@lix-js/sdk 0.6.0-preview.1 → 0.6.0-preview.3
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/SKILL.md +304 -320
- package/dist/engine-wasm/wasm/lix_engine.d.ts +5 -0
- package/dist/engine-wasm/wasm/lix_engine.js +9 -13
- package/dist/engine-wasm/wasm/lix_engine.wasm +0 -0
- package/dist/engine-wasm/wasm/lix_engine.wasm.d.ts +1 -0
- package/dist/generated/builtin-schemas.d.ts +87 -162
- package/dist/generated/builtin-schemas.js +139 -236
- package/dist/open-lix.d.ts +103 -14
- package/dist/open-lix.js +3 -0
- package/dist/sqlite/index.js +99 -22
- package/dist-engine-src/README.md +18 -0
- package/dist-engine-src/src/backend/kv.rs +358 -0
- package/dist-engine-src/src/backend/mod.rs +12 -0
- package/dist-engine-src/src/backend/testing.rs +658 -0
- package/dist-engine-src/src/backend/types.rs +96 -0
- package/dist-engine-src/src/binary_cas/chunking.rs +31 -0
- package/dist-engine-src/src/binary_cas/codec.rs +346 -0
- package/dist-engine-src/src/binary_cas/context.rs +139 -0
- package/dist-engine-src/src/binary_cas/kv.rs +1063 -0
- package/dist-engine-src/src/binary_cas/mod.rs +11 -0
- package/dist-engine-src/src/binary_cas/types.rs +121 -0
- package/dist-engine-src/src/catalog/context.rs +412 -0
- package/dist-engine-src/src/catalog/mod.rs +10 -0
- package/dist-engine-src/src/catalog/schema.rs +4 -0
- package/dist-engine-src/src/catalog/snapshot.rs +1114 -0
- package/dist-engine-src/src/cel/context.rs +86 -0
- package/dist-engine-src/src/cel/error.rs +19 -0
- package/dist-engine-src/src/cel/mod.rs +8 -0
- package/dist-engine-src/src/cel/provider.rs +9 -0
- package/dist-engine-src/src/cel/runtime.rs +167 -0
- package/dist-engine-src/src/cel/value.rs +50 -0
- package/dist-engine-src/src/commit_graph/context.rs +901 -0
- package/dist-engine-src/src/commit_graph/mod.rs +11 -0
- package/dist-engine-src/src/commit_graph/types.rs +109 -0
- package/dist-engine-src/src/commit_graph/walker.rs +756 -0
- package/dist-engine-src/src/commit_store/codec.rs +887 -0
- package/dist-engine-src/src/commit_store/context.rs +944 -0
- package/dist-engine-src/src/commit_store/materialization.rs +84 -0
- package/dist-engine-src/src/commit_store/mod.rs +16 -0
- package/dist-engine-src/src/commit_store/storage.rs +600 -0
- package/dist-engine-src/src/commit_store/types.rs +215 -0
- package/dist-engine-src/src/common/error.rs +313 -0
- package/dist-engine-src/src/common/fingerprint.rs +3 -0
- package/dist-engine-src/src/common/fs_path.rs +1336 -0
- package/dist-engine-src/src/common/identity.rs +145 -0
- package/dist-engine-src/src/common/json_pointer.rs +67 -0
- package/dist-engine-src/src/common/metadata.rs +40 -0
- package/dist-engine-src/src/common/mod.rs +23 -0
- package/dist-engine-src/src/common/types.rs +105 -0
- package/dist-engine-src/src/common/wire.rs +222 -0
- package/dist-engine-src/src/domain.rs +324 -0
- package/dist-engine-src/src/engine.rs +225 -0
- package/dist-engine-src/src/entity_identity.rs +405 -0
- package/dist-engine-src/src/functions/context.rs +292 -0
- package/dist-engine-src/src/functions/deterministic.rs +113 -0
- package/dist-engine-src/src/functions/mod.rs +18 -0
- package/dist-engine-src/src/functions/provider.rs +130 -0
- package/dist-engine-src/src/functions/state.rs +336 -0
- package/dist-engine-src/src/functions/types.rs +37 -0
- package/dist-engine-src/src/init.rs +558 -0
- package/dist-engine-src/src/json_store/compression.rs +77 -0
- package/dist-engine-src/src/json_store/context.rs +423 -0
- package/dist-engine-src/src/json_store/encoded.rs +15 -0
- package/dist-engine-src/src/json_store/mod.rs +12 -0
- package/dist-engine-src/src/json_store/store.rs +1109 -0
- package/dist-engine-src/src/json_store/types.rs +217 -0
- package/dist-engine-src/src/lib.rs +62 -0
- package/dist-engine-src/src/live_state/context.rs +2019 -0
- package/dist-engine-src/src/live_state/mod.rs +15 -0
- package/dist-engine-src/src/live_state/overlay.rs +75 -0
- package/dist-engine-src/src/live_state/reader.rs +23 -0
- package/dist-engine-src/src/live_state/types.rs +222 -0
- package/dist-engine-src/src/live_state/visibility.rs +223 -0
- package/dist-engine-src/src/plugin/archive.rs +438 -0
- package/dist-engine-src/src/plugin/component.rs +183 -0
- package/dist-engine-src/src/plugin/install.rs +619 -0
- package/dist-engine-src/src/plugin/manifest.rs +516 -0
- package/dist-engine-src/src/plugin/materializer.rs +477 -0
- package/dist-engine-src/src/plugin/mod.rs +33 -0
- package/dist-engine-src/src/plugin/plugin_manifest.json +118 -0
- package/dist-engine-src/src/plugin/storage.rs +74 -0
- package/dist-engine-src/src/schema/annotations/defaults.rs +275 -0
- package/dist-engine-src/src/schema/annotations/mod.rs +1 -0
- package/dist-engine-src/src/schema/builtin/lix_account.json +21 -0
- package/dist-engine-src/src/schema/builtin/lix_active_account.json +29 -0
- package/dist-engine-src/src/schema/builtin/lix_binary_blob_ref.json +29 -0
- package/dist-engine-src/src/schema/builtin/lix_change.json +63 -0
- package/dist-engine-src/src/schema/builtin/lix_change_author.json +45 -0
- package/dist-engine-src/src/schema/builtin/lix_commit.json +24 -0
- package/dist-engine-src/src/schema/builtin/lix_commit_edge.json +53 -0
- package/dist-engine-src/src/schema/builtin/lix_directory_descriptor.json +52 -0
- package/dist-engine-src/src/schema/builtin/lix_file_descriptor.json +52 -0
- package/dist-engine-src/src/schema/builtin/lix_key_value.json +40 -0
- package/dist-engine-src/src/schema/builtin/lix_label.json +29 -0
- package/dist-engine-src/src/schema/builtin/lix_label_assignment.json +74 -0
- package/dist-engine-src/src/schema/builtin/lix_registered_schema.json +25 -0
- package/dist-engine-src/src/schema/builtin/lix_version_descriptor.json +34 -0
- package/dist-engine-src/src/schema/builtin/lix_version_ref.json +48 -0
- package/dist-engine-src/src/schema/builtin/mod.rs +222 -0
- package/dist-engine-src/src/schema/compatibility.rs +787 -0
- package/dist-engine-src/src/schema/definition.json +187 -0
- package/dist-engine-src/src/schema/definition.rs +742 -0
- package/dist-engine-src/src/schema/key.rs +138 -0
- package/dist-engine-src/src/schema/mod.rs +20 -0
- package/dist-engine-src/src/schema/seed.rs +14 -0
- package/dist-engine-src/src/schema/tests.rs +780 -0
- package/dist-engine-src/src/session/context.rs +364 -0
- package/dist-engine-src/src/session/create_version.rs +88 -0
- package/dist-engine-src/src/session/execute.rs +478 -0
- package/dist-engine-src/src/session/merge/analysis.rs +102 -0
- package/dist-engine-src/src/session/merge/apply.rs +23 -0
- package/dist-engine-src/src/session/merge/conflicts.rs +63 -0
- package/dist-engine-src/src/session/merge/mod.rs +11 -0
- package/dist-engine-src/src/session/merge/stats.rs +65 -0
- package/dist-engine-src/src/session/merge/version.rs +427 -0
- package/dist-engine-src/src/session/mod.rs +27 -0
- package/dist-engine-src/src/session/optimization9_sql2_bench.rs +100 -0
- package/dist-engine-src/src/session/switch_version.rs +109 -0
- package/dist-engine-src/src/sql2/change_provider.rs +331 -0
- package/dist-engine-src/src/sql2/classify.rs +182 -0
- package/dist-engine-src/src/sql2/context.rs +311 -0
- package/dist-engine-src/src/sql2/directory_history_provider.rs +631 -0
- package/dist-engine-src/src/sql2/directory_provider.rs +2453 -0
- package/dist-engine-src/src/sql2/dml.rs +148 -0
- package/dist-engine-src/src/sql2/entity_history_provider.rs +440 -0
- package/dist-engine-src/src/sql2/entity_provider.rs +3211 -0
- package/dist-engine-src/src/sql2/error.rs +216 -0
- package/dist-engine-src/src/sql2/execute.rs +3440 -0
- package/dist-engine-src/src/sql2/file_history_provider.rs +910 -0
- package/dist-engine-src/src/sql2/file_provider.rs +3679 -0
- package/dist-engine-src/src/sql2/filesystem_planner.rs +1490 -0
- package/dist-engine-src/src/sql2/filesystem_predicates.rs +159 -0
- package/dist-engine-src/src/sql2/filesystem_visibility.rs +383 -0
- package/dist-engine-src/src/sql2/history_projection.rs +56 -0
- package/dist-engine-src/src/sql2/history_provider.rs +412 -0
- package/dist-engine-src/src/sql2/history_route.rs +657 -0
- package/dist-engine-src/src/sql2/lix_state_provider.rs +2512 -0
- package/dist-engine-src/src/sql2/mod.rs +46 -0
- package/dist-engine-src/src/sql2/predicate_typecheck.rs +246 -0
- package/dist-engine-src/src/sql2/public_bind/assignment.rs +46 -0
- package/dist-engine-src/src/sql2/public_bind/capability.rs +41 -0
- package/dist-engine-src/src/sql2/public_bind/dml.rs +166 -0
- package/dist-engine-src/src/sql2/public_bind/mod.rs +25 -0
- package/dist-engine-src/src/sql2/public_bind/table.rs +168 -0
- package/dist-engine-src/src/sql2/read_only.rs +63 -0
- package/dist-engine-src/src/sql2/record_batch.rs +17 -0
- package/dist-engine-src/src/sql2/result_metadata.rs +29 -0
- package/dist-engine-src/src/sql2/runtime.rs +60 -0
- package/dist-engine-src/src/sql2/session.rs +132 -0
- package/dist-engine-src/src/sql2/udfs/common.rs +295 -0
- package/dist-engine-src/src/sql2/udfs/lix_active_version_commit_id.rs +53 -0
- package/dist-engine-src/src/sql2/udfs/lix_empty_blob.rs +47 -0
- package/dist-engine-src/src/sql2/udfs/lix_json.rs +100 -0
- package/dist-engine-src/src/sql2/udfs/lix_json_get.rs +99 -0
- package/dist-engine-src/src/sql2/udfs/lix_json_get_text.rs +99 -0
- package/dist-engine-src/src/sql2/udfs/lix_text_decode.rs +82 -0
- package/dist-engine-src/src/sql2/udfs/lix_text_encode.rs +85 -0
- package/dist-engine-src/src/sql2/udfs/lix_timestamp.rs +76 -0
- package/dist-engine-src/src/sql2/udfs/lix_uuid_v7.rs +76 -0
- package/dist-engine-src/src/sql2/udfs/mod.rs +89 -0
- package/dist-engine-src/src/sql2/udfs/public_call.rs +211 -0
- package/dist-engine-src/src/sql2/version_provider.rs +1202 -0
- package/dist-engine-src/src/sql2/version_scope.rs +394 -0
- package/dist-engine-src/src/sql2/write_normalization.rs +345 -0
- package/dist-engine-src/src/storage/context.rs +356 -0
- package/dist-engine-src/src/storage/mod.rs +14 -0
- package/dist-engine-src/src/storage/read_scope.rs +88 -0
- package/dist-engine-src/src/storage/types.rs +501 -0
- package/dist-engine-src/src/storage_bench.rs +4863 -0
- package/dist-engine-src/src/test_support.rs +228 -0
- package/dist-engine-src/src/tracked_state/by_file_index.rs +98 -0
- package/dist-engine-src/src/tracked_state/codec.rs +2085 -0
- package/dist-engine-src/src/tracked_state/context.rs +1867 -0
- package/dist-engine-src/src/tracked_state/diff.rs +686 -0
- package/dist-engine-src/src/tracked_state/materialization.rs +403 -0
- package/dist-engine-src/src/tracked_state/materializer.rs +488 -0
- package/dist-engine-src/src/tracked_state/merge.rs +492 -0
- package/dist-engine-src/src/tracked_state/mod.rs +32 -0
- package/dist-engine-src/src/tracked_state/storage.rs +375 -0
- package/dist-engine-src/src/tracked_state/tree.rs +3187 -0
- package/dist-engine-src/src/tracked_state/types.rs +231 -0
- package/dist-engine-src/src/transaction/commit.rs +1484 -0
- package/dist-engine-src/src/transaction/context.rs +1548 -0
- package/dist-engine-src/src/transaction/live_state_overlay.rs +35 -0
- package/dist-engine-src/src/transaction/mod.rs +13 -0
- package/dist-engine-src/src/transaction/normalization.rs +890 -0
- package/dist-engine-src/src/transaction/prep.rs +37 -0
- package/dist-engine-src/src/transaction/schema_resolver.rs +149 -0
- package/dist-engine-src/src/transaction/staging.rs +1731 -0
- package/dist-engine-src/src/transaction/types.rs +460 -0
- package/dist-engine-src/src/transaction/validation.rs +5830 -0
- package/dist-engine-src/src/untracked_state/codec.rs +307 -0
- package/dist-engine-src/src/untracked_state/context.rs +98 -0
- package/dist-engine-src/src/untracked_state/materialization.rs +63 -0
- package/dist-engine-src/src/untracked_state/mod.rs +15 -0
- package/dist-engine-src/src/untracked_state/storage.rs +396 -0
- package/dist-engine-src/src/untracked_state/types.rs +146 -0
- package/dist-engine-src/src/version/context.rs +40 -0
- package/dist-engine-src/src/version/lifecycle.rs +221 -0
- package/dist-engine-src/src/version/mod.rs +13 -0
- package/dist-engine-src/src/version/refs.rs +330 -0
- package/dist-engine-src/src/version/stage_rows.rs +67 -0
- package/dist-engine-src/src/version/types.rs +21 -0
- package/dist-engine-src/src/wasm/mod.rs +60 -0
- package/package.json +68 -64
|
@@ -0,0 +1,295 @@
|
|
|
1
|
+
use std::sync::Arc;
|
|
2
|
+
|
|
3
|
+
use datafusion::arrow::array::{
|
|
4
|
+
Array, ArrayRef, BinaryArray, BooleanArray, Float32Array, Float64Array, Int16Array, Int32Array,
|
|
5
|
+
Int64Array, Int8Array, LargeBinaryArray, LargeStringArray, StringArray, UInt16Array,
|
|
6
|
+
UInt32Array, UInt64Array, UInt8Array,
|
|
7
|
+
};
|
|
8
|
+
use datafusion::common::{plan_err, DataFusionError, Result};
|
|
9
|
+
use datafusion::logical_expr::ColumnarValue;
|
|
10
|
+
use serde_json::Value as JsonValue;
|
|
11
|
+
|
|
12
|
+
pub(super) fn scalar_inputs(args: &[ColumnarValue]) -> bool {
|
|
13
|
+
args.iter()
|
|
14
|
+
.all(|value| matches!(value, ColumnarValue::Scalar(_)))
|
|
15
|
+
}
|
|
16
|
+
|
|
17
|
+
pub(super) fn json_value_to_serde(array: &dyn Array, row: usize) -> Result<Option<JsonValue>> {
|
|
18
|
+
let Some(raw) = text_like_value(array, row)? else {
|
|
19
|
+
return Ok(None);
|
|
20
|
+
};
|
|
21
|
+
serde_json::from_str::<JsonValue>(&raw)
|
|
22
|
+
.map(Some)
|
|
23
|
+
.map_err(|error| {
|
|
24
|
+
DataFusionError::Execution(format!(
|
|
25
|
+
"JSON function expected valid JSON text in its first argument, got error: {error}"
|
|
26
|
+
))
|
|
27
|
+
})
|
|
28
|
+
}
|
|
29
|
+
|
|
30
|
+
pub(super) fn text_like_value(array: &dyn Array, row: usize) -> Result<Option<String>> {
|
|
31
|
+
if let Some(array) = array.as_any().downcast_ref::<StringArray>() {
|
|
32
|
+
return Ok((!array.is_null(row)).then(|| array.value(row).to_string()));
|
|
33
|
+
}
|
|
34
|
+
if let Some(array) = array.as_any().downcast_ref::<LargeStringArray>() {
|
|
35
|
+
return Ok((!array.is_null(row)).then(|| array.value(row).to_string()));
|
|
36
|
+
}
|
|
37
|
+
if let Some(value) = numeric_value(array, row)? {
|
|
38
|
+
return Ok(Some(value));
|
|
39
|
+
}
|
|
40
|
+
if let Some(array) = array.as_any().downcast_ref::<BooleanArray>() {
|
|
41
|
+
return Ok((!array.is_null(row)).then(|| {
|
|
42
|
+
if array.value(row) {
|
|
43
|
+
"true".to_string()
|
|
44
|
+
} else {
|
|
45
|
+
"false".to_string()
|
|
46
|
+
}
|
|
47
|
+
}));
|
|
48
|
+
}
|
|
49
|
+
if let Some(array) = array.as_any().downcast_ref::<BinaryArray>() {
|
|
50
|
+
return Ok(
|
|
51
|
+
(!array.is_null(row)).then(|| String::from_utf8_lossy(array.value(row)).to_string())
|
|
52
|
+
);
|
|
53
|
+
}
|
|
54
|
+
if let Some(array) = array.as_any().downcast_ref::<LargeBinaryArray>() {
|
|
55
|
+
return Ok(
|
|
56
|
+
(!array.is_null(row)).then(|| String::from_utf8_lossy(array.value(row)).to_string())
|
|
57
|
+
);
|
|
58
|
+
}
|
|
59
|
+
Err(DataFusionError::Execution(format!(
|
|
60
|
+
"unsupported argument type for JSON/text function: {:?}",
|
|
61
|
+
array.data_type()
|
|
62
|
+
)))
|
|
63
|
+
}
|
|
64
|
+
|
|
65
|
+
pub(super) fn numeric_value(array: &dyn Array, row: usize) -> Result<Option<String>> {
|
|
66
|
+
macro_rules! numeric_array {
|
|
67
|
+
($ty:ty) => {
|
|
68
|
+
if let Some(array) = array.as_any().downcast_ref::<$ty>() {
|
|
69
|
+
return Ok((!array.is_null(row)).then(|| array.value(row).to_string()));
|
|
70
|
+
}
|
|
71
|
+
};
|
|
72
|
+
}
|
|
73
|
+
|
|
74
|
+
numeric_array!(Int8Array);
|
|
75
|
+
numeric_array!(Int16Array);
|
|
76
|
+
numeric_array!(Int32Array);
|
|
77
|
+
numeric_array!(Int64Array);
|
|
78
|
+
numeric_array!(UInt8Array);
|
|
79
|
+
numeric_array!(UInt16Array);
|
|
80
|
+
numeric_array!(UInt32Array);
|
|
81
|
+
numeric_array!(UInt64Array);
|
|
82
|
+
numeric_array!(Float32Array);
|
|
83
|
+
numeric_array!(Float64Array);
|
|
84
|
+
Ok(None)
|
|
85
|
+
}
|
|
86
|
+
|
|
87
|
+
pub(super) fn decode_utf8_value(array: &dyn Array, row: usize) -> Result<Option<String>> {
|
|
88
|
+
if let Some(array) = array.as_any().downcast_ref::<BinaryArray>() {
|
|
89
|
+
return (!array.is_null(row))
|
|
90
|
+
.then(|| String::from_utf8(array.value(row).to_vec()))
|
|
91
|
+
.transpose()
|
|
92
|
+
.map_err(|error| {
|
|
93
|
+
DataFusionError::Execution(format!(
|
|
94
|
+
"lix_text_decode() expected valid UTF8 bytes: {error}"
|
|
95
|
+
))
|
|
96
|
+
});
|
|
97
|
+
}
|
|
98
|
+
if let Some(array) = array.as_any().downcast_ref::<LargeBinaryArray>() {
|
|
99
|
+
return (!array.is_null(row))
|
|
100
|
+
.then(|| String::from_utf8(array.value(row).to_vec()))
|
|
101
|
+
.transpose()
|
|
102
|
+
.map_err(|error| {
|
|
103
|
+
DataFusionError::Execution(format!(
|
|
104
|
+
"lix_text_decode() expected valid UTF8 bytes: {error}"
|
|
105
|
+
))
|
|
106
|
+
});
|
|
107
|
+
}
|
|
108
|
+
if let Some(array) = array.as_any().downcast_ref::<StringArray>() {
|
|
109
|
+
return Ok((!array.is_null(row)).then(|| array.value(row).to_string()));
|
|
110
|
+
}
|
|
111
|
+
if let Some(array) = array.as_any().downcast_ref::<LargeStringArray>() {
|
|
112
|
+
return Ok((!array.is_null(row)).then(|| array.value(row).to_string()));
|
|
113
|
+
}
|
|
114
|
+
Err(DataFusionError::Execution(format!(
|
|
115
|
+
"lix_text_decode() expected Binary or Utf8, got {:?}",
|
|
116
|
+
array.data_type()
|
|
117
|
+
)))
|
|
118
|
+
}
|
|
119
|
+
|
|
120
|
+
pub(super) fn encode_utf8_value(array: &dyn Array, row: usize) -> Result<Option<Vec<u8>>> {
|
|
121
|
+
if let Some(array) = array.as_any().downcast_ref::<StringArray>() {
|
|
122
|
+
return Ok((!array.is_null(row)).then(|| array.value(row).as_bytes().to_vec()));
|
|
123
|
+
}
|
|
124
|
+
if let Some(array) = array.as_any().downcast_ref::<LargeStringArray>() {
|
|
125
|
+
return Ok((!array.is_null(row)).then(|| array.value(row).as_bytes().to_vec()));
|
|
126
|
+
}
|
|
127
|
+
if let Some(array) = array.as_any().downcast_ref::<BinaryArray>() {
|
|
128
|
+
return Ok((!array.is_null(row)).then(|| array.value(row).to_vec()));
|
|
129
|
+
}
|
|
130
|
+
if let Some(array) = array.as_any().downcast_ref::<LargeBinaryArray>() {
|
|
131
|
+
return Ok((!array.is_null(row)).then(|| array.value(row).to_vec()));
|
|
132
|
+
}
|
|
133
|
+
Err(DataFusionError::Execution(format!(
|
|
134
|
+
"lix_text_encode() expected Utf8 or Binary, got {:?}",
|
|
135
|
+
array.data_type()
|
|
136
|
+
)))
|
|
137
|
+
}
|
|
138
|
+
|
|
139
|
+
pub(super) fn validate_utf8_encoding_arg(
|
|
140
|
+
fn_name: &str,
|
|
141
|
+
encoding: Option<&ColumnarValue>,
|
|
142
|
+
) -> Result<()> {
|
|
143
|
+
let Some(encoding) = encoding else {
|
|
144
|
+
return Ok(());
|
|
145
|
+
};
|
|
146
|
+
let arrays = ColumnarValue::values_to_arrays(std::slice::from_ref(encoding))?;
|
|
147
|
+
let array = &arrays[0];
|
|
148
|
+
if array.len() == 0 {
|
|
149
|
+
return Ok(());
|
|
150
|
+
}
|
|
151
|
+
let Some(value) = text_like_value(array.as_ref(), 0)? else {
|
|
152
|
+
return Ok(());
|
|
153
|
+
};
|
|
154
|
+
let normalized = value.trim().to_ascii_uppercase().replace('-', "");
|
|
155
|
+
if normalized == "UTF8" {
|
|
156
|
+
Ok(())
|
|
157
|
+
} else {
|
|
158
|
+
plan_err!("{fn_name}() only supports UTF8 encoding, got '{value}'")
|
|
159
|
+
}
|
|
160
|
+
}
|
|
161
|
+
|
|
162
|
+
pub(super) fn extract_json_path(
|
|
163
|
+
fn_name: &str,
|
|
164
|
+
arrays: &[ArrayRef],
|
|
165
|
+
row: usize,
|
|
166
|
+
) -> Result<Option<JsonValue>> {
|
|
167
|
+
let Some(mut current) = json_value_to_serde(arrays[0].as_ref(), row)? else {
|
|
168
|
+
return Ok(None);
|
|
169
|
+
};
|
|
170
|
+
|
|
171
|
+
for path in &arrays[1..] {
|
|
172
|
+
let Some(segment) = json_path_segment(fn_name, path.as_ref(), row)? else {
|
|
173
|
+
return Ok(None);
|
|
174
|
+
};
|
|
175
|
+
let next = match segment {
|
|
176
|
+
JsonPathSegment::Key(key) => current.get(&key).cloned(),
|
|
177
|
+
JsonPathSegment::Index(index) => current
|
|
178
|
+
.as_array()
|
|
179
|
+
.and_then(|values| values.get(index))
|
|
180
|
+
.cloned(),
|
|
181
|
+
};
|
|
182
|
+
let Some(value) = next else {
|
|
183
|
+
return Ok(None);
|
|
184
|
+
};
|
|
185
|
+
current = value;
|
|
186
|
+
}
|
|
187
|
+
|
|
188
|
+
Ok(Some(current))
|
|
189
|
+
}
|
|
190
|
+
|
|
191
|
+
pub(super) fn json_text_value(value: &JsonValue) -> Result<String> {
|
|
192
|
+
match value {
|
|
193
|
+
JsonValue::String(text) => Ok(text.clone()),
|
|
194
|
+
JsonValue::Number(number) => Ok(number.to_string()),
|
|
195
|
+
JsonValue::Bool(boolean) => Ok(if *boolean {
|
|
196
|
+
"true".to_string()
|
|
197
|
+
} else {
|
|
198
|
+
"false".to_string()
|
|
199
|
+
}),
|
|
200
|
+
JsonValue::Array(_) | JsonValue::Object(_) => {
|
|
201
|
+
serde_json::to_string(value).map_err(|error| {
|
|
202
|
+
DataFusionError::Execution(format!(
|
|
203
|
+
"lix_json_get_text() could not render JSON value: {error}"
|
|
204
|
+
))
|
|
205
|
+
})
|
|
206
|
+
}
|
|
207
|
+
JsonValue::Null => Ok("null".to_string()),
|
|
208
|
+
}
|
|
209
|
+
}
|
|
210
|
+
|
|
211
|
+
pub(super) fn json_json_value(value: &JsonValue) -> Result<String> {
|
|
212
|
+
serde_json::to_string(value).map_err(|error| {
|
|
213
|
+
DataFusionError::Execution(format!(
|
|
214
|
+
"lix_json_get() could not render JSON value: {error}"
|
|
215
|
+
))
|
|
216
|
+
})
|
|
217
|
+
}
|
|
218
|
+
|
|
219
|
+
enum JsonPathSegment {
|
|
220
|
+
Key(String),
|
|
221
|
+
Index(usize),
|
|
222
|
+
}
|
|
223
|
+
|
|
224
|
+
fn json_path_segment(
|
|
225
|
+
fn_name: &str,
|
|
226
|
+
array: &dyn Array,
|
|
227
|
+
row: usize,
|
|
228
|
+
) -> Result<Option<JsonPathSegment>> {
|
|
229
|
+
if let Some(array) = array.as_any().downcast_ref::<StringArray>() {
|
|
230
|
+
if array.is_null(row) {
|
|
231
|
+
return Ok(None);
|
|
232
|
+
}
|
|
233
|
+
let value = array.value(row).to_string();
|
|
234
|
+
validate_json_path_key_segment(fn_name, &value)?;
|
|
235
|
+
return Ok(Some(JsonPathSegment::Key(value)));
|
|
236
|
+
}
|
|
237
|
+
if let Some(array) = array.as_any().downcast_ref::<LargeStringArray>() {
|
|
238
|
+
if array.is_null(row) {
|
|
239
|
+
return Ok(None);
|
|
240
|
+
}
|
|
241
|
+
let value = array.value(row).to_string();
|
|
242
|
+
validate_json_path_key_segment(fn_name, &value)?;
|
|
243
|
+
return Ok(Some(JsonPathSegment::Key(value)));
|
|
244
|
+
}
|
|
245
|
+
macro_rules! index_array {
|
|
246
|
+
($ty:ty) => {
|
|
247
|
+
if let Some(array) = array.as_any().downcast_ref::<$ty>() {
|
|
248
|
+
if array.is_null(row) {
|
|
249
|
+
return Ok(None);
|
|
250
|
+
}
|
|
251
|
+
let value = array.value(row);
|
|
252
|
+
let index = usize::try_from(value).map_err(|_| {
|
|
253
|
+
DataFusionError::Execution(format!(
|
|
254
|
+
"{fn_name}() path indexes must be non-negative integers"
|
|
255
|
+
))
|
|
256
|
+
})?;
|
|
257
|
+
return Ok(Some(JsonPathSegment::Index(index)));
|
|
258
|
+
}
|
|
259
|
+
};
|
|
260
|
+
}
|
|
261
|
+
index_array!(UInt8Array);
|
|
262
|
+
index_array!(UInt16Array);
|
|
263
|
+
index_array!(UInt32Array);
|
|
264
|
+
index_array!(UInt64Array);
|
|
265
|
+
index_array!(Int8Array);
|
|
266
|
+
index_array!(Int16Array);
|
|
267
|
+
index_array!(Int32Array);
|
|
268
|
+
index_array!(Int64Array);
|
|
269
|
+
Err(DataFusionError::Execution(format!(
|
|
270
|
+
"{fn_name}() path arguments must be strings or non-negative integers, got {:?}",
|
|
271
|
+
array.data_type()
|
|
272
|
+
)))
|
|
273
|
+
}
|
|
274
|
+
|
|
275
|
+
fn validate_json_path_key_segment(fn_name: &str, value: &str) -> Result<()> {
|
|
276
|
+
if value == "$" || value.starts_with("$.") || value.starts_with("$[") || value.starts_with('/')
|
|
277
|
+
{
|
|
278
|
+
return Err(DataFusionError::Execution(format!(
|
|
279
|
+
"{fn_name}() uses variadic path segments, not JSONPath or JSON Pointer; got '{value}'"
|
|
280
|
+
)));
|
|
281
|
+
}
|
|
282
|
+
Ok(())
|
|
283
|
+
}
|
|
284
|
+
|
|
285
|
+
pub(super) fn binary_array_from_owned(values: &[Option<Vec<u8>>]) -> BinaryArray {
|
|
286
|
+
let refs = values
|
|
287
|
+
.iter()
|
|
288
|
+
.map(|value| value.as_deref())
|
|
289
|
+
.collect::<Vec<_>>();
|
|
290
|
+
BinaryArray::from(refs)
|
|
291
|
+
}
|
|
292
|
+
|
|
293
|
+
pub(super) fn array_ref<T: Array + 'static>(array: T) -> ArrayRef {
|
|
294
|
+
Arc::new(array)
|
|
295
|
+
}
|
|
@@ -0,0 +1,53 @@
|
|
|
1
|
+
use std::any::Any;
|
|
2
|
+
|
|
3
|
+
use datafusion::arrow::datatypes::DataType;
|
|
4
|
+
use datafusion::common::{plan_err, Result, ScalarValue};
|
|
5
|
+
use datafusion::logical_expr::{
|
|
6
|
+
ColumnarValue, ScalarFunctionArgs, ScalarUDFImpl, Signature, Volatility,
|
|
7
|
+
};
|
|
8
|
+
|
|
9
|
+
#[derive(Clone, PartialEq, Eq, Hash)]
|
|
10
|
+
pub(super) struct LixActiveVersionCommitId {
|
|
11
|
+
commit_id: Option<String>,
|
|
12
|
+
}
|
|
13
|
+
|
|
14
|
+
impl LixActiveVersionCommitId {
|
|
15
|
+
pub(super) fn new(commit_id: Option<String>) -> Self {
|
|
16
|
+
Self { commit_id }
|
|
17
|
+
}
|
|
18
|
+
}
|
|
19
|
+
|
|
20
|
+
impl std::fmt::Debug for LixActiveVersionCommitId {
|
|
21
|
+
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
|
|
22
|
+
f.debug_struct("LixActiveVersionCommitId").finish()
|
|
23
|
+
}
|
|
24
|
+
}
|
|
25
|
+
|
|
26
|
+
impl ScalarUDFImpl for LixActiveVersionCommitId {
|
|
27
|
+
fn as_any(&self) -> &dyn Any {
|
|
28
|
+
self
|
|
29
|
+
}
|
|
30
|
+
|
|
31
|
+
fn name(&self) -> &str {
|
|
32
|
+
"lix_active_version_commit_id"
|
|
33
|
+
}
|
|
34
|
+
|
|
35
|
+
fn signature(&self) -> &Signature {
|
|
36
|
+
static SIGNATURE: std::sync::LazyLock<Signature> =
|
|
37
|
+
std::sync::LazyLock::new(|| Signature::nullary(Volatility::Stable));
|
|
38
|
+
&SIGNATURE
|
|
39
|
+
}
|
|
40
|
+
|
|
41
|
+
fn return_type(&self, _arg_types: &[DataType]) -> Result<DataType> {
|
|
42
|
+
Ok(DataType::Utf8)
|
|
43
|
+
}
|
|
44
|
+
|
|
45
|
+
fn invoke_with_args(&self, args: ScalarFunctionArgs) -> Result<ColumnarValue> {
|
|
46
|
+
if !args.args.is_empty() {
|
|
47
|
+
return plan_err!("lix_active_version_commit_id requires no arguments");
|
|
48
|
+
}
|
|
49
|
+
Ok(ColumnarValue::Scalar(ScalarValue::Utf8(
|
|
50
|
+
self.commit_id.clone(),
|
|
51
|
+
)))
|
|
52
|
+
}
|
|
53
|
+
}
|
|
@@ -0,0 +1,47 @@
|
|
|
1
|
+
use std::any::Any;
|
|
2
|
+
|
|
3
|
+
use datafusion::arrow::datatypes::DataType;
|
|
4
|
+
use datafusion::common::{Result, ScalarValue};
|
|
5
|
+
use datafusion::logical_expr::{
|
|
6
|
+
ColumnarValue, ScalarFunctionArgs, ScalarUDFImpl, Signature, Volatility,
|
|
7
|
+
};
|
|
8
|
+
|
|
9
|
+
#[derive(Debug, Clone, PartialEq, Eq, Hash)]
|
|
10
|
+
pub(super) struct LixEmptyBlob;
|
|
11
|
+
|
|
12
|
+
impl ScalarUDFImpl for LixEmptyBlob {
|
|
13
|
+
fn as_any(&self) -> &dyn Any {
|
|
14
|
+
self
|
|
15
|
+
}
|
|
16
|
+
|
|
17
|
+
fn name(&self) -> &str {
|
|
18
|
+
"lix_empty_blob"
|
|
19
|
+
}
|
|
20
|
+
|
|
21
|
+
fn signature(&self) -> &Signature {
|
|
22
|
+
static SIGNATURE: std::sync::LazyLock<Signature> =
|
|
23
|
+
std::sync::LazyLock::new(|| Signature::nullary(Volatility::Immutable));
|
|
24
|
+
&SIGNATURE
|
|
25
|
+
}
|
|
26
|
+
|
|
27
|
+
fn return_type(&self, _arg_types: &[DataType]) -> Result<DataType> {
|
|
28
|
+
Ok(DataType::Binary)
|
|
29
|
+
}
|
|
30
|
+
|
|
31
|
+
fn invoke_with_args(&self, _args: ScalarFunctionArgs) -> Result<ColumnarValue> {
|
|
32
|
+
Ok(ColumnarValue::Scalar(ScalarValue::Binary(Some(Vec::new()))))
|
|
33
|
+
}
|
|
34
|
+
}
|
|
35
|
+
|
|
36
|
+
#[cfg(test)]
|
|
37
|
+
mod tests {
|
|
38
|
+
use super::super::test_support::single_binary;
|
|
39
|
+
|
|
40
|
+
#[tokio::test]
|
|
41
|
+
async fn returns_empty_binary_value() {
|
|
42
|
+
assert_eq!(
|
|
43
|
+
single_binary("SELECT lix_empty_blob()").await,
|
|
44
|
+
Some(Vec::new())
|
|
45
|
+
);
|
|
46
|
+
}
|
|
47
|
+
}
|
|
@@ -0,0 +1,100 @@
|
|
|
1
|
+
use std::any::Any;
|
|
2
|
+
use std::sync::Arc;
|
|
3
|
+
|
|
4
|
+
use datafusion::arrow::array::{Array, StringArray};
|
|
5
|
+
use datafusion::arrow::datatypes::{DataType, FieldRef};
|
|
6
|
+
use datafusion::common::{plan_err, DataFusionError, Result, ScalarValue};
|
|
7
|
+
use datafusion::logical_expr::{
|
|
8
|
+
ColumnarValue, ReturnFieldArgs, ScalarFunctionArgs, ScalarUDFImpl, Signature, Volatility,
|
|
9
|
+
};
|
|
10
|
+
use serde_json::Value as JsonValue;
|
|
11
|
+
|
|
12
|
+
use crate::sql2::result_metadata::json_field;
|
|
13
|
+
|
|
14
|
+
use super::common::{scalar_inputs, text_like_value};
|
|
15
|
+
|
|
16
|
+
#[derive(Debug, Clone, PartialEq, Eq, Hash)]
|
|
17
|
+
pub(super) struct LixJson;
|
|
18
|
+
|
|
19
|
+
impl ScalarUDFImpl for LixJson {
|
|
20
|
+
fn as_any(&self) -> &dyn Any {
|
|
21
|
+
self
|
|
22
|
+
}
|
|
23
|
+
|
|
24
|
+
fn name(&self) -> &str {
|
|
25
|
+
"lix_json"
|
|
26
|
+
}
|
|
27
|
+
|
|
28
|
+
fn signature(&self) -> &Signature {
|
|
29
|
+
static SIGNATURE: std::sync::LazyLock<Signature> =
|
|
30
|
+
std::sync::LazyLock::new(|| Signature::any(1, Volatility::Immutable));
|
|
31
|
+
&SIGNATURE
|
|
32
|
+
}
|
|
33
|
+
|
|
34
|
+
fn return_type(&self, _arg_types: &[DataType]) -> Result<DataType> {
|
|
35
|
+
Ok(DataType::Utf8)
|
|
36
|
+
}
|
|
37
|
+
|
|
38
|
+
fn return_field_from_args(&self, _args: ReturnFieldArgs) -> Result<FieldRef> {
|
|
39
|
+
Ok(Arc::new(json_field(self.name(), true)))
|
|
40
|
+
}
|
|
41
|
+
|
|
42
|
+
fn invoke_with_args(&self, args: ScalarFunctionArgs) -> Result<ColumnarValue> {
|
|
43
|
+
if args.args.len() != 1 {
|
|
44
|
+
return plan_err!("lix_json requires exactly 1 argument");
|
|
45
|
+
}
|
|
46
|
+
let scalar_inputs = scalar_inputs(&args.args);
|
|
47
|
+
let arrays = ColumnarValue::values_to_arrays(&args.args)?;
|
|
48
|
+
let input = &arrays[0];
|
|
49
|
+
let len = input.len();
|
|
50
|
+
let mut values = Vec::with_capacity(len);
|
|
51
|
+
for row in 0..len {
|
|
52
|
+
values.push(json_value(input.as_ref(), row)?);
|
|
53
|
+
}
|
|
54
|
+
if scalar_inputs {
|
|
55
|
+
Ok(ColumnarValue::Scalar(ScalarValue::Utf8(
|
|
56
|
+
values.into_iter().next().flatten(),
|
|
57
|
+
)))
|
|
58
|
+
} else {
|
|
59
|
+
Ok(ColumnarValue::Array(Arc::new(StringArray::from(values))))
|
|
60
|
+
}
|
|
61
|
+
}
|
|
62
|
+
}
|
|
63
|
+
|
|
64
|
+
fn json_value(array: &dyn Array, row: usize) -> Result<Option<String>> {
|
|
65
|
+
if matches!(array.data_type(), DataType::Null) {
|
|
66
|
+
return Ok(Some("null".to_string()));
|
|
67
|
+
}
|
|
68
|
+
let Some(raw) = text_like_value(array, row)? else {
|
|
69
|
+
return Ok(Some("null".to_string()));
|
|
70
|
+
};
|
|
71
|
+
let parsed = serde_json::from_str::<JsonValue>(&raw).map_err(|error| {
|
|
72
|
+
DataFusionError::Execution(format!(
|
|
73
|
+
"lix_json() expected valid JSON text, got error: {error}"
|
|
74
|
+
))
|
|
75
|
+
})?;
|
|
76
|
+
Ok(Some(serde_json::to_string(&parsed).map_err(|error| {
|
|
77
|
+
DataFusionError::Execution(format!("lix_json() could not render JSON: {error}"))
|
|
78
|
+
})?))
|
|
79
|
+
}
|
|
80
|
+
|
|
81
|
+
#[cfg(test)]
|
|
82
|
+
mod tests {
|
|
83
|
+
use super::super::test_support::single_text;
|
|
84
|
+
|
|
85
|
+
#[tokio::test]
|
|
86
|
+
async fn canonicalizes_json_text() {
|
|
87
|
+
assert_eq!(
|
|
88
|
+
single_text("SELECT lix_json('{ \"name\" : \"Ada\" }')").await,
|
|
89
|
+
Some("{\"name\":\"Ada\"}".to_string())
|
|
90
|
+
);
|
|
91
|
+
}
|
|
92
|
+
|
|
93
|
+
#[tokio::test]
|
|
94
|
+
async fn null_input_returns_json_null() {
|
|
95
|
+
assert_eq!(
|
|
96
|
+
single_text("SELECT lix_json(NULL)").await,
|
|
97
|
+
Some("null".to_string())
|
|
98
|
+
);
|
|
99
|
+
}
|
|
100
|
+
}
|
|
@@ -0,0 +1,99 @@
|
|
|
1
|
+
use std::any::Any;
|
|
2
|
+
use std::sync::Arc;
|
|
3
|
+
|
|
4
|
+
use datafusion::arrow::array::StringArray;
|
|
5
|
+
use datafusion::arrow::datatypes::{DataType, FieldRef};
|
|
6
|
+
use datafusion::common::{plan_err, Result, ScalarValue};
|
|
7
|
+
use datafusion::logical_expr::{
|
|
8
|
+
ColumnarValue, ReturnFieldArgs, ScalarFunctionArgs, ScalarUDFImpl, Signature, Volatility,
|
|
9
|
+
};
|
|
10
|
+
use serde_json::Value as JsonValue;
|
|
11
|
+
|
|
12
|
+
use crate::sql2::result_metadata::json_field;
|
|
13
|
+
|
|
14
|
+
use super::common::{extract_json_path, json_json_value, scalar_inputs};
|
|
15
|
+
|
|
16
|
+
#[derive(Debug, Clone, PartialEq, Eq, Hash)]
|
|
17
|
+
pub(super) struct LixJsonGet {
|
|
18
|
+
signature: Signature,
|
|
19
|
+
}
|
|
20
|
+
|
|
21
|
+
impl LixJsonGet {
|
|
22
|
+
pub(super) fn new() -> Self {
|
|
23
|
+
Self {
|
|
24
|
+
signature: Signature::variadic_any(Volatility::Immutable),
|
|
25
|
+
}
|
|
26
|
+
}
|
|
27
|
+
}
|
|
28
|
+
|
|
29
|
+
impl ScalarUDFImpl for LixJsonGet {
|
|
30
|
+
fn as_any(&self) -> &dyn Any {
|
|
31
|
+
self
|
|
32
|
+
}
|
|
33
|
+
|
|
34
|
+
fn name(&self) -> &str {
|
|
35
|
+
"lix_json_get"
|
|
36
|
+
}
|
|
37
|
+
|
|
38
|
+
fn signature(&self) -> &Signature {
|
|
39
|
+
&self.signature
|
|
40
|
+
}
|
|
41
|
+
|
|
42
|
+
fn return_type(&self, _arg_types: &[DataType]) -> Result<DataType> {
|
|
43
|
+
Ok(DataType::Utf8)
|
|
44
|
+
}
|
|
45
|
+
|
|
46
|
+
fn return_field_from_args(&self, _args: ReturnFieldArgs) -> Result<FieldRef> {
|
|
47
|
+
Ok(Arc::new(json_field(self.name(), true)))
|
|
48
|
+
}
|
|
49
|
+
|
|
50
|
+
fn invoke_with_args(&self, args: ScalarFunctionArgs) -> Result<ColumnarValue> {
|
|
51
|
+
if args.args.len() < 2 {
|
|
52
|
+
return plan_err!("lix_json_get requires at least 2 arguments");
|
|
53
|
+
}
|
|
54
|
+
|
|
55
|
+
let scalar_inputs = scalar_inputs(&args.args);
|
|
56
|
+
let arrays = ColumnarValue::values_to_arrays(&args.args)?;
|
|
57
|
+
let len = arrays.first().map(|array| array.len()).unwrap_or(1);
|
|
58
|
+
|
|
59
|
+
let mut values = Vec::with_capacity(len);
|
|
60
|
+
for row in 0..len {
|
|
61
|
+
values.push(match extract_json_path(self.name(), &arrays, row)? {
|
|
62
|
+
None | Some(JsonValue::Null) => None,
|
|
63
|
+
Some(other) => Some(json_json_value(&other)?),
|
|
64
|
+
});
|
|
65
|
+
}
|
|
66
|
+
if scalar_inputs {
|
|
67
|
+
Ok(ColumnarValue::Scalar(ScalarValue::Utf8(
|
|
68
|
+
values.into_iter().next().flatten(),
|
|
69
|
+
)))
|
|
70
|
+
} else {
|
|
71
|
+
Ok(ColumnarValue::Array(Arc::new(StringArray::from(values))))
|
|
72
|
+
}
|
|
73
|
+
}
|
|
74
|
+
}
|
|
75
|
+
|
|
76
|
+
#[cfg(test)]
|
|
77
|
+
mod tests {
|
|
78
|
+
use super::super::test_support::single_text;
|
|
79
|
+
|
|
80
|
+
#[tokio::test]
|
|
81
|
+
async fn returns_json_representation() {
|
|
82
|
+
assert_eq!(
|
|
83
|
+
single_text("SELECT lix_json_get('{\"name\":\"Ada\"}', 'name')").await,
|
|
84
|
+
Some("\"Ada\"".to_string())
|
|
85
|
+
);
|
|
86
|
+
assert_eq!(
|
|
87
|
+
single_text("SELECT lix_json_get('{\"tags\":[\"db\"]}', 'tags')").await,
|
|
88
|
+
Some("[\"db\"]".to_string())
|
|
89
|
+
);
|
|
90
|
+
}
|
|
91
|
+
|
|
92
|
+
#[tokio::test]
|
|
93
|
+
async fn missing_path_returns_null() {
|
|
94
|
+
assert_eq!(
|
|
95
|
+
single_text("SELECT lix_json_get('{\"name\":\"Ada\"}', 'missing')").await,
|
|
96
|
+
None
|
|
97
|
+
);
|
|
98
|
+
}
|
|
99
|
+
}
|
|
@@ -0,0 +1,99 @@
|
|
|
1
|
+
use std::any::Any;
|
|
2
|
+
use std::sync::Arc;
|
|
3
|
+
|
|
4
|
+
use datafusion::arrow::array::StringArray;
|
|
5
|
+
use datafusion::arrow::datatypes::DataType;
|
|
6
|
+
use datafusion::common::{plan_err, Result, ScalarValue};
|
|
7
|
+
use datafusion::logical_expr::{
|
|
8
|
+
ColumnarValue, ScalarFunctionArgs, ScalarUDFImpl, Signature, Volatility,
|
|
9
|
+
};
|
|
10
|
+
use serde_json::Value as JsonValue;
|
|
11
|
+
|
|
12
|
+
use super::common::{extract_json_path, json_text_value, scalar_inputs};
|
|
13
|
+
|
|
14
|
+
#[derive(Debug, Clone, PartialEq, Eq, Hash)]
|
|
15
|
+
pub(super) struct LixJsonGetText {
|
|
16
|
+
signature: Signature,
|
|
17
|
+
}
|
|
18
|
+
|
|
19
|
+
impl LixJsonGetText {
|
|
20
|
+
pub(super) fn new() -> Self {
|
|
21
|
+
Self {
|
|
22
|
+
signature: Signature::variadic_any(Volatility::Immutable),
|
|
23
|
+
}
|
|
24
|
+
}
|
|
25
|
+
}
|
|
26
|
+
|
|
27
|
+
impl ScalarUDFImpl for LixJsonGetText {
|
|
28
|
+
fn as_any(&self) -> &dyn Any {
|
|
29
|
+
self
|
|
30
|
+
}
|
|
31
|
+
|
|
32
|
+
fn name(&self) -> &str {
|
|
33
|
+
"lix_json_get_text"
|
|
34
|
+
}
|
|
35
|
+
|
|
36
|
+
fn signature(&self) -> &Signature {
|
|
37
|
+
&self.signature
|
|
38
|
+
}
|
|
39
|
+
|
|
40
|
+
fn return_type(&self, _arg_types: &[DataType]) -> Result<DataType> {
|
|
41
|
+
Ok(DataType::Utf8)
|
|
42
|
+
}
|
|
43
|
+
|
|
44
|
+
fn invoke_with_args(&self, args: ScalarFunctionArgs) -> Result<ColumnarValue> {
|
|
45
|
+
if args.args.len() < 2 {
|
|
46
|
+
return plan_err!("lix_json_get_text requires at least 2 arguments");
|
|
47
|
+
}
|
|
48
|
+
|
|
49
|
+
let scalar_inputs = scalar_inputs(&args.args);
|
|
50
|
+
let arrays = ColumnarValue::values_to_arrays(&args.args)?;
|
|
51
|
+
let len = arrays.first().map(|array| array.len()).unwrap_or(1);
|
|
52
|
+
|
|
53
|
+
let mut values = Vec::with_capacity(len);
|
|
54
|
+
for row in 0..len {
|
|
55
|
+
values.push(match extract_json_path(self.name(), &arrays, row)? {
|
|
56
|
+
None | Some(JsonValue::Null) => None,
|
|
57
|
+
Some(JsonValue::Bool(value)) => Some(if value {
|
|
58
|
+
"true".to_string()
|
|
59
|
+
} else {
|
|
60
|
+
"false".to_string()
|
|
61
|
+
}),
|
|
62
|
+
Some(JsonValue::String(value)) => Some(value),
|
|
63
|
+
Some(other) => Some(json_text_value(&other)?),
|
|
64
|
+
});
|
|
65
|
+
}
|
|
66
|
+
if scalar_inputs {
|
|
67
|
+
Ok(ColumnarValue::Scalar(ScalarValue::Utf8(
|
|
68
|
+
values.into_iter().next().flatten(),
|
|
69
|
+
)))
|
|
70
|
+
} else {
|
|
71
|
+
Ok(ColumnarValue::Array(Arc::new(StringArray::from(values))))
|
|
72
|
+
}
|
|
73
|
+
}
|
|
74
|
+
}
|
|
75
|
+
|
|
76
|
+
#[cfg(test)]
|
|
77
|
+
mod tests {
|
|
78
|
+
use super::super::test_support::single_text;
|
|
79
|
+
|
|
80
|
+
#[tokio::test]
|
|
81
|
+
async fn returns_unwrapped_text() {
|
|
82
|
+
assert_eq!(
|
|
83
|
+
single_text("SELECT lix_json_get_text('{\"name\":\"Ada\"}', 'name')").await,
|
|
84
|
+
Some("Ada".to_string())
|
|
85
|
+
);
|
|
86
|
+
assert_eq!(
|
|
87
|
+
single_text("SELECT lix_json_get_text('{\"active\":true}', 'active')").await,
|
|
88
|
+
Some("true".to_string())
|
|
89
|
+
);
|
|
90
|
+
}
|
|
91
|
+
|
|
92
|
+
#[tokio::test]
|
|
93
|
+
async fn missing_path_returns_null() {
|
|
94
|
+
assert_eq!(
|
|
95
|
+
single_text("SELECT lix_json_get_text('{\"name\":\"Ada\"}', 'missing')").await,
|
|
96
|
+
None
|
|
97
|
+
);
|
|
98
|
+
}
|
|
99
|
+
}
|