feat: classify sensitive columns locally
This commit is contained in:
@@ -35,10 +35,10 @@ import {
|
||||
type DescriptionGenerationRun,
|
||||
type DescriptionGenerationRunUpdate,
|
||||
type DescriptionGenerationScope,
|
||||
type SensitiveDataSuggestionEvent,
|
||||
type SensitiveDataSuggestionRun,
|
||||
type SensitiveDataSuggestionRunUpdate,
|
||||
type SensitiveDataSuggestionScope,
|
||||
type SensitivityAnalysisEvent,
|
||||
type SensitivityAnalysisRun,
|
||||
type SensitivityAnalysisRunUpdate,
|
||||
type SensitivityAnalysisScope,
|
||||
type TableSyncRepositoryResult,
|
||||
type WorkspaceDatabase,
|
||||
} from "./types.js";
|
||||
@@ -56,8 +56,8 @@ export class MemoryCatalogRepository implements CatalogRepository {
|
||||
private readonly logicalRelationships = new Map<string, CatalogLogicalRelationship>();
|
||||
private readonly descriptionGenerationRuns = new Map<string, DescriptionGenerationRun>();
|
||||
private readonly descriptionGenerationEvents = new Map<string, DescriptionGenerationEvent[]>();
|
||||
private readonly sensitiveDataSuggestionRuns = new Map<string, SensitiveDataSuggestionRun>();
|
||||
private readonly sensitiveDataSuggestionEvents = new Map<string, SensitiveDataSuggestionEvent[]>();
|
||||
private readonly sensitivityAnalysisRuns = new Map<string, SensitivityAnalysisRun>();
|
||||
private readonly sensitivityAnalysisEvents = new Map<string, SensitivityAnalysisEvent[]>();
|
||||
private readonly syncRuns = new Map<string, CatalogSyncRun>();
|
||||
private readonly syncEvents = new Map<string, CatalogSyncEvent[]>();
|
||||
|
||||
@@ -183,10 +183,10 @@ export class MemoryCatalogRepository implements CatalogRepository {
|
||||
this.descriptionGenerationRuns.delete(runId);
|
||||
this.descriptionGenerationEvents.delete(runId);
|
||||
}
|
||||
for (const [runId, run] of this.sensitiveDataSuggestionRuns) {
|
||||
for (const [runId, run] of this.sensitivityAnalysisRuns) {
|
||||
if (run.databaseId !== id) continue;
|
||||
this.sensitiveDataSuggestionRuns.delete(runId);
|
||||
this.sensitiveDataSuggestionEvents.delete(runId);
|
||||
this.sensitivityAnalysisRuns.delete(runId);
|
||||
this.sensitivityAnalysisEvents.delete(runId);
|
||||
}
|
||||
return this.records.delete(id);
|
||||
}
|
||||
@@ -433,93 +433,96 @@ export class MemoryCatalogRepository implements CatalogRepository {
|
||||
.map((event) => structuredClone(event));
|
||||
}
|
||||
|
||||
async createSensitiveDataSuggestionRun(
|
||||
async createSensitivityAnalysisRun(
|
||||
databaseId: string,
|
||||
scope: SensitiveDataSuggestionScope,
|
||||
modelId: string,
|
||||
): Promise<SensitiveDataSuggestionRun> {
|
||||
scope: SensitivityAnalysisScope,
|
||||
origin: { engine: "llm"; modelId: string } | { engine: "local"; policyVersion: string },
|
||||
): Promise<SensitivityAnalysisRun> {
|
||||
const now = new Date().toISOString();
|
||||
const run: SensitiveDataSuggestionRun = {
|
||||
const run: SensitivityAnalysisRun = {
|
||||
id: randomUUID(),
|
||||
databaseId,
|
||||
scope,
|
||||
modelId,
|
||||
engine: origin.engine,
|
||||
modelId: origin.engine === "llm" ? origin.modelId : null,
|
||||
policyVersion: origin.engine === "local" ? origin.policyVersion : null,
|
||||
status: "running",
|
||||
total: 0,
|
||||
suggestedSensitive: 0,
|
||||
suggestedNonSensitive: 0,
|
||||
inputTokens: 0,
|
||||
cacheReadTokens: 0,
|
||||
outputTokens: 0,
|
||||
suggestedNonSensitive: 0,
|
||||
unknown: 0,
|
||||
inputTokens: 0,
|
||||
cacheReadTokens: 0,
|
||||
outputTokens: 0,
|
||||
createdAt: now,
|
||||
startedAt: now,
|
||||
updatedAt: now,
|
||||
finishedAt: null,
|
||||
errorSummary: null,
|
||||
};
|
||||
this.sensitiveDataSuggestionRuns.set(run.id, run);
|
||||
this.sensitivityAnalysisRuns.set(run.id, run);
|
||||
return structuredClone(run);
|
||||
}
|
||||
|
||||
async getSensitiveDataSuggestionRun(
|
||||
async getSensitivityAnalysisRun(
|
||||
runId: string,
|
||||
): Promise<SensitiveDataSuggestionRun | undefined> {
|
||||
const run = this.sensitiveDataSuggestionRuns.get(runId);
|
||||
): Promise<SensitivityAnalysisRun | undefined> {
|
||||
const run = this.sensitivityAnalysisRuns.get(runId);
|
||||
return run ? structuredClone(run) : undefined;
|
||||
}
|
||||
|
||||
async listSensitiveDataSuggestionRuns(limit = 50): Promise<SensitiveDataSuggestionRun[]> {
|
||||
return [...this.sensitiveDataSuggestionRuns.values()]
|
||||
async listSensitivityAnalysisRuns(limit = 50): Promise<SensitivityAnalysisRun[]> {
|
||||
return [...this.sensitivityAnalysisRuns.values()]
|
||||
.sort((a, b) => b.createdAt.localeCompare(a.createdAt) || b.id.localeCompare(a.id))
|
||||
.slice(0, limit)
|
||||
.map((run) => structuredClone(run));
|
||||
}
|
||||
|
||||
async interruptActiveSensitiveDataSuggestionRuns(
|
||||
async interruptActiveSensitivityAnalysisRuns(
|
||||
errorSummary: string,
|
||||
): Promise<SensitiveDataSuggestionRun[]> {
|
||||
const interrupted: SensitiveDataSuggestionRun[] = [];
|
||||
for (const run of this.sensitiveDataSuggestionRuns.values()) {
|
||||
): Promise<SensitivityAnalysisRun[]> {
|
||||
const interrupted: SensitivityAnalysisRun[] = [];
|
||||
for (const run of this.sensitivityAnalysisRuns.values()) {
|
||||
if (run.status !== "running") continue;
|
||||
const now = new Date().toISOString();
|
||||
const updated: SensitiveDataSuggestionRun = {
|
||||
const updated: SensitivityAnalysisRun = {
|
||||
...run,
|
||||
status: "interrupted",
|
||||
updatedAt: now,
|
||||
finishedAt: now,
|
||||
errorSummary,
|
||||
};
|
||||
this.sensitiveDataSuggestionRuns.set(run.id, updated);
|
||||
this.sensitivityAnalysisRuns.set(run.id, updated);
|
||||
interrupted.push(structuredClone(updated));
|
||||
}
|
||||
return interrupted;
|
||||
}
|
||||
|
||||
async updateSensitiveDataSuggestionRun(
|
||||
async updateSensitivityAnalysisRun(
|
||||
runId: string,
|
||||
update: SensitiveDataSuggestionRunUpdate,
|
||||
): Promise<SensitiveDataSuggestionRun | undefined> {
|
||||
const current = this.sensitiveDataSuggestionRuns.get(runId);
|
||||
update: SensitivityAnalysisRunUpdate,
|
||||
): Promise<SensitivityAnalysisRun | undefined> {
|
||||
const current = this.sensitivityAnalysisRuns.get(runId);
|
||||
if (!current) return undefined;
|
||||
const updated = {
|
||||
...current,
|
||||
...structuredClone(update),
|
||||
updatedAt: new Date().toISOString(),
|
||||
};
|
||||
this.sensitiveDataSuggestionRuns.set(runId, updated);
|
||||
this.sensitivityAnalysisRuns.set(runId, updated);
|
||||
return structuredClone(updated);
|
||||
}
|
||||
|
||||
async appendSensitiveDataSuggestionEvent(
|
||||
async appendSensitivityAnalysisEvent(
|
||||
runId: string,
|
||||
level: SensitiveDataSuggestionEvent["level"],
|
||||
level: SensitivityAnalysisEvent["level"],
|
||||
message: string,
|
||||
): Promise<SensitiveDataSuggestionEvent> {
|
||||
if (!this.sensitiveDataSuggestionRuns.has(runId)) {
|
||||
throw new CatalogConflictError("Sensitive Data Suggestion Run does not exist");
|
||||
): Promise<SensitivityAnalysisEvent> {
|
||||
if (!this.sensitivityAnalysisRuns.has(runId)) {
|
||||
throw new CatalogConflictError("Sensitivity Analysis Run does not exist");
|
||||
}
|
||||
const events = this.sensitiveDataSuggestionEvents.get(runId) ?? [];
|
||||
const event: SensitiveDataSuggestionEvent = {
|
||||
const events = this.sensitivityAnalysisEvents.get(runId) ?? [];
|
||||
const event: SensitivityAnalysisEvent = {
|
||||
runId,
|
||||
sequence: events.length + 1,
|
||||
level,
|
||||
@@ -527,15 +530,15 @@ export class MemoryCatalogRepository implements CatalogRepository {
|
||||
createdAt: new Date().toISOString(),
|
||||
};
|
||||
events.push(event);
|
||||
this.sensitiveDataSuggestionEvents.set(runId, events);
|
||||
this.sensitivityAnalysisEvents.set(runId, events);
|
||||
return structuredClone(event);
|
||||
}
|
||||
|
||||
async listSensitiveDataSuggestionEvents(
|
||||
async listSensitivityAnalysisEvents(
|
||||
runId: string,
|
||||
afterSequence = 0,
|
||||
): Promise<SensitiveDataSuggestionEvent[]> {
|
||||
return (this.sensitiveDataSuggestionEvents.get(runId) ?? [])
|
||||
): Promise<SensitivityAnalysisEvent[]> {
|
||||
return (this.sensitivityAnalysisEvents.get(runId) ?? [])
|
||||
.filter((event) => event.sequence > afterSequence)
|
||||
.map((event) => structuredClone(event));
|
||||
}
|
||||
|
||||
Reference in New Issue
Block a user