vectorwave 0.2.5__tar.gz → 0.2.6__tar.gz
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.
- {vectorwave-0.2.5 → vectorwave-0.2.6}/PKG-INFO +1 -1
- {vectorwave-0.2.5 → vectorwave-0.2.6}/crates/Cargo.lock +13 -85
- {vectorwave-0.2.5 → vectorwave-0.2.6}/crates/Cargo.toml +2 -3
- {vectorwave-0.2.5 → vectorwave-0.2.6}/crates/src/lib.rs +59 -14
- {vectorwave-0.2.5 → vectorwave-0.2.6}/pyproject.toml +1 -1
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/monitoring/tracer.py +9 -32
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/utils/replayer.py +2 -2
- {vectorwave-0.2.5 → vectorwave-0.2.6}/Readme.md +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/__init__.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/batch/__init__.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/batch/batch.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/core/__init__.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/core/auto_injector.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/core/core.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/core/decorator.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/core/generator.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/core/llm/__init__.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/core/llm/base.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/core/llm/factory.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/core/llm/openai_client.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/database/__init__.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/database/archiver.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/database/dataset.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/database/db.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/database/db_search.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/exception/__init__.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/exception/exceptions.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/models/__init__.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/models/db_config.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/monitoring/__init__.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/monitoring/alert/__init__.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/monitoring/alert/base.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/monitoring/alert/factory.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/monitoring/alert/null_alerter.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/monitoring/alert/webhook_alerter.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/monitoring/monitoring.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/prediction/__init__.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/prediction/predictor.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/search/__init__.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/search/execution_search.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/search/extended_search.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/search/rag_search.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/utils/__init__.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/utils/context.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/utils/function_cache.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/utils/healer.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/utils/replayer_semantic.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/utils/return_caching_utils.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/utils/status.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/vectorizer/__init__.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/vectorizer/base.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/vectorizer/factory.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/vectorizer/huggingface_vectorizer.py +0 -0
- {vectorwave-0.2.5 → vectorwave-0.2.6}/src/vectorwave/vectorizer/openai_vectorizer.py +0 -0
|
@@ -8,12 +8,6 @@ version = "1.5.0"
|
|
|
8
8
|
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
9
9
|
checksum = "c08606f8c3cbf4ce6ec8e28fb0014a2c086708fe954eaa885384a6165172e7e8"
|
|
10
10
|
|
|
11
|
-
[[package]]
|
|
12
|
-
name = "bitflags"
|
|
13
|
-
version = "2.10.0"
|
|
14
|
-
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
15
|
-
checksum = "812e12b5285cc515a9c72a5c1d3b6d46a19dac5acfef5265968c166106e31dd3"
|
|
16
|
-
|
|
17
11
|
[[package]]
|
|
18
12
|
name = "cfg-if"
|
|
19
13
|
version = "1.0.4"
|
|
@@ -37,9 +31,9 @@ checksum = "d0a5c400df2834b80a4c3327b3aad3a4c4cd4de0629063962b03235697506a28"
|
|
|
37
31
|
|
|
38
32
|
[[package]]
|
|
39
33
|
name = "heck"
|
|
40
|
-
version = "0.
|
|
34
|
+
version = "0.5.0"
|
|
41
35
|
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
42
|
-
checksum = "
|
|
36
|
+
checksum = "2304e00983f87ffb38b55b444b5e3b60a884b5d30c0fca7d82fe33449bbe55ea"
|
|
43
37
|
|
|
44
38
|
[[package]]
|
|
45
39
|
name = "indoc"
|
|
@@ -62,21 +56,6 @@ version = "0.2.178"
|
|
|
62
56
|
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
63
57
|
checksum = "37c93d8daa9d8a012fd8ab92f088405fb202ea0b6ab73ee2482ae66af4f42091"
|
|
64
58
|
|
|
65
|
-
[[package]]
|
|
66
|
-
name = "lock_api"
|
|
67
|
-
version = "0.4.14"
|
|
68
|
-
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
69
|
-
checksum = "224399e74b87b5f3557511d98dff8b14089b3dadafcab6bb93eab67d3aace965"
|
|
70
|
-
dependencies = [
|
|
71
|
-
"scopeguard",
|
|
72
|
-
]
|
|
73
|
-
|
|
74
|
-
[[package]]
|
|
75
|
-
name = "log"
|
|
76
|
-
version = "0.4.29"
|
|
77
|
-
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
78
|
-
checksum = "5e5032e24019045c762d3c0f28f5b6b8bbf38563a65908389bf7978758920897"
|
|
79
|
-
|
|
80
59
|
[[package]]
|
|
81
60
|
name = "memchr"
|
|
82
61
|
version = "2.7.6"
|
|
@@ -98,29 +77,6 @@ version = "1.21.3"
|
|
|
98
77
|
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
99
78
|
checksum = "42f5e15c9953c5e4ccceeb2e7382a716482c34515315f7b03532b8b4e8393d2d"
|
|
100
79
|
|
|
101
|
-
[[package]]
|
|
102
|
-
name = "parking_lot"
|
|
103
|
-
version = "0.12.5"
|
|
104
|
-
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
105
|
-
checksum = "93857453250e3077bd71ff98b6a65ea6621a19bb0f559a85248955ac12c45a1a"
|
|
106
|
-
dependencies = [
|
|
107
|
-
"lock_api",
|
|
108
|
-
"parking_lot_core",
|
|
109
|
-
]
|
|
110
|
-
|
|
111
|
-
[[package]]
|
|
112
|
-
name = "parking_lot_core"
|
|
113
|
-
version = "0.9.12"
|
|
114
|
-
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
115
|
-
checksum = "2621685985a2ebf1c516881c026032ac7deafcda1a2c9b7850dc81e3dfcb64c1"
|
|
116
|
-
dependencies = [
|
|
117
|
-
"cfg-if",
|
|
118
|
-
"libc",
|
|
119
|
-
"redox_syscall",
|
|
120
|
-
"smallvec",
|
|
121
|
-
"windows-link",
|
|
122
|
-
]
|
|
123
|
-
|
|
124
80
|
[[package]]
|
|
125
81
|
name = "portable-atomic"
|
|
126
82
|
version = "1.12.0"
|
|
@@ -138,15 +94,15 @@ dependencies = [
|
|
|
138
94
|
|
|
139
95
|
[[package]]
|
|
140
96
|
name = "pyo3"
|
|
141
|
-
version = "0.
|
|
97
|
+
version = "0.23.5"
|
|
142
98
|
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
143
|
-
checksum = "
|
|
99
|
+
checksum = "7778bffd85cf38175ac1f545509665d0b9b92a198ca7941f131f85f7a4f9a872"
|
|
144
100
|
dependencies = [
|
|
145
101
|
"cfg-if",
|
|
146
102
|
"indoc",
|
|
147
103
|
"libc",
|
|
148
104
|
"memoffset",
|
|
149
|
-
"
|
|
105
|
+
"once_cell",
|
|
150
106
|
"portable-atomic",
|
|
151
107
|
"pyo3-build-config",
|
|
152
108
|
"pyo3-ffi",
|
|
@@ -156,9 +112,9 @@ dependencies = [
|
|
|
156
112
|
|
|
157
113
|
[[package]]
|
|
158
114
|
name = "pyo3-build-config"
|
|
159
|
-
version = "0.
|
|
115
|
+
version = "0.23.5"
|
|
160
116
|
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
161
|
-
checksum = "
|
|
117
|
+
checksum = "94f6cbe86ef3bf18998d9df6e0f3fc1050a8c5efa409bf712e661a4366e010fb"
|
|
162
118
|
dependencies = [
|
|
163
119
|
"once_cell",
|
|
164
120
|
"target-lexicon",
|
|
@@ -166,9 +122,9 @@ dependencies = [
|
|
|
166
122
|
|
|
167
123
|
[[package]]
|
|
168
124
|
name = "pyo3-ffi"
|
|
169
|
-
version = "0.
|
|
125
|
+
version = "0.23.5"
|
|
170
126
|
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
171
|
-
checksum = "
|
|
127
|
+
checksum = "e9f1b4c431c0bb1c8fb0a338709859eed0d030ff6daa34368d3b152a63dfdd8d"
|
|
172
128
|
dependencies = [
|
|
173
129
|
"libc",
|
|
174
130
|
"pyo3-build-config",
|
|
@@ -176,9 +132,9 @@ dependencies = [
|
|
|
176
132
|
|
|
177
133
|
[[package]]
|
|
178
134
|
name = "pyo3-macros"
|
|
179
|
-
version = "0.
|
|
135
|
+
version = "0.23.5"
|
|
180
136
|
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
181
|
-
checksum = "
|
|
137
|
+
checksum = "fbc2201328f63c4710f68abdf653c89d8dbc2858b88c5d88b0ff38a75288a9da"
|
|
182
138
|
dependencies = [
|
|
183
139
|
"proc-macro2",
|
|
184
140
|
"pyo3-macros-backend",
|
|
@@ -188,9 +144,9 @@ dependencies = [
|
|
|
188
144
|
|
|
189
145
|
[[package]]
|
|
190
146
|
name = "pyo3-macros-backend"
|
|
191
|
-
version = "0.
|
|
147
|
+
version = "0.23.5"
|
|
192
148
|
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
193
|
-
checksum = "
|
|
149
|
+
checksum = "fca6726ad0f3da9c9de093d6f116a93c1a38e417ed73bf138472cf4064f72028"
|
|
194
150
|
dependencies = [
|
|
195
151
|
"heck",
|
|
196
152
|
"proc-macro2",
|
|
@@ -208,15 +164,6 @@ dependencies = [
|
|
|
208
164
|
"proc-macro2",
|
|
209
165
|
]
|
|
210
166
|
|
|
211
|
-
[[package]]
|
|
212
|
-
name = "redox_syscall"
|
|
213
|
-
version = "0.5.18"
|
|
214
|
-
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
215
|
-
checksum = "ed2bf2547551a7053d6fdfafda3f938979645c44812fbfcda098faae3f1a362d"
|
|
216
|
-
dependencies = [
|
|
217
|
-
"bitflags",
|
|
218
|
-
]
|
|
219
|
-
|
|
220
167
|
[[package]]
|
|
221
168
|
name = "rustversion"
|
|
222
169
|
version = "1.0.22"
|
|
@@ -229,12 +176,6 @@ version = "1.0.21"
|
|
|
229
176
|
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
230
177
|
checksum = "62049b2877bf12821e8f9ad256ee38fdc31db7387ec2d3b3f403024de2034aea"
|
|
231
178
|
|
|
232
|
-
[[package]]
|
|
233
|
-
name = "scopeguard"
|
|
234
|
-
version = "1.2.0"
|
|
235
|
-
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
236
|
-
checksum = "94143f37725109f92c262ed2cf5e59bce7498c01bcc1502d7b9afe439a4e9f49"
|
|
237
|
-
|
|
238
179
|
[[package]]
|
|
239
180
|
name = "serde"
|
|
240
181
|
version = "1.0.228"
|
|
@@ -278,12 +219,6 @@ dependencies = [
|
|
|
278
219
|
"serde_core",
|
|
279
220
|
]
|
|
280
221
|
|
|
281
|
-
[[package]]
|
|
282
|
-
name = "smallvec"
|
|
283
|
-
version = "1.15.1"
|
|
284
|
-
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
285
|
-
checksum = "67b1b7a3b5fe4f1376887184045fcf45c69e92af734b7aaddc05fb777b6fbd03"
|
|
286
|
-
|
|
287
222
|
[[package]]
|
|
288
223
|
name = "syn"
|
|
289
224
|
version = "2.0.111"
|
|
@@ -318,14 +253,7 @@ name = "vectorwave_core"
|
|
|
318
253
|
version = "0.1.0"
|
|
319
254
|
dependencies = [
|
|
320
255
|
"crossbeam-channel",
|
|
321
|
-
"log",
|
|
322
256
|
"pyo3",
|
|
323
257
|
"serde",
|
|
324
258
|
"serde_json",
|
|
325
259
|
]
|
|
326
|
-
|
|
327
|
-
[[package]]
|
|
328
|
-
name = "windows-link"
|
|
329
|
-
version = "0.2.1"
|
|
330
|
-
source = "registry+https://github.com/rust-lang/crates.io-index"
|
|
331
|
-
checksum = "f0805222e57f7521d6a62e36fa9163bc891acd422f971defe97d64e70d0a4fe5"
|
|
@@ -8,8 +8,7 @@ name = "vectorwave_core"
|
|
|
8
8
|
crate-type = ["cdylib"]
|
|
9
9
|
|
|
10
10
|
[dependencies]
|
|
11
|
-
pyo3 = { version = "0.
|
|
12
|
-
crossbeam-channel = "0.5"
|
|
11
|
+
pyo3 = { version = "0.23.3", features = ["extension-module"] }
|
|
13
12
|
serde = { version = "1.0", features = ["derive"] }
|
|
14
13
|
serde_json = "1.0"
|
|
15
|
-
|
|
14
|
+
crossbeam-channel = "0.5"
|
|
@@ -1,20 +1,21 @@
|
|
|
1
1
|
use pyo3::prelude::*;
|
|
2
|
-
use pyo3::types::{PyDict, PyList};
|
|
2
|
+
use pyo3::types::{PyDict, PyList, PyString, PyBool, PyFloat, PyInt};
|
|
3
|
+
use std::collections::HashSet;
|
|
3
4
|
use std::thread;
|
|
4
5
|
use std::time::{Duration, Instant};
|
|
5
6
|
use crossbeam_channel::{bounded, Receiver, Sender, TrySendError};
|
|
6
7
|
|
|
7
8
|
struct LogItem {
|
|
8
|
-
collection:
|
|
9
|
+
collection: PyObject,
|
|
9
10
|
properties: Py<PyDict>,
|
|
10
|
-
uuid: Option<
|
|
11
|
+
uuid: Option<PyObject>,
|
|
11
12
|
vector: Option<Vec<f32>>,
|
|
12
13
|
}
|
|
13
14
|
|
|
14
15
|
#[pyclass]
|
|
15
16
|
struct RustBatchManager {
|
|
16
17
|
sender: Sender<LogItem>,
|
|
17
|
-
flush_callback:
|
|
18
|
+
flush_callback: PyObject,
|
|
18
19
|
worker_handle: Option<thread::JoinHandle<()>>,
|
|
19
20
|
stop_signal: Sender<()>,
|
|
20
21
|
}
|
|
@@ -23,14 +24,15 @@ struct RustBatchManager {
|
|
|
23
24
|
impl RustBatchManager {
|
|
24
25
|
#[new]
|
|
25
26
|
fn new(
|
|
26
|
-
|
|
27
|
+
py: Python<'_>,
|
|
28
|
+
callback: PyObject,
|
|
27
29
|
batch_threshold: usize,
|
|
28
30
|
flush_interval_ms: u64,
|
|
29
31
|
) -> Self {
|
|
30
32
|
let (tx, rx) = bounded::<LogItem>(10000);
|
|
31
33
|
let (stop_tx, stop_rx) = bounded(1);
|
|
32
34
|
|
|
33
|
-
let worker_callback = callback.
|
|
35
|
+
let worker_callback = callback.clone_ref(py);
|
|
34
36
|
|
|
35
37
|
let handle = thread::spawn(move || {
|
|
36
38
|
Self::worker_loop(rx, stop_rx, worker_callback, batch_threshold, flush_interval_ms);
|
|
@@ -44,11 +46,12 @@ impl RustBatchManager {
|
|
|
44
46
|
}
|
|
45
47
|
}
|
|
46
48
|
|
|
49
|
+
#[pyo3(signature = (collection, properties, uuid=None, vector=None))]
|
|
47
50
|
fn add_object(
|
|
48
51
|
&self,
|
|
49
|
-
collection:
|
|
52
|
+
collection: PyObject,
|
|
50
53
|
properties: Py<PyDict>,
|
|
51
|
-
uuid: Option<
|
|
54
|
+
uuid: Option<PyObject>,
|
|
52
55
|
vector: Option<Vec<f32>>,
|
|
53
56
|
) {
|
|
54
57
|
let item = LogItem {
|
|
@@ -71,7 +74,6 @@ impl RustBatchManager {
|
|
|
71
74
|
|
|
72
75
|
fn shutdown(&self) {
|
|
73
76
|
let _ = self.stop_signal.send(());
|
|
74
|
-
|
|
75
77
|
}
|
|
76
78
|
}
|
|
77
79
|
|
|
@@ -79,7 +81,7 @@ impl RustBatchManager {
|
|
|
79
81
|
fn worker_loop(
|
|
80
82
|
rx: Receiver<LogItem>,
|
|
81
83
|
stop_rx: Receiver<()>,
|
|
82
|
-
callback:
|
|
84
|
+
callback: PyObject,
|
|
83
85
|
threshold: usize,
|
|
84
86
|
interval_ms: u64,
|
|
85
87
|
) {
|
|
@@ -109,12 +111,12 @@ impl RustBatchManager {
|
|
|
109
111
|
}
|
|
110
112
|
}
|
|
111
113
|
|
|
112
|
-
fn flush_buffer(buffer: &Vec<LogItem>, callback: &
|
|
114
|
+
fn flush_buffer(buffer: &Vec<LogItem>, callback: &PyObject) {
|
|
113
115
|
Python::with_gil(|py| {
|
|
114
|
-
let py_list = PyList::
|
|
116
|
+
let py_list = PyList::empty_bound(py);
|
|
115
117
|
|
|
116
118
|
for item in buffer {
|
|
117
|
-
let dict = PyDict::
|
|
119
|
+
let dict = PyDict::new_bound(py);
|
|
118
120
|
let _ = dict.set_item("collection", &item.collection);
|
|
119
121
|
let _ = dict.set_item("properties", &item.properties);
|
|
120
122
|
|
|
@@ -141,8 +143,51 @@ impl RustBatchManager {
|
|
|
141
143
|
}
|
|
142
144
|
}
|
|
143
145
|
|
|
146
|
+
|
|
147
|
+
fn process_recursive(py: Python, value: &Bound<'_, PyAny>, sensitive_set: &HashSet<String>) -> PyResult<PyObject> {
|
|
148
|
+
if let Ok(dict_obj) = value.downcast::<PyDict>() {
|
|
149
|
+
let new_dict = PyDict::new_bound(py);
|
|
150
|
+
for (k, v) in dict_obj {
|
|
151
|
+
let k_str = k.to_string().to_lowercase();
|
|
152
|
+
if sensitive_set.contains(&k_str) {
|
|
153
|
+
new_dict.set_item(k, "[MASKED]")?;
|
|
154
|
+
} else {
|
|
155
|
+
new_dict.set_item(k, process_recursive(py, &v, sensitive_set)?)?;
|
|
156
|
+
}
|
|
157
|
+
}
|
|
158
|
+
Ok(new_dict.into())
|
|
159
|
+
} else if let Ok(list_obj) = value.downcast::<PyList>() {
|
|
160
|
+
let new_list = PyList::empty_bound(py);
|
|
161
|
+
for item in list_obj {
|
|
162
|
+
new_list.append(process_recursive(py, &item, sensitive_set)?)?;
|
|
163
|
+
}
|
|
164
|
+
Ok(new_list.into())
|
|
165
|
+
} else {
|
|
166
|
+
if value.is_none()
|
|
167
|
+
|| value.is_instance_of::<PyBool>()
|
|
168
|
+
|| value.is_instance_of::<PyFloat>()
|
|
169
|
+
|| value.is_instance_of::<PyInt>()
|
|
170
|
+
|| value.is_instance_of::<PyString>() {
|
|
171
|
+
// [수정] unbind()를 사용하여 PyObject로 변환
|
|
172
|
+
Ok(value.clone().unbind())
|
|
173
|
+
} else {
|
|
174
|
+
match value.str() {
|
|
175
|
+
Ok(s) => Ok(s.into()),
|
|
176
|
+
Err(_) => Ok(PyString::new_bound(py, "[SERIALIZATION_ERROR]").into())
|
|
177
|
+
}
|
|
178
|
+
}
|
|
179
|
+
}
|
|
180
|
+
}
|
|
181
|
+
|
|
182
|
+
#[pyfunction]
|
|
183
|
+
fn mask_and_serialize(py: Python, data: &Bound<'_, PyAny>, sensitive_keys: Vec<String>) -> PyResult<PyObject> {
|
|
184
|
+
let sensitive_set: HashSet<String> = sensitive_keys.into_iter().map(|s| s.to_lowercase()).collect();
|
|
185
|
+
process_recursive(py, data, &sensitive_set)
|
|
186
|
+
}
|
|
187
|
+
|
|
144
188
|
#[pymodule]
|
|
145
|
-
fn vectorwave_core(
|
|
189
|
+
fn vectorwave_core(m: &Bound<'_, PyModule>) -> PyResult<()> {
|
|
146
190
|
m.add_class::<RustBatchManager>()?;
|
|
191
|
+
m.add_function(wrap_pyfunction!(mask_and_serialize, m)?)?;
|
|
147
192
|
Ok(())
|
|
148
193
|
}
|
|
@@ -9,6 +9,7 @@ from typing import Optional, List, Dict, Any, Callable
|
|
|
9
9
|
from uuid import uuid4
|
|
10
10
|
from datetime import datetime, timezone
|
|
11
11
|
|
|
12
|
+
import vectorwave.vectorwave_core as vectorwave_core
|
|
12
13
|
from .alert.base import BaseAlerter
|
|
13
14
|
from ..batch.batch import get_batch_manager
|
|
14
15
|
from ..models.db_config import get_weaviate_settings, WeaviateSettings
|
|
@@ -34,31 +35,6 @@ current_tracer_var: ContextVar[Optional[TraceCollector]] = ContextVar('current_t
|
|
|
34
35
|
current_span_id_var: ContextVar[Optional[str]] = ContextVar('current_span_id', default=None)
|
|
35
36
|
|
|
36
37
|
|
|
37
|
-
def _mask_and_serialize(data: Any, sensitive_keys: set) -> Any:
|
|
38
|
-
"""
|
|
39
|
-
Recursively masks sensitive data in dicts/lists and prepares for serialization.
|
|
40
|
-
"""
|
|
41
|
-
if isinstance(data, dict):
|
|
42
|
-
new_dict = {}
|
|
43
|
-
for k, v in data.items():
|
|
44
|
-
if str(k).lower() in sensitive_keys:
|
|
45
|
-
new_dict[k] = "[MASKED]"
|
|
46
|
-
else:
|
|
47
|
-
new_dict[k] = _mask_and_serialize(v, sensitive_keys)
|
|
48
|
-
return new_dict
|
|
49
|
-
|
|
50
|
-
if isinstance(data, list):
|
|
51
|
-
return [_mask_and_serialize(item, sensitive_keys) for item in data]
|
|
52
|
-
|
|
53
|
-
if not isinstance(data, (str, int, float, bool, type(None))):
|
|
54
|
-
try:
|
|
55
|
-
return str(data)
|
|
56
|
-
except Exception:
|
|
57
|
-
return "[SERIALIZATION_ERROR]"
|
|
58
|
-
|
|
59
|
-
return data
|
|
60
|
-
|
|
61
|
-
|
|
62
38
|
def _capture_span_attributes(
|
|
63
39
|
attributes_to_capture: Optional[List[str]],
|
|
64
40
|
kwargs: Dict[str, Any],
|
|
@@ -76,7 +52,8 @@ def _capture_span_attributes(
|
|
|
76
52
|
processed_value = "[MASKED]"
|
|
77
53
|
else:
|
|
78
54
|
raw_value = kwargs[attr_name]
|
|
79
|
-
|
|
55
|
+
# Rust extension expects a List[str] for sensitive_keys, not set
|
|
56
|
+
processed_value = vectorwave_core.mask_and_serialize(raw_value, list(sensitive_keys))
|
|
80
57
|
|
|
81
58
|
captured_attributes[attr_name] = processed_value
|
|
82
59
|
|
|
@@ -160,8 +137,8 @@ def _create_input_vector_data(
|
|
|
160
137
|
the core arguments and vectorize them
|
|
161
138
|
"""
|
|
162
139
|
# 1. Process/Mask positional and keyword arguments
|
|
163
|
-
processed_args =
|
|
164
|
-
processed_kwargs =
|
|
140
|
+
processed_args = vectorwave_core.mask_and_serialize(list(args), list(sensitive_keys))
|
|
141
|
+
processed_kwargs = vectorwave_core.mask_and_serialize(kwargs, list(sensitive_keys))
|
|
165
142
|
|
|
166
143
|
|
|
167
144
|
texts_for_vector = [f"Function Context: {func_name}"]
|
|
@@ -309,7 +286,7 @@ def trace_span(
|
|
|
309
286
|
result = await func(*args, **kwargs)
|
|
310
287
|
|
|
311
288
|
if capture_return_value:
|
|
312
|
-
processed_result =
|
|
289
|
+
processed_result = vectorwave_core.mask_and_serialize(result, list(tracer.settings.sensitive_keys))
|
|
313
290
|
try:
|
|
314
291
|
return_value_log = json.dumps(processed_result)
|
|
315
292
|
except TypeError:
|
|
@@ -442,7 +419,7 @@ def trace_span(
|
|
|
442
419
|
result = func(*args, **kwargs)
|
|
443
420
|
|
|
444
421
|
if capture_return_value:
|
|
445
|
-
processed_result =
|
|
422
|
+
processed_result = vectorwave_core.mask_and_serialize(result, list(tracer.settings.sensitive_keys))
|
|
446
423
|
try:
|
|
447
424
|
return_value_log = json.dumps(processed_result)
|
|
448
425
|
except TypeError:
|
|
@@ -498,7 +475,7 @@ def trace_span(
|
|
|
498
475
|
|
|
499
476
|
if is_drift:
|
|
500
477
|
drift_alert_props = span_properties.copy()
|
|
501
|
-
drift_alert_props["status"] = "WARNING"
|
|
478
|
+
drift_alert_props["status"] = "WARNING"
|
|
502
479
|
drift_alert_props["error_code"] = "SEMANTIC_DRIFT"
|
|
503
480
|
drift_alert_props["error_message"] = (
|
|
504
481
|
f"Anomaly detected in data distribution.\n"
|
|
@@ -535,4 +512,4 @@ def trace_span(
|
|
|
535
512
|
if _func is None:
|
|
536
513
|
return decorator
|
|
537
514
|
else:
|
|
538
|
-
return decorator(_func)
|
|
515
|
+
return decorator(_func)
|
|
@@ -12,7 +12,7 @@ import weaviate.classes.query as wvc_query
|
|
|
12
12
|
|
|
13
13
|
from ..database.db import get_cached_client
|
|
14
14
|
from ..models.db_config import get_weaviate_settings
|
|
15
|
-
|
|
15
|
+
import vectorwave.vectorwave_core as vectorwave_core
|
|
16
16
|
from .context import execution_source_context
|
|
17
17
|
|
|
18
18
|
logger = logging.getLogger(__name__)
|
|
@@ -238,7 +238,7 @@ class VectorWaveReplayer:
|
|
|
238
238
|
collection_name = self.golden_collection_name if is_golden else self.collection_name
|
|
239
239
|
collection = self.client.collections.get(collection_name)
|
|
240
240
|
|
|
241
|
-
processed_val =
|
|
241
|
+
processed_val = vectorwave_core.mask_and_serialize(new_value, [])
|
|
242
242
|
try:
|
|
243
243
|
val_str = json.dumps(processed_val)
|
|
244
244
|
except (TypeError, ValueError):
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|
|
File without changes
|