duckdb 0.4.1-dev71.0 → 0.4.1-dev710.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -0,0 +1,185 @@
1
+ #include "duckdb.hpp"
2
+ #include "duckdb_node.hpp"
3
+ #include "napi.h"
4
+
5
+ #include <thread>
6
+
7
+ namespace node_duckdb {
8
+
9
+ Napi::Array EncodeDataChunk(Napi::Env env, duckdb::DataChunk &chunk, bool with_types, bool with_data) {
10
+ Napi::Array col_descs(Napi::Array::New(env, chunk.ColumnCount()));
11
+ for (idx_t col_idx = 0; col_idx < chunk.ColumnCount(); col_idx++) {
12
+ auto col_desc = Napi::Object::New(env);
13
+
14
+ // Make sure we only have flat vectors hereafter (for now)
15
+ auto &chunk_vec = chunk.data[col_idx];
16
+ if (with_data) {
17
+ chunk_vec.Flatten(chunk.size());
18
+ }
19
+
20
+ // Do a post-order DFS traversal
21
+ std::vector<std::tuple<bool, duckdb::Vector *, Napi::Object, size_t, size_t>> pending;
22
+ pending.emplace_back(false, &chunk_vec, Napi::Object::New(env), 0, 0);
23
+
24
+ while (!pending.empty()) {
25
+ // Unpack DFS node
26
+ auto &back = pending.back();
27
+ auto &visited = std::get<0>(back);
28
+ auto &vec = std::get<1>(back);
29
+ auto &desc = std::get<2>(back);
30
+ auto &parent_idx = std::get<3>(back);
31
+ auto &idx_in_parent = std::get<4>(back);
32
+
33
+ // Already visited?
34
+ if (visited) {
35
+ if (pending.size() == 1) {
36
+ col_desc = desc;
37
+ break;
38
+ }
39
+ std::get<2>(pending[parent_idx]).Get("children").As<Napi::Array>().Set(idx_in_parent, desc);
40
+ pending.pop_back();
41
+ continue;
42
+ }
43
+ visited = true;
44
+ auto current_idx = pending.size() - 1;
45
+
46
+ // Store types
47
+ auto &vec_type = vec->GetType();
48
+ if (with_types) {
49
+ desc.Set("sqlType", vec_type.ToString());
50
+ desc.Set("physicalType", TypeIdToString(vec_type.InternalType()));
51
+ }
52
+
53
+ // Create validity vector
54
+ if (with_data) {
55
+ vec->Flatten(chunk.size());
56
+ auto &validity = duckdb::FlatVector::Validity(*vec);
57
+ auto validity_buffer = Napi::Uint8Array::New(env, chunk.size());
58
+ for (idx_t row_idx = 0; row_idx < chunk.size(); row_idx++) {
59
+ validity_buffer[row_idx] = validity.RowIsValid(row_idx);
60
+ }
61
+ desc.Set("validity", validity_buffer);
62
+ }
63
+
64
+ // Create data buffer
65
+ switch (vec_type.id()) {
66
+ case duckdb::LogicalTypeId::TINYINT: {
67
+ if (with_data) {
68
+ auto array = Napi::Int8Array::New(env, chunk.size());
69
+ auto data = duckdb::FlatVector::GetData<int8_t>(*vec);
70
+ for (size_t i = 0; i < chunk.size(); ++i) {
71
+ array[i] = data[i];
72
+ }
73
+ desc.Set("data", array);
74
+ }
75
+ break;
76
+ }
77
+ case duckdb::LogicalTypeId::SMALLINT: {
78
+ if (with_data) {
79
+ auto array = Napi::Int16Array::New(env, chunk.size());
80
+ auto data = duckdb::FlatVector::GetData<int16_t>(*vec);
81
+ for (size_t i = 0; i < chunk.size(); ++i) {
82
+ array[i] = data[i];
83
+ }
84
+ desc.Set("data", array);
85
+ }
86
+ break;
87
+ }
88
+ case duckdb::LogicalTypeId::INTEGER: {
89
+ if (with_data) {
90
+ auto array = Napi::Int32Array::New(env, chunk.size());
91
+ auto data = duckdb::FlatVector::GetData<int32_t>(*vec);
92
+ for (size_t i = 0; i < chunk.size(); ++i) {
93
+ array[i] = data[i];
94
+ }
95
+ desc.Set("data", array);
96
+ }
97
+ break;
98
+ }
99
+ case duckdb::LogicalTypeId::DOUBLE: {
100
+ if (with_data) {
101
+ auto array = Napi::Float64Array::New(env, chunk.size());
102
+ auto data = duckdb::FlatVector::GetData<double>(*vec);
103
+ for (size_t i = 0; i < chunk.size(); ++i) {
104
+ array[i] = data[i];
105
+ }
106
+ desc.Set("data", array);
107
+ }
108
+ break;
109
+ }
110
+ case duckdb::LogicalTypeId::BIGINT:
111
+ case duckdb::LogicalTypeId::TIME:
112
+ case duckdb::LogicalTypeId::TIME_TZ:
113
+ case duckdb::LogicalTypeId::TIMESTAMP_MS:
114
+ case duckdb::LogicalTypeId::TIMESTAMP_NS:
115
+ case duckdb::LogicalTypeId::TIMESTAMP_SEC:
116
+ case duckdb::LogicalTypeId::TIMESTAMP: {
117
+ if (with_data) {
118
+ #if NAPI_VERSION > 5
119
+ auto array = Napi::BigInt64Array::New(env, chunk.size());
120
+ auto data = duckdb::FlatVector::GetData<int64_t>(*vec);
121
+ #else
122
+ auto array = Napi::Float64Array::New(env, chunk.size());
123
+ auto data = duckdb::FlatVector::GetData<int64_t>(*vec);
124
+ #endif
125
+ for (size_t i = 0; i < chunk.size(); ++i) {
126
+ array[i] = data[i];
127
+ }
128
+ desc.Set("data", array);
129
+ }
130
+ break;
131
+ }
132
+ case duckdb::LogicalTypeId::UBIGINT: {
133
+ if (with_data) {
134
+ #if NAPI_VERSION > 5
135
+ auto array = Napi::BigUint64Array::New(env, chunk.size());
136
+ auto data = duckdb::FlatVector::GetData<uint64_t>(*vec);
137
+ #else
138
+ auto array = Napi::Float64Array::New(env, chunk.size());
139
+ auto data = duckdb::FlatVector::GetData<int64_t>(*vec);
140
+ #endif
141
+ for (size_t i = 0; i < chunk.size(); ++i) {
142
+ array[i] = data[i];
143
+ }
144
+ desc.Set("data", array);
145
+ }
146
+ break;
147
+ }
148
+ case duckdb::LogicalTypeId::BLOB:
149
+ case duckdb::LogicalTypeId::VARCHAR: {
150
+ if (with_data) {
151
+ auto array = Napi::Array::New(env, chunk.size());
152
+ auto data = duckdb::FlatVector::GetData<duckdb::string_t>(*vec);
153
+ for (size_t i = 0; i < chunk.size(); ++i) {
154
+ array.Set(i, data[i].GetString());
155
+ }
156
+ desc.Set("data", array);
157
+ }
158
+ break;
159
+ }
160
+ case duckdb::LogicalTypeId::STRUCT: {
161
+ auto child_count = duckdb::StructType::GetChildCount(vec_type);
162
+ auto &entries = duckdb::StructVector::GetEntries(*vec);
163
+ desc.Set("children", Napi::Array::New(env, child_count));
164
+ for (size_t i = 0; i < child_count; ++i) {
165
+ auto c = child_count - 1 - i;
166
+ auto &entry = entries[c];
167
+ auto desc = Napi::Object::New(env);
168
+ auto name = duckdb::StructType::GetChildName(vec_type, c);
169
+ desc.Set("name", name);
170
+ pending.emplace_back(false, entry.get(), desc, current_idx, i);
171
+ }
172
+ break;
173
+ }
174
+ default:
175
+ Napi::TypeError::New(env, "Unsupported UDF argument type " + vec->GetType().ToString())
176
+ .ThrowAsJavaScriptException();
177
+ break;
178
+ }
179
+ }
180
+ col_descs.Set(col_idx, col_desc);
181
+ }
182
+ return col_descs;
183
+ }
184
+
185
+ } // namespace node_duckdb
package/src/database.cpp CHANGED
@@ -22,17 +22,39 @@ Napi::Object Database::Init(Napi::Env env, Napi::Object exports) {
22
22
  }
23
23
 
24
24
  struct OpenTask : public Task {
25
- OpenTask(Database &database_, std::string filename_, bool read_only_, Napi::Function callback_)
26
- : Task(database_, callback_), filename(filename_), read_only(read_only_) {
25
+ OpenTask(Database &database_, std::string filename_, duckdb::AccessMode access_mode_, Napi::Object config_,
26
+ Napi::Function callback_)
27
+ : Task(database_, callback_), filename(filename_) {
28
+
29
+ duckdb_config.options.access_mode = access_mode_;
30
+ Napi::Env env = database_.Env();
31
+ Napi::HandleScope scope(env);
32
+
33
+ if (!config_.IsUndefined()) {
34
+ const Napi::Array config_names = config_.GetPropertyNames();
35
+
36
+ for (duckdb::idx_t config_idx = 0; config_idx < config_names.Length(); config_idx++) {
37
+ std::string key = config_names.Get(config_idx).As<Napi::String>();
38
+ std::string val = config_.Get(key).As<Napi::String>();
39
+ auto config_property = duckdb::DBConfig::GetOptionByName(key);
40
+ if (!config_property) {
41
+ Napi::TypeError::New(env, "Unrecognized configuration property" + key).ThrowAsJavaScriptException();
42
+ return;
43
+ }
44
+ try {
45
+ duckdb_config.SetOption(*config_property, duckdb::Value(val));
46
+ } catch (std::exception &e) {
47
+ Napi::TypeError::New(env, "Failed to set configuration option " + key + ": " + e.what())
48
+ .ThrowAsJavaScriptException();
49
+ return;
50
+ }
51
+ }
52
+ }
27
53
  }
28
54
 
29
55
  void DoWork() override {
30
56
  try {
31
- duckdb::DBConfig config;
32
- if (read_only) {
33
- config.access_mode = duckdb::AccessMode::READ_ONLY;
34
- }
35
- Get<Database>().database = duckdb::make_unique<duckdb::DuckDB>(filename, &config);
57
+ Get<Database>().database = duckdb::make_unique<duckdb::DuckDB>(filename, &duckdb_config);
36
58
  duckdb::ParquetExtension extension;
37
59
  extension.Load(*Get<Database>().database);
38
60
  success = true;
@@ -59,22 +81,34 @@ struct OpenTask : public Task {
59
81
  }
60
82
 
61
83
  std::string filename;
62
- bool read_only = false;
84
+ duckdb::DBConfig duckdb_config;
63
85
  std::string error = "";
64
86
  bool success = false;
65
87
  };
66
88
 
67
89
  Database::Database(const Napi::CallbackInfo &info) : Napi::ObjectWrap<Database>(info), task_inflight(false) {
68
90
  auto env = info.Env();
91
+
69
92
  if (info.Length() < 1 || !info[0].IsString()) {
70
93
  Napi::TypeError::New(env, "Database location expected").ThrowAsJavaScriptException();
71
94
  return;
72
95
  }
73
96
  std::string filename = info[0].As<Napi::String>();
74
97
  unsigned int pos = 1;
98
+
99
+ duckdb::AccessMode access_mode = duckdb::AccessMode::AUTOMATIC;
100
+
75
101
  int mode = 0;
76
102
  if (info.Length() >= pos && info[pos].IsNumber() && Utils::OtherIsInt(info[pos].As<Napi::Number>())) {
77
103
  mode = info[pos++].As<Napi::Number>().Int32Value();
104
+ if (mode == DUCKDB_NODEJS_READONLY) {
105
+ access_mode = duckdb::AccessMode::READ_ONLY;
106
+ }
107
+ }
108
+
109
+ Napi::Object config;
110
+ if (info.Length() >= pos && info[pos].IsObject() && !info[pos].IsFunction()) {
111
+ config = info[pos++].As<Napi::Object>();
78
112
  }
79
113
 
80
114
  Napi::Function callback;
@@ -82,7 +116,7 @@ Database::Database(const Napi::CallbackInfo &info) : Napi::ObjectWrap<Database>(
82
116
  callback = info[pos++].As<Napi::Function>();
83
117
  }
84
118
 
85
- Schedule(env, duckdb::make_unique<OpenTask>(*this, filename, mode == DUCKDB_NODEJS_READONLY, callback));
119
+ Schedule(env, duckdb::make_unique<OpenTask>(*this, filename, access_mode, config, callback));
86
120
  }
87
121
 
88
122
  void Database::Schedule(Napi::Env env, std::unique_ptr<Task> task) {
@@ -93,12 +127,12 @@ void Database::Schedule(Napi::Env env, std::unique_ptr<Task> task) {
93
127
  Process(env);
94
128
  }
95
129
 
96
- static void task_execute(napi_env e, void *data) {
130
+ static void TaskExecuteCallback(napi_env e, void *data) {
97
131
  auto holder = (TaskHolder *)data;
98
132
  holder->task->DoWork();
99
133
  }
100
134
 
101
- static void task_complete(napi_env e, napi_status status, void *data) {
135
+ static void TaskCompleteCallback(napi_env e, napi_status status, void *data) {
102
136
  std::unique_ptr<TaskHolder> holder((TaskHolder *)data);
103
137
  holder->db->TaskComplete(e);
104
138
  if (holder->task->callback.Value().IsFunction()) {
@@ -131,8 +165,8 @@ void Database::Process(Napi::Env env) {
131
165
  holder->task = move(task);
132
166
  holder->db = this;
133
167
 
134
- napi_create_async_work(env, NULL, Napi::String::New(env, "duckdb.Database.Task"), task_execute, task_complete,
135
- holder, &holder->request);
168
+ napi_create_async_work(env, nullptr, Napi::String::New(env, "duckdb.Database.Task"), TaskExecuteCallback,
169
+ TaskCompleteCallback, holder, &holder->request);
136
170
 
137
171
  napi_queue_async_work(env, holder->request);
138
172
  }
@@ -157,7 +191,7 @@ Napi::Value Database::Serialize(const Napi::CallbackInfo &info) {
157
191
  }
158
192
 
159
193
  struct WaitTask : public Task {
160
- WaitTask(Database &database_, Napi::Function callback_) : Task(database_, callback_) {
194
+ WaitTask(Database &database, Napi::Function callback) : Task(database, callback) {
161
195
  }
162
196
 
163
197
  void DoWork() override {
@@ -171,7 +205,7 @@ Napi::Value Database::Wait(const Napi::CallbackInfo &info) {
171
205
  }
172
206
 
173
207
  struct CloseTask : public Task {
174
- CloseTask(Database &database_, Napi::Function callback_) : Task(database_, callback_) {
208
+ CloseTask(Database &database, Napi::Function callback) : Task(database, callback) {
175
209
  }
176
210
 
177
211
  void DoWork() override {