should be done now, wrote a new batched logger for the last thing
This commit is contained in:
@@ -8,7 +8,7 @@
|
|||||||
#include <variant>
|
#include <variant>
|
||||||
#include "database_utils.hpp"
|
#include "database_utils.hpp"
|
||||||
namespace processing_utils {
|
namespace processing_utils {
|
||||||
|
constexpr std::size_t MAX_BATCH_ROWS = 500;
|
||||||
bool parse_date_from_filename(std::string_view path,
|
bool parse_date_from_filename(std::string_view path,
|
||||||
int& year, int& month, int& day)
|
int& year, int& month, int& day)
|
||||||
{
|
{
|
||||||
@@ -204,10 +204,12 @@ parse_chat(std::string_view line)
|
|||||||
activity_events.insert_as_batch(client); chat_events.insert_as_batch(client); client_events.insert_as_batch(client);
|
activity_events.insert_as_batch(client); chat_events.insert_as_batch(client); client_events.insert_as_batch(client);
|
||||||
current_line.clear();
|
current_line.clear();
|
||||||
}
|
}
|
||||||
void process_blocking(std::string path,
|
void process_blocking(std::string path,
|
||||||
std::shared_ptr<clickhouse::Client> client,
|
std::shared_ptr<clickhouse::Client> client,
|
||||||
std::string target_user_id)
|
std::string target_user_id)
|
||||||
{
|
{
|
||||||
|
constexpr uint64_t MAX_BATCH_ROWS = 500;
|
||||||
|
|
||||||
spdlog::info("Processing (blocking) file {}", path);
|
spdlog::info("Processing (blocking) file {}", path);
|
||||||
|
|
||||||
std::ifstream input_file(path);
|
std::ifstream input_file(path);
|
||||||
@@ -216,68 +218,123 @@ parse_chat(std::string_view line)
|
|||||||
return;
|
return;
|
||||||
}
|
}
|
||||||
|
|
||||||
// Seek to the end like tail -f
|
// tail -f semantics
|
||||||
input_file.seekg(0, std::ios::end);
|
input_file.seekg(0, std::ios::end);
|
||||||
|
|
||||||
std::string current_line;
|
std::string current_line;
|
||||||
db_utils::ClientEventsBatch client_events{};
|
|
||||||
db_utils::ChatEventsBatch chat_events{};
|
db_utils::ClientEventsBatch client_events;
|
||||||
db_utils::ActivityEventsBatch activity_events{};
|
db_utils::ChatEventsBatch chat_events;
|
||||||
|
db_utils::ActivityEventsBatch activity_events;
|
||||||
|
|
||||||
|
uint64_t row_count = 0;
|
||||||
|
|
||||||
while (true) {
|
while (true) {
|
||||||
// Read all available lines
|
bool read_any = false;
|
||||||
while (std::getline(input_file, current_line)) {
|
|
||||||
if (current_line.empty()) continue;
|
|
||||||
|
|
||||||
std::time_t log_timestamp = extract_timestamp_utc(path, current_line);
|
while (std::getline(input_file, current_line)) {
|
||||||
|
read_any = true;
|
||||||
|
if (current_line.empty())
|
||||||
|
continue;
|
||||||
|
|
||||||
|
std::time_t ts = extract_timestamp_utc(path, current_line);
|
||||||
|
if (ts == static_cast<std::time_t>(-1))
|
||||||
|
continue;
|
||||||
|
|
||||||
switch (classify_line(current_line)) {
|
switch (classify_line(current_line)) {
|
||||||
|
|
||||||
case Route::CLIENT_START:
|
case Route::CLIENT_START:
|
||||||
client_events.add_row(target_user_id, log_timestamp,
|
client_events.add_row(
|
||||||
std::variant<std::string, uint8_t>{"start"}, path);
|
target_user_id, ts,
|
||||||
|
std::variant<std::string, uint8_t>{"start"},
|
||||||
|
path
|
||||||
|
);
|
||||||
|
++row_count;
|
||||||
break;
|
break;
|
||||||
|
|
||||||
case Route::CLIENT_END:
|
case Route::CLIENT_END:
|
||||||
client_events.add_row(target_user_id, log_timestamp,
|
client_events.add_row(
|
||||||
std::variant<std::string, uint8_t>{"end"}, path);
|
target_user_id, ts,
|
||||||
|
std::variant<std::string, uint8_t>{"end"},
|
||||||
|
path
|
||||||
|
);
|
||||||
|
++row_count;
|
||||||
break;
|
break;
|
||||||
|
|
||||||
case Route::ACT_COMMAND:
|
case Route::ACT_COMMAND:
|
||||||
activity_events.add_row(target_user_id, log_timestamp, "command", 1.0f);
|
activity_events.add_row(
|
||||||
|
target_user_id, ts,
|
||||||
|
"command", 1.0f
|
||||||
|
);
|
||||||
|
++row_count;
|
||||||
break;
|
break;
|
||||||
|
|
||||||
case Route::ACT_DEATH:
|
case Route::ACT_DEATH:
|
||||||
activity_events.add_row(target_user_id, log_timestamp,
|
activity_events.add_row(
|
||||||
std::variant<std::string, uint8_t>{"death"}, 0.5f);
|
target_user_id, ts,
|
||||||
|
std::variant<std::string, uint8_t>{"death"},
|
||||||
|
0.5f
|
||||||
|
);
|
||||||
|
++row_count;
|
||||||
break;
|
break;
|
||||||
|
|
||||||
case Route::ACT_DIMENSION:
|
case Route::ACT_DIMENSION:
|
||||||
activity_events.add_row(target_user_id, log_timestamp,
|
activity_events.add_row(
|
||||||
std::variant<std::string, uint8_t>{"dimension"}, 0.3f);
|
target_user_id, ts,
|
||||||
break;
|
std::variant<std::string, uint8_t>{"dimension"},
|
||||||
case Route::ACT_OTHER:
|
0.3f
|
||||||
activity_events.add_row(target_user_id, log_timestamp,
|
);
|
||||||
std::variant<std::string, uint8_t>{"other"}, 0.2f);
|
++row_count;
|
||||||
break;
|
break;
|
||||||
|
|
||||||
case Route::CHAT: {
|
case Route::CHAT: {
|
||||||
auto chat_pair = parse_chat(current_line);
|
auto chat = parse_chat(current_line);
|
||||||
chat_events.add_row(target_user_id, log_timestamp,
|
if (!chat.second.empty()) {
|
||||||
chat_pair.first, chat_pair.second);
|
chat_events.add_row(
|
||||||
|
target_user_id, ts,
|
||||||
|
chat.first, chat.second
|
||||||
|
);
|
||||||
|
++row_count;
|
||||||
|
}
|
||||||
break;
|
break;
|
||||||
}
|
}
|
||||||
|
|
||||||
default:
|
default:
|
||||||
break;
|
break;
|
||||||
}
|
}
|
||||||
|
|
||||||
|
// HARD FLUSH CONDITION
|
||||||
|
if (row_count >= MAX_BATCH_ROWS) {
|
||||||
|
activity_events.insert_as_batch(client);
|
||||||
|
chat_events.insert_as_batch(client);
|
||||||
|
client_events.insert_as_batch(client);
|
||||||
|
|
||||||
|
// Destroy and recreate — ONLY valid reset
|
||||||
|
client_events = db_utils::ClientEventsBatch{};
|
||||||
|
chat_events = db_utils::ChatEventsBatch{};
|
||||||
|
activity_events = db_utils::ActivityEventsBatch{};
|
||||||
|
|
||||||
|
row_count = 0;
|
||||||
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
// Insert accumulated batches into ClickHouse
|
// Flush partial batch if we read anything
|
||||||
activity_events.insert_as_batch(client);
|
if (read_any && row_count > 0) {
|
||||||
chat_events.insert_as_batch(client);
|
activity_events.insert_as_batch(client);
|
||||||
client_events.insert_as_batch(client);
|
chat_events.insert_as_batch(client);
|
||||||
|
client_events.insert_as_batch(client);
|
||||||
|
|
||||||
current_line.clear();
|
client_events = db_utils::ClientEventsBatch{};
|
||||||
|
chat_events = db_utils::ChatEventsBatch{};
|
||||||
|
activity_events = db_utils::ActivityEventsBatch{};
|
||||||
|
|
||||||
// Clear EOF state so we can continue reading
|
row_count = 0;
|
||||||
input_file.clear();
|
}
|
||||||
|
|
||||||
// Sleep a bit before polling for new data
|
input_file.clear(); // clear EOF only
|
||||||
std::this_thread::sleep_for(std::chrono::milliseconds(200));
|
std::this_thread::sleep_for(std::chrono::milliseconds(200));
|
||||||
}
|
}
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
||||||
}
|
}
|
||||||
|
|||||||
Reference in New Issue
Block a user