orbitkit 0.8.57__tar.gz → 0.8.58__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.
- {orbitkit-0.8.57/orbitkit.egg-info → orbitkit-0.8.58}/PKG-INFO +1 -1
- orbitkit-0.8.58/orbitkit/VERSION +1 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/airflow_handler/data_preprocessing.py +4 -1
- {orbitkit-0.8.57 → orbitkit-0.8.58/orbitkit.egg-info}/PKG-INFO +1 -1
- orbitkit-0.8.57/orbitkit/VERSION +0 -1
- {orbitkit-0.8.57 → orbitkit-0.8.58}/LICENSE +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/MANIFEST.in +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/README.md +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/__init__.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/airflow_handler/__init__.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/airflow_handler/file_flow_entry_process.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/airflow_handler/file_flow_exit_process.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/airflow_handler/file_handler.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/airflow_handler/file_handler_v2.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/audio_transcoder/__init__.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/audio_transcoder/netmind_extract_v1.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/constant/__init__.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/constant/report_schema.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/id_srv/__init__.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/id_srv/id_gen.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/id_srv/id_perm_like.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/lark_send/__init__.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/lark_send/lark.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/llm_tools/__init__.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/llm_tools/quick_rag_chat.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/orbit_type/__init__.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/orbit_type/doc_4_compile_rule.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/orbit_type/orbit_type_simple.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/orbit_type/tools.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_embedding/__init__.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_embedding/pdf_txt_embedding.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_embedding/pdf_txt_embedding_v2.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor/__init__.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor/a_stock_extractor_v1.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor/exceptions.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_block_extractor_base.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_block_extractor_v1.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_block_extractor_v2.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_extractor_azure.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_extractor_minerU_v1.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_extractor_netmind_v1.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_extractor_netmind_v2.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_extractor_netmind_v3.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor/pdf_extractor_orbit.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor_simple/__init__.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor_simple/base.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor_simple/cloud_provider.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor_simple/core.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor_simple/exceptions.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor_simple/extractors.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_extractor_simple/utils.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_writer/__init__.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/pdf_writer/pdf_writer_simple.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/__init__.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/cache_asset_downloader.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/common.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/customize_regix_manager.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/secret_manager.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/universal_extractor.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/util_aliyun.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/util_aliyun_oss_simple.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/util_aws.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/util_aws_s3_wrapper.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/util_date.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/util_html.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/util_kafka.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/util_md5.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/util_selenium.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/util_simple_timer.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/util_str.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/util_type_mapping.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit/util/util_url.py +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit.egg-info/SOURCES.txt +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit.egg-info/dependency_links.txt +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit.egg-info/not-zip-safe +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit.egg-info/requires.txt +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/orbitkit.egg-info/top_level.txt +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/setup.cfg +0 -0
- {orbitkit-0.8.57 → orbitkit-0.8.58}/setup.py +0 -0
|
@@ -0,0 +1 @@
|
|
|
1
|
+
0.8.58
|
|
@@ -3,6 +3,7 @@ import datetime
|
|
|
3
3
|
from collections import defaultdict
|
|
4
4
|
from importlib.metadata import version
|
|
5
5
|
import googletrans
|
|
6
|
+
from dateutil.relativedelta import relativedelta
|
|
6
7
|
|
|
7
8
|
|
|
8
9
|
class DocumentProcessor:
|
|
@@ -122,11 +123,13 @@ class DocumentProcessor:
|
|
|
122
123
|
@staticmethod
|
|
123
124
|
def update_target_stage_by_reported_at(doc, target_stage):
|
|
124
125
|
date_str = doc.get('x_reported_at_utc_date', '1970-01-01')
|
|
126
|
+
now = datetime.datetime.now()
|
|
127
|
+
one_year_ago = now - relativedelta(years=1)
|
|
125
128
|
try:
|
|
126
129
|
reported_date = datetime.datetime.strptime(date_str, "%Y-%m-%d")
|
|
127
130
|
except ValueError:
|
|
128
131
|
reported_date = datetime.datetime(1970, 1, 1)
|
|
129
|
-
return "extract" if reported_date <
|
|
132
|
+
return "extract" if reported_date < one_year_ago else target_stage
|
|
130
133
|
|
|
131
134
|
@staticmethod
|
|
132
135
|
def update_target_stage_by_perm_match(doc, target_stage):
|
orbitkit-0.8.57/orbitkit/VERSION
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
0.8.57
|
|
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
|
|
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
|