documente_shared 0.1.183__tar.gz → 0.1.185b0__tar.gz

{documente_shared-0.1.183 → documente_shared-0.1.185b0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: documente_shared
-Version: 0.1.183
+Version: 0.1.185b0
 Summary: Shared utilities for LlamitAI projects
 Author-email: Tech <tech@llamitai.com>
 License: MIT

{documente_shared-0.1.183 → documente_shared-0.1.185b0}/documente_shared/application/retry_utils.py

@@ -2,11 +2,31 @@ import time
 import random
 from functools import wraps
 from typing import Optional, Callable
+from urllib3.util.retry import Retry
 import structlog
 
 logger = structlog.get_logger()
 
 
+class HttpRetry(Retry):
+    def increment(self, method=None, url=None, response=None, error=None, **kwargs):
+        if response is not None:
+            logger.warning(
+                "[shared.http.retry]",
+                method=method,
+                url=url,
+                status_code=response.status,
+            )
+        elif error is not None:
+            logger.warning(
+                "[shared.http.retry_connection_error]",
+                method=method,
+                url=url,
+                error=str(error),
+            )
+        return super().increment(method=method, url=url, response=response, error=error, **kwargs)
+
+
 def retry_on_size_integrity(base_delay: float = 0.8, max_retries: int = 3):
     def decorator(func: Callable):
         @wraps(func)

{documente_shared-0.1.183 → documente_shared-0.1.185b0}/documente_shared/domain/constants.py

@@ -7,4 +7,9 @@ la_paz_tz = ZoneInfo("America/La_Paz")
 DOCUMENTE_API_URL = environ.get("DOCUMENTE_API_URL")
 DOCUMENTE_API_KEY = environ.get("DOCUMENTE_API_KEY")
 
-DEFAULT_PAGINATION_PAGE_SIZE = 50
+DEFAULT_PAGINATION_PAGE_SIZE = 50
+
+HTTP_RETRY_TOTAL = 3
+HTTP_RETRY_BACKOFF_FACTOR = 0.5
+HTTP_RETRY_BACKOFF_JITTER = 0.25
+HTTP_RETRY_STATUS_FORCELIST = frozenset([429, 500, 502, 503, 504])

{documente_shared-0.1.183 → documente_shared-0.1.185b0}/documente_shared/domain/entities/document.py

@@ -36,6 +36,7 @@ class DocumentProcessing(object):
     processed_csv_path: Optional[str] = None
     processed_csv_bytes: Optional[bytes] = None
     processed_xlsx_path: Optional[str] = None
+    processed_digest: str | None = None
     processed_xlsx_bytes: Optional[bytes] = None
     processed_json_path: Optional[str] = None
     processed_json_bytes: Optional[bytes] = None
@@ -275,6 +276,7 @@ class DocumentProcessing(object):
             'locked_by_admin': self.locked_by_admin,
             'processed_csv_path': self.processed_csv_path,
             'processed_xlsx_path': self.processed_xlsx_path,
+            'processed_digest': self.processed_digest,
             'processed_json_path': self.processed_json_path,
             'processed_metadata_path': self.processed_metadata_path,
             'processing_time': (
@@ -343,6 +345,7 @@ class DocumentProcessing(object):
             'processed_csv_bytes',
             'processed_xlsx_path',
             'processed_xlsx_bytes',
+            'processed_digest',
             'processed_json_path',
             'processed_json_bytes',
             'processed_metadata_path',
@@ -358,7 +361,7 @@ class DocumentProcessing(object):
             'feedback',
             'metadata',
             'document_size',
-            'document_pages'
+            'document_pages',
             'completed_at',
         ]
         for _property in instance_properties:
@@ -390,6 +393,7 @@ class DocumentProcessing(object):
             locked_by_admin=data.get('locked_by_admin'),
             processed_csv_path=data.get('processed_csv_path'),
             processed_xlsx_path=data.get('processed_xlsx_path'),
+            processed_digest=data.get('processed_digest'),
             processed_json_path=data.get('processed_json_path'),
             processed_metadata_path=data.get('processed_metadata_path'),
             processing_time=(

{documente_shared-0.1.183 → documente_shared-0.1.185b0}/documente_shared/infrastructure/documente_client.py

@@ -1,9 +1,21 @@
 from contextlib import contextmanager
 from dataclasses import dataclass
 from requests import Response, Session
+from requests.adapters import HTTPAdapter
 from typing import Callable, Optional
 
+import structlog
+
 from documente_shared.application.payloads import camel_to_snake
+from documente_shared.application.retry_utils import HttpRetry
+from documente_shared.domain.constants import (
+    HTTP_RETRY_BACKOFF_FACTOR,
+    HTTP_RETRY_BACKOFF_JITTER,
+    HTTP_RETRY_STATUS_FORCELIST,
+    HTTP_RETRY_TOTAL,
+)
+
+logger = structlog.get_logger()
 
 
 @dataclass
@@ -19,6 +31,17 @@ class DocumenteClientMixin(object):
     def _get_session(self) -> Session:
         session = Session()
         session.headers.update(self.get_common_headers())
+        retry_strategy = HttpRetry(
+            total=HTTP_RETRY_TOTAL,
+            backoff_factor=HTTP_RETRY_BACKOFF_FACTOR,
+            backoff_jitter=HTTP_RETRY_BACKOFF_JITTER,
+            status_forcelist=HTTP_RETRY_STATUS_FORCELIST,
+            allowed_methods=["GET", "POST", "PUT", "PATCH", "DELETE"],
+            raise_on_status=False,
+        )
+        adapter = HTTPAdapter(max_retries=retry_strategy)
+        session.mount("http://", adapter)
+        session.mount("https://", adapter)
         return session
 
     def get_common_headers(self) -> dict:

{documente_shared-0.1.183 → documente_shared-0.1.185b0}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "documente_shared"
-version = "0.1.183"
+version = "0.1.185b"
 description = "Shared utilities for LlamitAI projects"
 authors = [{ name = "Tech", email = "tech@llamitai.com" }]
 license = { text = "MIT" }

{documente_shared-0.1.183 → documente_shared-0.1.185b0}/uv.lock

@@ -138,7 +138,7 @@ wheels = [
 
 [[package]]
 name = "documente-shared"
-version = "0.1.183"
+version = "0.1.185b0"
 source = { editable = "." }
 dependencies = [
     { name = "boto3" },