Database Models

Our RDBMS layer is very interesting: obviously it incorporates all of the traditional elements - all SQLAlchemy, but in order to be resiliant, we’ve a lot of JSON fields. These all have Pydantic models associated with them, so everything is deeply Pythonic.

class lbn.aidoc.models.Application

Bases: BaseModel

Structured JSON layout for a complete job application document set.

class CVProfile

Bases: BaseModel

keywords: List[str]

model_config = {'arbitrary_types_allowed': True, 'extra': 'allow'}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

required: List[str]

sections: List[str]

subject: str | None

title: str

class LetterProfile

Bases: BaseModel

freetext: str

marketplace: List[str]

model_config = {'arbitrary_types_allowed': True, 'extra': 'allow'}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

required: List[str]

sections: List[str]

title: str

__init__(**data: Any)

Surgical Constructor: Seamlessly intercepts input shapes, converting raw nested dictionaries into deep, dot-notation queryable sub-objects.

cv: Any

letter: Any

model_config = {'arbitrary_types_allowed': True, 'extra': 'allow'}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

url: HttpUrl | None

class lbn.aidoc.models.Comment

Bases: SQLModel

__init__(**data)

as_str()

category: str

content: str

id: int | None

model_config = {'from_attributes': True, 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

property parent

Automatically extracts the database session from itself to find the parent model.

property parent_class

parent_id: int

parent_type: str

timestamp: datetime

class lbn.aidoc.models.Company

Bases: BaseEntity

__init__(**data)

analytics: Any

as_str()

Abstract string serialization method.

Must be explicitly overridden by every child model to pass system tests.

code: ANZSICCode | None

property employees: List[Person]

The genuine internal team: Bosses, Peers, Talent, etc.

geo: str | None

id

property is_agency: bool

Dynamically classifies recruitment status by reading taxonomy depth.

jobs: Mapped[List[Job]]

linkedin: str | None

location: str | None

model_config = {'from_attributes': True, 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

name: str

property recruiters: List[Person]

The external agency partners linked to this company.

state: EntityStatus

stints: Mapped[List[EmploymentHistory]]

tier: CompanyTier | None

website: str | None

class lbn.aidoc.models.CompanyAlias

Bases: SQLModel

__init__(**data)

company_id: int

model_config = {'from_attributes': True, 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

name: str

class lbn.aidoc.models.EmploymentHistory

Bases: SQLModel

The Ledger: Tracks who works (or worked) where.

__init__(**data)

archetype: Archetype

company: Mapped[Company]

company_id: int

end_date: datetime | None

id: int | None

model_config = {'from_attributes': True, 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

person: Mapped[Person]

person_id: int

start_date: datetime | None

classmethod validate_styled_enums(v, info)

The ‘Inhale Handshake’: Forces Pydantic to use the StyledEnum constructor. This re-attaches the .style metadata lost during AI extraction.

class lbn.aidoc.models.Financials

Bases: SQLModel

Financial information, and state/processing of said information

__init__(**data)

analytics: Any

created_at: datetime

data: FinancialData

history: Mapped[List[History]]

id: int | None

initiative: Mapped[Initiative | None]

initiative_id: int

model_config = {'from_attributes': True, 'ignored_types': (<class 'sqlalchemy.ext.hybrid.hybrid_property'>,), 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

name: str

state: FinancialsStatus

updated_at: datetime

class lbn.aidoc.models.History

Bases: SQLModel

for state machine transition histories

__init__(**data)

id: int | None

model_config = {'from_attributes': True, 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

new_status: str | None

note: str

old_status: str | None

parent_id: int

parent_type: str

timestamp: datetime

class lbn.aidoc.models.Initiative

Bases: SQLModel

A project/initiative - bucket of financials, people, tasks

__init__(**data)

analytics: Any

as_str()

champion_id: int | None

created_at: datetime

description: str | None

end_date: date | None

financials: Mapped[List[Financials]]

get_financial_record(name: str) → Financials | None

Safely extracts a specific financial variant row matching the lookup name.

history: Mapped[List[History]]

id: int | None

initiative_type: InitiativeType

model_config = {'from_attributes': True, 'ignored_types': (<class 'sqlalchemy.ext.hybrid.hybrid_property'>,), 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

priority: Priority

start_date: date | None

state: InitiativeStatus

tasks: Mapped[List[Task]]

team: Mapped[List[Person]]

title: str

updated_at: datetime

classmethod validate_styled_enums(v, info)

The ‘Inhale Handshake’: Forces Pydantic to use the StyledEnum constructor. Re-attaches the color metadata attributes during extraction swaps.

class lbn.aidoc.models.Interview

Bases: BaseEntity

__init__(**data)

as_ical() → str

Surgical Export: Generates a valid iCalendar byte string using the icalendar library.

as_str()

Abstract string serialization method.

Must be explicitly overridden by every child model to pass system tests.

attendee_ids: List[int]

property attendees: List[Person]

Surgical Lookahead: Direct session query to resolve the JSON list of IDs. Returns a list of Person objects matching the internal array registry.

history: Mapped[List[History]]

id

job: Mapped[Job | None]

job_id: int | None

location: str

model_config = {'from_attributes': True, 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

predecessor: Mapped[Interview | None]

predecessor_id: int | None

predecessors() → List[Interview]

Walks backwards up the chain of predecessor links. Returns a list of all prior interview objects, sorted chronologically.

scheduled_at: datetime

state: MeetingStatus

title: str

classmethod validate_styled_enums(v, info)

The ‘Inhale Handshake’: Forces Pydantic to use the StyledEnum constructor. This re-attaches the .style metadata lost during AI extraction.

class lbn.aidoc.models.Job

Bases: SQLModel

__init__(**data)

analytics: Any

annual_equivalent

applicants: int | None

application: Any

as_str()

closes: datetime | None

company: Mapped[Company | None]

company_id: int | None

description: str | None

expired_at: datetime | None

external_id: str | None

geo: str | None

history: Mapped[List[History]]

id: int | None

ingested

interviews: Mapped[List[Interview]]

job_style: JobStyle

job_type: JobType

keywords: List[str]

listed: datetime | None

location: str | None

model_config = {'from_attributes': True, 'ignored_types': (<class 'sqlalchemy.ext.hybrid.hybrid_property'>,), 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

priority: Priority

recruiter: Mapped[Person | None]

recruiter_id: int | None

salary_currency: str

salary_max: int | None

salary_min: int | None

salary_period: SalaryPeriod

score: float | None

source: Mapped[Source | None]

source_id: int | None

state: JobStatus

title: str

url: str | None

classmethod validate_styled_enums(v, info)

The ‘Inhale Handshake’: Forces Pydantic to use the StyledEnum constructor. This re-attaches the .style metadata lost during AI extraction.

version: int

class lbn.aidoc.models.Person

Bases: BaseEntity

The Master Address Book: Purely digital presence. Independence: No relationships or back-links defined here.

__init__(**data)

analytics: Any

apikey: str | None

as_str()

Abstract string serialization method.

Must be explicitly overridden by every child model to pass system tests.

check_password(raw_password: str) → bool

Verify an incoming password attempt against the database hash.

colleagues(current_only: bool = True) → List[Person]

Returns a list of all distinct people who work (or worked) at the same companies.

Executes a native SQL self-join on the ledger, completely bypassing in-memory loops.

email: str | None

property employer: Company | None

Returns the current Company from the latest stint.

first_name()

generate_apikey() → str

Generate a secure, random API prefix string for headless CLI requests.

geo: str | None

id

jobtitle: str | None

last_name()

property latest_stint: EmploymentHistory | None

Returns the newest stint based on primary key ID fallback.

linkedin: str | None

location: str | None

model_config = {'from_attributes': True, 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

name: str

password_hash: str | None

phone: str | None

set_password(raw_password: str) → None

Hash a password securely using Python 3.13 standard libraries.

state: EntityStatus

stints: Mapped[List[EmploymentHistory]]

username: str | None

website: str | None

class lbn.aidoc.models.PersonAlias

Bases: SQLModel

__init__(**data)

model_config = {'from_attributes': True, 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

name: str

person_id: int

class lbn.aidoc.models.Prompt

Bases: SQLModel

an LLM instruction, with well defined inputs and outputs

__init__(**data)

compile(**kwargs) → List[Dict[str, str]]

The ‘Inhale’: Uses the input_model to resolve defaults and satisfy ‘required’ constraints before the slurp.

id: int | None

input_model() → Type[BaseModel]

The ‘Inhale Contractor’: Dynamic validation block for input parameters. Recursively compiles a genuine Pydantic model contract straight from the stored input_schema JSON dictionary.

input_schema: Dict[str, Any]

inputs() → Dict[str, List[str]]

Decomposes the input JSONSchema.

link(data: Any) → Any

The ‘Linker’: turns dicts, etc into our pydantic output schema structure

model_config = {'from_attributes': True, 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

name: str

output_model() → Type[BaseModel]

The ‘Inhale Contractor’: High-fidelity type mapping. Recursively compiles a genuine, deeply nested Pydantic model contract straight from the stored output_schema JSON dictionary.

output_schema: Dict[str, Any]

outputs() → Dict[str, List[str]]

Decomposes the output JSONSchema.

provider: str | None

rst_files: Mapped[List[RSTFile]]

system_message: str

temperature: float

ttl: int

validate_schema_integrity() → Prompt

Ensure every required field has a matching property definition.

weight: Weight

class lbn.aidoc.models.RSTFile

Bases: SQLModel

__init__(**data)

epigraph: str | None

epilog: str | None

id: int | None

is_active: bool

last_prompt_id: int | None

model_config = {'from_attributes': True, 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

name: str

prolog: str | None

prompt: Mapped[Prompt | None]

raw_content: str | None

seealso: str | None

target_path: str

class lbn.aidoc.models.Source

Bases: SQLModel

__init__(**data)

domain: str

driver: str

extras: Dict[str, Any]

classmethod get_by_url(session: Session, url: str) → Source

The Discovery Drill: Resolves Source by domain or falls back to ‘General’. PRINCIPAL: Guaranteed to return a Source object.

id: int | None

jobs: Mapped[List[Job]]

model_config = {'from_attributes': True, 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

name: str

password: str | None

searches: Dict[str, Any]

url: str | None

username: str | None

class lbn.aidoc.models.Tag

Bases: SQLModel

Surgical Metadata: Dynamic labels with built-in style.

__init__(**data)

as_str()

id: int | None

model_config = {'from_attributes': True, 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

name: str

style: str

class lbn.aidoc.models.TagLink

Bases: SQLModel

The Glue: Polymorphic Many-to-Many link with a standard ID.

__init__(**data)

id: int | None

model_config = {'from_attributes': True, 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

parent_id: int

parent_type: str

tag_id: int

class lbn.aidoc.models.Task

Bases: BaseEntity

TODO/Task List

__init__(**data)

as_todoist_payload() → dict

Surgical Export: Generates a clean Todoist payload schema mapping. Translates our StyledEnum priorities into Todoist’s 1-4 numeric envelope.

blocked_by_id: int | None

blocker_reason: str | None

created_at: datetime

description: str | None

due_date: datetime | None

history: Mapped[List[History]]

id

is_blocked

model_config = {'from_attributes': True, 'ignored_types': (<class 'sqlalchemy.ext.hybrid.hybrid_property'>,), 'read_from_attributes': True, 'read_with_orm_mode': True, 'registry': PydanticUndefined, 'table': True}

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

parent_id: int

parent_type: str

priority: Priority

state: TaskStatus

title: str

updated_at: datetime

classmethod validate_styled_enums(v, info)

The ‘Inhale Handshake’: Forces Pydantic to use the StyledEnum constructor. This re-attaches the .style metadata lost during database serialization.

lbn.aidoc.models.find_company(session: Session, search_name: str)

lbn.aidoc.models.find_person(session: Session, name: str = None, email: str = None)

The ‘Identity Drill’: Resolves Person via name/email or Name Alias. Surgically strikes only existing hardware columns. No redundant imports.

lbn.aidoc.models.get_comments_for(session, item)

Retrieves all narrative notes for the parent item.

lbn.aidoc.models.get_tags_for(session, item) → List[Tag]

Surgical Retrieval: Reaches into the TagLink silo for any arbitrary model.

lbn.aidoc.models.make_comment(item, content, category='note', timestamp=None)

Surgically injects a narrative comment against any hunting stakeholder model. Forces parent types to match get_comments_for extraction queries.