datamasque-python 1.2.0__tar.gz → 1.2.1__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.
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/HISTORY.rst +5 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/PKG-INFO +1 -1
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/__init__.py +9 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/discovery.py +198 -27
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/dmclient.py +2 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/exceptions.py +8 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/discovery.py +81 -2
- datamasque_python-1.2.1/datamasque/client/models/rg_config.py +49 -0
- datamasque_python-1.2.1/datamasque/client/rg_configs.py +155 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/ruleset_libraries.py +4 -1
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/docs/client.models.rst +8 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/docs/client.rst +8 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/pyproject.toml +1 -1
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/setup.cfg +1 -1
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_discovery.py +349 -0
- datamasque_python-1.2.1/tests/test_rg_configs.py +522 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/uv.lock +1 -1
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/.editorconfig +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/.github/workflows/ci.yml +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/.github/workflows/release-testpypi.yml +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/.github/workflows/release.yml +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/.gitignore +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/.readthedocs.yaml +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/CONTRIBUTING.rst +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/LICENSE +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/MANIFEST.in +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/Makefile +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/NOTICE +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/README.rst +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/base.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/connections.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/discovery_config_libraries.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/discovery_configs.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/files.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/ifm.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/license.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/__init__.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/connection.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/data_selection.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/discovery_config.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/discovery_config_library.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/dm_instance.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/files.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/git.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/ifm.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/license.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/pagination.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/ruleset.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/ruleset_library.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/runs.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/safe_data_preview.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/status.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/table_reference.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/models/user.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/py.typed +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/rulesets.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/runs.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/settings.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/spcs.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/table_references.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/datamasque/client/users.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/docs/Makefile +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/docs/conf.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/docs/contributing.rst +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/docs/history.rst +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/docs/index.rst +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/docs/installation.rst +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/docs/make.bat +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/docs/modules.rst +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/docs/readme.rst +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/docs/usage.rst +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/__init__.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/conftest.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/helpers.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_base.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_connections.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_discovery_config_libraries.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_discovery_configs.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_files.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_ifm.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_license.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_pagination.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_ruleset_library.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_rulesets.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_runs.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_safe_data_preview.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_settings.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_spcs.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_table_references.py +0 -0
- {datamasque_python-1.2.0 → datamasque_python-1.2.1}/tests/test_users.py +0 -0
|
@@ -1,6 +1,6 @@
|
|
|
1
1
|
Metadata-Version: 2.4
|
|
2
2
|
Name: datamasque-python
|
|
3
|
-
Version: 1.2.
|
|
3
|
+
Version: 1.2.1
|
|
4
4
|
Summary: Official Python client for the DataMasque data-masking API.
|
|
5
5
|
Project-URL: Homepage, https://datamasque.com/
|
|
6
6
|
Project-URL: Documentation, https://datamasque-python.readthedocs.io/
|
|
@@ -23,6 +23,7 @@ from datamasque.client.exceptions import (
|
|
|
23
23
|
InvalidDiscoveryConfigError,
|
|
24
24
|
InvalidLibraryError,
|
|
25
25
|
InvalidRulesetError,
|
|
26
|
+
RGConfigNotFoundError,
|
|
26
27
|
RunNotCancellableError,
|
|
27
28
|
)
|
|
28
29
|
from datamasque.client.ifm import DataMasqueIfmClient
|
|
@@ -67,11 +68,13 @@ from datamasque.client.models.discovery import (
|
|
|
67
68
|
FileFilter,
|
|
68
69
|
FileFilterMatchAgainst,
|
|
69
70
|
FileRulesetGenerationRequest,
|
|
71
|
+
FileRulesetGenerationWithRGConfigRequest,
|
|
70
72
|
ForeignKeyRef,
|
|
71
73
|
InDataDiscoveryConfig,
|
|
72
74
|
InDataDiscoveryRule,
|
|
73
75
|
ReferencingForeignKey,
|
|
74
76
|
RulesetGenerationRequest,
|
|
77
|
+
RulesetGenerationWithRGConfigRequest,
|
|
75
78
|
SchemaDiscoveryColumn,
|
|
76
79
|
SchemaDiscoveryFromConfigRequest,
|
|
77
80
|
SchemaDiscoveryPage,
|
|
@@ -105,6 +108,7 @@ from datamasque.client.models.ifm import (
|
|
|
105
108
|
RulesetPlanUpdateRequest,
|
|
106
109
|
)
|
|
107
110
|
from datamasque.client.models.license import LicenseInfo, SwitchableLicenseMetadata
|
|
111
|
+
from datamasque.client.models.rg_config import RGConfig, RGConfigId
|
|
108
112
|
from datamasque.client.models.ruleset import Ruleset, RulesetId, RulesetType
|
|
109
113
|
from datamasque.client.models.ruleset_library import RulesetLibrary, RulesetLibraryId
|
|
110
114
|
from datamasque.client.models.runs import (
|
|
@@ -214,6 +218,7 @@ __all__ = [
|
|
|
214
218
|
"FileId",
|
|
215
219
|
"FileOrContent",
|
|
216
220
|
"FileRulesetGenerationRequest",
|
|
221
|
+
"FileRulesetGenerationWithRGConfigRequest",
|
|
217
222
|
"FirstCharsStatistics",
|
|
218
223
|
"ForeignKeyRef",
|
|
219
224
|
"GitSnapshot",
|
|
@@ -252,9 +257,13 @@ __all__ = [
|
|
|
252
257
|
"PatternComposition",
|
|
253
258
|
"PatternEntry",
|
|
254
259
|
"PatternsStatistics",
|
|
260
|
+
"RGConfig",
|
|
261
|
+
"RGConfigId",
|
|
262
|
+
"RGConfigNotFoundError",
|
|
255
263
|
"ReferencingForeignKey",
|
|
256
264
|
"Ruleset",
|
|
257
265
|
"RulesetGenerationRequest",
|
|
266
|
+
"RulesetGenerationWithRGConfigRequest",
|
|
258
267
|
"RulesetId",
|
|
259
268
|
"RulesetLibrary",
|
|
260
269
|
"RulesetLibraryId",
|
|
@@ -13,6 +13,7 @@ from datamasque.client.exceptions import (
|
|
|
13
13
|
DiscoveryConfigNotFoundError,
|
|
14
14
|
FailedToStartError,
|
|
15
15
|
InvalidDiscoveryConfigError,
|
|
16
|
+
RGConfigNotFoundError,
|
|
16
17
|
)
|
|
17
18
|
from datamasque.client.models.connection import ConnectionId
|
|
18
19
|
from datamasque.client.models.data_selection import (
|
|
@@ -25,12 +26,15 @@ from datamasque.client.models.discovery import (
|
|
|
25
26
|
FileDataDiscoveryRequest,
|
|
26
27
|
FileDiscoveryResult,
|
|
27
28
|
FileRulesetGenerationRequest,
|
|
29
|
+
FileRulesetGenerationWithRGConfigRequest,
|
|
28
30
|
RulesetGenerationRequest,
|
|
31
|
+
RulesetGenerationWithRGConfigRequest,
|
|
29
32
|
SchemaDiscoveryFromConfigRequest,
|
|
30
33
|
SchemaDiscoveryPage,
|
|
31
34
|
SchemaDiscoveryRequest,
|
|
32
35
|
SchemaDiscoveryResult,
|
|
33
36
|
)
|
|
37
|
+
from datamasque.client.models.rg_config import RGConfig, RGConfigId, unwrap_rg_config_id
|
|
34
38
|
from datamasque.client.models.ruleset import Ruleset
|
|
35
39
|
from datamasque.client.models.runs import RunId
|
|
36
40
|
from datamasque.client.models.status import AsyncRulesetGenerationTaskStatus
|
|
@@ -41,21 +45,12 @@ logger = logging.getLogger(__name__)
|
|
|
41
45
|
class DiscoveryClient(BaseClient):
|
|
42
46
|
"""Schema-discovery and ruleset-generation API methods. Mixed into `DataMasqueClient`."""
|
|
43
47
|
|
|
44
|
-
|
|
45
|
-
|
|
46
|
-
|
|
47
|
-
|
|
48
|
-
If the connection is a database connection, `selected_data` should be of type `SelectedColumns`.
|
|
49
|
-
If the connection is a file connection, `selected_data` should be of type `SelectedFileData`.
|
|
50
|
-
|
|
51
|
-
Generation runs asynchronously on the server.
|
|
52
|
-
Poll `get_async_ruleset_generation_task_status` until it returns
|
|
53
|
-
`AsyncRulesetGenerationTaskStatus.finished`,
|
|
54
|
-
then call `get_generated_rulesets` to retrieve the resulting `Ruleset`.
|
|
55
|
-
"""
|
|
48
|
+
@staticmethod
|
|
49
|
+
def _selected_data_payload(selected_data: SelectedData) -> dict:
|
|
50
|
+
"""Build the async-generation request-body fields for a column or file selection, validating its shape."""
|
|
56
51
|
|
|
57
52
|
if not selected_data:
|
|
58
|
-
raise ValueError("`selected_data` is a required argument to
|
|
53
|
+
raise ValueError("`selected_data` is a required argument to async ruleset generation.")
|
|
59
54
|
|
|
60
55
|
data: dict = {}
|
|
61
56
|
if isinstance(selected_data, SelectedColumns):
|
|
@@ -75,12 +70,56 @@ class DiscoveryClient(BaseClient):
|
|
|
75
70
|
data["selected_data"] = [s.model_dump() for s in selected_data.user_selections]
|
|
76
71
|
else:
|
|
77
72
|
raise TypeError(
|
|
78
|
-
f"The
|
|
73
|
+
f"The `selected_data` argument to async ruleset generation was of an invalid type, "
|
|
79
74
|
f"expected `SelectedColumns` or `SelectedFileData`, got {type(selected_data)}."
|
|
80
75
|
)
|
|
81
76
|
|
|
77
|
+
return data
|
|
78
|
+
|
|
79
|
+
def start_async_ruleset_generation(self, connection_id: ConnectionId, selected_data: SelectedData) -> None:
|
|
80
|
+
"""
|
|
81
|
+
Starts async ruleset generation using the most recent discovery results on the given connection.
|
|
82
|
+
|
|
83
|
+
Masks are assigned from the server's default RG config;
|
|
84
|
+
use `start_async_ruleset_generation_with_rg_config` to generate with a saved RG config.
|
|
85
|
+
|
|
86
|
+
If the connection is a database connection, `selected_data` should be of type `SelectedColumns`.
|
|
87
|
+
If the connection is a file connection, `selected_data` should be of type `SelectedFileData`.
|
|
88
|
+
|
|
89
|
+
Generation runs asynchronously on the server.
|
|
90
|
+
Poll `get_async_ruleset_generation_task_status` until it returns `AsyncRulesetGenerationTaskStatus.finished`,
|
|
91
|
+
then call `get_generated_rulesets` to retrieve the resulting `Ruleset`.
|
|
92
|
+
"""
|
|
93
|
+
|
|
94
|
+
data = self._selected_data_payload(selected_data)
|
|
82
95
|
self.make_request(method="POST", path=f"/api/async-generate-ruleset/{connection_id}/", data=data)
|
|
83
96
|
|
|
97
|
+
def start_async_ruleset_generation_with_rg_config(
|
|
98
|
+
self,
|
|
99
|
+
connection_id: ConnectionId,
|
|
100
|
+
selected_data: SelectedData,
|
|
101
|
+
rg_config: Optional[Union[RGConfigId, RGConfig]],
|
|
102
|
+
) -> None:
|
|
103
|
+
"""
|
|
104
|
+
Starts async ruleset generation with a selected RG config mapping discovered labels to masks.
|
|
105
|
+
|
|
106
|
+
Like `start_async_ruleset_generation`,
|
|
107
|
+
but posts to the v2 endpoint with a required `rg_config`:
|
|
108
|
+
a saved RG config (`RGConfigId` or `RGConfig`),
|
|
109
|
+
or `None` for the server's default RG config.
|
|
110
|
+
|
|
111
|
+
Raises `RGConfigNotFoundError` if the selected RG config does not exist on the server.
|
|
112
|
+
|
|
113
|
+
Generation runs asynchronously on the server.
|
|
114
|
+
Poll `get_async_ruleset_generation_task_status` until it returns `AsyncRulesetGenerationTaskStatus.finished`,
|
|
115
|
+
then call `get_generated_rulesets` to retrieve the resulting `Ruleset`.
|
|
116
|
+
"""
|
|
117
|
+
|
|
118
|
+
data = self._selected_data_payload(selected_data)
|
|
119
|
+
# The server requires `rg_config` to be present; an explicit null selects the default RG config.
|
|
120
|
+
data["rg_config"] = unwrap_rg_config_id(rg_config)
|
|
121
|
+
self._post_with_rg_config(f"/api/async-generate-ruleset/v2/{connection_id}/", data)
|
|
122
|
+
|
|
84
123
|
def start_async_ruleset_generation_from_csv(
|
|
85
124
|
self,
|
|
86
125
|
connection_id: ConnectionId,
|
|
@@ -103,11 +142,57 @@ class DiscoveryClient(BaseClient):
|
|
|
103
142
|
otherwise it is uploaded as CSV.
|
|
104
143
|
|
|
105
144
|
Generation runs asynchronously on the server.
|
|
106
|
-
Poll `get_async_ruleset_generation_task_status` until it returns
|
|
107
|
-
`AsyncRulesetGenerationTaskStatus.finished`,
|
|
145
|
+
Poll `get_async_ruleset_generation_task_status` until it returns `AsyncRulesetGenerationTaskStatus.finished`,
|
|
108
146
|
then call `get_generated_rulesets` to retrieve the resulting `Ruleset` objects.
|
|
109
147
|
"""
|
|
110
148
|
|
|
149
|
+
self.make_request(
|
|
150
|
+
method="POST",
|
|
151
|
+
path=f"/api/async-generate-ruleset/{connection_id}/from-csv/",
|
|
152
|
+
data={"target_size_bytes": target_size_bytes} if target_size_bytes is not None else None,
|
|
153
|
+
files=self._csv_upload_files(csv_content),
|
|
154
|
+
)
|
|
155
|
+
|
|
156
|
+
def start_async_ruleset_generation_from_csv_with_rg_config(
|
|
157
|
+
self,
|
|
158
|
+
connection_id: ConnectionId,
|
|
159
|
+
csv_content: Union[str, bytes, TextIOBase, BufferedIOBase],
|
|
160
|
+
rg_config: Optional[Union[RGConfigId, RGConfig]],
|
|
161
|
+
target_size_bytes: Optional[int] = None,
|
|
162
|
+
) -> None:
|
|
163
|
+
"""
|
|
164
|
+
Generate ruleset(s) from a schema discovery CSV report with a selected RG config.
|
|
165
|
+
|
|
166
|
+
Like `start_async_ruleset_generation_from_csv`,
|
|
167
|
+
but posts to the v2 endpoint with a required `rg_config`:
|
|
168
|
+
a saved RG config (`RGConfigId` or `RGConfig`),
|
|
169
|
+
or `None` for the server's default RG config.
|
|
170
|
+
|
|
171
|
+
Raises `RGConfigNotFoundError` if the selected RG config does not exist on the server.
|
|
172
|
+
|
|
173
|
+
Generation runs asynchronously on the server.
|
|
174
|
+
Poll `get_async_ruleset_generation_task_status` until it returns `AsyncRulesetGenerationTaskStatus.finished`,
|
|
175
|
+
then call `get_generated_rulesets` to retrieve the resulting `Ruleset` objects.
|
|
176
|
+
"""
|
|
177
|
+
|
|
178
|
+
rg_config_id = unwrap_rg_config_id(rg_config)
|
|
179
|
+
# The upload is a multipart form, whose fields cannot carry a JSON null;
|
|
180
|
+
# the server reads an empty string as null for this nullable field,
|
|
181
|
+
# selecting the default RG config.
|
|
182
|
+
data: dict = {"rg_config": rg_config_id if rg_config_id is not None else ""}
|
|
183
|
+
if target_size_bytes is not None:
|
|
184
|
+
data["target_size_bytes"] = target_size_bytes
|
|
185
|
+
|
|
186
|
+
self._post_with_rg_config(
|
|
187
|
+
f"/api/async-generate-ruleset/v2/{connection_id}/from-csv/",
|
|
188
|
+
data,
|
|
189
|
+
files=self._csv_upload_files(csv_content),
|
|
190
|
+
)
|
|
191
|
+
|
|
192
|
+
@staticmethod
|
|
193
|
+
def _csv_upload_files(csv_content: Union[str, bytes, TextIOBase, BufferedIOBase]) -> list[UploadFile]:
|
|
194
|
+
"""Normalise CSV-or-zip report content into the `csv_or_zip_file` multipart upload."""
|
|
195
|
+
|
|
111
196
|
content: BufferedIOBase
|
|
112
197
|
if isinstance(csv_content, str):
|
|
113
198
|
content = BytesIO(csv_content.encode())
|
|
@@ -125,7 +210,7 @@ class DiscoveryClient(BaseClient):
|
|
|
125
210
|
filename = "ruleset.zip" if is_zip else "ruleset.csv"
|
|
126
211
|
content_type = "application/zip" if is_zip else "text/csv"
|
|
127
212
|
|
|
128
|
-
|
|
213
|
+
return [
|
|
129
214
|
UploadFile(
|
|
130
215
|
field_name="csv_or_zip_file",
|
|
131
216
|
filename=filename,
|
|
@@ -134,13 +219,6 @@ class DiscoveryClient(BaseClient):
|
|
|
134
219
|
),
|
|
135
220
|
]
|
|
136
221
|
|
|
137
|
-
self.make_request(
|
|
138
|
-
method="POST",
|
|
139
|
-
path=f"/api/async-generate-ruleset/{connection_id}/from-csv/",
|
|
140
|
-
data={"target_size_bytes": target_size_bytes} if target_size_bytes is not None else None,
|
|
141
|
-
files=files,
|
|
142
|
-
)
|
|
143
|
-
|
|
144
222
|
def get_async_ruleset_generation_task_status(self, connection_id: ConnectionId) -> AsyncRulesetGenerationTaskStatus:
|
|
145
223
|
"""Queries the status of an async ruleset generation task."""
|
|
146
224
|
|
|
@@ -381,7 +459,7 @@ class DiscoveryClient(BaseClient):
|
|
|
381
459
|
if not (errors := run_data.get(cls.DISCOVERY_CONFIG_ERROR_FIELD)):
|
|
382
460
|
return
|
|
383
461
|
|
|
384
|
-
detail = cls.
|
|
462
|
+
detail = cls._format_config_error(errors)
|
|
385
463
|
if cls.MISSING_DISCOVERY_CONFIG_SIGNATURE in detail:
|
|
386
464
|
raise DiscoveryConfigNotFoundError(
|
|
387
465
|
f"{run_kind} run failed to start: the referenced discovery config could not be found: {detail}",
|
|
@@ -394,14 +472,61 @@ class DiscoveryClient(BaseClient):
|
|
|
394
472
|
)
|
|
395
473
|
|
|
396
474
|
@staticmethod
|
|
397
|
-
def
|
|
398
|
-
"""Render the first server error, handling both string and `{message, ...}` dict items."""
|
|
475
|
+
def _format_config_error(errors: object) -> str:
|
|
476
|
+
"""Render the first server error for a config field, handling both string and `{message, ...}` dict items."""
|
|
399
477
|
first = errors[0] if isinstance(errors, list) and errors else errors
|
|
400
478
|
if isinstance(first, dict) and "message" in first:
|
|
401
479
|
return str(first["message"])
|
|
402
480
|
|
|
403
481
|
return str(first)
|
|
404
482
|
|
|
483
|
+
# Server key for an error that names the selected RG config.
|
|
484
|
+
RG_CONFIG_ERROR_FIELD = "rg_config"
|
|
485
|
+
|
|
486
|
+
# The phrase the server uses when the config id cannot be resolved
|
|
487
|
+
# (the config does not exist, or has been archived).
|
|
488
|
+
MISSING_RG_CONFIG_SIGNATURE = "object does not exist"
|
|
489
|
+
|
|
490
|
+
def _post_with_rg_config(
|
|
491
|
+
self,
|
|
492
|
+
path: str,
|
|
493
|
+
data: dict,
|
|
494
|
+
files: Optional[list[UploadFile]] = None,
|
|
495
|
+
) -> Response:
|
|
496
|
+
"""Post a generation request carrying an `rg_config`, classifying a rejected config on failure."""
|
|
497
|
+
|
|
498
|
+
response = self.make_request("POST", path, data=data, files=files, require_status_check=False)
|
|
499
|
+
self._maybe_raise_rg_config_not_found(response)
|
|
500
|
+
self._raise_for_status(response, request_data=data)
|
|
501
|
+
return response
|
|
502
|
+
|
|
503
|
+
@classmethod
|
|
504
|
+
def _maybe_raise_rg_config_not_found(cls, response: Response) -> None:
|
|
505
|
+
"""Raise `RGConfigNotFoundError` if the server's error body says the selected RG config does not exist."""
|
|
506
|
+
|
|
507
|
+
if response.ok:
|
|
508
|
+
return
|
|
509
|
+
|
|
510
|
+
try:
|
|
511
|
+
body = response.json()
|
|
512
|
+
except ValueError:
|
|
513
|
+
return
|
|
514
|
+
|
|
515
|
+
if not isinstance(body, dict):
|
|
516
|
+
return
|
|
517
|
+
|
|
518
|
+
if not (errors := body.get(cls.RG_CONFIG_ERROR_FIELD)):
|
|
519
|
+
return
|
|
520
|
+
|
|
521
|
+
detail = cls._format_config_error(errors)
|
|
522
|
+
if cls.MISSING_RG_CONFIG_SIGNATURE not in detail:
|
|
523
|
+
return
|
|
524
|
+
|
|
525
|
+
raise RGConfigNotFoundError(
|
|
526
|
+
f"The referenced RG config could not be found: {detail}",
|
|
527
|
+
response=response,
|
|
528
|
+
)
|
|
529
|
+
|
|
405
530
|
def iter_schema_discovery_results(self, run_id: RunId) -> Iterator[SchemaDiscoveryResult]:
|
|
406
531
|
"""Lazily iterate all schema discovery results for a run via the paginated v2 endpoint."""
|
|
407
532
|
|
|
@@ -433,6 +558,9 @@ class DiscoveryClient(BaseClient):
|
|
|
433
558
|
"""
|
|
434
559
|
Generates database-masking ruleset YAML from the most recent discovery run on the given connection.
|
|
435
560
|
|
|
561
|
+
Masks are assigned from the server's default RG config;
|
|
562
|
+
use `generate_ruleset_with_rg_config` to generate with a saved RG config.
|
|
563
|
+
|
|
436
564
|
`generation_request` is a `RulesetGenerationRequest`.
|
|
437
565
|
"""
|
|
438
566
|
|
|
@@ -440,10 +568,26 @@ class DiscoveryClient(BaseClient):
|
|
|
440
568
|
response = self.make_request("POST", "/api/generate-ruleset/v2/", data=data)
|
|
441
569
|
return response.content.decode("utf-8")
|
|
442
570
|
|
|
571
|
+
def generate_ruleset_with_rg_config(self, generation_request: RulesetGenerationWithRGConfigRequest) -> str:
|
|
572
|
+
"""
|
|
573
|
+
Generates database-masking ruleset YAML with a selected RG config mapping discovered labels to masks.
|
|
574
|
+
|
|
575
|
+
`generation_request` is a `RulesetGenerationWithRGConfigRequest`;
|
|
576
|
+
its required `rg_config` selects the saved RG config to use
|
|
577
|
+
(`None` for the server's default RG config).
|
|
578
|
+
|
|
579
|
+
Raises `RGConfigNotFoundError` if the selected RG config does not exist on the server.
|
|
580
|
+
"""
|
|
581
|
+
|
|
582
|
+
return self._generate_ruleset_with_rg_config(generation_request, "/api/generate-ruleset/v3/")
|
|
583
|
+
|
|
443
584
|
def generate_file_ruleset(self, generation_request: FileRulesetGenerationRequest) -> str:
|
|
444
585
|
"""
|
|
445
586
|
Generates file-masking ruleset YAML from the most recent file-data-discovery run on the given connection.
|
|
446
587
|
|
|
588
|
+
Masks are assigned from the server's default RG config;
|
|
589
|
+
use `generate_file_ruleset_with_rg_config` to generate with a saved RG config.
|
|
590
|
+
|
|
447
591
|
`generation_request` is a `FileRulesetGenerationRequest`.
|
|
448
592
|
"""
|
|
449
593
|
|
|
@@ -451,6 +595,33 @@ class DiscoveryClient(BaseClient):
|
|
|
451
595
|
response = self.make_request("POST", "/api/generate-file-ruleset/", data=data)
|
|
452
596
|
return response.content.decode("utf-8")
|
|
453
597
|
|
|
598
|
+
def generate_file_ruleset_with_rg_config(self, generation_request: FileRulesetGenerationWithRGConfigRequest) -> str:
|
|
599
|
+
"""
|
|
600
|
+
Generates file-masking ruleset YAML with a selected RG config mapping discovered labels to masks.
|
|
601
|
+
|
|
602
|
+
`generation_request` is a `FileRulesetGenerationWithRGConfigRequest`;
|
|
603
|
+
its required `rg_config` selects the saved RG config to use
|
|
604
|
+
(`None` for the server's default RG config).
|
|
605
|
+
|
|
606
|
+
Raises `RGConfigNotFoundError` if the selected RG config does not exist on the server.
|
|
607
|
+
"""
|
|
608
|
+
|
|
609
|
+
return self._generate_ruleset_with_rg_config(generation_request, "/api/generate-file-ruleset/v2/")
|
|
610
|
+
|
|
611
|
+
def _generate_ruleset_with_rg_config(
|
|
612
|
+
self,
|
|
613
|
+
generation_request: Union[RulesetGenerationWithRGConfigRequest, FileRulesetGenerationWithRGConfigRequest],
|
|
614
|
+
path: str,
|
|
615
|
+
) -> str:
|
|
616
|
+
"""Post a with-RG-config generation request and return the generated ruleset YAML."""
|
|
617
|
+
|
|
618
|
+
data = generation_request.model_dump(exclude_none=True, mode="json")
|
|
619
|
+
# The server requires `rg_config` to be present; a null selects its default RG config,
|
|
620
|
+
# so send it explicitly rather than letting `exclude_none` drop a None.
|
|
621
|
+
data.setdefault("rg_config", None)
|
|
622
|
+
response = self._post_with_rg_config(path, data)
|
|
623
|
+
return response.content.decode("utf-8")
|
|
624
|
+
|
|
454
625
|
def get_file_data_discovery_report(self, run_id: RunId) -> list[FileDiscoveryResult]:
|
|
455
626
|
"""Returns the file-data-discovery results for the specified run."""
|
|
456
627
|
|
|
@@ -5,6 +5,7 @@ from datamasque.client.discovery_config_libraries import DiscoveryConfigLibraryC
|
|
|
5
5
|
from datamasque.client.discovery_configs import DiscoveryConfigClient
|
|
6
6
|
from datamasque.client.files import FileClient
|
|
7
7
|
from datamasque.client.license import LicenseClient
|
|
8
|
+
from datamasque.client.rg_configs import RGConfigClient
|
|
8
9
|
from datamasque.client.ruleset_libraries import RulesetLibraryClient
|
|
9
10
|
from datamasque.client.rulesets import RulesetClient
|
|
10
11
|
from datamasque.client.runs import RunClient
|
|
@@ -26,6 +27,7 @@ class DataMasqueClient(
|
|
|
26
27
|
DiscoveryConfigClient,
|
|
27
28
|
DiscoveryConfigLibraryClient,
|
|
28
29
|
TableReferenceClient,
|
|
30
|
+
RGConfigClient,
|
|
29
31
|
UserClient,
|
|
30
32
|
SettingsClient,
|
|
31
33
|
):
|
|
@@ -58,6 +58,14 @@ class DiscoveryConfigNotFoundError(FailedToStartError):
|
|
|
58
58
|
"""
|
|
59
59
|
|
|
60
60
|
|
|
61
|
+
class RGConfigNotFoundError(DataMasqueApiError):
|
|
62
|
+
"""
|
|
63
|
+
Raised when ruleset generation references an RG config that cannot be found.
|
|
64
|
+
|
|
65
|
+
The config does not exist, or has been deleted.
|
|
66
|
+
"""
|
|
67
|
+
|
|
68
|
+
|
|
61
69
|
class DataMasqueTransportError(DataMasqueException):
|
|
62
70
|
"""
|
|
63
71
|
Raised when a request to the DataMasque server fails before any response is received.
|
|
@@ -9,6 +9,7 @@ from datamasque.client.models.connection import ConnectionConfig, ConnectionId,
|
|
|
9
9
|
from datamasque.client.models.data_selection import HashColumnsTableConfig, Locator, UserSelection
|
|
10
10
|
from datamasque.client.models.discovery_config import DiscoveryConfig, DiscoveryConfigId, unwrap_discovery_config_id
|
|
11
11
|
from datamasque.client.models.pagination import Page
|
|
12
|
+
from datamasque.client.models.rg_config import RGConfig, RGConfigId, unwrap_rg_config_id
|
|
12
13
|
from datamasque.client.models.runs import RunConnectionRef
|
|
13
14
|
from datamasque.client.models.safe_data_preview import (
|
|
14
15
|
SafeDataPreview,
|
|
@@ -106,12 +107,13 @@ class SchemaDiscoveryFromConfigRequest(BaseModel):
|
|
|
106
107
|
|
|
107
108
|
class RulesetGenerationRequest(BaseModel):
|
|
108
109
|
"""
|
|
109
|
-
Request body for `POST /api/generate-ruleset/v2
|
|
110
|
+
Request body for `POST /api/generate-ruleset/v2/` (generation with the default RG config).
|
|
110
111
|
|
|
111
112
|
`connection` accepts either a `ConnectionId` or a full `ConnectionConfig` returned by an earlier client call.
|
|
112
113
|
`selected_columns` is the same nested `schema -> table -> [column, ...]` mapping
|
|
113
114
|
used by `SelectedColumns.columns`,
|
|
114
115
|
and `hash_columns` follows the `HashColumnsTableConfig` shape.
|
|
116
|
+
This request does not accept an `rg_config`.
|
|
115
117
|
"""
|
|
116
118
|
|
|
117
119
|
model_config = ConfigDict(extra="forbid")
|
|
@@ -125,6 +127,45 @@ class RulesetGenerationRequest(BaseModel):
|
|
|
125
127
|
def _unwrap_connection(cls, value: Any) -> Any:
|
|
126
128
|
return unwrap_connection_id(value)
|
|
127
129
|
|
|
130
|
+
@model_validator(mode="before")
|
|
131
|
+
@classmethod
|
|
132
|
+
def _reject_rg_config(cls, data: Any) -> Any:
|
|
133
|
+
if isinstance(data, dict) and "rg_config" in data:
|
|
134
|
+
raise ValueError(
|
|
135
|
+
"`rg_config` is not accepted by `RulesetGenerationRequest`; "
|
|
136
|
+
"use `generate_ruleset_with_rg_config` with a `RulesetGenerationWithRGConfigRequest` "
|
|
137
|
+
"to generate with a selected RG config."
|
|
138
|
+
)
|
|
139
|
+
return data
|
|
140
|
+
|
|
141
|
+
|
|
142
|
+
class RulesetGenerationWithRGConfigRequest(BaseModel):
|
|
143
|
+
"""
|
|
144
|
+
Request body for `POST /api/generate-ruleset/v3/` (generation with a selected RG config).
|
|
145
|
+
|
|
146
|
+
`connection` accepts either a `ConnectionId` or a full `ConnectionConfig` returned by an earlier client call.
|
|
147
|
+
`selected_columns` and `hash_columns` follow the same shapes as `RulesetGenerationRequest`.
|
|
148
|
+
`rg_config` is required: pass an `RGConfigId`, a full `RGConfig`,
|
|
149
|
+
or `None` to generate with the server's default RG config.
|
|
150
|
+
"""
|
|
151
|
+
|
|
152
|
+
model_config = ConfigDict(extra="forbid")
|
|
153
|
+
|
|
154
|
+
connection: Union[ConnectionId, ConnectionConfig]
|
|
155
|
+
selected_columns: dict[str, dict[str, list[str]]]
|
|
156
|
+
rg_config: Optional[Union[RGConfigId, RGConfig]]
|
|
157
|
+
hash_columns: Optional[dict[str, dict[str, HashColumnsTableConfig]]] = None
|
|
158
|
+
|
|
159
|
+
@field_validator("connection", mode="before")
|
|
160
|
+
@classmethod
|
|
161
|
+
def _unwrap_connection(cls, value: Any) -> Any:
|
|
162
|
+
return unwrap_connection_id(value)
|
|
163
|
+
|
|
164
|
+
@field_validator("rg_config", mode="before")
|
|
165
|
+
@classmethod
|
|
166
|
+
def _unwrap_rg_config(cls, value: Any) -> Any:
|
|
167
|
+
return unwrap_rg_config_id(value)
|
|
168
|
+
|
|
128
169
|
|
|
129
170
|
class FileFilterMatchAgainst(Enum):
|
|
130
171
|
"""Which part of a file's path an `include`/`skip` filter is matched against."""
|
|
@@ -236,9 +277,10 @@ class FileDataDiscoveryFromConfigRequest(BaseModel):
|
|
|
236
277
|
|
|
237
278
|
class FileRulesetGenerationRequest(BaseModel):
|
|
238
279
|
"""
|
|
239
|
-
Request body for `POST /api/generate-file-ruleset
|
|
280
|
+
Request body for `POST /api/generate-file-ruleset/` (generation with the default RG config).
|
|
240
281
|
|
|
241
282
|
`connection` accepts either a `ConnectionId` or a full `ConnectionConfig` returned by an earlier client call.
|
|
283
|
+
This request does not accept an `rg_config`.
|
|
242
284
|
"""
|
|
243
285
|
|
|
244
286
|
model_config = ConfigDict(extra="forbid")
|
|
@@ -251,6 +293,43 @@ class FileRulesetGenerationRequest(BaseModel):
|
|
|
251
293
|
def _unwrap_connection(cls, value: Any) -> Any:
|
|
252
294
|
return unwrap_connection_id(value)
|
|
253
295
|
|
|
296
|
+
@model_validator(mode="before")
|
|
297
|
+
@classmethod
|
|
298
|
+
def _reject_rg_config(cls, data: Any) -> Any:
|
|
299
|
+
if isinstance(data, dict) and "rg_config" in data:
|
|
300
|
+
raise ValueError(
|
|
301
|
+
"`rg_config` is not accepted by `FileRulesetGenerationRequest`; "
|
|
302
|
+
"use `generate_file_ruleset_with_rg_config` with a `FileRulesetGenerationWithRGConfigRequest` "
|
|
303
|
+
"to generate with a selected RG config."
|
|
304
|
+
)
|
|
305
|
+
return data
|
|
306
|
+
|
|
307
|
+
|
|
308
|
+
class FileRulesetGenerationWithRGConfigRequest(BaseModel):
|
|
309
|
+
"""
|
|
310
|
+
Request body for `POST /api/generate-file-ruleset/v2/` (generation with a selected RG config).
|
|
311
|
+
|
|
312
|
+
`connection` accepts either a `ConnectionId` or a full `ConnectionConfig` returned by an earlier client call.
|
|
313
|
+
`rg_config` is required: pass an `RGConfigId`, a full `RGConfig`,
|
|
314
|
+
or `None` to generate with the server's default RG config.
|
|
315
|
+
"""
|
|
316
|
+
|
|
317
|
+
model_config = ConfigDict(extra="forbid")
|
|
318
|
+
|
|
319
|
+
connection: Union[ConnectionId, ConnectionConfig]
|
|
320
|
+
selected_data: list[UserSelection]
|
|
321
|
+
rg_config: Optional[Union[RGConfigId, RGConfig]]
|
|
322
|
+
|
|
323
|
+
@field_validator("connection", mode="before")
|
|
324
|
+
@classmethod
|
|
325
|
+
def _unwrap_connection(cls, value: Any) -> Any:
|
|
326
|
+
return unwrap_connection_id(value)
|
|
327
|
+
|
|
328
|
+
@field_validator("rg_config", mode="before")
|
|
329
|
+
@classmethod
|
|
330
|
+
def _unwrap_rg_config(cls, value: Any) -> Any:
|
|
331
|
+
return unwrap_rg_config_id(value)
|
|
332
|
+
|
|
254
333
|
|
|
255
334
|
class DiscoveryMatch(BaseModel):
|
|
256
335
|
"""A single match found by schema or file discovery."""
|
|
@@ -0,0 +1,49 @@
|
|
|
1
|
+
from datetime import datetime
|
|
2
|
+
from typing import Any, NewType, Optional
|
|
3
|
+
|
|
4
|
+
from pydantic import BaseModel, ConfigDict, Field
|
|
5
|
+
|
|
6
|
+
from datamasque.client.models.status import ValidationErrorDetails, ValidationStatus
|
|
7
|
+
|
|
8
|
+
RGConfigId = NewType("RGConfigId", str)
|
|
9
|
+
|
|
10
|
+
|
|
11
|
+
def unwrap_rg_config_id(value: Any) -> Any:
|
|
12
|
+
"""
|
|
13
|
+
Coerce an `RGConfig` to its `id`; pass other values through unchanged.
|
|
14
|
+
|
|
15
|
+
Used by request-model validators and the generate methods
|
|
16
|
+
that accept either an `RGConfigId` or a full `RGConfig` for user convenience.
|
|
17
|
+
Raises `ValueError` if the config has no `id`
|
|
18
|
+
(i.e. the caller hasn't yet created it on the server).
|
|
19
|
+
"""
|
|
20
|
+
|
|
21
|
+
if isinstance(value, RGConfig):
|
|
22
|
+
if value.id is None:
|
|
23
|
+
raise ValueError("RG config has not been created yet (id is None)")
|
|
24
|
+
return value.id
|
|
25
|
+
|
|
26
|
+
return value
|
|
27
|
+
|
|
28
|
+
|
|
29
|
+
class RGConfig(BaseModel):
|
|
30
|
+
"""
|
|
31
|
+
Represents a named, persisted YAML ruleset-generation (RG) configuration.
|
|
32
|
+
|
|
33
|
+
An RG config maps discovered labels to masks;
|
|
34
|
+
ruleset generation applies it to a discovery run's results.
|
|
35
|
+
Unlike discovery configs, RG configs are untyped —
|
|
36
|
+
the same config serves both database and file ruleset generation.
|
|
37
|
+
"""
|
|
38
|
+
|
|
39
|
+
model_config = ConfigDict(extra="allow", populate_by_name=True)
|
|
40
|
+
|
|
41
|
+
name: str
|
|
42
|
+
yaml: Optional[str] = Field(default=None, alias="config_yaml")
|
|
43
|
+
|
|
44
|
+
# Server-populated read-only fields, excluded from request bodies.
|
|
45
|
+
id: Optional[RGConfigId] = Field(default=None, exclude=True)
|
|
46
|
+
is_valid: Optional[ValidationStatus] = Field(default=None, exclude=True)
|
|
47
|
+
validation_errors: list[ValidationErrorDetails] = Field(default_factory=list, exclude=True)
|
|
48
|
+
created: Optional[datetime] = Field(default=None, exclude=True)
|
|
49
|
+
modified: Optional[datetime] = Field(default=None, exclude=True)
|