Spaces:
Sleeping
Sleeping
Create models.py
Browse files
models.py
ADDED
|
@@ -0,0 +1,101 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
Database models for HF Uploader Dashboard
|
| 3 |
+
"""
|
| 4 |
+
from datetime import datetime
|
| 5 |
+
from sqlalchemy import Column, Integer, String, Text, DateTime, Boolean, Enum
|
| 6 |
+
from sqlalchemy.ext.declarative import declarative_base
|
| 7 |
+
import enum
|
| 8 |
+
|
| 9 |
+
Base = declarative_base()
|
| 10 |
+
|
| 11 |
+
|
| 12 |
+
class HFConfig(Base):
|
| 13 |
+
"""Configuration for Hugging Face API and datasets"""
|
| 14 |
+
__tablename__ = "hf_config"
|
| 15 |
+
|
| 16 |
+
id = Column(Integer, primary_key=True, index=True)
|
| 17 |
+
hf_token = Column(Text, nullable=True)
|
| 18 |
+
source_all_repo = Column(String(255), default="samfred2/ALL", nullable=False)
|
| 19 |
+
source_ato_repo = Column(String(255), default="samfred2/ATO", nullable=False)
|
| 20 |
+
target_repo = Column(String(255), default="samfred2/ALL2", nullable=False)
|
| 21 |
+
upload_batch_size = Column(Integer, default=10, nullable=False)
|
| 22 |
+
max_uploads_per_hour = Column(Integer, default=128, nullable=False)
|
| 23 |
+
created_at = Column(DateTime, default=datetime.utcnow, nullable=False)
|
| 24 |
+
updated_at = Column(DateTime, default=datetime.utcnow, onupdate=datetime.utcnow, nullable=False)
|
| 25 |
+
|
| 26 |
+
|
| 27 |
+
class UploadStatusEnum(str, enum.Enum):
|
| 28 |
+
"""Upload status enumeration"""
|
| 29 |
+
PENDING = "pending"
|
| 30 |
+
UPLOADING = "uploading"
|
| 31 |
+
COMPLETED = "completed"
|
| 32 |
+
FAILED = "failed"
|
| 33 |
+
|
| 34 |
+
|
| 35 |
+
class UploadQueue(Base):
|
| 36 |
+
"""Upload queue tracking"""
|
| 37 |
+
__tablename__ = "upload_queue"
|
| 38 |
+
|
| 39 |
+
id = Column(Integer, primary_key=True, index=True)
|
| 40 |
+
file_name = Column(String(255), nullable=False)
|
| 41 |
+
file_path = Column(Text, nullable=False)
|
| 42 |
+
file_size = Column(Integer, nullable=False)
|
| 43 |
+
status = Column(Enum(UploadStatusEnum), default=UploadStatusEnum.PENDING, nullable=False)
|
| 44 |
+
uploaded_at = Column(DateTime, nullable=True)
|
| 45 |
+
failure_reason = Column(Text, nullable=True)
|
| 46 |
+
retry_count = Column(Integer, default=0, nullable=False)
|
| 47 |
+
max_retries = Column(Integer, default=3, nullable=False)
|
| 48 |
+
created_at = Column(DateTime, default=datetime.utcnow, nullable=False)
|
| 49 |
+
updated_at = Column(DateTime, default=datetime.utcnow, onupdate=datetime.utcnow, nullable=False)
|
| 50 |
+
|
| 51 |
+
|
| 52 |
+
class RateLimitLog(Base):
|
| 53 |
+
"""Rate limit tracking per hour"""
|
| 54 |
+
__tablename__ = "rate_limit_log"
|
| 55 |
+
|
| 56 |
+
id = Column(Integer, primary_key=True, index=True)
|
| 57 |
+
upload_count = Column(Integer, default=0, nullable=False)
|
| 58 |
+
hour_start = Column(DateTime, nullable=False)
|
| 59 |
+
hour_end = Column(DateTime, nullable=False)
|
| 60 |
+
limit_hit = Column(Boolean, default=False, nullable=False)
|
| 61 |
+
resume_time = Column(DateTime, nullable=True)
|
| 62 |
+
created_at = Column(DateTime, default=datetime.utcnow, nullable=False)
|
| 63 |
+
|
| 64 |
+
|
| 65 |
+
class UploadErrorLog(Base):
|
| 66 |
+
"""Error logging for uploads"""
|
| 67 |
+
__tablename__ = "upload_error_log"
|
| 68 |
+
|
| 69 |
+
id = Column(Integer, primary_key=True, index=True)
|
| 70 |
+
file_name = Column(String(255), nullable=False)
|
| 71 |
+
error_code = Column(String(50), nullable=True)
|
| 72 |
+
error_message = Column(Text, nullable=True)
|
| 73 |
+
status_code = Column(Integer, nullable=True)
|
| 74 |
+
retryable = Column(Boolean, default=True, nullable=False)
|
| 75 |
+
created_at = Column(DateTime, default=datetime.utcnow, nullable=False)
|
| 76 |
+
|
| 77 |
+
|
| 78 |
+
class ProcessingStatusEnum(str, enum.Enum):
|
| 79 |
+
"""Processing status enumeration"""
|
| 80 |
+
IDLE = "idle"
|
| 81 |
+
DOWNLOADING = "downloading"
|
| 82 |
+
MATCHING = "matching"
|
| 83 |
+
INTEGRATING = "integrating"
|
| 84 |
+
COMPLETED = "completed"
|
| 85 |
+
ERROR = "error"
|
| 86 |
+
|
| 87 |
+
|
| 88 |
+
class ProcessingState(Base):
|
| 89 |
+
"""File processing state tracking"""
|
| 90 |
+
__tablename__ = "processing_state"
|
| 91 |
+
|
| 92 |
+
id = Column(Integer, primary_key=True, index=True)
|
| 93 |
+
status = Column(Enum(ProcessingStatusEnum), default=ProcessingStatusEnum.IDLE, nullable=False)
|
| 94 |
+
total_files = Column(Integer, default=0, nullable=False)
|
| 95 |
+
processed_files = Column(Integer, default=0, nullable=False)
|
| 96 |
+
matched_pairs = Column(Integer, default=0, nullable=False)
|
| 97 |
+
error_message = Column(Text, nullable=True)
|
| 98 |
+
started_at = Column(DateTime, nullable=True)
|
| 99 |
+
completed_at = Column(DateTime, nullable=True)
|
| 100 |
+
created_at = Column(DateTime, default=datetime.utcnow, nullable=False)
|
| 101 |
+
updated_at = Column(DateTime, default=datetime.utcnow, onupdate=datetime.utcnow, nullable=False)
|