orbitkit 0.8.40__tar.gz → 0.8.41__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-0.8.40/orbitkit.egg-info → orbitkit-0.8.41}/PKG-INFO +1 -1
- orbitkit-0.8.41/orbitkit/VERSION +1 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/airflow_handler/data_preprocessing.py +5 -10
- {orbitkit-0.8.40 → orbitkit-0.8.41/orbitkit.egg-info}/PKG-INFO +1 -1
- orbitkit-0.8.40/orbitkit/VERSION +0 -1
- {orbitkit-0.8.40 → orbitkit-0.8.41}/LICENSE +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/MANIFEST.in +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/README.md +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/__init__.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/airflow_handler/__init__.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/airflow_handler/file_flow_entry_process.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/airflow_handler/file_flow_exit_process.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/airflow_handler/file_handler.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/airflow_handler/file_handler_v2.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/audio_transcoder/__init__.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/audio_transcoder/netmind_extract_v1.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/constant/__init__.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/constant/report_schema.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/id_srv/__init__.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/id_srv/id_gen.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/id_srv/id_perm_like.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/lark_send/__init__.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/lark_send/lark.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/llm_tools/__init__.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/llm_tools/quick_rag_chat.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/orbit_type/__init__.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/orbit_type/doc_4_compile_rule.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/orbit_type/orbit_type_simple.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/orbit_type/tools.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_embedding/__init__.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_embedding/pdf_txt_embedding.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_embedding/pdf_txt_embedding_v2.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor/__init__.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor/a_stock_extractor_v1.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor/exceptions.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_block_extractor_base.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_block_extractor_v1.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_block_extractor_v2.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_extractor_azure.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_extractor_minerU_v1.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_extractor_netmind_v1.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_extractor_netmind_v2.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_extractor_netmind_v3.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor/pdf_extractor_orbit.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor_simple/__init__.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor_simple/base.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor_simple/cloud_provider.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor_simple/core.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor_simple/exceptions.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor_simple/extractors.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_extractor_simple/utils.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_writer/__init__.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/pdf_writer/pdf_writer_simple.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/__init__.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/cache_asset_downloader.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/common.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/customize_regix_manager.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/secret_manager.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/util_aliyun.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/util_aliyun_oss_simple.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/util_aws.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/util_aws_s3_wrapper.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/util_date.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/util_html.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/util_kafka.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/util_md5.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/util_selenium.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/util_simple_timer.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/util_str.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/util_type_mapping.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit/util/util_url.py +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit.egg-info/SOURCES.txt +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit.egg-info/dependency_links.txt +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit.egg-info/not-zip-safe +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit.egg-info/requires.txt +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/orbitkit.egg-info/top_level.txt +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/setup.cfg +0 -0
- {orbitkit-0.8.40 → orbitkit-0.8.41}/setup.py +0 -0
|
@@ -0,0 +1 @@
|
|
|
1
|
+
0.8.41
|
|
@@ -1,12 +1,10 @@
|
|
|
1
1
|
import os
|
|
2
2
|
import datetime
|
|
3
3
|
from collections import defaultdict
|
|
4
|
-
import asyncio
|
|
5
4
|
from googletrans import Translator
|
|
6
5
|
|
|
7
6
|
|
|
8
7
|
class DocumentProcessor:
|
|
9
|
-
_translator = Translator()
|
|
10
8
|
|
|
11
9
|
AUDIO_SUFFIXES = [".mp3", ".wav", ".aac", ".wma", ".m4a"]
|
|
12
10
|
VIDEO_SUFFIXES = [".mp4", ".avi", ".mkv", ".mov", ".wmv", ".flv", ".webm", ".mpeg", ".3gp", ".hevc"]
|
|
@@ -21,15 +19,12 @@ class DocumentProcessor:
|
|
|
21
19
|
def get_file_suffix(file_path):
|
|
22
20
|
return f".{file_path.split('.')[-1]}".lower()
|
|
23
21
|
|
|
24
|
-
@
|
|
25
|
-
|
|
26
|
-
|
|
22
|
+
@staticmethod
|
|
23
|
+
def translate_text(text, dest='en'):
|
|
24
|
+
translator = Translator()
|
|
25
|
+
result = translator.translate(text, dest=dest)
|
|
27
26
|
return result.text
|
|
28
27
|
|
|
29
|
-
@classmethod
|
|
30
|
-
def translate_sync(cls, text, dest='en'):
|
|
31
|
-
return asyncio.run(cls.translate(text, dest))
|
|
32
|
-
|
|
33
28
|
@staticmethod
|
|
34
29
|
def create_xbrl_template():
|
|
35
30
|
return {
|
|
@@ -138,7 +133,7 @@ class DocumentProcessor:
|
|
|
138
133
|
if ext in cls.AUDIO_SUFFIXES or ext in cls.VIDEO_SUFFIXES:
|
|
139
134
|
add_extends = {
|
|
140
135
|
"original_title": doc['x_orbit_data']['report_title'],
|
|
141
|
-
"title": cls.
|
|
136
|
+
"title": cls.translate_text(doc['x_orbit_data']['report_title']),
|
|
142
137
|
"published": doc['x_reported_at_utc_date'],
|
|
143
138
|
"tickers": [],
|
|
144
139
|
"perm_id_list": doc['x_orbit_data']['perm_id_list'],
|
orbitkit-0.8.40/orbitkit/VERSION
DELETED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
0.8.40
|
|
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
|