ThothII/backend/src/catalog/logical-relationship-service.ts

import type {
  CatalogLogicalRelationship,
  CatalogLogicalRelationshipCandidate,
  CatalogLogicalRelationshipContext,
  CatalogLogicalRelationshipEndpoint,
  CatalogRelationship,
  CatalogRepository,
} from "./types.js";

export class LogicalRelationshipDatabaseNotFoundError extends Error {}
export class LogicalRelationshipDuplicateError extends Error {}
export class LogicalRelationshipNotFoundError extends Error {}
export class LogicalRelationshipReadOnlyError extends Error {}
export class LogicalRelationshipSchemaStaleError extends Error {}
export class LogicalRelationshipTargetNotUniqueError extends Error {}
export class LogicalRelationshipTypeIncompatibleError extends Error {}
export class LogicalRelationshipColumnNotFoundError extends Error {
  constructor(readonly field: "sourceColumnId" | "targetColumnId") {
    super(`Catalog column '${field}' was not found`);
  }
}

export interface RebuildGeneratedRelationshipsResult {
  added: number;
  alreadyPresent: number;
  excluded: number;
  ambiguous: number;
}

function identifierTokens(value: string): string[] {
  return value
    .replace(/([a-z0-9])([A-Z])/g, "$1_$2")
    .toLowerCase()
    .split(/[^a-z0-9]+/)
    .filter(Boolean);
}

function singularWord(value: string): string {
  if (value.length > 4 && value.endsWith("ies")) return `${value.slice(0, -3)}y`;
  if (value.length > 4 && /(ches|shes|xes|zes|ses)$/.test(value)) return value.slice(0, -2);
  if (value.length > 3 && value.endsWith("s") && !/(ss|us)$/.test(value)) return value.slice(0, -1);
  return value;
}

function tableAliases(tableName: string): string[] {
  const tokens = identifierTokens(tableName);
  if (tokens.length === 0) return [];
  const normalized = tokens.join("_");
  const singular = [...tokens];
  singular[singular.length - 1] = singularWord(singular[singular.length - 1]);
  return [...new Set([normalized, singular.join("_")])];
}

const GENERIC_PRIMARY_KEY_NAMES = new Set(["id", "key", "code", "pk"]);

function nameMatches(
  source: CatalogLogicalRelationshipEndpoint,
  target: CatalogLogicalRelationshipEndpoint,
): boolean {
  const sourceName = identifierTokens(source.columnName).join("_");
  const targetName = identifierTokens(target.columnName).join("_");
  if (!sourceName || !targetName) return false;
  const expected = new Set<string>();
  if (!GENERIC_PRIMARY_KEY_NAMES.has(targetName)) expected.add(targetName);
  for (const alias of tableAliases(target.tableName)) {
    expected.add(`${alias}_${targetName}`);
    expected.add(`${alias.replaceAll("_", "")}${targetName.replaceAll("_", "")}`);
    if (targetName === "id" || targetName === "pk") expected.add(alias);
  }
  return expected.has(sourceName);
}

function canonicalDataType(value: string): string {
  const normalized = value.trim().toLowerCase().replace(/\s+/g, " ");
  const arraySuffix = normalized.endsWith("[]") ? "[]" : "";
  const base = arraySuffix ? normalized.slice(0, -2) : normalized;
  const withoutModifier = base.replace(/\([^)]*\)/g, "").trim();
  const aliases: Record<string, string> = {
    int2: "smallint",
    smallserial: "smallint",
    int4: "integer",
    int: "integer",
    serial: "integer",
    int8: "bigint",
    bigserial: "bigint",
    decimal: "numeric",
    varchar: "text",
    "character varying": "text",
    bool: "boolean",
    "timestamp without time zone": "timestamp",
    "timestamp with time zone": "timestamptz",
    "time without time zone": "time",
    "time with time zone": "timetz",
  };
  return `${aliases[withoutModifier] ?? withoutModifier}${arraySuffix}`;
}

function typesCompatible(left: string, right: string): boolean {
  return canonicalDataType(left) === canonicalDataType(right);
}

function pairKey(sourceColumnId: string, targetColumnId: string): string {
  return `${sourceColumnId}\u0000${targetColumnId}`;
}

function relationshipPair(relationship: CatalogLogicalRelationship): CatalogLogicalRelationshipCandidate {
  return {
    sourceColumnId: relationship.columns[0].sourceColumnId,
    targetColumnId: relationship.columns[0].targetColumnId,
  };
}

function relationshipSortKey(relationship: CatalogRelationship): string {
  const sourceColumns = relationship.columns.map((column) => column.sourceColumnName).join(",");
  const targetColumns = relationship.columns.map((column) => column.targetColumnName).join(",");
  return [
    relationship.sourceTableName,
    sourceColumns,
    relationship.targetTableName,
    targetColumns,
    relationship.origin,
  ].join("\u0000");
}

export class CatalogLogicalRelationshipService {
  constructor(private readonly repository: CatalogRepository) {}

  async list(databaseId: string): Promise<CatalogRelationship[]> {
    if (!(await this.repository.get(databaseId))) throw new LogicalRelationshipDatabaseNotFoundError();
    const relationships: CatalogRelationship[] = [
      ...await this.repository.listRelationships(databaseId),
      ...await this.repository.listLogicalRelationships(databaseId),
    ];
    return relationships.sort((left, right) => relationshipSortKey(left).localeCompare(relationshipSortKey(right)));
  }

  async addManual(
    databaseId: string,
    sourceColumnId: string,
    targetColumnId: string,
  ): Promise<CatalogLogicalRelationship> {
    const context = await this.requiredContext(databaseId);
    const source = context.endpoints.find((endpoint) => endpoint.columnId === sourceColumnId);
    if (!source) throw new LogicalRelationshipColumnNotFoundError("sourceColumnId");
    const target = context.endpoints.find((endpoint) => endpoint.columnId === targetColumnId);
    if (!target) throw new LogicalRelationshipColumnNotFoundError("targetColumnId");
    if (sourceColumnId === targetColumnId
      || target.primaryKeyPosition === null
      || target.tablePrimaryKeyColumnCount !== 1) {
      throw new LogicalRelationshipTargetNotUniqueError();
    }
    if (!typesCompatible(source.dataType, target.dataType)) {
      throw new LogicalRelationshipTypeIncompatibleError();
    }
    const key = pairKey(sourceColumnId, targetColumnId);
    if (context.physicalPairs.some((pair) => pairKey(pair.sourceColumnId, pair.targetColumnId) === key)
      || context.logicalRelationships.some((relationship) => {
        const pair = relationshipPair(relationship);
        return pairKey(pair.sourceColumnId, pair.targetColumnId) === key;
      })) throw new LogicalRelationshipDuplicateError();
    const created = await this.repository.insertLogicalRelationship(
      databaseId,
      sourceColumnId,
      targetColumnId,
      false,
    );
    if (!created) throw new LogicalRelationshipDuplicateError();
    return created;
  }

  async rebuildGenerated(databaseId: string): Promise<RebuildGeneratedRelationshipsResult> {
    const context = await this.requiredContext(databaseId);
    const targets = context.endpoints.filter((endpoint) => (
      endpoint.primaryKeyPosition !== null && endpoint.tablePrimaryKeyColumnCount === 1
    ));
    const physical = new Set(context.physicalPairs.map((pair) => pairKey(pair.sourceColumnId, pair.targetColumnId)));
    const active = new Set<string>();
    const excluded = new Set<string>();
    for (const relationship of context.logicalRelationships) {
      const pair = relationshipPair(relationship);
      (relationship.status === "excluded" ? excluded : active)
        .add(pairKey(pair.sourceColumnId, pair.targetColumnId));
    }

    const pending: CatalogLogicalRelationshipCandidate[] = [];
    let alreadyPresent = 0;
    let excludedCount = 0;
    let ambiguous = 0;
    const dimTimeTargets = targets.filter((target) => (
      identifierTokens(target.tableName).join("_") === "dim_time"
    ));
    const sources = context.endpoints.filter((endpoint) => (
      endpoint.primaryKeyPosition === null || endpoint.tablePrimaryKeyColumnCount > 1
    ));
    for (const source of sources) {
      const sourceName = identifierTokens(source.columnName).join("_");
      const isTimeKey = sourceName.endsWith("time_key")
        && identifierTokens(source.tableName).join("_") !== "dim_time";
      const candidates = isTimeKey ? dimTimeTargets : targets;
      const matches = candidates.filter((target) => (
        target.columnId !== source.columnId
        && typesCompatible(source.dataType, target.dataType)
        && (isTimeKey || nameMatches(source, target))
      ));
      if (matches.length > 1) {
        ambiguous += 1;
        continue;
      }
      if (matches.length === 0) continue;
      const candidate = { sourceColumnId: source.columnId, targetColumnId: matches[0].columnId };
      const key = pairKey(candidate.sourceColumnId, candidate.targetColumnId);
      if (physical.has(key) || active.has(key)) {
        alreadyPresent += 1;
      } else if (excluded.has(key)) {
        excludedCount += 1;
      } else {
        pending.push(candidate);
      }
    }
    const added = await this.repository.insertGeneratedLogicalRelationships(databaseId, pending);
    alreadyPresent += pending.length - added;
    return { added, alreadyPresent, excluded: excludedCount, ambiguous };
  }

  async setStatus(
    databaseId: string,
    relationshipId: string,
    status: CatalogLogicalRelationship["status"],
  ): Promise<CatalogLogicalRelationship> {
    if (!(await this.repository.get(databaseId))) throw new LogicalRelationshipDatabaseNotFoundError();
    const updated = await this.repository.setLogicalRelationshipStatus(databaseId, relationshipId, status);
    if (updated) return updated;
    await this.assertNotPhysical(databaseId, relationshipId);
    throw new LogicalRelationshipNotFoundError();
  }

  async deletePermanently(databaseId: string, relationshipId: string): Promise<void> {
    if (!(await this.repository.get(databaseId))) throw new LogicalRelationshipDatabaseNotFoundError();
    if (await this.repository.deleteLogicalRelationship(databaseId, relationshipId)) return;
    await this.assertNotPhysical(databaseId, relationshipId);
    throw new LogicalRelationshipNotFoundError();
  }

  private async requiredContext(databaseId: string): Promise<CatalogLogicalRelationshipContext> {
    const database = await this.repository.get(databaseId);
    if (!database) throw new LogicalRelationshipDatabaseNotFoundError();
    if (database.schemaSyncedVersion !== database.version) {
      throw new LogicalRelationshipSchemaStaleError();
    }
    const context = await this.repository.getLogicalRelationshipContext(databaseId);
    if (!context) throw new LogicalRelationshipDatabaseNotFoundError();
    return context;
  }

  private async assertNotPhysical(databaseId: string, relationshipId: string): Promise<void> {
    if ((await this.repository.listRelationships(databaseId)).some((relationship) => relationship.id === relationshipId)) {
      throw new LogicalRelationshipReadOnlyError();
    }
  }
}