openparser-sdk 0.0.1__py3-none-any.whl
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.
- openparser/__init__.py +54 -0
- openparser/_files.py +43 -0
- openparser/_generated/__init__.py +8 -0
- openparser/_generated/api/__init__.py +1 -0
- openparser/_generated/api/extract/__init__.py +1 -0
- openparser/_generated/api/extract/extract_async.py +280 -0
- openparser/_generated/api/extract/extract_batch.py +284 -0
- openparser/_generated/api/extract/extract_sync.py +315 -0
- openparser/_generated/api/extract/suggest_schema.py +260 -0
- openparser/_generated/api/files/__init__.py +1 -0
- openparser/_generated/api/files/create_file.py +237 -0
- openparser/_generated/api/files/delete_file.py +200 -0
- openparser/_generated/api/files/get_file.py +200 -0
- openparser/_generated/api/files/get_file_content.py +198 -0
- openparser/_generated/api/jobs/__init__.py +1 -0
- openparser/_generated/api/jobs/get_job.py +261 -0
- openparser/_generated/api/jobs/get_job_result.py +305 -0
- openparser/_generated/api/jobs/get_job_source.py +233 -0
- openparser/_generated/api/jobs/list_jobs.py +340 -0
- openparser/_generated/api/models/__init__.py +1 -0
- openparser/_generated/api/models/list_llm_models.py +288 -0
- openparser/_generated/api/models/list_ocr_models.py +184 -0
- openparser/_generated/api/parse/__init__.py +1 -0
- openparser/_generated/api/parse/parse_async.py +316 -0
- openparser/_generated/api/parse/parse_batch.py +280 -0
- openparser/_generated/api/parse/parse_sync.py +369 -0
- openparser/_generated/api/pipelines/__init__.py +1 -0
- openparser/_generated/api/pipelines/create_extraction_pipeline.py +242 -0
- openparser/_generated/api/pipelines/delete_extraction_pipeline.py +212 -0
- openparser/_generated/api/pipelines/get_extraction_pipeline.py +212 -0
- openparser/_generated/api/pipelines/list_extraction_pipelines.py +180 -0
- openparser/_generated/api/pipelines/update_extraction_pipeline.py +258 -0
- openparser/_generated/client.py +270 -0
- openparser/_generated/errors.py +14 -0
- openparser/_generated/models/__init__.py +261 -0
- openparser/_generated/models/batch_child_page.py +97 -0
- openparser/_generated/models/batch_child_summary.py +195 -0
- openparser/_generated/models/batch_job_accepted.py +139 -0
- openparser/_generated/models/batch_job_accepted_operation.py +8 -0
- openparser/_generated/models/batch_summary_counts.py +97 -0
- openparser/_generated/models/bounding_box.py +82 -0
- openparser/_generated/models/chunk_provenance_span.py +97 -0
- openparser/_generated/models/content_kind.py +10 -0
- openparser/_generated/models/create_extraction_pipeline_request.py +177 -0
- openparser/_generated/models/create_extraction_pipeline_request_grounding.py +8 -0
- openparser/_generated/models/create_extraction_pipeline_request_llm_options.py +84 -0
- openparser/_generated/models/create_extraction_pipeline_request_llm_options_reasoning_effort_type_1.py +13 -0
- openparser/_generated/models/create_extraction_pipeline_request_ocr_options.py +78 -0
- openparser/_generated/models/create_extraction_pipeline_request_schema.py +68 -0
- openparser/_generated/models/create_file_body.py +78 -0
- openparser/_generated/models/delete_extraction_pipeline_response.py +66 -0
- openparser/_generated/models/delete_file_response.py +57 -0
- openparser/_generated/models/error_body.py +106 -0
- openparser/_generated/models/error_body_details.py +65 -0
- openparser/_generated/models/error_response.py +65 -0
- openparser/_generated/models/extract_async_body.py +136 -0
- openparser/_generated/models/extract_batch_body.py +143 -0
- openparser/_generated/models/extract_batch_item.py +207 -0
- openparser/_generated/models/extract_batch_item_grounding.py +8 -0
- openparser/_generated/models/extract_batch_item_llm_options.py +84 -0
- openparser/_generated/models/extract_batch_item_llm_options_reasoning_effort_type_1.py +13 -0
- openparser/_generated/models/extract_batch_item_ocr_options.py +78 -0
- openparser/_generated/models/extract_batch_item_schema.py +65 -0
- openparser/_generated/models/extract_batch_request.py +99 -0
- openparser/_generated/models/extract_batch_request_items_item.py +201 -0
- openparser/_generated/models/extract_batch_request_items_item_grounding.py +8 -0
- openparser/_generated/models/extract_batch_request_items_item_llm_options.py +84 -0
- openparser/_generated/models/extract_batch_request_items_item_llm_options_reasoning_effort_type_1.py +13 -0
- openparser/_generated/models/extract_batch_request_items_item_ocr_options.py +78 -0
- openparser/_generated/models/extract_batch_request_items_item_schema.py +65 -0
- openparser/_generated/models/extract_batch_request_output_format.py +8 -0
- openparser/_generated/models/extract_request.py +226 -0
- openparser/_generated/models/extract_request_grounding.py +8 -0
- openparser/_generated/models/extract_request_llm_options.py +84 -0
- openparser/_generated/models/extract_request_llm_options_reasoning_effort_type_1.py +13 -0
- openparser/_generated/models/extract_request_ocr_options.py +78 -0
- openparser/_generated/models/extract_request_output_format.py +8 -0
- openparser/_generated/models/extract_request_schema.py +65 -0
- openparser/_generated/models/extract_sync_body.py +136 -0
- openparser/_generated/models/extraction_attempt.py +122 -0
- openparser/_generated/models/extraction_attempt_kind.py +8 -0
- openparser/_generated/models/extraction_attempt_status.py +9 -0
- openparser/_generated/models/extraction_chunk.py +184 -0
- openparser/_generated/models/extraction_citation.py +169 -0
- openparser/_generated/models/extraction_citation_granularity.py +8 -0
- openparser/_generated/models/extraction_citation_source_type.py +16 -0
- openparser/_generated/models/extraction_grounding_field.py +102 -0
- openparser/_generated/models/extraction_grounding_mode.py +8 -0
- openparser/_generated/models/extraction_grounding_result.py +87 -0
- openparser/_generated/models/extraction_pipeline.py +201 -0
- openparser/_generated/models/extraction_pipeline_list_response.py +76 -0
- openparser/_generated/models/extraction_pipeline_llm_options.py +84 -0
- openparser/_generated/models/extraction_pipeline_llm_options_reasoning_effort_type_0_type_1.py +13 -0
- openparser/_generated/models/extraction_pipeline_ocr_options.py +73 -0
- openparser/_generated/models/extraction_pipeline_schema.py +67 -0
- openparser/_generated/models/extraction_terminal_result.py +206 -0
- openparser/_generated/models/extraction_terminal_result_reasoning_effort_type_0.py +13 -0
- openparser/_generated/models/extraction_usage_totals.py +80 -0
- openparser/_generated/models/job.py +437 -0
- openparser/_generated/models/job_accepted.py +131 -0
- openparser/_generated/models/job_accepted_operation.py +8 -0
- openparser/_generated/models/job_extraction_schema.py +65 -0
- openparser/_generated/models/job_failure.py +98 -0
- openparser/_generated/models/job_failure_details.py +65 -0
- openparser/_generated/models/job_list_response.py +91 -0
- openparser/_generated/models/job_operation.py +10 -0
- openparser/_generated/models/job_progress.py +68 -0
- openparser/_generated/models/job_related_extractions_item.py +145 -0
- openparser/_generated/models/job_status.py +11 -0
- openparser/_generated/models/job_summary.py +264 -0
- openparser/_generated/models/json_schema_object.py +66 -0
- openparser/_generated/models/list_llm_models_mode.py +8 -0
- openparser/_generated/models/ocr_llm_model_catalog_entry.py +257 -0
- openparser/_generated/models/ocr_llm_model_catalog_entry_pricing.py +76 -0
- openparser/_generated/models/ocr_llm_model_catalog_entry_reasoning_type_0.py +134 -0
- openparser/_generated/models/ocr_llm_model_catalog_entry_reasoning_type_0_default_effort_type_0.py +13 -0
- openparser/_generated/models/ocr_llm_model_catalog_entry_reasoning_type_0_supported_efforts_type_0_item.py +13 -0
- openparser/_generated/models/ocr_llm_model_catalog_entry_recommendation.py +8 -0
- openparser/_generated/models/ocr_llm_models_response.py +150 -0
- openparser/_generated/models/ocr_llm_models_response_mode.py +8 -0
- openparser/_generated/models/ocr_model.py +7 -0
- openparser/_generated/models/ocr_model_catalog_entry.py +131 -0
- openparser/_generated/models/ocr_model_catalog_entry_availability.py +9 -0
- openparser/_generated/models/ocr_model_catalog_entry_capabilities.py +97 -0
- openparser/_generated/models/ocr_model_catalog_entry_capabilities_options.py +73 -0
- openparser/_generated/models/ocr_model_catalog_entry_option_defaults.py +73 -0
- openparser/_generated/models/ocr_model_catalog_entry_pricing.py +68 -0
- openparser/_generated/models/ocr_models_response.py +76 -0
- openparser/_generated/models/ocr_output_format.py +8 -0
- openparser/_generated/models/paddle_raw_profile.py +76 -0
- openparser/_generated/models/paddle_raw_profile_options.py +115 -0
- openparser/_generated/models/page_block.py +332 -0
- openparser/_generated/models/page_block_kind.py +9 -0
- openparser/_generated/models/page_block_polygon_type_0_item.py +65 -0
- openparser/_generated/models/parse_async_body.py +133 -0
- openparser/_generated/models/parse_batch_body.py +143 -0
- openparser/_generated/models/parse_batch_item.py +109 -0
- openparser/_generated/models/parse_batch_item_ocr_options.py +78 -0
- openparser/_generated/models/parse_batch_request.py +100 -0
- openparser/_generated/models/parse_batch_request_items_item.py +108 -0
- openparser/_generated/models/parse_batch_request_items_item_ocr_options.py +78 -0
- openparser/_generated/models/parse_batch_request_output_format.py +8 -0
- openparser/_generated/models/parse_request.py +112 -0
- openparser/_generated/models/parse_request_ocr_options.py +78 -0
- openparser/_generated/models/parse_request_output_format.py +8 -0
- openparser/_generated/models/parse_sync_body.py +133 -0
- openparser/_generated/models/parsed_document.py +187 -0
- openparser/_generated/models/public_file.py +117 -0
- openparser/_generated/models/raw_parse_result.py +110 -0
- openparser/_generated/models/raw_parse_result_result.py +65 -0
- openparser/_generated/models/region.py +224 -0
- openparser/_generated/models/region_content.py +155 -0
- openparser/_generated/models/region_polygon_type_0_item.py +65 -0
- openparser/_generated/models/region_type.py +16 -0
- openparser/_generated/models/suggest_schema_request.py +69 -0
- openparser/_generated/models/suggest_schema_response.py +74 -0
- openparser/_generated/models/suggest_schema_response_schema.py +66 -0
- openparser/_generated/models/update_extraction_pipeline_request.py +227 -0
- openparser/_generated/models/update_extraction_pipeline_request_grounding.py +8 -0
- openparser/_generated/models/update_extraction_pipeline_request_llm_options_type_0.py +84 -0
- openparser/_generated/models/update_extraction_pipeline_request_llm_options_type_0_reasoning_effort_type_1.py +13 -0
- openparser/_generated/models/update_extraction_pipeline_request_ocr_options_type_0.py +78 -0
- openparser/_generated/models/update_extraction_pipeline_request_schema.py +67 -0
- openparser/_generated/py.typed +1 -0
- openparser/_generated/types.py +53 -0
- openparser/_telemetry.py +34 -0
- openparser/client.py +541 -0
- openparser/errors.py +172 -0
- openparser/py.typed +0 -0
- openparser_sdk-0.0.1.dist-info/METADATA +178 -0
- openparser_sdk-0.0.1.dist-info/RECORD +173 -0
- openparser_sdk-0.0.1.dist-info/WHEEL +4 -0
- openparser_sdk-0.0.1.dist-info/licenses/LICENSE +201 -0
|
@@ -0,0 +1,280 @@
|
|
|
1
|
+
from http import HTTPStatus
|
|
2
|
+
from typing import Any, cast
|
|
3
|
+
from urllib.parse import quote
|
|
4
|
+
|
|
5
|
+
import httpx
|
|
6
|
+
|
|
7
|
+
from ...client import AuthenticatedClient, Client
|
|
8
|
+
from ...types import Response, UNSET
|
|
9
|
+
from ... import errors
|
|
10
|
+
|
|
11
|
+
from ...models.batch_job_accepted import BatchJobAccepted
|
|
12
|
+
from ...models.error_response import ErrorResponse
|
|
13
|
+
from ...models.parse_batch_body import ParseBatchBody
|
|
14
|
+
from typing import cast
|
|
15
|
+
|
|
16
|
+
|
|
17
|
+
|
|
18
|
+
def _get_kwargs(
|
|
19
|
+
*,
|
|
20
|
+
body: ParseBatchBody,
|
|
21
|
+
idempotency_key: str,
|
|
22
|
+
|
|
23
|
+
) -> dict[str, Any]:
|
|
24
|
+
headers: dict[str, Any] = {}
|
|
25
|
+
headers["Idempotency-Key"] = idempotency_key
|
|
26
|
+
|
|
27
|
+
|
|
28
|
+
|
|
29
|
+
|
|
30
|
+
|
|
31
|
+
|
|
32
|
+
|
|
33
|
+
_kwargs: dict[str, Any] = {
|
|
34
|
+
"method": "post",
|
|
35
|
+
"url": "/parse/batch",
|
|
36
|
+
}
|
|
37
|
+
|
|
38
|
+
_kwargs["files"] = body.to_multipart()
|
|
39
|
+
|
|
40
|
+
|
|
41
|
+
|
|
42
|
+
_kwargs["headers"] = headers
|
|
43
|
+
return _kwargs
|
|
44
|
+
|
|
45
|
+
|
|
46
|
+
|
|
47
|
+
def _parse_response(*, client: AuthenticatedClient | Client, response: httpx.Response) -> BatchJobAccepted | ErrorResponse | None:
|
|
48
|
+
if response.status_code == 202:
|
|
49
|
+
response_202 = BatchJobAccepted.from_dict(response.json())
|
|
50
|
+
|
|
51
|
+
|
|
52
|
+
|
|
53
|
+
return response_202
|
|
54
|
+
|
|
55
|
+
if response.status_code == 400:
|
|
56
|
+
response_400 = ErrorResponse.from_dict(response.json())
|
|
57
|
+
|
|
58
|
+
|
|
59
|
+
|
|
60
|
+
return response_400
|
|
61
|
+
|
|
62
|
+
if response.status_code == 401:
|
|
63
|
+
response_401 = ErrorResponse.from_dict(response.json())
|
|
64
|
+
|
|
65
|
+
|
|
66
|
+
|
|
67
|
+
return response_401
|
|
68
|
+
|
|
69
|
+
if response.status_code == 402:
|
|
70
|
+
response_402 = ErrorResponse.from_dict(response.json())
|
|
71
|
+
|
|
72
|
+
|
|
73
|
+
|
|
74
|
+
return response_402
|
|
75
|
+
|
|
76
|
+
if response.status_code == 403:
|
|
77
|
+
response_403 = ErrorResponse.from_dict(response.json())
|
|
78
|
+
|
|
79
|
+
|
|
80
|
+
|
|
81
|
+
return response_403
|
|
82
|
+
|
|
83
|
+
if response.status_code == 409:
|
|
84
|
+
response_409 = ErrorResponse.from_dict(response.json())
|
|
85
|
+
|
|
86
|
+
|
|
87
|
+
|
|
88
|
+
return response_409
|
|
89
|
+
|
|
90
|
+
if response.status_code == 413:
|
|
91
|
+
response_413 = ErrorResponse.from_dict(response.json())
|
|
92
|
+
|
|
93
|
+
|
|
94
|
+
|
|
95
|
+
return response_413
|
|
96
|
+
|
|
97
|
+
if response.status_code == 415:
|
|
98
|
+
response_415 = ErrorResponse.from_dict(response.json())
|
|
99
|
+
|
|
100
|
+
|
|
101
|
+
|
|
102
|
+
return response_415
|
|
103
|
+
|
|
104
|
+
if response.status_code == 422:
|
|
105
|
+
response_422 = ErrorResponse.from_dict(response.json())
|
|
106
|
+
|
|
107
|
+
|
|
108
|
+
|
|
109
|
+
return response_422
|
|
110
|
+
|
|
111
|
+
if response.status_code == 429:
|
|
112
|
+
response_429 = ErrorResponse.from_dict(response.json())
|
|
113
|
+
|
|
114
|
+
|
|
115
|
+
|
|
116
|
+
return response_429
|
|
117
|
+
|
|
118
|
+
if response.status_code == 503:
|
|
119
|
+
response_503 = ErrorResponse.from_dict(response.json())
|
|
120
|
+
|
|
121
|
+
|
|
122
|
+
|
|
123
|
+
return response_503
|
|
124
|
+
|
|
125
|
+
if client.raise_on_unexpected_status:
|
|
126
|
+
raise errors.UnexpectedStatus(response.status_code, response.content)
|
|
127
|
+
else:
|
|
128
|
+
return None
|
|
129
|
+
|
|
130
|
+
|
|
131
|
+
def _build_response(*, client: AuthenticatedClient | Client, response: httpx.Response) -> Response[BatchJobAccepted | ErrorResponse]:
|
|
132
|
+
return Response(
|
|
133
|
+
status_code=HTTPStatus(response.status_code),
|
|
134
|
+
content=response.content,
|
|
135
|
+
headers=response.headers,
|
|
136
|
+
parsed=_parse_response(client=client, response=response),
|
|
137
|
+
)
|
|
138
|
+
|
|
139
|
+
|
|
140
|
+
def sync_detailed(
|
|
141
|
+
*,
|
|
142
|
+
client: AuthenticatedClient | Client,
|
|
143
|
+
body: ParseBatchBody,
|
|
144
|
+
idempotency_key: str,
|
|
145
|
+
|
|
146
|
+
) -> Response[BatchJobAccepted | ErrorResponse]:
|
|
147
|
+
""" Parse multiple documents in one batch
|
|
148
|
+
|
|
149
|
+
Admit up to 100 parse child jobs in one durable batch. Every uploaded file must be mapped
|
|
150
|
+
exactly once by `file_index`. Partial child failures are preserved in batch results.
|
|
151
|
+
Parent status derivation for mixed outcomes is defined on `JobStatus`.
|
|
152
|
+
|
|
153
|
+
Args:
|
|
154
|
+
idempotency_key (str):
|
|
155
|
+
body (ParseBatchBody):
|
|
156
|
+
|
|
157
|
+
Raises:
|
|
158
|
+
errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
|
|
159
|
+
httpx.TimeoutException: If the request takes longer than Client.timeout.
|
|
160
|
+
|
|
161
|
+
Returns:
|
|
162
|
+
Response[BatchJobAccepted | ErrorResponse]
|
|
163
|
+
"""
|
|
164
|
+
|
|
165
|
+
|
|
166
|
+
kwargs = _get_kwargs(
|
|
167
|
+
body=body,
|
|
168
|
+
idempotency_key=idempotency_key,
|
|
169
|
+
|
|
170
|
+
)
|
|
171
|
+
|
|
172
|
+
response = client.get_httpx_client().request(
|
|
173
|
+
**kwargs,
|
|
174
|
+
)
|
|
175
|
+
|
|
176
|
+
return _build_response(client=client, response=response)
|
|
177
|
+
|
|
178
|
+
def sync(
|
|
179
|
+
*,
|
|
180
|
+
client: AuthenticatedClient | Client,
|
|
181
|
+
body: ParseBatchBody,
|
|
182
|
+
idempotency_key: str,
|
|
183
|
+
|
|
184
|
+
) -> BatchJobAccepted | ErrorResponse | None:
|
|
185
|
+
""" Parse multiple documents in one batch
|
|
186
|
+
|
|
187
|
+
Admit up to 100 parse child jobs in one durable batch. Every uploaded file must be mapped
|
|
188
|
+
exactly once by `file_index`. Partial child failures are preserved in batch results.
|
|
189
|
+
Parent status derivation for mixed outcomes is defined on `JobStatus`.
|
|
190
|
+
|
|
191
|
+
Args:
|
|
192
|
+
idempotency_key (str):
|
|
193
|
+
body (ParseBatchBody):
|
|
194
|
+
|
|
195
|
+
Raises:
|
|
196
|
+
errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
|
|
197
|
+
httpx.TimeoutException: If the request takes longer than Client.timeout.
|
|
198
|
+
|
|
199
|
+
Returns:
|
|
200
|
+
BatchJobAccepted | ErrorResponse
|
|
201
|
+
"""
|
|
202
|
+
|
|
203
|
+
|
|
204
|
+
return sync_detailed(
|
|
205
|
+
client=client,
|
|
206
|
+
body=body,
|
|
207
|
+
idempotency_key=idempotency_key,
|
|
208
|
+
|
|
209
|
+
).parsed
|
|
210
|
+
|
|
211
|
+
async def asyncio_detailed(
|
|
212
|
+
*,
|
|
213
|
+
client: AuthenticatedClient | Client,
|
|
214
|
+
body: ParseBatchBody,
|
|
215
|
+
idempotency_key: str,
|
|
216
|
+
|
|
217
|
+
) -> Response[BatchJobAccepted | ErrorResponse]:
|
|
218
|
+
""" Parse multiple documents in one batch
|
|
219
|
+
|
|
220
|
+
Admit up to 100 parse child jobs in one durable batch. Every uploaded file must be mapped
|
|
221
|
+
exactly once by `file_index`. Partial child failures are preserved in batch results.
|
|
222
|
+
Parent status derivation for mixed outcomes is defined on `JobStatus`.
|
|
223
|
+
|
|
224
|
+
Args:
|
|
225
|
+
idempotency_key (str):
|
|
226
|
+
body (ParseBatchBody):
|
|
227
|
+
|
|
228
|
+
Raises:
|
|
229
|
+
errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
|
|
230
|
+
httpx.TimeoutException: If the request takes longer than Client.timeout.
|
|
231
|
+
|
|
232
|
+
Returns:
|
|
233
|
+
Response[BatchJobAccepted | ErrorResponse]
|
|
234
|
+
"""
|
|
235
|
+
|
|
236
|
+
|
|
237
|
+
kwargs = _get_kwargs(
|
|
238
|
+
body=body,
|
|
239
|
+
idempotency_key=idempotency_key,
|
|
240
|
+
|
|
241
|
+
)
|
|
242
|
+
|
|
243
|
+
response = await client.get_async_httpx_client().request(
|
|
244
|
+
**kwargs
|
|
245
|
+
)
|
|
246
|
+
|
|
247
|
+
return _build_response(client=client, response=response)
|
|
248
|
+
|
|
249
|
+
async def asyncio(
|
|
250
|
+
*,
|
|
251
|
+
client: AuthenticatedClient | Client,
|
|
252
|
+
body: ParseBatchBody,
|
|
253
|
+
idempotency_key: str,
|
|
254
|
+
|
|
255
|
+
) -> BatchJobAccepted | ErrorResponse | None:
|
|
256
|
+
""" Parse multiple documents in one batch
|
|
257
|
+
|
|
258
|
+
Admit up to 100 parse child jobs in one durable batch. Every uploaded file must be mapped
|
|
259
|
+
exactly once by `file_index`. Partial child failures are preserved in batch results.
|
|
260
|
+
Parent status derivation for mixed outcomes is defined on `JobStatus`.
|
|
261
|
+
|
|
262
|
+
Args:
|
|
263
|
+
idempotency_key (str):
|
|
264
|
+
body (ParseBatchBody):
|
|
265
|
+
|
|
266
|
+
Raises:
|
|
267
|
+
errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
|
|
268
|
+
httpx.TimeoutException: If the request takes longer than Client.timeout.
|
|
269
|
+
|
|
270
|
+
Returns:
|
|
271
|
+
BatchJobAccepted | ErrorResponse
|
|
272
|
+
"""
|
|
273
|
+
|
|
274
|
+
|
|
275
|
+
return (await asyncio_detailed(
|
|
276
|
+
client=client,
|
|
277
|
+
body=body,
|
|
278
|
+
idempotency_key=idempotency_key,
|
|
279
|
+
|
|
280
|
+
)).parsed
|
|
@@ -0,0 +1,369 @@
|
|
|
1
|
+
from http import HTTPStatus
|
|
2
|
+
from typing import Any, cast
|
|
3
|
+
from urllib.parse import quote
|
|
4
|
+
|
|
5
|
+
import httpx
|
|
6
|
+
|
|
7
|
+
from ...client import AuthenticatedClient, Client
|
|
8
|
+
from ...types import Response, UNSET
|
|
9
|
+
from ... import errors
|
|
10
|
+
|
|
11
|
+
from ...models.error_response import ErrorResponse
|
|
12
|
+
from ...models.job_accepted import JobAccepted
|
|
13
|
+
from ...models.parse_sync_body import ParseSyncBody
|
|
14
|
+
from ...models.parsed_document import ParsedDocument
|
|
15
|
+
from ...models.raw_parse_result import RawParseResult
|
|
16
|
+
from typing import cast
|
|
17
|
+
|
|
18
|
+
|
|
19
|
+
|
|
20
|
+
def _get_kwargs(
|
|
21
|
+
*,
|
|
22
|
+
body: ParseSyncBody,
|
|
23
|
+
idempotency_key: str,
|
|
24
|
+
|
|
25
|
+
) -> dict[str, Any]:
|
|
26
|
+
headers: dict[str, Any] = {}
|
|
27
|
+
headers["Idempotency-Key"] = idempotency_key
|
|
28
|
+
|
|
29
|
+
|
|
30
|
+
|
|
31
|
+
|
|
32
|
+
|
|
33
|
+
|
|
34
|
+
|
|
35
|
+
_kwargs: dict[str, Any] = {
|
|
36
|
+
"method": "post",
|
|
37
|
+
"url": "/parse",
|
|
38
|
+
}
|
|
39
|
+
|
|
40
|
+
_kwargs["files"] = body.to_multipart()
|
|
41
|
+
|
|
42
|
+
|
|
43
|
+
|
|
44
|
+
_kwargs["headers"] = headers
|
|
45
|
+
return _kwargs
|
|
46
|
+
|
|
47
|
+
|
|
48
|
+
|
|
49
|
+
def _parse_response(*, client: AuthenticatedClient | Client, response: httpx.Response) -> ErrorResponse | JobAccepted | ParsedDocument | RawParseResult | None:
|
|
50
|
+
if response.status_code == 200:
|
|
51
|
+
def _parse_response_200(data: object) -> ParsedDocument | RawParseResult:
|
|
52
|
+
try:
|
|
53
|
+
if not isinstance(data, dict):
|
|
54
|
+
raise TypeError()
|
|
55
|
+
componentsschemas_parse_result_type_0 = ParsedDocument.from_dict(data)
|
|
56
|
+
|
|
57
|
+
|
|
58
|
+
|
|
59
|
+
return componentsschemas_parse_result_type_0
|
|
60
|
+
except (TypeError, ValueError, AttributeError, KeyError):
|
|
61
|
+
pass
|
|
62
|
+
if not isinstance(data, dict):
|
|
63
|
+
raise TypeError()
|
|
64
|
+
componentsschemas_parse_result_type_1 = RawParseResult.from_dict(data)
|
|
65
|
+
|
|
66
|
+
|
|
67
|
+
|
|
68
|
+
return componentsschemas_parse_result_type_1
|
|
69
|
+
|
|
70
|
+
response_200 = _parse_response_200(response.json())
|
|
71
|
+
|
|
72
|
+
return response_200
|
|
73
|
+
|
|
74
|
+
if response.status_code == 202:
|
|
75
|
+
response_202 = JobAccepted.from_dict(response.json())
|
|
76
|
+
|
|
77
|
+
|
|
78
|
+
|
|
79
|
+
return response_202
|
|
80
|
+
|
|
81
|
+
if response.status_code == 400:
|
|
82
|
+
response_400 = ErrorResponse.from_dict(response.json())
|
|
83
|
+
|
|
84
|
+
|
|
85
|
+
|
|
86
|
+
return response_400
|
|
87
|
+
|
|
88
|
+
if response.status_code == 401:
|
|
89
|
+
response_401 = ErrorResponse.from_dict(response.json())
|
|
90
|
+
|
|
91
|
+
|
|
92
|
+
|
|
93
|
+
return response_401
|
|
94
|
+
|
|
95
|
+
if response.status_code == 402:
|
|
96
|
+
response_402 = ErrorResponse.from_dict(response.json())
|
|
97
|
+
|
|
98
|
+
|
|
99
|
+
|
|
100
|
+
return response_402
|
|
101
|
+
|
|
102
|
+
if response.status_code == 403:
|
|
103
|
+
response_403 = ErrorResponse.from_dict(response.json())
|
|
104
|
+
|
|
105
|
+
|
|
106
|
+
|
|
107
|
+
return response_403
|
|
108
|
+
|
|
109
|
+
if response.status_code == 409:
|
|
110
|
+
response_409 = ErrorResponse.from_dict(response.json())
|
|
111
|
+
|
|
112
|
+
|
|
113
|
+
|
|
114
|
+
return response_409
|
|
115
|
+
|
|
116
|
+
if response.status_code == 413:
|
|
117
|
+
response_413 = ErrorResponse.from_dict(response.json())
|
|
118
|
+
|
|
119
|
+
|
|
120
|
+
|
|
121
|
+
return response_413
|
|
122
|
+
|
|
123
|
+
if response.status_code == 415:
|
|
124
|
+
response_415 = ErrorResponse.from_dict(response.json())
|
|
125
|
+
|
|
126
|
+
|
|
127
|
+
|
|
128
|
+
return response_415
|
|
129
|
+
|
|
130
|
+
if response.status_code == 422:
|
|
131
|
+
response_422 = ErrorResponse.from_dict(response.json())
|
|
132
|
+
|
|
133
|
+
|
|
134
|
+
|
|
135
|
+
return response_422
|
|
136
|
+
|
|
137
|
+
if response.status_code == 429:
|
|
138
|
+
response_429 = ErrorResponse.from_dict(response.json())
|
|
139
|
+
|
|
140
|
+
|
|
141
|
+
|
|
142
|
+
return response_429
|
|
143
|
+
|
|
144
|
+
if response.status_code == 503:
|
|
145
|
+
response_503 = ErrorResponse.from_dict(response.json())
|
|
146
|
+
|
|
147
|
+
|
|
148
|
+
|
|
149
|
+
return response_503
|
|
150
|
+
|
|
151
|
+
if response.status_code == 504:
|
|
152
|
+
response_504 = ErrorResponse.from_dict(response.json())
|
|
153
|
+
|
|
154
|
+
|
|
155
|
+
|
|
156
|
+
return response_504
|
|
157
|
+
|
|
158
|
+
if client.raise_on_unexpected_status:
|
|
159
|
+
raise errors.UnexpectedStatus(response.status_code, response.content)
|
|
160
|
+
else:
|
|
161
|
+
return None
|
|
162
|
+
|
|
163
|
+
|
|
164
|
+
def _build_response(*, client: AuthenticatedClient | Client, response: httpx.Response) -> Response[ErrorResponse | JobAccepted | ParsedDocument | RawParseResult]:
|
|
165
|
+
return Response(
|
|
166
|
+
status_code=HTTPStatus(response.status_code),
|
|
167
|
+
content=response.content,
|
|
168
|
+
headers=response.headers,
|
|
169
|
+
parsed=_parse_response(client=client, response=response),
|
|
170
|
+
)
|
|
171
|
+
|
|
172
|
+
|
|
173
|
+
def sync_detailed(
|
|
174
|
+
*,
|
|
175
|
+
client: AuthenticatedClient | Client,
|
|
176
|
+
body: ParseSyncBody,
|
|
177
|
+
idempotency_key: str,
|
|
178
|
+
|
|
179
|
+
) -> Response[ErrorResponse | JobAccepted | ParsedDocument | RawParseResult]:
|
|
180
|
+
r""" Parse a document synchronously
|
|
181
|
+
|
|
182
|
+
Admit a parse job, wait up to the sync wait limit, and return the selected terminal parse
|
|
183
|
+
representation when ready. `output_format` defaults to the versioned provider-neutral
|
|
184
|
+
`openparser@1`; `raw` returns a stable provider envelope around the untouched successful
|
|
185
|
+
Paddle result. Terminal `failed` within the wait window returns `422` `ErrorResponse`;
|
|
186
|
+
terminal `indeterminate` returns `504` `ErrorResponse`. Returns `202` with a durable job
|
|
187
|
+
reference and `Location` if the wait limit expires first.
|
|
188
|
+
|
|
189
|
+
Example (`multipart/form-data`):
|
|
190
|
+
|
|
191
|
+
```bash
|
|
192
|
+
curl -X POST 'https://api.openparser.dev/parse' \
|
|
193
|
+
-H 'Authorization: Bearer YOUR_API_KEY' \
|
|
194
|
+
-H \"Idempotency-Key: $(uuidgen 2>/dev/null || openssl rand -hex 16)\" \
|
|
195
|
+
-F 'request={\"ocr_model\":\"paddleocr-
|
|
196
|
+
vl-1.6\",\"output_format\":\"openparser@1\"};type=application/json' \
|
|
197
|
+
-F 'file=@./document.pdf'
|
|
198
|
+
```
|
|
199
|
+
|
|
200
|
+
Args:
|
|
201
|
+
idempotency_key (str):
|
|
202
|
+
body (ParseSyncBody):
|
|
203
|
+
|
|
204
|
+
Raises:
|
|
205
|
+
errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
|
|
206
|
+
httpx.TimeoutException: If the request takes longer than Client.timeout.
|
|
207
|
+
|
|
208
|
+
Returns:
|
|
209
|
+
Response[ErrorResponse | JobAccepted | ParsedDocument | RawParseResult]
|
|
210
|
+
"""
|
|
211
|
+
|
|
212
|
+
|
|
213
|
+
kwargs = _get_kwargs(
|
|
214
|
+
body=body,
|
|
215
|
+
idempotency_key=idempotency_key,
|
|
216
|
+
|
|
217
|
+
)
|
|
218
|
+
|
|
219
|
+
response = client.get_httpx_client().request(
|
|
220
|
+
**kwargs,
|
|
221
|
+
)
|
|
222
|
+
|
|
223
|
+
return _build_response(client=client, response=response)
|
|
224
|
+
|
|
225
|
+
def sync(
|
|
226
|
+
*,
|
|
227
|
+
client: AuthenticatedClient | Client,
|
|
228
|
+
body: ParseSyncBody,
|
|
229
|
+
idempotency_key: str,
|
|
230
|
+
|
|
231
|
+
) -> ErrorResponse | JobAccepted | ParsedDocument | RawParseResult | None:
|
|
232
|
+
r""" Parse a document synchronously
|
|
233
|
+
|
|
234
|
+
Admit a parse job, wait up to the sync wait limit, and return the selected terminal parse
|
|
235
|
+
representation when ready. `output_format` defaults to the versioned provider-neutral
|
|
236
|
+
`openparser@1`; `raw` returns a stable provider envelope around the untouched successful
|
|
237
|
+
Paddle result. Terminal `failed` within the wait window returns `422` `ErrorResponse`;
|
|
238
|
+
terminal `indeterminate` returns `504` `ErrorResponse`. Returns `202` with a durable job
|
|
239
|
+
reference and `Location` if the wait limit expires first.
|
|
240
|
+
|
|
241
|
+
Example (`multipart/form-data`):
|
|
242
|
+
|
|
243
|
+
```bash
|
|
244
|
+
curl -X POST 'https://api.openparser.dev/parse' \
|
|
245
|
+
-H 'Authorization: Bearer YOUR_API_KEY' \
|
|
246
|
+
-H \"Idempotency-Key: $(uuidgen 2>/dev/null || openssl rand -hex 16)\" \
|
|
247
|
+
-F 'request={\"ocr_model\":\"paddleocr-
|
|
248
|
+
vl-1.6\",\"output_format\":\"openparser@1\"};type=application/json' \
|
|
249
|
+
-F 'file=@./document.pdf'
|
|
250
|
+
```
|
|
251
|
+
|
|
252
|
+
Args:
|
|
253
|
+
idempotency_key (str):
|
|
254
|
+
body (ParseSyncBody):
|
|
255
|
+
|
|
256
|
+
Raises:
|
|
257
|
+
errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
|
|
258
|
+
httpx.TimeoutException: If the request takes longer than Client.timeout.
|
|
259
|
+
|
|
260
|
+
Returns:
|
|
261
|
+
ErrorResponse | JobAccepted | ParsedDocument | RawParseResult
|
|
262
|
+
"""
|
|
263
|
+
|
|
264
|
+
|
|
265
|
+
return sync_detailed(
|
|
266
|
+
client=client,
|
|
267
|
+
body=body,
|
|
268
|
+
idempotency_key=idempotency_key,
|
|
269
|
+
|
|
270
|
+
).parsed
|
|
271
|
+
|
|
272
|
+
async def asyncio_detailed(
|
|
273
|
+
*,
|
|
274
|
+
client: AuthenticatedClient | Client,
|
|
275
|
+
body: ParseSyncBody,
|
|
276
|
+
idempotency_key: str,
|
|
277
|
+
|
|
278
|
+
) -> Response[ErrorResponse | JobAccepted | ParsedDocument | RawParseResult]:
|
|
279
|
+
r""" Parse a document synchronously
|
|
280
|
+
|
|
281
|
+
Admit a parse job, wait up to the sync wait limit, and return the selected terminal parse
|
|
282
|
+
representation when ready. `output_format` defaults to the versioned provider-neutral
|
|
283
|
+
`openparser@1`; `raw` returns a stable provider envelope around the untouched successful
|
|
284
|
+
Paddle result. Terminal `failed` within the wait window returns `422` `ErrorResponse`;
|
|
285
|
+
terminal `indeterminate` returns `504` `ErrorResponse`. Returns `202` with a durable job
|
|
286
|
+
reference and `Location` if the wait limit expires first.
|
|
287
|
+
|
|
288
|
+
Example (`multipart/form-data`):
|
|
289
|
+
|
|
290
|
+
```bash
|
|
291
|
+
curl -X POST 'https://api.openparser.dev/parse' \
|
|
292
|
+
-H 'Authorization: Bearer YOUR_API_KEY' \
|
|
293
|
+
-H \"Idempotency-Key: $(uuidgen 2>/dev/null || openssl rand -hex 16)\" \
|
|
294
|
+
-F 'request={\"ocr_model\":\"paddleocr-
|
|
295
|
+
vl-1.6\",\"output_format\":\"openparser@1\"};type=application/json' \
|
|
296
|
+
-F 'file=@./document.pdf'
|
|
297
|
+
```
|
|
298
|
+
|
|
299
|
+
Args:
|
|
300
|
+
idempotency_key (str):
|
|
301
|
+
body (ParseSyncBody):
|
|
302
|
+
|
|
303
|
+
Raises:
|
|
304
|
+
errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
|
|
305
|
+
httpx.TimeoutException: If the request takes longer than Client.timeout.
|
|
306
|
+
|
|
307
|
+
Returns:
|
|
308
|
+
Response[ErrorResponse | JobAccepted | ParsedDocument | RawParseResult]
|
|
309
|
+
"""
|
|
310
|
+
|
|
311
|
+
|
|
312
|
+
kwargs = _get_kwargs(
|
|
313
|
+
body=body,
|
|
314
|
+
idempotency_key=idempotency_key,
|
|
315
|
+
|
|
316
|
+
)
|
|
317
|
+
|
|
318
|
+
response = await client.get_async_httpx_client().request(
|
|
319
|
+
**kwargs
|
|
320
|
+
)
|
|
321
|
+
|
|
322
|
+
return _build_response(client=client, response=response)
|
|
323
|
+
|
|
324
|
+
async def asyncio(
|
|
325
|
+
*,
|
|
326
|
+
client: AuthenticatedClient | Client,
|
|
327
|
+
body: ParseSyncBody,
|
|
328
|
+
idempotency_key: str,
|
|
329
|
+
|
|
330
|
+
) -> ErrorResponse | JobAccepted | ParsedDocument | RawParseResult | None:
|
|
331
|
+
r""" Parse a document synchronously
|
|
332
|
+
|
|
333
|
+
Admit a parse job, wait up to the sync wait limit, and return the selected terminal parse
|
|
334
|
+
representation when ready. `output_format` defaults to the versioned provider-neutral
|
|
335
|
+
`openparser@1`; `raw` returns a stable provider envelope around the untouched successful
|
|
336
|
+
Paddle result. Terminal `failed` within the wait window returns `422` `ErrorResponse`;
|
|
337
|
+
terminal `indeterminate` returns `504` `ErrorResponse`. Returns `202` with a durable job
|
|
338
|
+
reference and `Location` if the wait limit expires first.
|
|
339
|
+
|
|
340
|
+
Example (`multipart/form-data`):
|
|
341
|
+
|
|
342
|
+
```bash
|
|
343
|
+
curl -X POST 'https://api.openparser.dev/parse' \
|
|
344
|
+
-H 'Authorization: Bearer YOUR_API_KEY' \
|
|
345
|
+
-H \"Idempotency-Key: $(uuidgen 2>/dev/null || openssl rand -hex 16)\" \
|
|
346
|
+
-F 'request={\"ocr_model\":\"paddleocr-
|
|
347
|
+
vl-1.6\",\"output_format\":\"openparser@1\"};type=application/json' \
|
|
348
|
+
-F 'file=@./document.pdf'
|
|
349
|
+
```
|
|
350
|
+
|
|
351
|
+
Args:
|
|
352
|
+
idempotency_key (str):
|
|
353
|
+
body (ParseSyncBody):
|
|
354
|
+
|
|
355
|
+
Raises:
|
|
356
|
+
errors.UnexpectedStatus: If the server returns an undocumented status code and Client.raise_on_unexpected_status is True.
|
|
357
|
+
httpx.TimeoutException: If the request takes longer than Client.timeout.
|
|
358
|
+
|
|
359
|
+
Returns:
|
|
360
|
+
ErrorResponse | JobAccepted | ParsedDocument | RawParseResult
|
|
361
|
+
"""
|
|
362
|
+
|
|
363
|
+
|
|
364
|
+
return (await asyncio_detailed(
|
|
365
|
+
client=client,
|
|
366
|
+
body=body,
|
|
367
|
+
idempotency_key=idempotency_key,
|
|
368
|
+
|
|
369
|
+
)).parsed
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
""" Contains endpoint functions for accessing the API """
|