@transcend-io/cli 8.26.2 → 8.27.0

package/dist/{chunk-5HNRVLH2.cjs.map → chunk-NXWJFTTG.cjs.map}

@@ -1 +1 @@
-{"version":3,"sources":["/home/runner/work/cli/cli/dist/chunk-5HNRVLH2.cjs","../src/lib/data-inventory/pullAllDatapoints.ts","../src/lib/data-inventory/pullUnstructuredSubDataPointRecommendations.ts"],"names":["pullSubDatapoints","client","dataSiloIds","includeGuessedCategories","includeAttributes","parentCategories","subCategories","pageSize","subDataPoints","t0","progressBar","cliProgress","filterBy","SubDataPointDataSubCategoryGuessStatus","totalCount","makeGraphQLRequest","SUB_DATA_POINTS_COUNT","logger","colors","total","shouldContinue","cursor","offset","nodes","gql","err"],"mappings":"AAAA,quBAAqE,wDAAyC,wDAA8D,2DCKrK,qGACiB,iDACJ,gFACD,oCAWO,MAuE1B,SAAeA,CAAAA,CACbC,CAAAA,CACA,CACE,WAAA,CAAAC,CAAAA,CAAc,CAAC,CAAA,CACf,wBAAA,CAAAC,CAAAA,CACA,iBAAA,CAAAC,CAAAA,CACA,gBAAA,CAAAC,CAAAA,CAAmB,CAAC,CAAA,CACpB,aAAA,CAAAC,CAAAA,CAAgB,CAAC,CAAA,CACjB,QAAA,CAAAC,CAAAA,CAAW,GACb,CAAA,CAGI,CAAC,CAAA,CAC8B,CACnC,IAAMC,CAAAA,CAA0C,CAAC,CAAA,CAG3CC,CAAAA,CAAK,IAAI,IAAA,CAAK,CAAA,CAAE,OAAA,CAAQ,CAAA,CAGxBC,CAAAA,CAAc,IAAIC,qBAAAA,CAAY,SAAA,CAClC,CAAC,CAAA,CACDA,qBAAAA,CAAY,OAAA,CAAQ,cACtB,CAAA,CAGMC,CAAAA,CAAW,CACf,GAAIP,CAAAA,CAAiB,MAAA,CAAS,CAAA,CAAI,CAAE,QAAA,CAAUA,CAAiB,CAAA,CAAI,CAAC,CAAA,CACpE,GAAIC,CAAAA,CAAc,MAAA,CAAS,CAAA,CAAI,CAAE,cAAA,CAAgBA,CAAc,CAAA,CAAI,CAAC,CAAA,CAEpE,GAAID,CAAAA,CAAiB,MAAA,CAASC,CAAAA,CAAc,MAAA,CAAS,CAAA,EACrD,CAACH,CAAAA,CAEG,CAAE,MAAA,CAAQU,oDAAAA,CAAuC,QAAS,CAAA,CAC1D,CAAC,CAAA,CACL,GAAIX,CAAAA,CAAY,MAAA,CAAS,CAAA,CAAI,CAAE,SAAA,CAAWA,CAAY,CAAA,CAAI,CAAC,CAC7D,CAAA,CAGM,CACJ,aAAA,CAAe,CAAE,UAAA,CAAAY,CAAW,CAC9B,CAAA,CAAI,MAAMC,kCAAAA,CAMPd,CAAQe,mBAAAA,CAAuB,CAChC,QAAA,CAAAJ,CACF,CAAC,CAAA,CAEDK,mBAAAA,CAAO,IAAA,CAAKC,gBAAAA,CAAO,OAAA,CAAQ,yCAAyC,CAAC,CAAA,CAErER,CAAAA,CAAY,KAAA,CAAMI,CAAAA,CAAY,CAAC,CAAA,CAC/B,IAAIK,CAAAA,CAAQ,CAAA,CACRC,CAAAA,CAAiB,CAAA,CAAA,CACjBC,CAAAA,CACAC,CAAAA,CAAS,CAAA,CACb,GACE,GAAI,CACF,GAAM,CACJ,aAAA,CAAe,CAAE,KAAA,CAAAC,CAAM,CACzB,CAAA,CAAI,MAAMR,kCAAAA,CAORd,CACAuB,mBAAAA,CAAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,gBAAA,EA2BUrB,CAAAA,CACI,CAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,iBAAA,CAAA,CAQA,EACN,CAAA;AAAA,gBAAA,EAEEC,CAAAA,CACI,CAAA;AAAA;AAAA;AAAA;AAAA;AAAA,iBAAA,CAAA,CAMA,EACN,CAAA;AAAA;AAAA;AAAA;AAAA,QAAA,CAAA,CAKR,CACE,KAAA,CAAOG,CAAAA,CACP,MAAA,CAAAe,CAAAA,CACA,QAAA,CAAU,CACR,GAAGV,CAGL,CACF,CACF,CAAA,CAEAS,CAAAA,iBAASE,CAAAA,qBAAMA,CAAAA,CAAM,MAAA,CAAS,CAAC,CAAA,6BAAG,IAAA,CAClCf,CAAAA,CAAc,IAAA,CAAK,GAAGe,CAAK,CAAA,CAC3BH,CAAAA,CAAiBG,CAAAA,CAAM,MAAA,GAAWhB,CAAAA,CAClCY,CAAAA,EAASI,CAAAA,CAAM,MAAA,CACfD,CAAAA,EAAUC,CAAAA,CAAM,MAAA,CAChBb,CAAAA,CAAY,MAAA,CAAOS,CAAK,CAC1B,CAAA,KAAA,CAASM,CAAAA,CAAK,CACZ,MAAAR,mBAAAA,CAAO,KAAA,CACLC,gBAAAA,CAAO,GAAA,CACL,CAAA,2CAAA,EAA8CG,CAAM,CAAA,YAAA,EAAeC,CAAM,CAAA,CAAA;AC7G3E;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAiBgD,gBAAA;AACU,gBAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AA+BiB,QAAA","file":"/home/runner/work/cli/cli/dist/chunk-5HNRVLH2.cjs","sourcesContent":[null,"/* eslint-disable max-lines */\nimport { keyBy, uniq, chunk, sortBy } from 'lodash-es';\nimport {\n  type DataCategoryType,\n  SubDataPointDataSubCategoryGuessStatus,\n} from '@transcend-io/privacy-types';\nimport cliProgress from 'cli-progress';\nimport { gql } from 'graphql-request';\nimport colors from 'colors';\nimport type { GraphQLClient } from 'graphql-request';\nimport {\n  DATAPOINT_EXPORT,\n  DATA_SILO_EXPORT,\n  type DataSiloAttributeValue,\n  SUB_DATA_POINTS_COUNT,\n  makeGraphQLRequest,\n} from '../graphql';\nimport { logger } from '../../logger';\nimport type { DataCategoryInput, ProcessingPurposeInput } from '../../codecs';\nimport { mapSeries } from 'bluebird';\n\nexport interface DataSiloCsvPreview {\n  /** ID of dataSilo */\n  id: string;\n  /** Name of dataSilo */\n  title: string;\n}\n\nexport interface DataPointCsvPreview {\n  /** ID of dataPoint */\n  id: string;\n  /** The path to this data point */\n  path: string[];\n  /** Description */\n  description: {\n    /** Default message */\n    defaultMessage: string;\n  };\n  /** Name */\n  name: string;\n}\n\nexport interface SubDataPointCsvPreview {\n  /** ID of subDatapoint */\n  id: string;\n  /** Name (or key) of the subdatapoint */\n  name: string;\n  /** The description */\n  description?: string;\n  /** Personal data category */\n  categories: DataCategoryInput[];\n  /** Data point ID */\n  dataPointId: string;\n  /** The data silo ID */\n  dataSiloId: string;\n  /** The processing purpose for this sub datapoint */\n  purposes: ProcessingPurposeInput[];\n  /** Attribute attached to subdatapoint */\n  attributeValues?: DataSiloAttributeValue[];\n  /** Data category guesses that are output by the classifier */\n  pendingCategoryGuesses?: {\n    /** Data category being guessed */\n    category: DataCategoryInput;\n    /** Status of guess */\n    status: SubDataPointDataSubCategoryGuessStatus;\n    /** classifier version that produced the guess */\n    classifierVersion: number;\n  }[];\n}\n\nexport interface DatapointFilterOptions {\n  /** IDs of data silos to filter down */\n  dataSiloIds?: string[];\n  /** Whether to include guessed categories, defaults to only approved categories */\n  includeGuessedCategories?: boolean;\n  /** Whether or not to include attributes */\n  includeAttributes?: boolean;\n  /** Parent categories to filter down for */\n  parentCategories?: DataCategoryType[];\n  /** Sub categories to filter down for */\n  subCategories?: string[]; // TODO: https://transcend.height.app/T-40482 - do by name not ID\n}\n\n/**\n * Pull subdatapoint information\n *\n * @param client - Client to use for the request\n * @param options - Options\n * @returns The subdatapoints\n */\nasync function pullSubDatapoints(\n  client: GraphQLClient,\n  {\n    dataSiloIds = [],\n    includeGuessedCategories,\n    includeAttributes,\n    parentCategories = [],\n    subCategories = [],\n    pageSize = 1000,\n  }: DatapointFilterOptions & {\n    /** Page size to pull in */\n    pageSize?: number;\n  } = {},\n): Promise<SubDataPointCsvPreview[]> {\n  const subDataPoints: SubDataPointCsvPreview[] = [];\n\n  // Time duration\n  const t0 = new Date().getTime();\n\n  // create a new progress bar instance and use shades_classic theme\n  const progressBar = new cliProgress.SingleBar(\n    {},\n    cliProgress.Presets.shades_classic,\n  );\n\n  // Filters\n  const filterBy = {\n    ...(parentCategories.length > 0 ? { category: parentCategories } : {}),\n    ...(subCategories.length > 0 ? { subCategoryIds: subCategories } : {}),\n    // if parentCategories or subCategories and not includeGuessedCategories\n    ...(parentCategories.length + subCategories.length > 0 &&\n    !includeGuessedCategories\n      ? // then only show data points with approved data categories\n        { status: SubDataPointDataSubCategoryGuessStatus.Approved }\n      : {}),\n    ...(dataSiloIds.length > 0 ? { dataSilos: dataSiloIds } : {}),\n  };\n\n  // Build a GraphQL client\n  const {\n    subDataPoints: { totalCount },\n  } = await makeGraphQLRequest<{\n    /** Query response */\n    subDataPoints: {\n      /** Count */\n      totalCount: number;\n    };\n  }>(client, SUB_DATA_POINTS_COUNT, {\n    filterBy,\n  });\n\n  logger.info(colors.magenta('[Step 1/3] Pulling in all subdatapoints'));\n\n  progressBar.start(totalCount, 0);\n  let total = 0;\n  let shouldContinue = false;\n  let cursor: string | undefined;\n  let offset = 0;\n  do {\n    try {\n      const {\n        subDataPoints: { nodes },\n      } = await makeGraphQLRequest<{\n        /** Query response */\n        subDataPoints: {\n          /** List of matches */\n          nodes: SubDataPointCsvPreview[];\n        };\n      }>(\n        client,\n        gql`\n          query TranscendCliSubDataPointCsvExport(\n            $filterBy: SubDataPointFiltersInput\n            $first: Int!\n            $offset: Int!\n          ) {\n            subDataPoints(\n              filterBy: $filterBy\n              first: $first\n              offset: $offset\n              useMaster: false\n            ) {\n              nodes {\n                id\n                name\n                description\n                dataPointId\n                dataSiloId\n                purposes {\n                  name\n                  purpose\n                }\n                categories {\n                  name\n                  category\n                }\n                ${\n                  includeGuessedCategories\n                    ? `pendingCategoryGuesses {\n                  category {\n                    name\n                    category\n                  }\n                  status\n                  classifierVersion\n                }`\n                    : ''\n                }\n                ${\n                  includeAttributes\n                    ? `attributeValues {\n                  attributeKey {\n                    name\n                  }\n                  name\n                }`\n                    : ''\n                }\n              }\n            }\n          }\n        `,\n        {\n          first: pageSize,\n          offset,\n          filterBy: {\n            ...filterBy,\n            // TODO: https://transcend.height.app/T-40484 - add cursor support\n            // ...(cursor ? { cursor: { id: cursor } } : {}),\n          },\n        },\n      );\n\n      cursor = nodes[nodes.length - 1]?.id as string;\n      subDataPoints.push(...nodes);\n      shouldContinue = nodes.length === pageSize;\n      total += nodes.length;\n      offset += nodes.length;\n      progressBar.update(total);\n    } catch (err) {\n      logger.error(\n        colors.red(\n          `An error fetching subdatapoints for cursor ${cursor} and offset ${offset}`,\n        ),\n      );\n      throw err;\n    }\n  } while (shouldContinue);\n\n  progressBar.stop();\n  const t1 = new Date().getTime();\n  const totalTime = t1 - t0;\n\n  const sorted = sortBy(subDataPoints, 'name');\n\n  logger.info(\n    colors.green(\n      `Successfully pulled in ${sorted.length} subdatapoints in ${\n        totalTime / 1000\n      } seconds!`,\n    ),\n  );\n  return sorted;\n}\n\n/**\n * Pull datapoint information\n *\n * @param client - Client to use for the request\n * @param options - Options\n * @returns The datapoints\n */\nasync function pullDatapoints(\n  client: GraphQLClient,\n  {\n    dataPointIds = [],\n    pageSize = 100,\n  }: {\n    /** IDs of data points to filter down */\n    dataPointIds: string[];\n    /** Page size to pull in */\n    pageSize?: number;\n  },\n): Promise<DataPointCsvPreview[]> {\n  const dataPoints: DataPointCsvPreview[] = [];\n\n  // Time duration\n  const t0 = new Date().getTime();\n\n  // create a new progress bar instance and use shades_classic theme\n  const progressBar = new cliProgress.SingleBar(\n    {},\n    cliProgress.Presets.shades_classic,\n  );\n\n  logger.info(\n    colors.magenta(\n      `[Step 2/3] Fetching metadata for ${dataPointIds.length} datapoints`,\n    ),\n  );\n\n  // Group by 100\n  const dataPointsGrouped = chunk(dataPointIds, pageSize);\n\n  progressBar.start(dataPointIds.length, 0);\n  let total = 0;\n  await mapSeries(dataPointsGrouped, async (dataPointIdsGroup) => {\n    try {\n      const {\n        dataPoints: { nodes },\n      } = await makeGraphQLRequest<{\n        /** Query response */\n        dataPoints: {\n          /** List of matches */\n          nodes: DataPointCsvPreview[];\n        };\n      }>(client, DATAPOINT_EXPORT, {\n        first: pageSize,\n        filterBy: {\n          ids: dataPointIdsGroup,\n        },\n      });\n\n      dataPoints.push(...nodes);\n      total += dataPointIdsGroup.length;\n      progressBar.update(total);\n    } catch (err) {\n      logger.error(\n        colors.red(\n          `An error fetching subdatapoints for IDs ${dataPointIdsGroup.join(\n            ', ',\n          )}`,\n        ),\n      );\n      throw err;\n    }\n  });\n\n  progressBar.stop();\n  const t1 = new Date().getTime();\n  const totalTime = t1 - t0;\n\n  logger.info(\n    colors.green(\n      `Successfully pulled in ${dataPoints.length} dataPoints in ${\n        totalTime / 1000\n      } seconds!`,\n    ),\n  );\n  return dataPoints;\n}\n\n/**\n * Pull data silo information\n *\n * @param client - Client to use for the request\n * @param options - Options\n * @returns The data silos\n */\nasync function pullDataSilos(\n  client: GraphQLClient,\n  {\n    dataSiloIds = [],\n    pageSize = 100,\n  }: {\n    /** IDs of data silos to filter down */\n    dataSiloIds: string[];\n    /** Page size to pull in */\n    pageSize?: number;\n  },\n): Promise<DataSiloCsvPreview[]> {\n  const dataSilos: DataSiloCsvPreview[] = [];\n\n  // Time duration\n  const t0 = new Date().getTime();\n\n  // create a new progress bar instance and use shades_classic theme\n  const progressBar = new cliProgress.SingleBar(\n    {},\n    cliProgress.Presets.shades_classic,\n  );\n\n  logger.info(\n    colors.magenta(\n      `[Step 3/3] Fetching metadata for ${dataSiloIds.length} data silos`,\n    ),\n  );\n\n  // Group by 100\n  const dataSilosGrouped = chunk(dataSiloIds, pageSize);\n\n  progressBar.start(dataSiloIds.length, 0);\n  let total = 0;\n  await mapSeries(dataSilosGrouped, async (dataSiloIdsGroup) => {\n    try {\n      const {\n        dataSilos: { nodes },\n      } = await makeGraphQLRequest<{\n        /** Query response */\n        dataSilos: {\n          /** List of matches */\n          nodes: DataSiloCsvPreview[];\n        };\n      }>(client, DATA_SILO_EXPORT, {\n        first: pageSize,\n        filterBy: {\n          ids: dataSiloIdsGroup,\n        },\n      });\n\n      dataSilos.push(...nodes);\n      total += dataSiloIdsGroup.length;\n      progressBar.update(total);\n    } catch (err) {\n      logger.error(\n        colors.red(\n          `An error fetching data silos for IDs ${dataSiloIdsGroup.join(', ')}`,\n        ),\n      );\n      throw err;\n    }\n  });\n\n  progressBar.stop();\n  const t1 = new Date().getTime();\n  const totalTime = t1 - t0;\n\n  logger.info(\n    colors.green(\n      `Successfully pulled in ${dataSilos.length} data silos in ${\n        totalTime / 1000\n      } seconds!`,\n    ),\n  );\n  return dataSilos;\n}\n\n/**\n * Pull all datapoints from the data inventory.\n *\n * @param client - Client to use for the request\n * @param options - Options\n * @returns The datapoints and data silos\n */\nexport async function pullAllDatapoints(\n  client: GraphQLClient,\n  {\n    dataSiloIds = [],\n    includeGuessedCategories,\n    includeAttributes,\n    parentCategories = [],\n    subCategories = [],\n    pageSize = 1000,\n  }: DatapointFilterOptions & {\n    /** Page size to pull in */\n    pageSize?: number;\n  } = {},\n): Promise<\n  (SubDataPointCsvPreview & {\n    /** Data point information */\n    dataPoint: DataPointCsvPreview;\n    /** Data silo information */\n    dataSilo: DataSiloCsvPreview;\n  })[]\n> {\n  // Subdatapoint information\n  const subDatapoints = await pullSubDatapoints(client, {\n    dataSiloIds,\n    includeGuessedCategories,\n    includeAttributes,\n    parentCategories,\n    subCategories,\n    pageSize,\n  });\n\n  // The datapoint ids to grab\n  const dataPointIds = uniq(subDatapoints.map((point) => point.dataPointId));\n  const dataPoints = await pullDatapoints(client, {\n    dataPointIds,\n  });\n  const dataPointById = keyBy(dataPoints, 'id');\n\n  // The data silo IDs to grab\n  const allDataSiloIds = uniq(subDatapoints.map((point) => point.dataSiloId));\n  const dataSilos = await pullDataSilos(client, {\n    dataSiloIds: allDataSiloIds,\n  });\n  const dataSiloById = keyBy(dataSilos, 'id');\n\n  return subDatapoints.map((subDataPoint) => ({\n    ...subDataPoint,\n    dataPoint: dataPointById[subDataPoint.dataPointId],\n    dataSilo: dataSiloById[subDataPoint.dataSiloId],\n  }));\n}\n/* eslint-enable max-lines */\n","import type { UnstructuredSubDataPointRecommendationStatus } from '@transcend-io/privacy-types';\nimport cliProgress from 'cli-progress';\nimport colors from 'colors';\nimport { gql, type GraphQLClient } from 'graphql-request';\nimport { sortBy } from 'lodash-es';\nimport type { DataCategoryInput } from '../../codecs';\nimport { ENTRY_COUNT, makeGraphQLRequest } from '../graphql';\nimport { logger } from '../../logger';\n\ninterface UnstructuredSubDataPointRecommendationCsvPreview {\n  /** ID of subDatapoint */\n  id: string;\n  /** Entry or Named Entity recognized by the classifier */\n  name: string;\n  /** Context snippet including entry */\n  contextSnippet: string;\n  /** Scanned object ID */\n  scannedObjectId: string;\n  /** Scanned object path ID */\n  scannedObjectPathId: string;\n  /** The data silo ID */\n  dataSiloId: string;\n  /** Personal data category */\n  dataSubCategory: DataCategoryInput;\n  /** Classification Status */\n  status: UnstructuredSubDataPointRecommendationStatus;\n  /** Confidence */\n  confidence: number;\n  /** Classification method */\n  classificationMethod: string;\n  /** Classifier version */\n  classifierVersion: string;\n}\n\ninterface EntryFilterOptions {\n  /** IDs of data silos to filter down */\n  dataSiloIds?: string[];\n  /** Parent categories to filter down for */\n  status?: UnstructuredSubDataPointRecommendationStatus[];\n  /** Sub categories to filter down for */\n  subCategories?: string[]; // TODO: https://transcend.height.app/T-40482 - do by name not ID\n  /** Include entry and snippet */\n  includeEncryptedSnippets?: boolean;\n  /** Include encryptedSamplesS3Key */\n  includeEncryptedSamplesS3Key?: boolean;\n}\n/**\n * Pull unstructured subdatapoint information\n *\n * @param client - Client to use for the request\n * @param options - Options\n * @param options.dataSiloIds - IDs of data silos to filter down\n * @param options.status - Parent categories to filter down for\n * @param options.subCategories - Sub categories to filter down for\n * @param options.includeEncryptedSnippets - Include entry and snippet\n * @param options.includeEncryptedSamplesS3Key - Include encryptedSamplesS3Key\n * @param options.pageSize - Page size to pull in\n * @returns A promise that resolves to an array of unstructured subdatapoint recommendations\n */\nexport async function pullUnstructuredSubDataPointRecommendations(\n  client: GraphQLClient,\n  {\n    dataSiloIds = [],\n    status,\n    subCategories = [],\n    includeEncryptedSnippets,\n    pageSize = 100,\n  }: EntryFilterOptions & {\n    /** Page size to pull in */\n    pageSize?: number;\n  } = {},\n): Promise<UnstructuredSubDataPointRecommendationCsvPreview[]> {\n  const unstructuredSubDataPointRecommendations: UnstructuredSubDataPointRecommendationCsvPreview[] =\n    [];\n\n  // Time duration\n  const t0 = new Date().getTime();\n\n  // create a new progress bar instance and use shades_classic theme\n  const progressBar = new cliProgress.SingleBar(\n    {},\n    cliProgress.Presets.shades_classic,\n  );\n\n  // Filters\n  const filterBy = {\n    ...(subCategories.length > 0 ? { subCategoryIds: subCategories } : {}),\n    ...(status ? { status } : {}),\n    ...(dataSiloIds.length > 0 ? { dataSilos: dataSiloIds } : {}),\n  };\n\n  // Build a GraphQL client\n  const {\n    unstructuredSubDataPointRecommendations: { totalCount },\n  } = await makeGraphQLRequest<{\n    /** Query response */\n    unstructuredSubDataPointRecommendations: {\n      /** Count */\n      totalCount: number;\n    };\n  }>(client, ENTRY_COUNT, {\n    filterBy,\n  });\n\n  logger.info(colors.magenta('[Step 1/3] Pulling in all subdatapoints'));\n\n  progressBar.start(totalCount, 0);\n  let total = 0;\n  let shouldContinue = false;\n  let cursor: string | undefined;\n  let offset = 0;\n  do {\n    try {\n      const {\n        unstructuredSubDataPointRecommendations: { nodes },\n      } = await makeGraphQLRequest<{\n        /** Query response */\n        unstructuredSubDataPointRecommendations: {\n          /** List of matches */\n          nodes: UnstructuredSubDataPointRecommendationCsvPreview[];\n        };\n      }>(\n        client,\n        gql`\n          query TranscendCliUnstructuredSubDataPointRecommendationCsvExport(\n            $filterBy: UnstructuredSubDataPointRecommendationsFilterInput\n            $first: Int!\n            $offset: Int!\n          ) {\n            unstructuredSubDataPointRecommendations(\n              filterBy: $filterBy\n              first: $first\n              offset: $offset\n              useMaster: false\n            ) {\n              nodes {\n                id\n                dataSiloId\n                scannedObjectPathId\n                scannedObjectId\n                ${includeEncryptedSnippets ? 'name' : ''}\n                ${includeEncryptedSnippets ? 'contextSnippet' : ''}\n                dataSubCategory {\n                  name\n                  category\n                }\n                status\n                confidence\n                classificationMethod\n                classifierVersion\n              }\n            }\n          }\n        `,\n        {\n          first: pageSize,\n          offset,\n          filterBy: {\n            ...filterBy,\n          },\n        },\n      );\n\n      cursor = nodes[nodes.length - 1]?.id as string;\n      unstructuredSubDataPointRecommendations.push(...nodes);\n      shouldContinue = nodes.length === pageSize;\n      total += nodes.length;\n      offset += nodes.length;\n      progressBar.update(total);\n    } catch (err) {\n      logger.error(\n        colors.red(\n          `An error fetching subdatapoints for cursor ${cursor} and offset ${offset}`,\n        ),\n      );\n      throw err;\n    }\n  } while (shouldContinue);\n\n  progressBar.stop();\n  const t1 = new Date().getTime();\n  const totalTime = t1 - t0;\n\n  const sorted = sortBy(unstructuredSubDataPointRecommendations, 'name');\n\n  logger.info(\n    colors.green(\n      `Successfully pulled in ${sorted.length} subdatapoints in ${\n        totalTime / 1000\n      } seconds!`,\n    ),\n  );\n  return sorted;\n}\n"]}
+{"version":3,"sources":["/home/runner/work/cli/cli/dist/chunk-NXWJFTTG.cjs","../src/lib/data-inventory/pullAllDatapoints.ts","../src/lib/data-inventory/pullUnstructuredSubDataPointRecommendations.ts"],"names":["pullSubDatapoints","client","dataSiloIds","includeGuessedCategories","includeAttributes","parentCategories","subCategories","pageSize","subDataPoints","t0","progressBar","cliProgress","filterBy","SubDataPointDataSubCategoryGuessStatus","totalCount","makeGraphQLRequest","SUB_DATA_POINTS_COUNT","logger","colors","total","shouldContinue","cursor","offset","nodes","gql","err"],"mappings":"AAAA,quBAAqE,wDAAyC,wDAA8D,2DCKrK,qGACiB,iDACJ,gFACD,oCAWO,MAuE1B,SAAeA,CAAAA,CACbC,CAAAA,CACA,CACE,WAAA,CAAAC,CAAAA,CAAc,CAAC,CAAA,CACf,wBAAA,CAAAC,CAAAA,CACA,iBAAA,CAAAC,CAAAA,CACA,gBAAA,CAAAC,CAAAA,CAAmB,CAAC,CAAA,CACpB,aAAA,CAAAC,CAAAA,CAAgB,CAAC,CAAA,CACjB,QAAA,CAAAC,CAAAA,CAAW,GACb,CAAA,CAGI,CAAC,CAAA,CAC8B,CACnC,IAAMC,CAAAA,CAA0C,CAAC,CAAA,CAG3CC,CAAAA,CAAK,IAAI,IAAA,CAAK,CAAA,CAAE,OAAA,CAAQ,CAAA,CAGxBC,CAAAA,CAAc,IAAIC,qBAAAA,CAAY,SAAA,CAClC,CAAC,CAAA,CACDA,qBAAAA,CAAY,OAAA,CAAQ,cACtB,CAAA,CAGMC,CAAAA,CAAW,CACf,GAAIP,CAAAA,CAAiB,MAAA,CAAS,CAAA,CAAI,CAAE,QAAA,CAAUA,CAAiB,CAAA,CAAI,CAAC,CAAA,CACpE,GAAIC,CAAAA,CAAc,MAAA,CAAS,CAAA,CAAI,CAAE,cAAA,CAAgBA,CAAc,CAAA,CAAI,CAAC,CAAA,CAEpE,GAAID,CAAAA,CAAiB,MAAA,CAASC,CAAAA,CAAc,MAAA,CAAS,CAAA,EACrD,CAACH,CAAAA,CAEG,CAAE,MAAA,CAAQU,oDAAAA,CAAuC,QAAS,CAAA,CAC1D,CAAC,CAAA,CACL,GAAIX,CAAAA,CAAY,MAAA,CAAS,CAAA,CAAI,CAAE,SAAA,CAAWA,CAAY,CAAA,CAAI,CAAC,CAC7D,CAAA,CAGM,CACJ,aAAA,CAAe,CAAE,UAAA,CAAAY,CAAW,CAC9B,CAAA,CAAI,MAAMC,kCAAAA,CAMPd,CAAQe,mBAAAA,CAAuB,CAChC,QAAA,CAAAJ,CACF,CAAC,CAAA,CAEDK,mBAAAA,CAAO,IAAA,CAAKC,gBAAAA,CAAO,OAAA,CAAQ,yCAAyC,CAAC,CAAA,CAErER,CAAAA,CAAY,KAAA,CAAMI,CAAAA,CAAY,CAAC,CAAA,CAC/B,IAAIK,CAAAA,CAAQ,CAAA,CACRC,CAAAA,CAAiB,CAAA,CAAA,CACjBC,CAAAA,CACAC,CAAAA,CAAS,CAAA,CACb,GACE,GAAI,CACF,GAAM,CACJ,aAAA,CAAe,CAAE,KAAA,CAAAC,CAAM,CACzB,CAAA,CAAI,MAAMR,kCAAAA,CAORd,CACAuB,mBAAAA,CAAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,gBAAA,EA2BUrB,CAAAA,CACI,CAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA,iBAAA,CAAA,CAQA,EACN,CAAA;AAAA,gBAAA,EAEEC,CAAAA,CACI,CAAA;AAAA;AAAA;AAAA;AAAA;AAAA,iBAAA,CAAA,CAMA,EACN,CAAA;AAAA;AAAA;AAAA;AAAA,QAAA,CAAA,CAKR,CACE,KAAA,CAAOG,CAAAA,CACP,MAAA,CAAAe,CAAAA,CACA,QAAA,CAAU,CACR,GAAGV,CAGL,CACF,CACF,CAAA,CAEAS,CAAAA,iBAASE,CAAAA,qBAAMA,CAAAA,CAAM,MAAA,CAAS,CAAC,CAAA,6BAAG,IAAA,CAClCf,CAAAA,CAAc,IAAA,CAAK,GAAGe,CAAK,CAAA,CAC3BH,CAAAA,CAAiBG,CAAAA,CAAM,MAAA,GAAWhB,CAAAA,CAClCY,CAAAA,EAASI,CAAAA,CAAM,MAAA,CACfD,CAAAA,EAAUC,CAAAA,CAAM,MAAA,CAChBb,CAAAA,CAAY,MAAA,CAAOS,CAAK,CAC1B,CAAA,KAAA,CAASM,CAAAA,CAAK,CACZ,MAAAR,mBAAAA,CAAO,KAAA,CACLC,gBAAAA,CAAO,GAAA,CACL,CAAA,2CAAA,EAA8CG,CAAM,CAAA,YAAA,EAAeC,CAAM,CAAA,CAAA;AC7G3E;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAiBgD,gBAAA;AACU,gBAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AAAA;AA+BiB,QAAA","file":"/home/runner/work/cli/cli/dist/chunk-NXWJFTTG.cjs","sourcesContent":[null,"/* eslint-disable max-lines */\nimport { keyBy, uniq, chunk, sortBy } from 'lodash-es';\nimport {\n  type DataCategoryType,\n  SubDataPointDataSubCategoryGuessStatus,\n} from '@transcend-io/privacy-types';\nimport cliProgress from 'cli-progress';\nimport { gql } from 'graphql-request';\nimport colors from 'colors';\nimport type { GraphQLClient } from 'graphql-request';\nimport {\n  DATAPOINT_EXPORT,\n  DATA_SILO_EXPORT,\n  type DataSiloAttributeValue,\n  SUB_DATA_POINTS_COUNT,\n  makeGraphQLRequest,\n} from '../graphql';\nimport { logger } from '../../logger';\nimport type { DataCategoryInput, ProcessingPurposeInput } from '../../codecs';\nimport { mapSeries } from 'bluebird';\n\nexport interface DataSiloCsvPreview {\n  /** ID of dataSilo */\n  id: string;\n  /** Name of dataSilo */\n  title: string;\n}\n\nexport interface DataPointCsvPreview {\n  /** ID of dataPoint */\n  id: string;\n  /** The path to this data point */\n  path: string[];\n  /** Description */\n  description: {\n    /** Default message */\n    defaultMessage: string;\n  };\n  /** Name */\n  name: string;\n}\n\nexport interface SubDataPointCsvPreview {\n  /** ID of subDatapoint */\n  id: string;\n  /** Name (or key) of the subdatapoint */\n  name: string;\n  /** The description */\n  description?: string;\n  /** Personal data category */\n  categories: DataCategoryInput[];\n  /** Data point ID */\n  dataPointId: string;\n  /** The data silo ID */\n  dataSiloId: string;\n  /** The processing purpose for this sub datapoint */\n  purposes: ProcessingPurposeInput[];\n  /** Attribute attached to subdatapoint */\n  attributeValues?: DataSiloAttributeValue[];\n  /** Data category guesses that are output by the classifier */\n  pendingCategoryGuesses?: {\n    /** Data category being guessed */\n    category: DataCategoryInput;\n    /** Status of guess */\n    status: SubDataPointDataSubCategoryGuessStatus;\n    /** classifier version that produced the guess */\n    classifierVersion: number;\n  }[];\n}\n\nexport interface DatapointFilterOptions {\n  /** IDs of data silos to filter down */\n  dataSiloIds?: string[];\n  /** Whether to include guessed categories, defaults to only approved categories */\n  includeGuessedCategories?: boolean;\n  /** Whether or not to include attributes */\n  includeAttributes?: boolean;\n  /** Parent categories to filter down for */\n  parentCategories?: DataCategoryType[];\n  /** Sub categories to filter down for */\n  subCategories?: string[]; // TODO: https://transcend.height.app/T-40482 - do by name not ID\n}\n\n/**\n * Pull subdatapoint information\n *\n * @param client - Client to use for the request\n * @param options - Options\n * @returns The subdatapoints\n */\nasync function pullSubDatapoints(\n  client: GraphQLClient,\n  {\n    dataSiloIds = [],\n    includeGuessedCategories,\n    includeAttributes,\n    parentCategories = [],\n    subCategories = [],\n    pageSize = 1000,\n  }: DatapointFilterOptions & {\n    /** Page size to pull in */\n    pageSize?: number;\n  } = {},\n): Promise<SubDataPointCsvPreview[]> {\n  const subDataPoints: SubDataPointCsvPreview[] = [];\n\n  // Time duration\n  const t0 = new Date().getTime();\n\n  // create a new progress bar instance and use shades_classic theme\n  const progressBar = new cliProgress.SingleBar(\n    {},\n    cliProgress.Presets.shades_classic,\n  );\n\n  // Filters\n  const filterBy = {\n    ...(parentCategories.length > 0 ? { category: parentCategories } : {}),\n    ...(subCategories.length > 0 ? { subCategoryIds: subCategories } : {}),\n    // if parentCategories or subCategories and not includeGuessedCategories\n    ...(parentCategories.length + subCategories.length > 0 &&\n    !includeGuessedCategories\n      ? // then only show data points with approved data categories\n        { status: SubDataPointDataSubCategoryGuessStatus.Approved }\n      : {}),\n    ...(dataSiloIds.length > 0 ? { dataSilos: dataSiloIds } : {}),\n  };\n\n  // Build a GraphQL client\n  const {\n    subDataPoints: { totalCount },\n  } = await makeGraphQLRequest<{\n    /** Query response */\n    subDataPoints: {\n      /** Count */\n      totalCount: number;\n    };\n  }>(client, SUB_DATA_POINTS_COUNT, {\n    filterBy,\n  });\n\n  logger.info(colors.magenta('[Step 1/3] Pulling in all subdatapoints'));\n\n  progressBar.start(totalCount, 0);\n  let total = 0;\n  let shouldContinue = false;\n  let cursor: string | undefined;\n  let offset = 0;\n  do {\n    try {\n      const {\n        subDataPoints: { nodes },\n      } = await makeGraphQLRequest<{\n        /** Query response */\n        subDataPoints: {\n          /** List of matches */\n          nodes: SubDataPointCsvPreview[];\n        };\n      }>(\n        client,\n        gql`\n          query TranscendCliSubDataPointCsvExport(\n            $filterBy: SubDataPointFiltersInput\n            $first: Int!\n            $offset: Int!\n          ) {\n            subDataPoints(\n              filterBy: $filterBy\n              first: $first\n              offset: $offset\n              useMaster: false\n            ) {\n              nodes {\n                id\n                name\n                description\n                dataPointId\n                dataSiloId\n                purposes {\n                  name\n                  purpose\n                }\n                categories {\n                  name\n                  category\n                }\n                ${\n                  includeGuessedCategories\n                    ? `pendingCategoryGuesses {\n                  category {\n                    name\n                    category\n                  }\n                  status\n                  classifierVersion\n                }`\n                    : ''\n                }\n                ${\n                  includeAttributes\n                    ? `attributeValues {\n                  attributeKey {\n                    name\n                  }\n                  name\n                }`\n                    : ''\n                }\n              }\n            }\n          }\n        `,\n        {\n          first: pageSize,\n          offset,\n          filterBy: {\n            ...filterBy,\n            // TODO: https://transcend.height.app/T-40484 - add cursor support\n            // ...(cursor ? { cursor: { id: cursor } } : {}),\n          },\n        },\n      );\n\n      cursor = nodes[nodes.length - 1]?.id as string;\n      subDataPoints.push(...nodes);\n      shouldContinue = nodes.length === pageSize;\n      total += nodes.length;\n      offset += nodes.length;\n      progressBar.update(total);\n    } catch (err) {\n      logger.error(\n        colors.red(\n          `An error fetching subdatapoints for cursor ${cursor} and offset ${offset}`,\n        ),\n      );\n      throw err;\n    }\n  } while (shouldContinue);\n\n  progressBar.stop();\n  const t1 = new Date().getTime();\n  const totalTime = t1 - t0;\n\n  const sorted = sortBy(subDataPoints, 'name');\n\n  logger.info(\n    colors.green(\n      `Successfully pulled in ${sorted.length} subdatapoints in ${\n        totalTime / 1000\n      } seconds!`,\n    ),\n  );\n  return sorted;\n}\n\n/**\n * Pull datapoint information\n *\n * @param client - Client to use for the request\n * @param options - Options\n * @returns The datapoints\n */\nasync function pullDatapoints(\n  client: GraphQLClient,\n  {\n    dataPointIds = [],\n    pageSize = 100,\n  }: {\n    /** IDs of data points to filter down */\n    dataPointIds: string[];\n    /** Page size to pull in */\n    pageSize?: number;\n  },\n): Promise<DataPointCsvPreview[]> {\n  const dataPoints: DataPointCsvPreview[] = [];\n\n  // Time duration\n  const t0 = new Date().getTime();\n\n  // create a new progress bar instance and use shades_classic theme\n  const progressBar = new cliProgress.SingleBar(\n    {},\n    cliProgress.Presets.shades_classic,\n  );\n\n  logger.info(\n    colors.magenta(\n      `[Step 2/3] Fetching metadata for ${dataPointIds.length} datapoints`,\n    ),\n  );\n\n  // Group by 100\n  const dataPointsGrouped = chunk(dataPointIds, pageSize);\n\n  progressBar.start(dataPointIds.length, 0);\n  let total = 0;\n  await mapSeries(dataPointsGrouped, async (dataPointIdsGroup) => {\n    try {\n      const {\n        dataPoints: { nodes },\n      } = await makeGraphQLRequest<{\n        /** Query response */\n        dataPoints: {\n          /** List of matches */\n          nodes: DataPointCsvPreview[];\n        };\n      }>(client, DATAPOINT_EXPORT, {\n        first: pageSize,\n        filterBy: {\n          ids: dataPointIdsGroup,\n        },\n      });\n\n      dataPoints.push(...nodes);\n      total += dataPointIdsGroup.length;\n      progressBar.update(total);\n    } catch (err) {\n      logger.error(\n        colors.red(\n          `An error fetching subdatapoints for IDs ${dataPointIdsGroup.join(\n            ', ',\n          )}`,\n        ),\n      );\n      throw err;\n    }\n  });\n\n  progressBar.stop();\n  const t1 = new Date().getTime();\n  const totalTime = t1 - t0;\n\n  logger.info(\n    colors.green(\n      `Successfully pulled in ${dataPoints.length} dataPoints in ${\n        totalTime / 1000\n      } seconds!`,\n    ),\n  );\n  return dataPoints;\n}\n\n/**\n * Pull data silo information\n *\n * @param client - Client to use for the request\n * @param options - Options\n * @returns The data silos\n */\nasync function pullDataSilos(\n  client: GraphQLClient,\n  {\n    dataSiloIds = [],\n    pageSize = 100,\n  }: {\n    /** IDs of data silos to filter down */\n    dataSiloIds: string[];\n    /** Page size to pull in */\n    pageSize?: number;\n  },\n): Promise<DataSiloCsvPreview[]> {\n  const dataSilos: DataSiloCsvPreview[] = [];\n\n  // Time duration\n  const t0 = new Date().getTime();\n\n  // create a new progress bar instance and use shades_classic theme\n  const progressBar = new cliProgress.SingleBar(\n    {},\n    cliProgress.Presets.shades_classic,\n  );\n\n  logger.info(\n    colors.magenta(\n      `[Step 3/3] Fetching metadata for ${dataSiloIds.length} data silos`,\n    ),\n  );\n\n  // Group by 100\n  const dataSilosGrouped = chunk(dataSiloIds, pageSize);\n\n  progressBar.start(dataSiloIds.length, 0);\n  let total = 0;\n  await mapSeries(dataSilosGrouped, async (dataSiloIdsGroup) => {\n    try {\n      const {\n        dataSilos: { nodes },\n      } = await makeGraphQLRequest<{\n        /** Query response */\n        dataSilos: {\n          /** List of matches */\n          nodes: DataSiloCsvPreview[];\n        };\n      }>(client, DATA_SILO_EXPORT, {\n        first: pageSize,\n        filterBy: {\n          ids: dataSiloIdsGroup,\n        },\n      });\n\n      dataSilos.push(...nodes);\n      total += dataSiloIdsGroup.length;\n      progressBar.update(total);\n    } catch (err) {\n      logger.error(\n        colors.red(\n          `An error fetching data silos for IDs ${dataSiloIdsGroup.join(', ')}`,\n        ),\n      );\n      throw err;\n    }\n  });\n\n  progressBar.stop();\n  const t1 = new Date().getTime();\n  const totalTime = t1 - t0;\n\n  logger.info(\n    colors.green(\n      `Successfully pulled in ${dataSilos.length} data silos in ${\n        totalTime / 1000\n      } seconds!`,\n    ),\n  );\n  return dataSilos;\n}\n\n/**\n * Pull all datapoints from the data inventory.\n *\n * @param client - Client to use for the request\n * @param options - Options\n * @returns The datapoints and data silos\n */\nexport async function pullAllDatapoints(\n  client: GraphQLClient,\n  {\n    dataSiloIds = [],\n    includeGuessedCategories,\n    includeAttributes,\n    parentCategories = [],\n    subCategories = [],\n    pageSize = 1000,\n  }: DatapointFilterOptions & {\n    /** Page size to pull in */\n    pageSize?: number;\n  } = {},\n): Promise<\n  (SubDataPointCsvPreview & {\n    /** Data point information */\n    dataPoint: DataPointCsvPreview;\n    /** Data silo information */\n    dataSilo: DataSiloCsvPreview;\n  })[]\n> {\n  // Subdatapoint information\n  const subDatapoints = await pullSubDatapoints(client, {\n    dataSiloIds,\n    includeGuessedCategories,\n    includeAttributes,\n    parentCategories,\n    subCategories,\n    pageSize,\n  });\n\n  // The datapoint ids to grab\n  const dataPointIds = uniq(subDatapoints.map((point) => point.dataPointId));\n  const dataPoints = await pullDatapoints(client, {\n    dataPointIds,\n  });\n  const dataPointById = keyBy(dataPoints, 'id');\n\n  // The data silo IDs to grab\n  const allDataSiloIds = uniq(subDatapoints.map((point) => point.dataSiloId));\n  const dataSilos = await pullDataSilos(client, {\n    dataSiloIds: allDataSiloIds,\n  });\n  const dataSiloById = keyBy(dataSilos, 'id');\n\n  return subDatapoints.map((subDataPoint) => ({\n    ...subDataPoint,\n    dataPoint: dataPointById[subDataPoint.dataPointId],\n    dataSilo: dataSiloById[subDataPoint.dataSiloId],\n  }));\n}\n/* eslint-enable max-lines */\n","import type { UnstructuredSubDataPointRecommendationStatus } from '@transcend-io/privacy-types';\nimport cliProgress from 'cli-progress';\nimport colors from 'colors';\nimport { gql, type GraphQLClient } from 'graphql-request';\nimport { sortBy } from 'lodash-es';\nimport type { DataCategoryInput } from '../../codecs';\nimport { ENTRY_COUNT, makeGraphQLRequest } from '../graphql';\nimport { logger } from '../../logger';\n\ninterface UnstructuredSubDataPointRecommendationCsvPreview {\n  /** ID of subDatapoint */\n  id: string;\n  /** Entry or Named Entity recognized by the classifier */\n  name: string;\n  /** Context snippet including entry */\n  contextSnippet: string;\n  /** Scanned object ID */\n  scannedObjectId: string;\n  /** Scanned object path ID */\n  scannedObjectPathId: string;\n  /** The data silo ID */\n  dataSiloId: string;\n  /** Personal data category */\n  dataSubCategory: DataCategoryInput;\n  /** Classification Status */\n  status: UnstructuredSubDataPointRecommendationStatus;\n  /** Confidence */\n  confidence: number;\n  /** Classification method */\n  classificationMethod: string;\n  /** Classifier version */\n  classifierVersion: string;\n}\n\ninterface EntryFilterOptions {\n  /** IDs of data silos to filter down */\n  dataSiloIds?: string[];\n  /** Parent categories to filter down for */\n  status?: UnstructuredSubDataPointRecommendationStatus[];\n  /** Sub categories to filter down for */\n  subCategories?: string[]; // TODO: https://transcend.height.app/T-40482 - do by name not ID\n  /** Include entry and snippet */\n  includeEncryptedSnippets?: boolean;\n  /** Include encryptedSamplesS3Key */\n  includeEncryptedSamplesS3Key?: boolean;\n}\n/**\n * Pull unstructured subdatapoint information\n *\n * @param client - Client to use for the request\n * @param options - Options\n * @param options.dataSiloIds - IDs of data silos to filter down\n * @param options.status - Parent categories to filter down for\n * @param options.subCategories - Sub categories to filter down for\n * @param options.includeEncryptedSnippets - Include entry and snippet\n * @param options.includeEncryptedSamplesS3Key - Include encryptedSamplesS3Key\n * @param options.pageSize - Page size to pull in\n * @returns A promise that resolves to an array of unstructured subdatapoint recommendations\n */\nexport async function pullUnstructuredSubDataPointRecommendations(\n  client: GraphQLClient,\n  {\n    dataSiloIds = [],\n    status,\n    subCategories = [],\n    includeEncryptedSnippets,\n    pageSize = 100,\n  }: EntryFilterOptions & {\n    /** Page size to pull in */\n    pageSize?: number;\n  } = {},\n): Promise<UnstructuredSubDataPointRecommendationCsvPreview[]> {\n  const unstructuredSubDataPointRecommendations: UnstructuredSubDataPointRecommendationCsvPreview[] =\n    [];\n\n  // Time duration\n  const t0 = new Date().getTime();\n\n  // create a new progress bar instance and use shades_classic theme\n  const progressBar = new cliProgress.SingleBar(\n    {},\n    cliProgress.Presets.shades_classic,\n  );\n\n  // Filters\n  const filterBy = {\n    ...(subCategories.length > 0 ? { subCategoryIds: subCategories } : {}),\n    ...(status ? { status } : {}),\n    ...(dataSiloIds.length > 0 ? { dataSilos: dataSiloIds } : {}),\n  };\n\n  // Build a GraphQL client\n  const {\n    unstructuredSubDataPointRecommendations: { totalCount },\n  } = await makeGraphQLRequest<{\n    /** Query response */\n    unstructuredSubDataPointRecommendations: {\n      /** Count */\n      totalCount: number;\n    };\n  }>(client, ENTRY_COUNT, {\n    filterBy,\n  });\n\n  logger.info(colors.magenta('[Step 1/3] Pulling in all subdatapoints'));\n\n  progressBar.start(totalCount, 0);\n  let total = 0;\n  let shouldContinue = false;\n  let cursor: string | undefined;\n  let offset = 0;\n  do {\n    try {\n      const {\n        unstructuredSubDataPointRecommendations: { nodes },\n      } = await makeGraphQLRequest<{\n        /** Query response */\n        unstructuredSubDataPointRecommendations: {\n          /** List of matches */\n          nodes: UnstructuredSubDataPointRecommendationCsvPreview[];\n        };\n      }>(\n        client,\n        gql`\n          query TranscendCliUnstructuredSubDataPointRecommendationCsvExport(\n            $filterBy: UnstructuredSubDataPointRecommendationsFilterInput\n            $first: Int!\n            $offset: Int!\n          ) {\n            unstructuredSubDataPointRecommendations(\n              filterBy: $filterBy\n              first: $first\n              offset: $offset\n              useMaster: false\n            ) {\n              nodes {\n                id\n                dataSiloId\n                scannedObjectPathId\n                scannedObjectId\n                ${includeEncryptedSnippets ? 'name' : ''}\n                ${includeEncryptedSnippets ? 'contextSnippet' : ''}\n                dataSubCategory {\n                  name\n                  category\n                }\n                status\n                confidence\n                classificationMethod\n                classifierVersion\n              }\n            }\n          }\n        `,\n        {\n          first: pageSize,\n          offset,\n          filterBy: {\n            ...filterBy,\n          },\n        },\n      );\n\n      cursor = nodes[nodes.length - 1]?.id as string;\n      unstructuredSubDataPointRecommendations.push(...nodes);\n      shouldContinue = nodes.length === pageSize;\n      total += nodes.length;\n      offset += nodes.length;\n      progressBar.update(total);\n    } catch (err) {\n      logger.error(\n        colors.red(\n          `An error fetching subdatapoints for cursor ${cursor} and offset ${offset}`,\n        ),\n      );\n      throw err;\n    }\n  } while (shouldContinue);\n\n  progressBar.stop();\n  const t1 = new Date().getTime();\n  const totalTime = t1 - t0;\n\n  const sorted = sortBy(unstructuredSubDataPointRecommendations, 'name');\n\n  logger.info(\n    colors.green(\n      `Successfully pulled in ${sorted.length} subdatapoints in ${\n        totalTime / 1000\n      } seconds!`,\n    ),\n  );\n  return sorted;\n}\n"]}