Files
govoplan-organizations/src/govoplan_organizations/backend/directory.py

915 lines
32 KiB
Python

from __future__ import annotations
from collections import deque
from collections.abc import Callable, Iterator, Sequence
from contextlib import contextmanager
from dataclasses import dataclass
from datetime import datetime, timezone
from sqlalchemy import or_
from sqlalchemy.orm import Session
from govoplan_core.core.organizations import (
OrganizationDirectory,
OrganizationFunctionRef,
OrganizationFunctionTypeRef,
OrganizationFunctionTypeResolution,
OrganizationHierarchyDirection,
OrganizationHierarchyDirectory,
OrganizationHierarchyEdgeRef,
OrganizationHierarchyMatchRef,
OrganizationHierarchyPathResolution,
OrganizationHierarchyResolution,
OrganizationRelationTypeRef,
OrganizationResolutionStatus,
OrganizationStructureRef,
OrganizationUnitRef,
OrganizationUnitTypeRef,
OrganizationUnitTypeResolution,
)
from govoplan_core.db.session import get_database
from govoplan_organizations.backend.db.models import (
OrganizationFunction,
OrganizationFunctionType,
OrganizationRelation,
OrganizationRelationType,
OrganizationStructure,
OrganizationUnit,
OrganizationUnitType,
)
MAX_DIRECTORY_BATCH = 500
MAX_HIERARCHY_DEPTH = 100
def _status(active: bool) -> str:
return "active" if active else "inactive"
def _unit_ref(item: OrganizationUnit) -> OrganizationUnitRef:
return OrganizationUnitRef(
id=item.id,
tenant_id=item.tenant_id,
slug=item.slug,
name=item.name,
unit_type_id=item.unit_type_id,
parent_id=item.parent_id,
description=item.description,
status=_status(item.is_active), # type: ignore[arg-type]
)
def _unit_type_ref(item: OrganizationUnitType) -> OrganizationUnitTypeRef:
return OrganizationUnitTypeRef(
id=item.id,
tenant_id=item.tenant_id,
slug=item.slug,
name=item.name,
description=item.description,
status=_status(item.is_active), # type: ignore[arg-type]
)
def _function_ref(item: OrganizationFunction) -> OrganizationFunctionRef:
return OrganizationFunctionRef(
id=item.id,
tenant_id=item.tenant_id,
organization_unit_id=item.organization_unit_id,
slug=item.slug,
name=item.name,
function_type_id=item.function_type_id,
description=item.description,
delegable=item.delegable,
act_in_place_allowed=item.act_in_place_allowed,
status=_status(item.is_active), # type: ignore[arg-type]
)
def _function_type_ref(
item: OrganizationFunctionType,
) -> OrganizationFunctionTypeRef:
return OrganizationFunctionTypeRef(
id=item.id,
tenant_id=item.tenant_id,
slug=item.slug,
name=item.name,
organization_unit_type_id=item.organization_unit_type_id,
description=item.description,
delegable=item.delegable,
act_in_place_allowed=item.act_in_place_allowed,
status=_status(item.is_active), # type: ignore[arg-type]
)
def _structure_ref(item: OrganizationStructure) -> OrganizationStructureRef:
return OrganizationStructureRef(
id=item.id,
tenant_id=item.tenant_id,
slug=item.slug,
name=item.name,
structure_kind=item.structure_kind,
description=item.description,
status=_status(item.is_active), # type: ignore[arg-type]
)
def _relation_type_ref(
item: OrganizationRelationType,
) -> OrganizationRelationTypeRef:
return OrganizationRelationTypeRef(
id=item.id,
tenant_id=item.tenant_id,
slug=item.slug,
name=item.name,
structure_id=item.structure_id,
source_unit_type_id=item.source_unit_type_id,
target_unit_type_id=item.target_unit_type_id,
is_hierarchical=item.is_hierarchical,
allow_cycles=item.allow_cycles,
description=item.description,
status=_status(item.is_active), # type: ignore[arg-type]
)
def _unique_ids(values: Sequence[str], *, label: str) -> tuple[str, ...]:
result = tuple(dict.fromkeys(str(value).strip() for value in values))
if any(not value for value in result):
raise ValueError(f"{label} must not contain empty IDs.")
if len(result) > MAX_DIRECTORY_BATCH:
raise ValueError(
f"{label} is limited to {MAX_DIRECTORY_BATCH} entries."
)
return result
def _validated_direction(
value: OrganizationHierarchyDirection,
) -> OrganizationHierarchyDirection:
if value not in {"ancestors", "descendants"}:
raise ValueError(
"Hierarchy direction must be ancestors or descendants."
)
return value
def _validated_depth(value: int) -> int:
depth = int(value)
if depth < 1 or depth > MAX_HIERARCHY_DEPTH:
raise ValueError(
f"Hierarchy depth must be between 1 and {MAX_HIERARCHY_DEPTH}."
)
return depth
@dataclass(slots=True)
class _HierarchyGraph:
status: OrganizationResolutionStatus
structure_id: str
structure: OrganizationStructureRef | None
relation_type_ids: tuple[str, ...]
units: dict[str, OrganizationUnitRef]
outgoing: dict[str, tuple[OrganizationHierarchyEdgeRef, ...]]
incoming: dict[str, tuple[OrganizationHierarchyEdgeRef, ...]]
diagnostics: tuple[str, ...]
class SqlOrganizationDirectory(
OrganizationDirectory,
OrganizationHierarchyDirectory,
):
def __init__(
self,
*,
session_factory: Callable[[], Session] | None = None,
) -> None:
self._session_factory = session_factory
@contextmanager
def _session(self) -> Iterator[Session]:
if self._session_factory is None:
with get_database().session() as session:
yield session
return
session = self._session_factory()
try:
yield session
finally:
session.close()
def get_organization_unit(
self,
organization_unit_id: str,
) -> OrganizationUnitRef | None:
with self._session() as session:
item = session.get(OrganizationUnit, organization_unit_id)
return _unit_ref(item) if item is not None else None
def organization_units_for_tenant(
self,
tenant_id: str,
) -> tuple[OrganizationUnitRef, ...]:
with self._session() as session:
rows = (
session.query(OrganizationUnit)
.filter(OrganizationUnit.tenant_id == tenant_id)
.order_by(OrganizationUnit.name.asc(), OrganizationUnit.id)
.all()
)
return tuple(_unit_ref(item) for item in rows)
def get_function(
self,
function_id: str,
) -> OrganizationFunctionRef | None:
with self._session() as session:
item = session.get(OrganizationFunction, function_id)
return _function_ref(item) if item is not None else None
def functions_for_organization_unit(
self,
organization_unit_id: str,
*,
include_subunits: bool = False,
) -> tuple[OrganizationFunctionRef, ...]:
with self._session() as session:
unit = session.get(OrganizationUnit, organization_unit_id)
if unit is None:
return ()
unit_ids = {organization_unit_id}
if include_subunits:
pending = [organization_unit_id]
while pending:
parent_id = pending.pop()
child_ids = [
row[0]
for row in session.query(OrganizationUnit.id)
.filter(
OrganizationUnit.tenant_id == unit.tenant_id,
OrganizationUnit.parent_id == parent_id,
)
.all()
]
for child_id in child_ids:
if child_id not in unit_ids:
unit_ids.add(child_id)
pending.append(child_id)
rows = (
session.query(OrganizationFunction)
.filter(
OrganizationFunction.tenant_id == unit.tenant_id,
OrganizationFunction.organization_unit_id.in_(unit_ids),
)
.order_by(
OrganizationFunction.name.asc(),
OrganizationFunction.id,
)
.all()
)
return tuple(_function_ref(item) for item in rows)
def get_unit_type(
self,
tenant_id: str,
unit_type_id: str,
) -> OrganizationUnitTypeRef | None:
with self._session() as session:
item = session.get(OrganizationUnitType, unit_type_id)
if item is None or item.tenant_id != tenant_id:
return None
return _unit_type_ref(item)
def get_function_type(
self,
tenant_id: str,
function_type_id: str,
) -> OrganizationFunctionTypeRef | None:
with self._session() as session:
item = session.get(OrganizationFunctionType, function_type_id)
if item is None or item.tenant_id != tenant_id:
return None
return _function_type_ref(item)
def resolve_functions_by_type(
self,
tenant_id: str,
function_type_id: str,
*,
organization_unit_ids: Sequence[str] = (),
) -> OrganizationFunctionTypeResolution:
requested_ids = _unique_ids(
organization_unit_ids,
label="Organization unit scope",
)
with self._session() as session:
function_type = session.get(
OrganizationFunctionType,
function_type_id,
)
if (
function_type is None
or function_type.tenant_id != tenant_id
):
return OrganizationFunctionTypeResolution(
tenant_id=tenant_id,
function_type_id=function_type_id,
requested_unit_ids=requested_ids,
status="missing",
diagnostics=("function_type_missing",),
)
units: dict[str, OrganizationUnit] = {}
if requested_ids:
units = {
item.id: item
for item in session.query(OrganizationUnit)
.filter(
OrganizationUnit.tenant_id == tenant_id,
OrganizationUnit.id.in_(requested_ids),
)
.all()
}
query = session.query(OrganizationFunction).filter(
OrganizationFunction.tenant_id == tenant_id,
OrganizationFunction.function_type_id == function_type_id,
)
if requested_ids:
query = query.filter(
OrganizationFunction.organization_unit_id.in_(
tuple(units)
)
)
rows = query.order_by(
OrganizationFunction.organization_unit_id,
OrganizationFunction.name,
OrganizationFunction.id,
).all()
relevant_unit_ids = (
requested_ids
if requested_ids
else tuple(
dict.fromkeys(
item.organization_unit_id for item in rows
)
)
)
if not requested_ids and relevant_unit_ids:
units = {
item.id: item
for item in session.query(OrganizationUnit)
.filter(
OrganizationUnit.tenant_id == tenant_id,
OrganizationUnit.id.in_(relevant_unit_ids),
)
.all()
}
missing = tuple(
item_id
for item_id in relevant_unit_ids
if item_id not in units
)
inactive = tuple(
item_id
for item_id in relevant_unit_ids
if item_id in units and not units[item_id].is_active
)
diagnostics = []
if missing:
diagnostics.append("organization_units_missing")
if inactive:
diagnostics.append("organization_units_inactive")
if not rows:
diagnostics.append("no_matching_functions")
return OrganizationFunctionTypeResolution(
tenant_id=tenant_id,
function_type_id=function_type_id,
requested_unit_ids=requested_ids,
status=(
"active" if function_type.is_active else "inactive"
),
function_type=_function_type_ref(function_type),
matches=tuple(_function_ref(item) for item in rows),
missing_unit_ids=missing,
inactive_unit_ids=inactive,
diagnostics=tuple(diagnostics),
)
def resolve_units_by_type(
self,
tenant_id: str,
unit_type_id: str,
*,
structure_id: str | None = None,
root_unit_id: str | None = None,
relation_type_ids: Sequence[str] = (),
direction: OrganizationHierarchyDirection = "descendants",
max_depth: int = 10,
) -> OrganizationUnitTypeResolution:
relation_ids = _unique_ids(
relation_type_ids,
label="Relation type filter",
)
direction = _validated_direction(direction)
depth = _validated_depth(max_depth)
if bool(structure_id) != bool(root_unit_id):
return OrganizationUnitTypeResolution(
tenant_id=tenant_id,
unit_type_id=unit_type_id,
status="invalid",
structure_id=structure_id,
root_unit_id=root_unit_id,
diagnostics=(
"structure_and_root_must_be_supplied_together",
),
)
with self._session() as session:
unit_type = session.get(OrganizationUnitType, unit_type_id)
if unit_type is None or unit_type.tenant_id != tenant_id:
return OrganizationUnitTypeResolution(
tenant_id=tenant_id,
unit_type_id=unit_type_id,
status="missing",
structure_id=structure_id,
root_unit_id=root_unit_id,
diagnostics=("unit_type_missing",),
)
rows = (
session.query(OrganizationUnit)
.filter(
OrganizationUnit.tenant_id == tenant_id,
OrganizationUnit.unit_type_id == unit_type_id,
)
.order_by(OrganizationUnit.name, OrganizationUnit.id)
.all()
)
diagnostics: tuple[str, ...] = ()
status: OrganizationResolutionStatus = (
"active" if unit_type.is_active else "inactive"
)
if structure_id and root_unit_id:
hierarchy = self.resolve_hierarchy_relatives(
tenant_id,
(root_unit_id,),
structure_id=structure_id,
relation_type_ids=relation_ids,
direction=direction,
max_depth=depth,
)[0]
scope_ids = {
root_unit_id,
*(match.unit.id for match in hierarchy.matches),
}
rows = [item for item in rows if item.id in scope_ids]
diagnostics = hierarchy.diagnostics
if hierarchy.status in {"missing", "invalid"}:
status = hierarchy.status
elif hierarchy.status == "inactive" and status == "active":
status = "inactive"
return OrganizationUnitTypeResolution(
tenant_id=tenant_id,
unit_type_id=unit_type_id,
status=status,
unit_type=_unit_type_ref(unit_type),
structure_id=structure_id,
root_unit_id=root_unit_id,
matches=tuple(_unit_ref(item) for item in rows),
diagnostics=diagnostics,
)
def resolve_hierarchy_relatives(
self,
tenant_id: str,
organization_unit_ids: Sequence[str],
*,
structure_id: str,
relation_type_ids: Sequence[str] = (),
direction: OrganizationHierarchyDirection = "ancestors",
max_depth: int = 10,
) -> tuple[OrganizationHierarchyResolution, ...]:
root_ids = _unique_ids(
organization_unit_ids,
label="Organization hierarchy roots",
)
direction = _validated_direction(direction)
depth = _validated_depth(max_depth)
relation_ids = _unique_ids(
relation_type_ids,
label="Relation type filter",
)
with self._session() as session:
graph = self._hierarchy_graph(
session,
tenant_id=tenant_id,
structure_id=structure_id,
relation_type_ids=relation_ids,
required_unit_ids=root_ids,
)
return tuple(
self._relative_resolution(
graph,
tenant_id=tenant_id,
root_unit_id=root_id,
direction=direction,
max_depth=depth,
)
for root_id in root_ids
)
def resolve_hierarchy_paths(
self,
tenant_id: str,
unit_pairs: Sequence[tuple[str, str]],
*,
structure_id: str,
relation_type_ids: Sequence[str] = (),
direction: OrganizationHierarchyDirection = "descendants",
max_depth: int = 10,
) -> tuple[OrganizationHierarchyPathResolution, ...]:
if len(unit_pairs) > MAX_DIRECTORY_BATCH:
raise ValueError(
f"Organization path batches are limited to "
f"{MAX_DIRECTORY_BATCH} entries."
)
pairs = tuple(
(str(source).strip(), str(target).strip())
for source, target in unit_pairs
)
if any(not source or not target for source, target in pairs):
raise ValueError("Organization paths require source and target IDs.")
direction = _validated_direction(direction)
depth = _validated_depth(max_depth)
relation_ids = _unique_ids(
relation_type_ids,
label="Relation type filter",
)
required_ids = tuple(
dict.fromkeys(
item
for pair in pairs
for item in pair
)
)
with self._session() as session:
graph = self._hierarchy_graph(
session,
tenant_id=tenant_id,
structure_id=structure_id,
relation_type_ids=relation_ids,
required_unit_ids=required_ids,
)
cache: dict[str, OrganizationHierarchyResolution] = {}
results: list[OrganizationHierarchyPathResolution] = []
for source_id, target_id in pairs:
if source_id not in cache:
cache[source_id] = self._relative_resolution(
graph,
tenant_id=tenant_id,
root_unit_id=source_id,
direction=direction,
max_depth=depth,
)
relative = cache[source_id]
target = graph.units.get(target_id)
match = next(
(
item
for item in relative.matches
if item.unit.id == target_id
),
None,
)
if source_id == target_id and relative.root is not None:
status: OrganizationResolutionStatus = (
"active"
if relative.root.status == "active"
else "inactive"
)
path: tuple[OrganizationHierarchyEdgeRef, ...] = ()
elif relative.root is None or target is None:
status = "missing"
path = ()
elif match is None:
status = (
"invalid"
if relative.status == "invalid"
else "unreachable"
)
path = ()
else:
status = (
"active"
if (
relative.root.status == "active"
and target.status == "active"
)
else "inactive"
)
path = match.path
results.append(
OrganizationHierarchyPathResolution(
tenant_id=tenant_id,
source_unit_id=source_id,
target_unit_id=target_id,
direction=direction,
structure_id=structure_id,
relation_type_ids=graph.relation_type_ids,
max_depth=depth,
status=status,
source=relative.root,
target=target,
path=path,
cycle_detected=relative.cycle_detected,
depth_limited=relative.depth_limited,
diagnostics=relative.diagnostics,
)
)
return tuple(results)
def _hierarchy_graph(
self,
session: Session,
*,
tenant_id: str,
structure_id: str,
relation_type_ids: tuple[str, ...],
required_unit_ids: tuple[str, ...],
) -> _HierarchyGraph:
structure = session.get(OrganizationStructure, structure_id)
if structure is None or structure.tenant_id != tenant_id:
return _HierarchyGraph(
status="missing",
structure_id=structure_id,
structure=None,
relation_type_ids=relation_type_ids,
units={},
outgoing={},
incoming={},
diagnostics=("structure_missing",),
)
relation_query = session.query(OrganizationRelationType).filter(
OrganizationRelationType.tenant_id == tenant_id,
OrganizationRelationType.is_hierarchical.is_(True),
or_(
OrganizationRelationType.structure_id.is_(None),
OrganizationRelationType.structure_id == structure_id,
),
)
if relation_type_ids:
relation_query = relation_query.filter(
OrganizationRelationType.id.in_(relation_type_ids)
)
relation_types = relation_query.all()
relation_type_by_id = {
item.id: item
for item in relation_types
if item.is_active
}
diagnostics: list[str] = []
invalid_relation_filter = False
if relation_type_ids:
missing_relation_ids = tuple(
item
for item in relation_type_ids
if item not in {row.id for row in relation_types}
)
inactive_relation_ids = tuple(
item.id for item in relation_types if not item.is_active
)
if missing_relation_ids:
diagnostics.append("relation_types_missing_or_invalid")
invalid_relation_filter = True
if inactive_relation_ids:
diagnostics.append("relation_types_inactive")
invalid_relation_filter = True
selected_ids = tuple(sorted(relation_type_by_id))
now = datetime.now(timezone.utc)
relations = []
if selected_ids:
relations = (
session.query(OrganizationRelation)
.filter(
OrganizationRelation.tenant_id == tenant_id,
OrganizationRelation.structure_id == structure_id,
OrganizationRelation.relation_type_id.in_(selected_ids),
OrganizationRelation.is_active.is_(True),
)
.order_by(OrganizationRelation.id)
.all()
)
relations = [
item
for item in relations
if (
(
item.valid_from is None
or _utc_datetime(item.valid_from) <= now
)
and (
item.valid_until is None
or _utc_datetime(item.valid_until) > now
)
)
]
unit_ids = {
*required_unit_ids,
*(item.source_unit_id for item in relations),
*(item.target_unit_id for item in relations),
}
units = {
item.id: _unit_ref(item)
for item in session.query(OrganizationUnit)
.filter(
OrganizationUnit.tenant_id == tenant_id,
OrganizationUnit.id.in_(unit_ids),
)
.all()
} if unit_ids else {}
structure_ref = _structure_ref(structure)
outgoing_lists: dict[
str,
list[OrganizationHierarchyEdgeRef],
] = {}
incoming_lists: dict[
str,
list[OrganizationHierarchyEdgeRef],
] = {}
for relation in relations:
relation_type = relation_type_by_id[relation.relation_type_id]
if (
relation.source_unit_id not in units
or relation.target_unit_id not in units
):
diagnostics.append(
f"relation_unit_missing:{relation.id}"
)
continue
edge = OrganizationHierarchyEdgeRef(
id=relation.id,
tenant_id=tenant_id,
structure=structure_ref,
relation_type=_relation_type_ref(relation_type),
source_unit_id=relation.source_unit_id,
target_unit_id=relation.target_unit_id,
valid_from=relation.valid_from,
valid_until=relation.valid_until,
)
outgoing_lists.setdefault(edge.source_unit_id, []).append(edge)
incoming_lists.setdefault(edge.target_unit_id, []).append(edge)
return _HierarchyGraph(
status=(
"inactive"
if not structure.is_active
else "invalid"
if invalid_relation_filter
else "active"
),
structure_id=structure_id,
structure=structure_ref,
relation_type_ids=selected_ids,
units=units,
outgoing={
key: tuple(sorted(value, key=lambda item: item.id))
for key, value in outgoing_lists.items()
},
incoming={
key: tuple(sorted(value, key=lambda item: item.id))
for key, value in incoming_lists.items()
},
diagnostics=tuple(dict.fromkeys(diagnostics)),
)
def _relative_resolution(
self,
graph: _HierarchyGraph,
*,
tenant_id: str,
root_unit_id: str,
direction: OrganizationHierarchyDirection,
max_depth: int,
) -> OrganizationHierarchyResolution:
root = graph.units.get(root_unit_id)
if root is None:
return OrganizationHierarchyResolution(
tenant_id=tenant_id,
root_unit_id=root_unit_id,
direction=direction,
structure_id=(
graph.structure_id
),
relation_type_ids=graph.relation_type_ids,
max_depth=max_depth,
status="missing",
diagnostics=tuple(
dict.fromkeys(
(*graph.diagnostics, "root_unit_missing")
)
),
)
if graph.status in {"missing", "invalid"}:
return OrganizationHierarchyResolution(
tenant_id=tenant_id,
root_unit_id=root_unit_id,
direction=direction,
structure_id=(
graph.structure_id
),
relation_type_ids=graph.relation_type_ids,
max_depth=max_depth,
status=graph.status,
root=root,
diagnostics=graph.diagnostics,
)
adjacency = (
graph.incoming
if direction == "ancestors"
else graph.outgoing
)
pending = deque(
[(root_unit_id, 0, (), frozenset({root_unit_id}))]
)
best_depth = {root_unit_id: 0}
matches: dict[str, OrganizationHierarchyMatchRef] = {}
diagnostics = list(graph.diagnostics)
cycle_detected = False
depth_limited = False
while pending:
current_id, current_depth, path, path_units = pending.popleft()
edges = adjacency.get(current_id, ())
if current_depth >= max_depth:
if edges:
depth_limited = True
continue
for edge in edges:
next_id = (
edge.source_unit_id
if direction == "ancestors"
else edge.target_unit_id
)
if next_id in path_units:
cycle_detected = True
diagnostics.append(f"cycle_detected:{edge.id}")
continue
unit = graph.units.get(next_id)
if unit is None:
diagnostics.append(f"unit_missing:{next_id}")
continue
next_depth = current_depth + 1
previous_depth = best_depth.get(next_id)
if (
previous_depth is not None
and previous_depth <= next_depth
):
continue
next_path = (*path, edge)
best_depth[next_id] = next_depth
matches[next_id] = OrganizationHierarchyMatchRef(
unit=unit,
depth=next_depth,
path=next_path,
)
pending.append(
(
next_id,
next_depth,
next_path,
path_units | {next_id},
)
)
if depth_limited:
diagnostics.append("maximum_depth_reached")
return OrganizationHierarchyResolution(
tenant_id=tenant_id,
root_unit_id=root_unit_id,
direction=direction,
structure_id=graph.structure_id,
relation_type_ids=graph.relation_type_ids,
max_depth=max_depth,
status=(
"active"
if root.status == "active" and graph.status == "active"
else "inactive"
),
root=root,
matches=tuple(
sorted(
matches.values(),
key=lambda item: (
item.depth,
item.unit.name.casefold(),
item.unit.id,
),
)
),
cycle_detected=cycle_detected,
depth_limited=depth_limited,
diagnostics=tuple(dict.fromkeys(diagnostics)),
)
def _utc_datetime(value: datetime) -> datetime:
if value.tzinfo is None:
return value.replace(tzinfo=timezone.utc)
return value.astimezone(timezone.utc)
__all__ = ["SqlOrganizationDirectory"]