event sizes: " << origSizes; LI << "Original event sizes: " << origSizes;
LI << "New event sizes: " << compressedSizes; LI << "New event sizes: " << compressedSizes;
} else if (args["decompress"].asBool()) { } else if (args["decompress"].asBool()) {
auto txn = env.txn_rw(); txn.abort();
txn = env.txn_rw();
uint64_t pendingFlush = 0; uint64_t pendingFlush = 0;
uint64_t processed = 0; uint64_t processed = 0;
for (auto levId : levIds) { for (auto levId : levIds) {
auto orig = getEventJson(txn, decomp, levId); std::string_view orig;
try {
orig = getEventJson(txn, decomp, levId);
} catch (std::exception &e) {
continue;
}
std::string newVal; std::string newVal;