@databricks/appkit 0.44.0 → 0.45.0
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.
- package/dist/appkit/package.js +1 -1
- package/dist/connectors/sql-warehouse/arrow-schema.js +283 -0
- package/dist/connectors/sql-warehouse/arrow-schema.js.map +1 -0
- package/dist/connectors/sql-warehouse/client.js +166 -60
- package/dist/connectors/sql-warehouse/client.js.map +1 -1
- package/dist/errors/base.d.ts +18 -0
- package/dist/errors/base.d.ts.map +1 -1
- package/dist/errors/base.js +20 -0
- package/dist/errors/base.js.map +1 -1
- package/dist/errors/execution.d.ts +34 -2
- package/dist/errors/execution.d.ts.map +1 -1
- package/dist/errors/execution.js +40 -5
- package/dist/errors/execution.js.map +1 -1
- package/dist/plugins/agents/agents.d.ts.map +1 -1
- package/dist/plugins/agents/agents.js.map +1 -1
- package/dist/plugins/analytics/analytics.d.ts +84 -8
- package/dist/plugins/analytics/analytics.d.ts.map +1 -1
- package/dist/plugins/analytics/analytics.js +296 -61
- package/dist/plugins/analytics/analytics.js.map +1 -1
- package/dist/plugins/analytics/index.js +1 -1
- package/dist/plugins/analytics/result-delivery.js +287 -0
- package/dist/plugins/analytics/result-delivery.js.map +1 -0
- package/dist/plugins/analytics/types.d.ts +8 -0
- package/dist/plugins/analytics/types.d.ts.map +1 -1
- package/dist/plugins/analytics/types.js.map +1 -1
- package/dist/plugins/files/plugin.js +1 -1
- package/dist/plugins/genie/genie.js +1 -1
- package/dist/plugins/jobs/types.d.ts +1 -1
- package/dist/plugins/lakebase/lakebase.js +1 -1
- package/dist/shared/src/index.d.ts +1 -0
- package/dist/shared/src/sse/analytics.d.ts +1 -0
- package/dist/shared/src/sse/analytics.js +40 -0
- package/dist/shared/src/sse/analytics.js.map +1 -0
- package/dist/stream/arrow-stream-processor.js +93 -117
- package/dist/stream/arrow-stream-processor.js.map +1 -1
- package/dist/stream/defaults.js +1 -1
- package/dist/stream/defaults.js.map +1 -1
- package/dist/stream/sse-writer.js +3 -2
- package/dist/stream/sse-writer.js.map +1 -1
- package/dist/stream/stream-manager.d.ts.map +1 -1
- package/dist/stream/stream-manager.js +15 -9
- package/dist/stream/stream-manager.js.map +1 -1
- package/dist/stream/types.js.map +1 -1
- package/dist/type-generator/query-registry.js +32 -2
- package/dist/type-generator/query-registry.js.map +1 -1
- package/docs/api/appkit/Class.AppKitError.md +40 -8
- package/docs/api/appkit/Class.AuthenticationError.md +56 -16
- package/docs/api/appkit/Class.ConfigurationError.md +57 -17
- package/docs/api/appkit/Class.ConnectionError.md +57 -17
- package/docs/api/appkit/Class.ExecutionError.md +80 -22
- package/docs/api/appkit/Class.InitializationError.md +55 -15
- package/docs/api/appkit/Class.ServerError.md +55 -15
- package/docs/api/appkit/Class.TunnelError.md +56 -16
- package/docs/api/appkit/Class.ValidationError.md +55 -15
- package/package.json +1 -1
- package/sbom.cdx.json +1 -1
package/dist/appkit/package.js
CHANGED
|
@@ -0,0 +1,283 @@
|
|
|
1
|
+
import { Binary, Bool, DateDay, Decimal, DurationMicrosecond, Field, Float32, Float64, Int16, Int32, Int64, Int8, IntervalYearMonth, List, Map_, Null, Schema, Struct, Table, TimestampMicrosecond, Utf8, tableToIPC } from "apache-arrow";
|
|
2
|
+
|
|
3
|
+
//#region src/connectors/sql-warehouse/arrow-schema.ts
|
|
4
|
+
/**
|
|
5
|
+
* Parse a Databricks SQL type text (the value returned by the Statement
|
|
6
|
+
* Execution API in `ColumnInfo.type_text`) into an Apache Arrow DataType.
|
|
7
|
+
*
|
|
8
|
+
* Supports:
|
|
9
|
+
* - All scalar types (STRING, INT, BIGINT, DECIMAL, TIMESTAMP, etc.)
|
|
10
|
+
* - Parameterized scalars: DECIMAL(p,s), VARCHAR(n), CHAR(n)
|
|
11
|
+
* - Nested types: ARRAY<T>, MAP<K,V>, STRUCT<f1:T1, f2:T2 COMMENT '...', ...>
|
|
12
|
+
* - INTERVAL year-month and day-time variants
|
|
13
|
+
* - Backtick-quoted struct field names with embedded `` `` `` escapes
|
|
14
|
+
*
|
|
15
|
+
* Unknown or unparseable types fall back to Utf8 — empty-Table consumers
|
|
16
|
+
* still see a column with the right name; only the inner type is degraded.
|
|
17
|
+
*/
|
|
18
|
+
function parseDatabricksType(typeText) {
|
|
19
|
+
const parser = new TypeParser(typeText);
|
|
20
|
+
const result = parser.parseType();
|
|
21
|
+
parser.expectEnd();
|
|
22
|
+
return result;
|
|
23
|
+
}
|
|
24
|
+
/**
|
|
25
|
+
* Build an empty Arrow IPC stream (base64-encoded) matching the column schema
|
|
26
|
+
* returned by the warehouse. Used so ARROW_STREAM responses with no rows still
|
|
27
|
+
* deliver a real Arrow Table to the client, preserving the hook's typed
|
|
28
|
+
* contract.
|
|
29
|
+
*/
|
|
30
|
+
function buildEmptyArrowIPCBase64(columns) {
|
|
31
|
+
const ipc = tableToIPC(new Table(new Schema(columns.map((col, index) => {
|
|
32
|
+
const typeText = col.type_text ?? col.type_name ?? "STRING";
|
|
33
|
+
let dataType;
|
|
34
|
+
try {
|
|
35
|
+
dataType = parseDatabricksType(typeText);
|
|
36
|
+
} catch {
|
|
37
|
+
dataType = new Utf8();
|
|
38
|
+
}
|
|
39
|
+
return new Field(col.name && col.name.length > 0 ? col.name : `column_${index}`, dataType, true);
|
|
40
|
+
}))), "stream");
|
|
41
|
+
return Buffer.from(ipc).toString("base64");
|
|
42
|
+
}
|
|
43
|
+
var TypeParser = class {
|
|
44
|
+
input;
|
|
45
|
+
pos = 0;
|
|
46
|
+
constructor(input) {
|
|
47
|
+
this.input = input;
|
|
48
|
+
}
|
|
49
|
+
parseType() {
|
|
50
|
+
this.skipWs();
|
|
51
|
+
let name;
|
|
52
|
+
if (this.peek() === "`") name = this.consumeBacktickIdent();
|
|
53
|
+
else name = this.consumeIdent();
|
|
54
|
+
const upper = name.toUpperCase();
|
|
55
|
+
this.skipWs();
|
|
56
|
+
if (upper === "INTERVAL") return this.parseInterval();
|
|
57
|
+
if (this.peek() === "(") {
|
|
58
|
+
this.consume("(");
|
|
59
|
+
const args = this.parseNumberArgs();
|
|
60
|
+
this.consume(")");
|
|
61
|
+
this.skipWs();
|
|
62
|
+
return this.makeParameterized(upper, args);
|
|
63
|
+
}
|
|
64
|
+
if (this.peek() === "<") {
|
|
65
|
+
this.consume("<");
|
|
66
|
+
const result = this.makeGeneric(upper);
|
|
67
|
+
this.skipWs();
|
|
68
|
+
this.consume(">");
|
|
69
|
+
return result;
|
|
70
|
+
}
|
|
71
|
+
return this.makeScalar(upper);
|
|
72
|
+
}
|
|
73
|
+
expectEnd() {
|
|
74
|
+
this.skipWs();
|
|
75
|
+
if (this.pos < this.input.length) throw new Error(`Unexpected trailing input at position ${this.pos}: "${this.input.slice(this.pos)}"`);
|
|
76
|
+
}
|
|
77
|
+
makeScalar(upper) {
|
|
78
|
+
switch (upper) {
|
|
79
|
+
case "STRING":
|
|
80
|
+
case "VARIANT": return new Utf8();
|
|
81
|
+
case "VARCHAR":
|
|
82
|
+
case "CHAR": return new Utf8();
|
|
83
|
+
case "BINARY":
|
|
84
|
+
case "GEOGRAPHY":
|
|
85
|
+
case "GEOMETRY": return new Binary();
|
|
86
|
+
case "BOOLEAN":
|
|
87
|
+
case "BOOL": return new Bool();
|
|
88
|
+
case "TINYINT":
|
|
89
|
+
case "BYTE": return new Int8();
|
|
90
|
+
case "SMALLINT":
|
|
91
|
+
case "SHORT": return new Int16();
|
|
92
|
+
case "INT":
|
|
93
|
+
case "INTEGER": return new Int32();
|
|
94
|
+
case "BIGINT":
|
|
95
|
+
case "LONG": return new Int64();
|
|
96
|
+
case "FLOAT":
|
|
97
|
+
case "REAL": return new Float32();
|
|
98
|
+
case "DOUBLE": return new Float64();
|
|
99
|
+
case "DECIMAL":
|
|
100
|
+
case "NUMERIC":
|
|
101
|
+
case "DEC": return new Decimal(0, 10, 128);
|
|
102
|
+
case "DATE": return new DateDay();
|
|
103
|
+
case "TIMESTAMP":
|
|
104
|
+
case "TIMESTAMP_LTZ": return new TimestampMicrosecond("UTC");
|
|
105
|
+
case "TIMESTAMP_NTZ": return new TimestampMicrosecond();
|
|
106
|
+
case "VOID":
|
|
107
|
+
case "NULL": return new Null();
|
|
108
|
+
default: return new Utf8();
|
|
109
|
+
}
|
|
110
|
+
}
|
|
111
|
+
makeParameterized(upper, args) {
|
|
112
|
+
switch (upper) {
|
|
113
|
+
case "DECIMAL":
|
|
114
|
+
case "NUMERIC":
|
|
115
|
+
case "DEC": {
|
|
116
|
+
const precision = args[0] ?? 10;
|
|
117
|
+
return new Decimal(args[1] ?? 0, precision, 128);
|
|
118
|
+
}
|
|
119
|
+
case "VARCHAR":
|
|
120
|
+
case "CHAR": return new Utf8();
|
|
121
|
+
default: return new Utf8();
|
|
122
|
+
}
|
|
123
|
+
}
|
|
124
|
+
makeGeneric(upper) {
|
|
125
|
+
switch (upper) {
|
|
126
|
+
case "ARRAY": return new List(new Field("item", this.parseType(), true));
|
|
127
|
+
case "MAP": {
|
|
128
|
+
const keyType = this.parseType();
|
|
129
|
+
this.skipWs();
|
|
130
|
+
this.consume(",");
|
|
131
|
+
this.skipWs();
|
|
132
|
+
const valueType = this.parseType();
|
|
133
|
+
return new Map_(new Field("entries", new Struct([new Field("key", keyType, false), new Field("value", valueType, true)]), false), false);
|
|
134
|
+
}
|
|
135
|
+
case "STRUCT": return this.parseStructFields();
|
|
136
|
+
default:
|
|
137
|
+
this.skipBalancedAngles();
|
|
138
|
+
return new Utf8();
|
|
139
|
+
}
|
|
140
|
+
}
|
|
141
|
+
parseStructFields() {
|
|
142
|
+
const fields = [];
|
|
143
|
+
while (true) {
|
|
144
|
+
this.skipWs();
|
|
145
|
+
if (this.peek() === ">") break;
|
|
146
|
+
let name;
|
|
147
|
+
if (this.peek() === "`") name = this.consumeBacktickIdent();
|
|
148
|
+
else name = this.consumeIdent();
|
|
149
|
+
this.skipWs();
|
|
150
|
+
this.consume(":");
|
|
151
|
+
this.skipWs();
|
|
152
|
+
const type = this.parseType();
|
|
153
|
+
this.skipWs();
|
|
154
|
+
while (this.peekKeyword("NOT")) {
|
|
155
|
+
this.consumeIdent();
|
|
156
|
+
this.skipWs();
|
|
157
|
+
if (this.peekKeyword("NULL")) this.consumeIdent();
|
|
158
|
+
this.skipWs();
|
|
159
|
+
}
|
|
160
|
+
if (this.peekKeyword("COMMENT")) {
|
|
161
|
+
this.consumeIdent();
|
|
162
|
+
this.skipWs();
|
|
163
|
+
this.consumeStringLiteral();
|
|
164
|
+
this.skipWs();
|
|
165
|
+
}
|
|
166
|
+
fields.push(new Field(name, type, true));
|
|
167
|
+
this.skipWs();
|
|
168
|
+
if (this.peek() === ",") this.consume(",");
|
|
169
|
+
else break;
|
|
170
|
+
}
|
|
171
|
+
return new Struct(fields);
|
|
172
|
+
}
|
|
173
|
+
parseInterval() {
|
|
174
|
+
const seen = [];
|
|
175
|
+
while (this.pos < this.input.length) {
|
|
176
|
+
this.skipWs();
|
|
177
|
+
const c = this.peek();
|
|
178
|
+
if (c === "" || c === "," || c === ">" || c === ")") break;
|
|
179
|
+
const word = this.consumeIdent().toUpperCase();
|
|
180
|
+
seen.push(word);
|
|
181
|
+
}
|
|
182
|
+
return seen.some((w) => w === "YEAR" || w === "MONTH") ? new IntervalYearMonth() : new DurationMicrosecond();
|
|
183
|
+
}
|
|
184
|
+
parseNumberArgs() {
|
|
185
|
+
const args = [];
|
|
186
|
+
while (true) {
|
|
187
|
+
this.skipWs();
|
|
188
|
+
if (this.peek() === ")") break;
|
|
189
|
+
args.push(this.consumeNumber());
|
|
190
|
+
this.skipWs();
|
|
191
|
+
if (this.peek() === ",") this.consume(",");
|
|
192
|
+
else break;
|
|
193
|
+
}
|
|
194
|
+
return args;
|
|
195
|
+
}
|
|
196
|
+
peek() {
|
|
197
|
+
return this.input[this.pos] ?? "";
|
|
198
|
+
}
|
|
199
|
+
peekKeyword(word) {
|
|
200
|
+
if (this.input.slice(this.pos, this.pos + word.length).toUpperCase() !== word.toUpperCase()) return false;
|
|
201
|
+
const next = this.input[this.pos + word.length] ?? "";
|
|
202
|
+
return !/[A-Za-z0-9_]/.test(next);
|
|
203
|
+
}
|
|
204
|
+
consume(expected) {
|
|
205
|
+
if (this.peek() !== expected) throw new Error(`Expected "${expected}" at position ${this.pos}, got "${this.peek()}" in "${this.input}"`);
|
|
206
|
+
this.pos++;
|
|
207
|
+
}
|
|
208
|
+
skipWs() {
|
|
209
|
+
while (this.pos < this.input.length && /\s/.test(this.input[this.pos] ?? "")) this.pos++;
|
|
210
|
+
}
|
|
211
|
+
consumeIdent() {
|
|
212
|
+
const start = this.pos;
|
|
213
|
+
while (this.pos < this.input.length && /[A-Za-z0-9_]/.test(this.input[this.pos] ?? "")) this.pos++;
|
|
214
|
+
if (this.pos === start) throw new Error(`Expected identifier at position ${this.pos}, got "${this.peek()}" in "${this.input}"`);
|
|
215
|
+
return this.input.slice(start, this.pos);
|
|
216
|
+
}
|
|
217
|
+
consumeBacktickIdent() {
|
|
218
|
+
this.consume("`");
|
|
219
|
+
let value = "";
|
|
220
|
+
while (this.pos < this.input.length) {
|
|
221
|
+
if (this.input[this.pos] === "`") {
|
|
222
|
+
if (this.input[this.pos + 1] === "`") {
|
|
223
|
+
value += "`";
|
|
224
|
+
this.pos += 2;
|
|
225
|
+
continue;
|
|
226
|
+
}
|
|
227
|
+
break;
|
|
228
|
+
}
|
|
229
|
+
value += this.input[this.pos];
|
|
230
|
+
this.pos++;
|
|
231
|
+
}
|
|
232
|
+
this.consume("`");
|
|
233
|
+
return value;
|
|
234
|
+
}
|
|
235
|
+
consumeNumber() {
|
|
236
|
+
const start = this.pos;
|
|
237
|
+
while (this.pos < this.input.length && /[0-9]/.test(this.input[this.pos] ?? "")) this.pos++;
|
|
238
|
+
if (this.pos === start) throw new Error(`Expected number at position ${this.pos}, got "${this.peek()}" in "${this.input}"`);
|
|
239
|
+
return Number.parseInt(this.input.slice(start, this.pos), 10);
|
|
240
|
+
}
|
|
241
|
+
consumeStringLiteral() {
|
|
242
|
+
const quote = this.peek();
|
|
243
|
+
if (quote !== "'" && quote !== "\"") throw new Error(`Expected string literal at position ${this.pos}, got "${quote}" in "${this.input}"`);
|
|
244
|
+
this.pos++;
|
|
245
|
+
let value = "";
|
|
246
|
+
while (this.pos < this.input.length) {
|
|
247
|
+
const c = this.input[this.pos];
|
|
248
|
+
if (c === "\\") {
|
|
249
|
+
const next = this.input[this.pos + 1];
|
|
250
|
+
if (next !== void 0) {
|
|
251
|
+
value += next;
|
|
252
|
+
this.pos += 2;
|
|
253
|
+
continue;
|
|
254
|
+
}
|
|
255
|
+
this.pos++;
|
|
256
|
+
continue;
|
|
257
|
+
}
|
|
258
|
+
if (c === quote) {
|
|
259
|
+
this.pos++;
|
|
260
|
+
return value;
|
|
261
|
+
}
|
|
262
|
+
value += c;
|
|
263
|
+
this.pos++;
|
|
264
|
+
}
|
|
265
|
+
throw new Error(`Unterminated string literal in "${this.input}"`);
|
|
266
|
+
}
|
|
267
|
+
skipBalancedAngles() {
|
|
268
|
+
let depth = 1;
|
|
269
|
+
while (this.pos < this.input.length && depth > 0) {
|
|
270
|
+
const c = this.peek();
|
|
271
|
+
if (c === "<") depth++;
|
|
272
|
+
else if (c === ">") {
|
|
273
|
+
depth--;
|
|
274
|
+
if (depth === 0) return;
|
|
275
|
+
}
|
|
276
|
+
this.pos++;
|
|
277
|
+
}
|
|
278
|
+
}
|
|
279
|
+
};
|
|
280
|
+
|
|
281
|
+
//#endregion
|
|
282
|
+
export { buildEmptyArrowIPCBase64 };
|
|
283
|
+
//# sourceMappingURL=arrow-schema.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"arrow-schema.js","names":[],"sources":["../../../src/connectors/sql-warehouse/arrow-schema.ts"],"sourcesContent":["import {\n Binary,\n Bool,\n type DataType,\n DateDay,\n Decimal,\n DurationMicrosecond,\n Field,\n Float32,\n Float64,\n Int8,\n Int16,\n Int32,\n Int64,\n IntervalYearMonth,\n List,\n Map_,\n Null,\n Schema,\n Struct,\n Table,\n TimestampMicrosecond,\n tableToIPC,\n Utf8,\n} from \"apache-arrow\";\n\n/**\n * Parse a Databricks SQL type text (the value returned by the Statement\n * Execution API in `ColumnInfo.type_text`) into an Apache Arrow DataType.\n *\n * Supports:\n * - All scalar types (STRING, INT, BIGINT, DECIMAL, TIMESTAMP, etc.)\n * - Parameterized scalars: DECIMAL(p,s), VARCHAR(n), CHAR(n)\n * - Nested types: ARRAY<T>, MAP<K,V>, STRUCT<f1:T1, f2:T2 COMMENT '...', ...>\n * - INTERVAL year-month and day-time variants\n * - Backtick-quoted struct field names with embedded `` `` `` escapes\n *\n * Unknown or unparseable types fall back to Utf8 — empty-Table consumers\n * still see a column with the right name; only the inner type is degraded.\n */\nexport function parseDatabricksType(typeText: string): DataType {\n const parser = new TypeParser(typeText);\n const result = parser.parseType();\n parser.expectEnd();\n return result;\n}\n\n/**\n * Build an empty Arrow IPC stream (base64-encoded) matching the column schema\n * returned by the warehouse. Used so ARROW_STREAM responses with no rows still\n * deliver a real Arrow Table to the client, preserving the hook's typed\n * contract.\n */\nexport function buildEmptyArrowIPCBase64(\n columns: Array<{\n name?: string;\n type_text?: string;\n type_name?: string;\n }>,\n): string {\n const fields = columns.map((col, index) => {\n const typeText = col.type_text ?? col.type_name ?? \"STRING\";\n let dataType: DataType;\n try {\n dataType = parseDatabricksType(typeText);\n } catch {\n dataType = new Utf8();\n }\n const name = col.name && col.name.length > 0 ? col.name : `column_${index}`;\n return new Field(name, dataType, true);\n });\n const schema = new Schema(fields);\n const table = new Table(schema);\n const ipc = tableToIPC(table, \"stream\");\n return Buffer.from(ipc).toString(\"base64\");\n}\n\n// ============================================================================\n// Recursive-descent parser\n// ============================================================================\n\nclass TypeParser {\n private readonly input: string;\n private pos = 0;\n\n constructor(input: string) {\n this.input = input;\n }\n\n parseType(): DataType {\n this.skipWs();\n\n let name: string;\n if (this.peek() === \"`\") {\n name = this.consumeBacktickIdent();\n } else {\n name = this.consumeIdent();\n }\n const upper = name.toUpperCase();\n\n this.skipWs();\n\n if (upper === \"INTERVAL\") {\n return this.parseInterval();\n }\n\n if (this.peek() === \"(\") {\n this.consume(\"(\");\n const args = this.parseNumberArgs();\n this.consume(\")\");\n this.skipWs();\n return this.makeParameterized(upper, args);\n }\n\n if (this.peek() === \"<\") {\n this.consume(\"<\");\n const result = this.makeGeneric(upper);\n this.skipWs();\n this.consume(\">\");\n return result;\n }\n\n return this.makeScalar(upper);\n }\n\n expectEnd(): void {\n this.skipWs();\n if (this.pos < this.input.length) {\n throw new Error(\n `Unexpected trailing input at position ${this.pos}: \"${this.input.slice(this.pos)}\"`,\n );\n }\n }\n\n // ─── Type constructors ───────────────────────────────────\n\n private makeScalar(upper: string): DataType {\n switch (upper) {\n case \"STRING\":\n case \"VARIANT\":\n return new Utf8();\n case \"VARCHAR\":\n case \"CHAR\":\n return new Utf8();\n case \"BINARY\":\n case \"GEOGRAPHY\":\n case \"GEOMETRY\":\n return new Binary();\n case \"BOOLEAN\":\n case \"BOOL\":\n return new Bool();\n case \"TINYINT\":\n case \"BYTE\":\n return new Int8();\n case \"SMALLINT\":\n case \"SHORT\":\n return new Int16();\n case \"INT\":\n case \"INTEGER\":\n return new Int32();\n case \"BIGINT\":\n case \"LONG\":\n return new Int64();\n case \"FLOAT\":\n case \"REAL\":\n return new Float32();\n case \"DOUBLE\":\n return new Float64();\n case \"DECIMAL\":\n case \"NUMERIC\":\n case \"DEC\":\n return new Decimal(0, 10, 128);\n case \"DATE\":\n return new DateDay();\n case \"TIMESTAMP\":\n case \"TIMESTAMP_LTZ\":\n return new TimestampMicrosecond(\"UTC\");\n case \"TIMESTAMP_NTZ\":\n return new TimestampMicrosecond();\n case \"VOID\":\n case \"NULL\":\n return new Null();\n default:\n return new Utf8();\n }\n }\n\n private makeParameterized(upper: string, args: number[]): DataType {\n switch (upper) {\n case \"DECIMAL\":\n case \"NUMERIC\":\n case \"DEC\": {\n const precision = args[0] ?? 10;\n const scale = args[1] ?? 0;\n // Arrow JS Decimal constructor signature is (scale, precision, bitWidth).\n return new Decimal(scale, precision, 128);\n }\n case \"VARCHAR\":\n case \"CHAR\":\n return new Utf8();\n default:\n return new Utf8();\n }\n }\n\n private makeGeneric(upper: string): DataType {\n switch (upper) {\n case \"ARRAY\": {\n const inner = this.parseType();\n return new List(new Field(\"item\", inner, true));\n }\n case \"MAP\": {\n const keyType = this.parseType();\n this.skipWs();\n this.consume(\",\");\n this.skipWs();\n const valueType = this.parseType();\n const entriesStruct = new Struct([\n new Field(\"key\", keyType, false),\n new Field(\"value\", valueType, true),\n ]);\n return new Map_(new Field(\"entries\", entriesStruct, false), false);\n }\n case \"STRUCT\":\n return this.parseStructFields();\n default:\n // Unknown generic — skip to matching '>' and fall back.\n this.skipBalancedAngles();\n return new Utf8();\n }\n }\n\n private parseStructFields(): DataType {\n const fields: Field[] = [];\n while (true) {\n this.skipWs();\n if (this.peek() === \">\") break;\n\n let name: string;\n if (this.peek() === \"`\") {\n name = this.consumeBacktickIdent();\n } else {\n name = this.consumeIdent();\n }\n\n this.skipWs();\n this.consume(\":\");\n this.skipWs();\n\n const type = this.parseType();\n\n // Optional `NOT NULL` and `COMMENT '...'`. Both are accepted by\n // Databricks DDL and may appear in `type_text`.\n this.skipWs();\n while (this.peekKeyword(\"NOT\")) {\n this.consumeIdent();\n this.skipWs();\n if (this.peekKeyword(\"NULL\")) {\n this.consumeIdent();\n }\n this.skipWs();\n }\n if (this.peekKeyword(\"COMMENT\")) {\n this.consumeIdent();\n this.skipWs();\n this.consumeStringLiteral();\n this.skipWs();\n }\n\n fields.push(new Field(name, type, true));\n\n this.skipWs();\n if (this.peek() === \",\") {\n this.consume(\",\");\n } else {\n break;\n }\n }\n return new Struct(fields);\n }\n\n private parseInterval(): DataType {\n // Grammar: INTERVAL <unit> [TO <unit>]\n // YEAR / MONTH variants -> IntervalYearMonth\n // DAY / HOUR / MINUTE / SECOND variants -> Duration(microsecond)\n const seen: string[] = [];\n while (this.pos < this.input.length) {\n this.skipWs();\n const c = this.peek();\n if (c === \"\" || c === \",\" || c === \">\" || c === \")\") break;\n const word = this.consumeIdent().toUpperCase();\n seen.push(word);\n }\n const isYearMonth = seen.some((w) => w === \"YEAR\" || w === \"MONTH\");\n return isYearMonth ? new IntervalYearMonth() : new DurationMicrosecond();\n }\n\n private parseNumberArgs(): number[] {\n const args: number[] = [];\n while (true) {\n this.skipWs();\n if (this.peek() === \")\") break;\n args.push(this.consumeNumber());\n this.skipWs();\n if (this.peek() === \",\") {\n this.consume(\",\");\n } else {\n break;\n }\n }\n return args;\n }\n\n // ─── Token utilities ─────────────────────────────────────\n\n private peek(): string {\n return this.input[this.pos] ?? \"\";\n }\n\n private peekKeyword(word: string): boolean {\n const slice = this.input.slice(this.pos, this.pos + word.length);\n if (slice.toUpperCase() !== word.toUpperCase()) return false;\n // Must be followed by a non-identifier character (boundary check).\n const next = this.input[this.pos + word.length] ?? \"\";\n return !/[A-Za-z0-9_]/.test(next);\n }\n\n private consume(expected: string): void {\n if (this.peek() !== expected) {\n throw new Error(\n `Expected \"${expected}\" at position ${this.pos}, got \"${this.peek()}\" in \"${this.input}\"`,\n );\n }\n this.pos++;\n }\n\n private skipWs(): void {\n while (\n this.pos < this.input.length &&\n /\\s/.test(this.input[this.pos] ?? \"\")\n ) {\n this.pos++;\n }\n }\n\n private consumeIdent(): string {\n const start = this.pos;\n while (\n this.pos < this.input.length &&\n /[A-Za-z0-9_]/.test(this.input[this.pos] ?? \"\")\n ) {\n this.pos++;\n }\n if (this.pos === start) {\n throw new Error(\n `Expected identifier at position ${this.pos}, got \"${this.peek()}\" in \"${this.input}\"`,\n );\n }\n return this.input.slice(start, this.pos);\n }\n\n private consumeBacktickIdent(): string {\n this.consume(\"`\");\n let value = \"\";\n while (this.pos < this.input.length) {\n if (this.input[this.pos] === \"`\") {\n if (this.input[this.pos + 1] === \"`\") {\n value += \"`\";\n this.pos += 2;\n continue;\n }\n break;\n }\n value += this.input[this.pos];\n this.pos++;\n }\n this.consume(\"`\");\n return value;\n }\n\n private consumeNumber(): number {\n const start = this.pos;\n while (\n this.pos < this.input.length &&\n /[0-9]/.test(this.input[this.pos] ?? \"\")\n ) {\n this.pos++;\n }\n if (this.pos === start) {\n throw new Error(\n `Expected number at position ${this.pos}, got \"${this.peek()}\" in \"${this.input}\"`,\n );\n }\n return Number.parseInt(this.input.slice(start, this.pos), 10);\n }\n\n private consumeStringLiteral(): string {\n const quote = this.peek();\n if (quote !== \"'\" && quote !== '\"') {\n throw new Error(\n `Expected string literal at position ${this.pos}, got \"${quote}\" in \"${this.input}\"`,\n );\n }\n this.pos++;\n let value = \"\";\n while (this.pos < this.input.length) {\n const c = this.input[this.pos];\n if (c === \"\\\\\") {\n // Escape sequence: keep the next char verbatim.\n const next = this.input[this.pos + 1];\n if (next !== undefined) {\n value += next;\n this.pos += 2;\n continue;\n }\n this.pos++;\n continue;\n }\n if (c === quote) {\n this.pos++;\n return value;\n }\n value += c;\n this.pos++;\n }\n throw new Error(`Unterminated string literal in \"${this.input}\"`);\n }\n\n private skipBalancedAngles(): void {\n let depth = 1;\n while (this.pos < this.input.length && depth > 0) {\n const c = this.peek();\n if (c === \"<\") depth++;\n else if (c === \">\") {\n depth--;\n if (depth === 0) return;\n }\n this.pos++;\n }\n }\n}\n"],"mappings":";;;;;;;;;;;;;;;;;AAwCA,SAAgB,oBAAoB,UAA4B;CAC9D,MAAM,SAAS,IAAI,WAAW,SAAS;CACvC,MAAM,SAAS,OAAO,WAAW;AACjC,QAAO,WAAW;AAClB,QAAO;;;;;;;;AAST,SAAgB,yBACd,SAKQ;CAcR,MAAM,MAAM,WADE,IAAI,MADH,IAAI,OAXJ,QAAQ,KAAK,KAAK,UAAU;EACzC,MAAM,WAAW,IAAI,aAAa,IAAI,aAAa;EACnD,IAAI;AACJ,MAAI;AACF,cAAW,oBAAoB,SAAS;UAClC;AACN,cAAW,IAAI,MAAM;;AAGvB,SAAO,IAAI,MADE,IAAI,QAAQ,IAAI,KAAK,SAAS,IAAI,IAAI,OAAO,UAAU,SAC7C,UAAU,KAAK;GACtC,CAC+B,CACF,EACD,SAAS;AACvC,QAAO,OAAO,KAAK,IAAI,CAAC,SAAS,SAAS;;AAO5C,IAAM,aAAN,MAAiB;CACf,AAAiB;CACjB,AAAQ,MAAM;CAEd,YAAY,OAAe;AACzB,OAAK,QAAQ;;CAGf,YAAsB;AACpB,OAAK,QAAQ;EAEb,IAAI;AACJ,MAAI,KAAK,MAAM,KAAK,IAClB,QAAO,KAAK,sBAAsB;MAElC,QAAO,KAAK,cAAc;EAE5B,MAAM,QAAQ,KAAK,aAAa;AAEhC,OAAK,QAAQ;AAEb,MAAI,UAAU,WACZ,QAAO,KAAK,eAAe;AAG7B,MAAI,KAAK,MAAM,KAAK,KAAK;AACvB,QAAK,QAAQ,IAAI;GACjB,MAAM,OAAO,KAAK,iBAAiB;AACnC,QAAK,QAAQ,IAAI;AACjB,QAAK,QAAQ;AACb,UAAO,KAAK,kBAAkB,OAAO,KAAK;;AAG5C,MAAI,KAAK,MAAM,KAAK,KAAK;AACvB,QAAK,QAAQ,IAAI;GACjB,MAAM,SAAS,KAAK,YAAY,MAAM;AACtC,QAAK,QAAQ;AACb,QAAK,QAAQ,IAAI;AACjB,UAAO;;AAGT,SAAO,KAAK,WAAW,MAAM;;CAG/B,YAAkB;AAChB,OAAK,QAAQ;AACb,MAAI,KAAK,MAAM,KAAK,MAAM,OACxB,OAAM,IAAI,MACR,yCAAyC,KAAK,IAAI,KAAK,KAAK,MAAM,MAAM,KAAK,IAAI,CAAC,GACnF;;CAML,AAAQ,WAAW,OAAyB;AAC1C,UAAQ,OAAR;GACE,KAAK;GACL,KAAK,UACH,QAAO,IAAI,MAAM;GACnB,KAAK;GACL,KAAK,OACH,QAAO,IAAI,MAAM;GACnB,KAAK;GACL,KAAK;GACL,KAAK,WACH,QAAO,IAAI,QAAQ;GACrB,KAAK;GACL,KAAK,OACH,QAAO,IAAI,MAAM;GACnB,KAAK;GACL,KAAK,OACH,QAAO,IAAI,MAAM;GACnB,KAAK;GACL,KAAK,QACH,QAAO,IAAI,OAAO;GACpB,KAAK;GACL,KAAK,UACH,QAAO,IAAI,OAAO;GACpB,KAAK;GACL,KAAK,OACH,QAAO,IAAI,OAAO;GACpB,KAAK;GACL,KAAK,OACH,QAAO,IAAI,SAAS;GACtB,KAAK,SACH,QAAO,IAAI,SAAS;GACtB,KAAK;GACL,KAAK;GACL,KAAK,MACH,QAAO,IAAI,QAAQ,GAAG,IAAI,IAAI;GAChC,KAAK,OACH,QAAO,IAAI,SAAS;GACtB,KAAK;GACL,KAAK,gBACH,QAAO,IAAI,qBAAqB,MAAM;GACxC,KAAK,gBACH,QAAO,IAAI,sBAAsB;GACnC,KAAK;GACL,KAAK,OACH,QAAO,IAAI,MAAM;GACnB,QACE,QAAO,IAAI,MAAM;;;CAIvB,AAAQ,kBAAkB,OAAe,MAA0B;AACjE,UAAQ,OAAR;GACE,KAAK;GACL,KAAK;GACL,KAAK,OAAO;IACV,MAAM,YAAY,KAAK,MAAM;AAG7B,WAAO,IAAI,QAFG,KAAK,MAAM,GAEC,WAAW,IAAI;;GAE3C,KAAK;GACL,KAAK,OACH,QAAO,IAAI,MAAM;GACnB,QACE,QAAO,IAAI,MAAM;;;CAIvB,AAAQ,YAAY,OAAyB;AAC3C,UAAQ,OAAR;GACE,KAAK,QAEH,QAAO,IAAI,KAAK,IAAI,MAAM,QADZ,KAAK,WAAW,EACW,KAAK,CAAC;GAEjD,KAAK,OAAO;IACV,MAAM,UAAU,KAAK,WAAW;AAChC,SAAK,QAAQ;AACb,SAAK,QAAQ,IAAI;AACjB,SAAK,QAAQ;IACb,MAAM,YAAY,KAAK,WAAW;AAKlC,WAAO,IAAI,KAAK,IAAI,MAAM,WAJJ,IAAI,OAAO,CAC/B,IAAI,MAAM,OAAO,SAAS,MAAM,EAChC,IAAI,MAAM,SAAS,WAAW,KAAK,CACpC,CAAC,EACkD,MAAM,EAAE,MAAM;;GAEpE,KAAK,SACH,QAAO,KAAK,mBAAmB;GACjC;AAEE,SAAK,oBAAoB;AACzB,WAAO,IAAI,MAAM;;;CAIvB,AAAQ,oBAA8B;EACpC,MAAM,SAAkB,EAAE;AAC1B,SAAO,MAAM;AACX,QAAK,QAAQ;AACb,OAAI,KAAK,MAAM,KAAK,IAAK;GAEzB,IAAI;AACJ,OAAI,KAAK,MAAM,KAAK,IAClB,QAAO,KAAK,sBAAsB;OAElC,QAAO,KAAK,cAAc;AAG5B,QAAK,QAAQ;AACb,QAAK,QAAQ,IAAI;AACjB,QAAK,QAAQ;GAEb,MAAM,OAAO,KAAK,WAAW;AAI7B,QAAK,QAAQ;AACb,UAAO,KAAK,YAAY,MAAM,EAAE;AAC9B,SAAK,cAAc;AACnB,SAAK,QAAQ;AACb,QAAI,KAAK,YAAY,OAAO,CAC1B,MAAK,cAAc;AAErB,SAAK,QAAQ;;AAEf,OAAI,KAAK,YAAY,UAAU,EAAE;AAC/B,SAAK,cAAc;AACnB,SAAK,QAAQ;AACb,SAAK,sBAAsB;AAC3B,SAAK,QAAQ;;AAGf,UAAO,KAAK,IAAI,MAAM,MAAM,MAAM,KAAK,CAAC;AAExC,QAAK,QAAQ;AACb,OAAI,KAAK,MAAM,KAAK,IAClB,MAAK,QAAQ,IAAI;OAEjB;;AAGJ,SAAO,IAAI,OAAO,OAAO;;CAG3B,AAAQ,gBAA0B;EAIhC,MAAM,OAAiB,EAAE;AACzB,SAAO,KAAK,MAAM,KAAK,MAAM,QAAQ;AACnC,QAAK,QAAQ;GACb,MAAM,IAAI,KAAK,MAAM;AACrB,OAAI,MAAM,MAAM,MAAM,OAAO,MAAM,OAAO,MAAM,IAAK;GACrD,MAAM,OAAO,KAAK,cAAc,CAAC,aAAa;AAC9C,QAAK,KAAK,KAAK;;AAGjB,SADoB,KAAK,MAAM,MAAM,MAAM,UAAU,MAAM,QAAQ,GAC9C,IAAI,mBAAmB,GAAG,IAAI,qBAAqB;;CAG1E,AAAQ,kBAA4B;EAClC,MAAM,OAAiB,EAAE;AACzB,SAAO,MAAM;AACX,QAAK,QAAQ;AACb,OAAI,KAAK,MAAM,KAAK,IAAK;AACzB,QAAK,KAAK,KAAK,eAAe,CAAC;AAC/B,QAAK,QAAQ;AACb,OAAI,KAAK,MAAM,KAAK,IAClB,MAAK,QAAQ,IAAI;OAEjB;;AAGJ,SAAO;;CAKT,AAAQ,OAAe;AACrB,SAAO,KAAK,MAAM,KAAK,QAAQ;;CAGjC,AAAQ,YAAY,MAAuB;AAEzC,MADc,KAAK,MAAM,MAAM,KAAK,KAAK,KAAK,MAAM,KAAK,OAAO,CACtD,aAAa,KAAK,KAAK,aAAa,CAAE,QAAO;EAEvD,MAAM,OAAO,KAAK,MAAM,KAAK,MAAM,KAAK,WAAW;AACnD,SAAO,CAAC,eAAe,KAAK,KAAK;;CAGnC,AAAQ,QAAQ,UAAwB;AACtC,MAAI,KAAK,MAAM,KAAK,SAClB,OAAM,IAAI,MACR,aAAa,SAAS,gBAAgB,KAAK,IAAI,SAAS,KAAK,MAAM,CAAC,QAAQ,KAAK,MAAM,GACxF;AAEH,OAAK;;CAGP,AAAQ,SAAe;AACrB,SACE,KAAK,MAAM,KAAK,MAAM,UACtB,KAAK,KAAK,KAAK,MAAM,KAAK,QAAQ,GAAG,CAErC,MAAK;;CAIT,AAAQ,eAAuB;EAC7B,MAAM,QAAQ,KAAK;AACnB,SACE,KAAK,MAAM,KAAK,MAAM,UACtB,eAAe,KAAK,KAAK,MAAM,KAAK,QAAQ,GAAG,CAE/C,MAAK;AAEP,MAAI,KAAK,QAAQ,MACf,OAAM,IAAI,MACR,mCAAmC,KAAK,IAAI,SAAS,KAAK,MAAM,CAAC,QAAQ,KAAK,MAAM,GACrF;AAEH,SAAO,KAAK,MAAM,MAAM,OAAO,KAAK,IAAI;;CAG1C,AAAQ,uBAA+B;AACrC,OAAK,QAAQ,IAAI;EACjB,IAAI,QAAQ;AACZ,SAAO,KAAK,MAAM,KAAK,MAAM,QAAQ;AACnC,OAAI,KAAK,MAAM,KAAK,SAAS,KAAK;AAChC,QAAI,KAAK,MAAM,KAAK,MAAM,OAAO,KAAK;AACpC,cAAS;AACT,UAAK,OAAO;AACZ;;AAEF;;AAEF,YAAS,KAAK,MAAM,KAAK;AACzB,QAAK;;AAEP,OAAK,QAAQ,IAAI;AACjB,SAAO;;CAGT,AAAQ,gBAAwB;EAC9B,MAAM,QAAQ,KAAK;AACnB,SACE,KAAK,MAAM,KAAK,MAAM,UACtB,QAAQ,KAAK,KAAK,MAAM,KAAK,QAAQ,GAAG,CAExC,MAAK;AAEP,MAAI,KAAK,QAAQ,MACf,OAAM,IAAI,MACR,+BAA+B,KAAK,IAAI,SAAS,KAAK,MAAM,CAAC,QAAQ,KAAK,MAAM,GACjF;AAEH,SAAO,OAAO,SAAS,KAAK,MAAM,MAAM,OAAO,KAAK,IAAI,EAAE,GAAG;;CAG/D,AAAQ,uBAA+B;EACrC,MAAM,QAAQ,KAAK,MAAM;AACzB,MAAI,UAAU,OAAO,UAAU,KAC7B,OAAM,IAAI,MACR,uCAAuC,KAAK,IAAI,SAAS,MAAM,QAAQ,KAAK,MAAM,GACnF;AAEH,OAAK;EACL,IAAI,QAAQ;AACZ,SAAO,KAAK,MAAM,KAAK,MAAM,QAAQ;GACnC,MAAM,IAAI,KAAK,MAAM,KAAK;AAC1B,OAAI,MAAM,MAAM;IAEd,MAAM,OAAO,KAAK,MAAM,KAAK,MAAM;AACnC,QAAI,SAAS,QAAW;AACtB,cAAS;AACT,UAAK,OAAO;AACZ;;AAEF,SAAK;AACL;;AAEF,OAAI,MAAM,OAAO;AACf,SAAK;AACL,WAAO;;AAET,YAAS;AACT,QAAK;;AAEP,QAAM,IAAI,MAAM,mCAAmC,KAAK,MAAM,GAAG;;CAGnE,AAAQ,qBAA2B;EACjC,IAAI,QAAQ;AACZ,SAAO,KAAK,MAAM,KAAK,MAAM,UAAU,QAAQ,GAAG;GAChD,MAAM,IAAI,KAAK,MAAM;AACrB,OAAI,MAAM,IAAK;YACN,MAAM,KAAK;AAClB;AACA,QAAI,UAAU,EAAG;;AAEnB,QAAK"}
|
|
@@ -8,6 +8,7 @@ import "../../errors/index.js";
|
|
|
8
8
|
import { TelemetryManager } from "../../telemetry/telemetry-manager.js";
|
|
9
9
|
import { SpanKind, SpanStatusCode } from "../../telemetry/index.js";
|
|
10
10
|
import { ArrowStreamProcessor } from "../../stream/arrow-stream-processor.js";
|
|
11
|
+
import { buildEmptyArrowIPCBase64 } from "./arrow-schema.js";
|
|
11
12
|
import { executeStatementDefaults } from "./defaults.js";
|
|
12
13
|
import { WarehousePollBackoff } from "./warehouse-poll-backoff.js";
|
|
13
14
|
import { WarehouseStatusEmitter } from "./warehouse-status-emitter.js";
|
|
@@ -16,6 +17,35 @@ import { Context } from "@databricks/sdk-experimental";
|
|
|
16
17
|
//#region src/connectors/sql-warehouse/client.ts
|
|
17
18
|
const logger = createLogger("connectors:sql-warehouse");
|
|
18
19
|
/**
|
|
20
|
+
* Real column names from a statement's result manifest (positional
|
|
21
|
+
* `column_N` fallback for any blank ones). Databricks encodes ARROW_STREAM
|
|
22
|
+
* schemas positionally (`col_0`, …); these names let consumers relabel an
|
|
23
|
+
* Arrow result to match the JSON path. Returns `undefined` when the manifest
|
|
24
|
+
* carries no columns.
|
|
25
|
+
*/
|
|
26
|
+
function arrowColumnNames(response) {
|
|
27
|
+
const cols = response.manifest?.schema?.columns;
|
|
28
|
+
if (!cols || cols.length === 0) return void 0;
|
|
29
|
+
return cols.map((c, i) => c.name && c.name.length > 0 ? c.name : `column_${i}`);
|
|
30
|
+
}
|
|
31
|
+
/**
|
|
32
|
+
* Maximum size for inline Arrow IPC attachments (25 MiB decoded — the
|
|
33
|
+
* Databricks Statement Execution API hard cap on INLINE responses).
|
|
34
|
+
*
|
|
35
|
+
* The analytics route streams the decoded attachment straight back on the
|
|
36
|
+
* query response body; this cap bounds the single in-memory copy. Larger
|
|
37
|
+
* results fall through to `disposition: "EXTERNAL_LINKS"`, which streams
|
|
38
|
+
* chunk-by-chunk and is not bound by this limit.
|
|
39
|
+
*/
|
|
40
|
+
const MAX_INLINE_ATTACHMENT_BYTES = 25 * 1024 * 1024;
|
|
41
|
+
/**
|
|
42
|
+
* Safety cap on how many additional EXTERNAL_LINKS chunks
|
|
43
|
+
* {@link SQLWarehouseConnector._resolveAllExternalLinks} will follow when the
|
|
44
|
+
* manifest omits `total_chunk_count`. High enough to cover any real result;
|
|
45
|
+
* only bounds a misbehaving warehouse with a cyclic `next_chunk_index`.
|
|
46
|
+
*/
|
|
47
|
+
const MAX_EXTERNAL_CHUNK_FOLLOWS = 1e4;
|
|
48
|
+
/**
|
|
19
49
|
* Default ceiling for how long {@link SQLWarehouseConnector.ensureWarehouseRunning}
|
|
20
50
|
* will wait for a warehouse to reach the RUNNING state before giving up.
|
|
21
51
|
*
|
|
@@ -66,7 +96,6 @@ var SQLWarehouseConnector = class {
|
|
|
66
96
|
get arrowProcessor() {
|
|
67
97
|
if (!this._arrowProcessor) this._arrowProcessor = new ArrowStreamProcessor({
|
|
68
98
|
timeout: this.config.timeout || executeStatementDefaults.timeout,
|
|
69
|
-
maxConcurrentDownloads: ArrowStreamProcessor.DEFAULT_MAX_CONCURRENT_DOWNLOADS,
|
|
70
99
|
retries: ArrowStreamProcessor.DEFAULT_RETRIES
|
|
71
100
|
});
|
|
72
101
|
return this._arrowProcessor;
|
|
@@ -135,9 +164,9 @@ var SQLWarehouseConnector = class {
|
|
|
135
164
|
result = await this._pollForStatementResult(workspaceClient, statementId, this.config.timeout, signal);
|
|
136
165
|
break;
|
|
137
166
|
case "SUCCEEDED":
|
|
138
|
-
result = this._transformDataArray(response);
|
|
167
|
+
result = await this._transformDataArray(response, workspaceClient, signal);
|
|
139
168
|
break;
|
|
140
|
-
case "FAILED": throw ExecutionError.statementFailed(status.error?.message);
|
|
169
|
+
case "FAILED": throw ExecutionError.statementFailed(status.error?.message, status.error?.error_code);
|
|
141
170
|
case "CANCELED": throw ExecutionError.canceled();
|
|
142
171
|
case "CLOSED": throw ExecutionError.resultsClosed();
|
|
143
172
|
default: throw ExecutionError.unknownState(String(status?.state ?? "unknown"));
|
|
@@ -163,9 +192,7 @@ var SQLWarehouseConnector = class {
|
|
|
163
192
|
});
|
|
164
193
|
logger.error("Statement execution failed: %s", error instanceof Error ? error.message : String(error));
|
|
165
194
|
}
|
|
166
|
-
|
|
167
|
-
if (error instanceof AppKitError) throw error;
|
|
168
|
-
throw ExecutionError.statementFailed(error instanceof Error ? error.message : String(error));
|
|
195
|
+
this._rethrowStatementError(error);
|
|
169
196
|
} finally {
|
|
170
197
|
if (abortHandler && signal) signal.removeEventListener("abort", abortHandler);
|
|
171
198
|
const duration = Date.now() - startTime;
|
|
@@ -462,8 +489,8 @@ var SQLWarehouseConnector = class {
|
|
|
462
489
|
"poll.duration_ms": elapsedTime
|
|
463
490
|
});
|
|
464
491
|
span.setStatus({ code: SpanStatusCode.OK });
|
|
465
|
-
return this._transformDataArray(response);
|
|
466
|
-
case "FAILED": throw ExecutionError.statementFailed(status.error?.message);
|
|
492
|
+
return this._transformDataArray(response, workspaceClient, signal);
|
|
493
|
+
case "FAILED": throw ExecutionError.statementFailed(status.error?.message, status.error?.error_code);
|
|
467
494
|
case "CANCELED": throw ExecutionError.canceled();
|
|
468
495
|
case "CLOSED": throw ExecutionError.resultsClosed();
|
|
469
496
|
default: throw ExecutionError.unknownState(String(status?.state ?? "unknown"));
|
|
@@ -477,9 +504,7 @@ var SQLWarehouseConnector = class {
|
|
|
477
504
|
code: SpanStatusCode.ERROR,
|
|
478
505
|
message: error instanceof Error ? error.message : String(error)
|
|
479
506
|
});
|
|
480
|
-
|
|
481
|
-
if (error instanceof AppKitError) throw error;
|
|
482
|
-
throw ExecutionError.statementFailed(error instanceof Error ? error.message : String(error));
|
|
507
|
+
this._rethrowStatementError(error);
|
|
483
508
|
} finally {
|
|
484
509
|
span.end();
|
|
485
510
|
}
|
|
@@ -488,8 +513,22 @@ var SQLWarehouseConnector = class {
|
|
|
488
513
|
includePrefix: true
|
|
489
514
|
});
|
|
490
515
|
}
|
|
491
|
-
_transformDataArray(response) {
|
|
492
|
-
if (response.manifest?.format === "ARROW_STREAM")
|
|
516
|
+
async _transformDataArray(response, workspaceClient, signal) {
|
|
517
|
+
if (response.manifest?.format === "ARROW_STREAM") {
|
|
518
|
+
const result = response.result;
|
|
519
|
+
if (result?.attachment) return this._validateArrowAttachment(response, result.attachment);
|
|
520
|
+
if (result?.external_links && result.external_links.length > 0) return this.updateWithArrowStatus(response, workspaceClient, signal);
|
|
521
|
+
if ((!result?.data_array || Array.isArray(result.data_array) && result.data_array.length === 0) && response.manifest?.schema?.columns) {
|
|
522
|
+
const synthesized = buildEmptyArrowIPCBase64(response.manifest.schema.columns);
|
|
523
|
+
return {
|
|
524
|
+
...response,
|
|
525
|
+
result: {
|
|
526
|
+
...result ?? {},
|
|
527
|
+
attachment: synthesized
|
|
528
|
+
}
|
|
529
|
+
};
|
|
530
|
+
}
|
|
531
|
+
}
|
|
493
532
|
if (!response.result?.data_array || !response.manifest?.schema?.columns) return response;
|
|
494
533
|
const columns = response.manifest.schema.columns;
|
|
495
534
|
const transformedData = response.result.data_array.map((row) => {
|
|
@@ -515,59 +554,126 @@ var SQLWarehouseConnector = class {
|
|
|
515
554
|
}
|
|
516
555
|
};
|
|
517
556
|
}
|
|
518
|
-
|
|
557
|
+
/**
|
|
558
|
+
* Validate a base64 Arrow IPC attachment (INLINE ARROW_STREAM) and attach
|
|
559
|
+
* the real column names from the result manifest.
|
|
560
|
+
*
|
|
561
|
+
* Databricks warehouses encode the Arrow schema positionally (`col_0`, …);
|
|
562
|
+
* the real, aliased names live only in `manifest.schema.columns`. Rather
|
|
563
|
+
* than decode/re-encode the payload server-side, we carry the names on the
|
|
564
|
+
* result (`columnNames`) so the route can hand them to the client in a
|
|
565
|
+
* response header and the client relabels the decoded Table — the one
|
|
566
|
+
* mechanism used for both INLINE and EXTERNAL_LINKS.
|
|
567
|
+
*/
|
|
568
|
+
_validateArrowAttachment(response, attachment) {
|
|
569
|
+
const stripped = attachment.replace(/\s+/g, "");
|
|
570
|
+
const padding = stripped.endsWith("==") ? 2 : stripped.endsWith("=") ? 1 : 0;
|
|
571
|
+
const decodedSize = Math.floor(stripped.length * 3 / 4) - padding;
|
|
572
|
+
if (decodedSize > MAX_INLINE_ATTACHMENT_BYTES) throw ExecutionError.statementFailed(`Inline Arrow attachment exceeds maximum size (${decodedSize} > ${MAX_INLINE_ATTACHMENT_BYTES} bytes)`);
|
|
573
|
+
const columnNames = arrowColumnNames(response);
|
|
574
|
+
if (columnNames) return {
|
|
575
|
+
...response,
|
|
576
|
+
result: {
|
|
577
|
+
...response.result,
|
|
578
|
+
statement_id: response.statement_id,
|
|
579
|
+
columnNames
|
|
580
|
+
}
|
|
581
|
+
};
|
|
582
|
+
return response;
|
|
583
|
+
}
|
|
584
|
+
async updateWithArrowStatus(response, workspaceClient, signal) {
|
|
585
|
+
const statementId = response.statement_id;
|
|
519
586
|
return { result: {
|
|
520
|
-
statement_id:
|
|
587
|
+
statement_id: statementId,
|
|
521
588
|
status: {
|
|
522
589
|
state: response.status?.state,
|
|
523
590
|
error: response.status?.error
|
|
524
|
-
}
|
|
591
|
+
},
|
|
592
|
+
columnNames: arrowColumnNames(response),
|
|
593
|
+
external_links: await this._resolveAllExternalLinks(workspaceClient, statementId, response, signal),
|
|
594
|
+
refreshChunkLink: this._makeChunkLinkRefresher(workspaceClient, statementId)
|
|
525
595
|
} };
|
|
526
596
|
}
|
|
527
|
-
|
|
528
|
-
|
|
529
|
-
|
|
530
|
-
|
|
531
|
-
|
|
532
|
-
|
|
533
|
-
|
|
534
|
-
|
|
535
|
-
|
|
536
|
-
|
|
537
|
-
|
|
538
|
-
|
|
539
|
-
|
|
540
|
-
|
|
541
|
-
|
|
542
|
-
|
|
543
|
-
|
|
544
|
-
|
|
545
|
-
|
|
546
|
-
|
|
547
|
-
|
|
548
|
-
|
|
549
|
-
|
|
550
|
-
|
|
551
|
-
|
|
552
|
-
|
|
553
|
-
|
|
554
|
-
|
|
555
|
-
|
|
556
|
-
|
|
557
|
-
|
|
558
|
-
|
|
559
|
-
|
|
560
|
-
|
|
561
|
-
|
|
562
|
-
|
|
563
|
-
|
|
564
|
-
|
|
565
|
-
|
|
566
|
-
|
|
567
|
-
|
|
568
|
-
|
|
569
|
-
|
|
570
|
-
|
|
597
|
+
/**
|
|
598
|
+
* Resolve pre-signed links for EVERY chunk of an EXTERNAL_LINKS result.
|
|
599
|
+
*
|
|
600
|
+
* The execute/getStatement response carries only the first chunk's links
|
|
601
|
+
* (each link, except the last, exposes `next_chunk_index`); the remaining
|
|
602
|
+
* chunks are fetched with `getStatementResultChunkN`. Runs in the caller's
|
|
603
|
+
* identity context (user creds for `.obo.sql`), so there is no cross-identity
|
|
604
|
+
* fetch. Only the tiny link metadata is resolved eagerly — the bytes still
|
|
605
|
+
* stream one chunk at a time downstream. Without this a multi-chunk result
|
|
606
|
+
* would be silently truncated to its first chunk.
|
|
607
|
+
*/
|
|
608
|
+
async _resolveAllExternalLinks(workspaceClient, statementId, response, signal) {
|
|
609
|
+
const first = response.result?.external_links;
|
|
610
|
+
if (!first || first.length === 0) return first;
|
|
611
|
+
const links = [...first];
|
|
612
|
+
const maxFetches = response.manifest?.total_chunk_count ?? MAX_EXTERNAL_CHUNK_FOLLOWS;
|
|
613
|
+
let next = this._nextChunkIndex(first);
|
|
614
|
+
for (let fetches = 0; next != null && fetches < maxFetches; fetches++) {
|
|
615
|
+
if (signal?.aborted) throw ExecutionError.canceled();
|
|
616
|
+
const chunkLinks = (await workspaceClient.statementExecution.getStatementResultChunkN({
|
|
617
|
+
statement_id: statementId,
|
|
618
|
+
chunk_index: next
|
|
619
|
+
}, this._createContext(signal))).external_links ?? [];
|
|
620
|
+
if (chunkLinks.length === 0) break;
|
|
621
|
+
links.push(...chunkLinks);
|
|
622
|
+
next = this._nextChunkIndex(chunkLinks);
|
|
623
|
+
}
|
|
624
|
+
return links;
|
|
625
|
+
}
|
|
626
|
+
/** The `next_chunk_index` advertised by a chunk's links, if any. */
|
|
627
|
+
_nextChunkIndex(links) {
|
|
628
|
+
for (const link of links) if (link.next_chunk_index != null) return link.next_chunk_index;
|
|
629
|
+
}
|
|
630
|
+
/**
|
|
631
|
+
* A closure that re-mints a single chunk's pre-signed link via
|
|
632
|
+
* `getStatementResultChunkN`, bound to the caller's workspace client +
|
|
633
|
+
* statement id. Created here (in the caller's identity context) so the
|
|
634
|
+
* streamer — which runs outside that context — can refresh an expired link
|
|
635
|
+
* for `.obo.sql` statements without a cross-identity `getStatement`.
|
|
636
|
+
*/
|
|
637
|
+
_makeChunkLinkRefresher(workspaceClient, statementId) {
|
|
638
|
+
return async (chunkIndex, signal) => {
|
|
639
|
+
return (await workspaceClient.statementExecution.getStatementResultChunkN({
|
|
640
|
+
statement_id: statementId,
|
|
641
|
+
chunk_index: chunkIndex
|
|
642
|
+
}, this._createContext(signal))).external_links?.find((l) => l.chunk_index === chunkIndex);
|
|
643
|
+
};
|
|
644
|
+
}
|
|
645
|
+
/**
|
|
646
|
+
* Stream already-resolved EXTERNAL_LINKS chunks as raw Arrow IPC bytes, one
|
|
647
|
+
* chunk in memory at a time. Takes the `external_links` the caller already
|
|
648
|
+
* has from `executeStatement` (see {@link updateWithArrowStatus}), so there
|
|
649
|
+
* is no extra `getStatement` round-trip and no execution-context mismatch —
|
|
650
|
+
* the pre-signed URLs need no auth to download.
|
|
651
|
+
*/
|
|
652
|
+
streamExternalLinks(chunks, signal, refresh) {
|
|
653
|
+
return this.arrowProcessor.streamChunks(chunks, signal, refresh);
|
|
654
|
+
}
|
|
655
|
+
/**
|
|
656
|
+
* Fetch just the real column names for a completed statement from its result
|
|
657
|
+
* manifest — no result data. Backs the `/columns/:statementId` fallback the
|
|
658
|
+
* client uses when a very wide schema's names exceed the response-header
|
|
659
|
+
* size limit. Stateless: re-derives from the warehouse, no server cache.
|
|
660
|
+
*/
|
|
661
|
+
async getColumnNames(workspaceClient, jobId, signal) {
|
|
662
|
+
return arrowColumnNames(await workspaceClient.statementExecution.getStatement({ statement_id: jobId }, this._createContext(signal)));
|
|
663
|
+
}
|
|
664
|
+
/**
|
|
665
|
+
* Normalize an error caught during statement execution/polling and rethrow.
|
|
666
|
+
* Abort and already-structured `AppKitError`s pass through untouched;
|
|
667
|
+
* anything else is wrapped as an `ExecutionError`, preserving the SDK's
|
|
668
|
+
* structured `ApiError.errorCode` (e.g. "INVALID_PARAMETER_VALUE",
|
|
669
|
+
* "BAD_REQUEST") so callers can branch on a stable identifier rather than
|
|
670
|
+
* substring-matching the message.
|
|
671
|
+
*/
|
|
672
|
+
_rethrowStatementError(error) {
|
|
673
|
+
if (error instanceof Error && error.name === "AbortError") throw error;
|
|
674
|
+
if (error instanceof AppKitError) throw error;
|
|
675
|
+
const sdkErrorCode = error && typeof error === "object" && "errorCode" in error ? error.errorCode : void 0;
|
|
676
|
+
throw ExecutionError.statementFailed(error instanceof Error ? error.message : String(error), typeof sdkErrorCode === "string" ? sdkErrorCode : void 0);
|
|
571
677
|
}
|
|
572
678
|
_createContext(signal) {
|
|
573
679
|
return new Context({ cancellationToken: {
|