update app/db_models.py
Browse files- app/db_models.py +230 -0
app/db_models.py
ADDED
|
@@ -0,0 +1,230 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from sqlalchemy import (
|
| 2 |
+
Column,
|
| 3 |
+
Integer,
|
| 4 |
+
String,
|
| 5 |
+
Boolean,
|
| 6 |
+
Float,
|
| 7 |
+
Text,
|
| 8 |
+
DateTime,
|
| 9 |
+
ForeignKey,
|
| 10 |
+
Index,
|
| 11 |
+
JSON,
|
| 12 |
+
)
|
| 13 |
+
from sqlalchemy.orm import relationship
|
| 14 |
+
from sqlalchemy.sql import func
|
| 15 |
+
from app.database import Base
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
class User(Base):
|
| 19 |
+
__tablename__ = "users"
|
| 20 |
+
|
| 21 |
+
id = Column(Integer, primary_key=True, index=True)
|
| 22 |
+
oauth_provider = Column(String(20), nullable=False)
|
| 23 |
+
oauth_id = Column(String(100), nullable=False, unique=True, index=True)
|
| 24 |
+
email = Column(String(255), nullable=False)
|
| 25 |
+
username = Column(String(100), nullable=False)
|
| 26 |
+
avatar_url = Column(Text, nullable=True)
|
| 27 |
+
role = Column(String(20), default="user", nullable=False)
|
| 28 |
+
created_at = Column(DateTime(timezone=True), server_default=func.now())
|
| 29 |
+
last_login = Column(DateTime(timezone=True), nullable=True)
|
| 30 |
+
|
| 31 |
+
sources = relationship(
|
| 32 |
+
"ProxySource", back_populates="owner", cascade="all, delete-orphan"
|
| 33 |
+
)
|
| 34 |
+
|
| 35 |
+
__table_args__ = (
|
| 36 |
+
Index("idx_oauth_provider_id", "oauth_provider", "oauth_id", unique=True),
|
| 37 |
+
)
|
| 38 |
+
|
| 39 |
+
|
| 40 |
+
class ProxySource(Base):
|
| 41 |
+
__tablename__ = "proxy_sources"
|
| 42 |
+
|
| 43 |
+
id = Column(Integer, primary_key=True, index=True)
|
| 44 |
+
user_id = Column(Integer, ForeignKey("users.id", ondelete="CASCADE"), nullable=True)
|
| 45 |
+
url = Column(Text, nullable=False, unique=True, index=True)
|
| 46 |
+
type = Column(String(50), nullable=False)
|
| 47 |
+
name = Column(String(200), nullable=True)
|
| 48 |
+
description = Column(Text, nullable=True)
|
| 49 |
+
is_paid = Column(Boolean, default=False)
|
| 50 |
+
enabled = Column(Boolean, default=True, index=True)
|
| 51 |
+
validated = Column(Boolean, default=False)
|
| 52 |
+
validation_error = Column(Text, nullable=True)
|
| 53 |
+
created_at = Column(DateTime(timezone=True), server_default=func.now())
|
| 54 |
+
updated_at = Column(
|
| 55 |
+
DateTime(timezone=True), server_default=func.now(), onupdate=func.now()
|
| 56 |
+
)
|
| 57 |
+
last_scraped = Column(DateTime(timezone=True), nullable=True)
|
| 58 |
+
total_scraped = Column(Integer, default=0)
|
| 59 |
+
success_rate = Column(Float, default=0.0)
|
| 60 |
+
is_admin_source = Column(Boolean, default=False)
|
| 61 |
+
|
| 62 |
+
owner = relationship("User", back_populates="sources")
|
| 63 |
+
proxies = relationship("Proxy", back_populates="source")
|
| 64 |
+
|
| 65 |
+
|
| 66 |
+
class Proxy(Base):
|
| 67 |
+
__tablename__ = "proxies"
|
| 68 |
+
|
| 69 |
+
id = Column(Integer, primary_key=True, index=True)
|
| 70 |
+
source_id = Column(
|
| 71 |
+
Integer, ForeignKey("proxy_sources.id", ondelete="SET NULL"), nullable=True
|
| 72 |
+
)
|
| 73 |
+
url = Column(Text, nullable=False, unique=True, index=True)
|
| 74 |
+
protocol = Column(String(50), nullable=False, index=True)
|
| 75 |
+
ip = Column(String(50), nullable=True)
|
| 76 |
+
port = Column(Integer, nullable=True)
|
| 77 |
+
country_code = Column(String(2), nullable=True, index=True)
|
| 78 |
+
country_name = Column(String(100), nullable=True)
|
| 79 |
+
state = Column(String(100), nullable=True)
|
| 80 |
+
city = Column(String(100), nullable=True)
|
| 81 |
+
|
| 82 |
+
latency_ms = Column(Integer, nullable=True, index=True)
|
| 83 |
+
speed_mbps = Column(Float, nullable=True)
|
| 84 |
+
uptime_percent = Column(Float, nullable=True)
|
| 85 |
+
|
| 86 |
+
anonymity = Column(String(20), nullable=True, index=True)
|
| 87 |
+
proxy_type = Column(String(20), nullable=True)
|
| 88 |
+
can_access_google = Column(Boolean, nullable=True)
|
| 89 |
+
quality_score = Column(Integer, nullable=True, index=True)
|
| 90 |
+
|
| 91 |
+
validation_status = Column(
|
| 92 |
+
String(20), default="pending", nullable=False, index=True
|
| 93 |
+
) # pending, validating, validated, failed
|
| 94 |
+
last_validated = Column(DateTime(timezone=True), nullable=True)
|
| 95 |
+
validation_failures = Column(Integer, default=0)
|
| 96 |
+
is_working = Column(Boolean, default=True, index=True)
|
| 97 |
+
|
| 98 |
+
created_at = Column(DateTime(timezone=True), server_default=func.now())
|
| 99 |
+
updated_at = Column(
|
| 100 |
+
DateTime(timezone=True), server_default=func.now(), onupdate=func.now()
|
| 101 |
+
)
|
| 102 |
+
first_seen = Column(DateTime(timezone=True), server_default=func.now())
|
| 103 |
+
last_seen = Column(DateTime(timezone=True), server_default=func.now())
|
| 104 |
+
|
| 105 |
+
source = relationship("ProxySource", back_populates="proxies")
|
| 106 |
+
validation_history = relationship(
|
| 107 |
+
"ValidationHistory", back_populates="proxy", cascade="all, delete-orphan"
|
| 108 |
+
)
|
| 109 |
+
|
| 110 |
+
__table_args__ = (
|
| 111 |
+
Index(
|
| 112 |
+
"idx_proxy_working_status_quality",
|
| 113 |
+
"is_working",
|
| 114 |
+
"validation_status",
|
| 115 |
+
"quality_score",
|
| 116 |
+
),
|
| 117 |
+
)
|
| 118 |
+
|
| 119 |
+
|
| 120 |
+
class ValidationHistory(Base):
|
| 121 |
+
__tablename__ = "validation_history"
|
| 122 |
+
|
| 123 |
+
id = Column(Integer, primary_key=True, index=True)
|
| 124 |
+
proxy_id = Column(
|
| 125 |
+
Integer,
|
| 126 |
+
ForeignKey("proxies.id", ondelete="CASCADE"),
|
| 127 |
+
nullable=False,
|
| 128 |
+
index=True,
|
| 129 |
+
)
|
| 130 |
+
validated_at = Column(
|
| 131 |
+
DateTime(timezone=True), server_default=func.now(), index=True
|
| 132 |
+
)
|
| 133 |
+
latency_ms = Column(Integer, nullable=True)
|
| 134 |
+
anonymity = Column(String(20), nullable=True)
|
| 135 |
+
can_access_google = Column(Boolean, nullable=True)
|
| 136 |
+
success = Column(Boolean, nullable=False)
|
| 137 |
+
error_message = Column(Text, nullable=True)
|
| 138 |
+
|
| 139 |
+
proxy = relationship("Proxy", back_populates="validation_history")
|
| 140 |
+
|
| 141 |
+
|
| 142 |
+
class CandidateSource(Base):
|
| 143 |
+
__tablename__ = "candidate_sources"
|
| 144 |
+
|
| 145 |
+
id = Column(Integer, primary_key=True, index=True)
|
| 146 |
+
url = Column(Text, nullable=False, unique=True, index=True)
|
| 147 |
+
domain = Column(String(255), nullable=True, index=True)
|
| 148 |
+
discovery_method = Column(String(50), nullable=False) # github, search, ai
|
| 149 |
+
status = Column(
|
| 150 |
+
String(20), default="pending", nullable=False, index=True
|
| 151 |
+
) # pending, validating, approved, rejected
|
| 152 |
+
confidence_score = Column(Integer, default=0, index=True)
|
| 153 |
+
proxies_found_count = Column(Integer, default=0)
|
| 154 |
+
last_checked_at = Column(DateTime(timezone=True), nullable=True)
|
| 155 |
+
created_at = Column(DateTime(timezone=True), server_default=func.now())
|
| 156 |
+
fail_count = Column(Integer, default=0)
|
| 157 |
+
meta_data = Column(JSON, nullable=True)
|
| 158 |
+
|
| 159 |
+
__table_args__ = (
|
| 160 |
+
Index("idx_candidate_status_score", "status", "confidence_score"),
|
| 161 |
+
)
|
| 162 |
+
|
| 163 |
+
|
| 164 |
+
class Notification(Base):
|
| 165 |
+
__tablename__ = "notifications"
|
| 166 |
+
|
| 167 |
+
id = Column(Integer, primary_key=True, index=True)
|
| 168 |
+
user_id = Column(
|
| 169 |
+
Integer, ForeignKey("users.id", ondelete="CASCADE"), nullable=False, index=True
|
| 170 |
+
)
|
| 171 |
+
type = Column(String(50), nullable=False, index=True)
|
| 172 |
+
title = Column(String(255), nullable=False)
|
| 173 |
+
message = Column(Text, nullable=False)
|
| 174 |
+
severity = Column(String(20), default="info", nullable=False)
|
| 175 |
+
created_at = Column(DateTime(timezone=True), server_default=func.now(), index=True)
|
| 176 |
+
read = Column(Boolean, default=False, index=True)
|
| 177 |
+
|
| 178 |
+
user = relationship("User", backref="notifications")
|
| 179 |
+
|
| 180 |
+
|
| 181 |
+
class ScrapingSession(Base):
|
| 182 |
+
__tablename__ = "scraping_sessions"
|
| 183 |
+
|
| 184 |
+
id = Column(Integer, primary_key=True, index=True)
|
| 185 |
+
source_id = Column(
|
| 186 |
+
Integer,
|
| 187 |
+
ForeignKey("proxy_sources.id", ondelete="CASCADE"),
|
| 188 |
+
nullable=False,
|
| 189 |
+
index=True,
|
| 190 |
+
)
|
| 191 |
+
scraping_type = Column(String(50), default="scheduled", nullable=False)
|
| 192 |
+
status = Column(String(20), default="running", nullable=False, index=True)
|
| 193 |
+
proxies_found = Column(Integer, default=0)
|
| 194 |
+
proxies_valid = Column(Integer, default=0)
|
| 195 |
+
config = Column(JSON, nullable=True)
|
| 196 |
+
error_message = Column(Text, nullable=True)
|
| 197 |
+
initiated_by = Column(
|
| 198 |
+
Integer, ForeignKey("users.id", ondelete="SET NULL"), nullable=True, index=True
|
| 199 |
+
)
|
| 200 |
+
started_at = Column(DateTime(timezone=True), server_default=func.now(), index=True)
|
| 201 |
+
finished_at = Column(DateTime(timezone=True), nullable=True)
|
| 202 |
+
|
| 203 |
+
source = relationship("ProxySource")
|
| 204 |
+
initiator = relationship("User")
|
| 205 |
+
|
| 206 |
+
__table_args__ = (
|
| 207 |
+
Index("idx_scraping_session_status_source", "status", "source_id"),
|
| 208 |
+
)
|
| 209 |
+
|
| 210 |
+
|
| 211 |
+
class BackgroundTask(Base):
|
| 212 |
+
__tablename__ = "background_tasks"
|
| 213 |
+
|
| 214 |
+
id = Column(Integer, primary_key=True, index=True)
|
| 215 |
+
task_type = Column(String(100), nullable=False, index=True)
|
| 216 |
+
task_data = Column(JSON, nullable=False)
|
| 217 |
+
status = Column(String(20), default="pending", nullable=False, index=True)
|
| 218 |
+
result = Column(JSON, nullable=True)
|
| 219 |
+
error_message = Column(Text, nullable=True)
|
| 220 |
+
retry_count = Column(Integer, default=0)
|
| 221 |
+
max_retries = Column(Integer, default=3)
|
| 222 |
+
scheduled_for = Column(
|
| 223 |
+
DateTime(timezone=True), server_default=func.now(), index=True
|
| 224 |
+
)
|
| 225 |
+
started_at = Column(DateTime(timezone=True), nullable=True)
|
| 226 |
+
completed_at = Column(DateTime(timezone=True), nullable=True)
|
| 227 |
+
|
| 228 |
+
__table_args__ = (
|
| 229 |
+
Index("idx_background_task_status_scheduled", "status", "scheduled_for"),
|
| 230 |
+
)
|