Metadata APIs

Use METADATA_BASE_URL=https://<host>/metadata and /base for system lookup. Locate systems and sources before searching objects and reading definitions, fields, and history. Paths below include service prefixes.

Core interfaces

Method and complete path	Purpose	Key inputs
POST /base/modelCategory/getModelCategories	Systems	No required body; array with categoryId, categoryName
POST /base/modelCategory/getModelCategoryById	System detail	Required Long query categoryId
POST /metadata/models/fromre/page	Harvested sources	Body Long categoryId, modelName, currentPage, pageSize; query Boolean showLogicalModel, default false
GET /metadata/models/{modelId}/schemas	Schemas	Required Long source ID; returns strings
POST /metadata/entities/searchMetadata	Objects	Body modelIds, schemas, types, keyword, currentPage, pageSize
GET /metadata/entities/{objectId}/get	Definition	Required Long object ID
GET /metadata/entities/{objectId}/columns	Table fields	Table ID; optional query keyword
GET /metadata/entities/{objectId}/history	Object revisions	Required Long object ID
POST /metadata/dataReport/search	Reports	Body keyword, String array path, currentPage, pageSize
GET /metadata/dataReport/{dataReportId}	Report and associations	Required Long report ID

Search criteria

modelIds is a Long array; schemas is a String array. Common types are TABLE, COLUMN, VIEW, STORED_PROCEDURE, FUNCTION. Files, APIs, and metamodels also use SHARE_FILE, API, META_MODEL, and typeIds. Follow actual metamodel types.

{
  "currentPage": 1,
  "pageSize": 20,
  "modelIds": [1001],
  "schemas": ["demo_sales"],
  "types": ["TABLE", "VIEW"],
  "keyword": "customer"
}

Replace fictional source and schema. Responses contain totalItems, currentPage, pageSize, and content. Records include id, name, type, definition, and path; use the returned id for detail lookup rather than a display name as a unique key.

Detail fields

Fields	Meaning
objectId, modelId, schema, typeId	Source, granularity, type
physicalName, logicalName, definition	Physical and business names, description
domains, standardCodes, tags	Applicable standards, reference codes, tags
owner, udps, properties	Ownership, custom and technical properties
Field type, notNull, defaultValue, ordinal	Data type, nullability, default, order

The schema spells precision percision; preserve that actual key. History includes version, createTime, changeHistory, and source details. It describes metadata changes, not database row change logs.

Select the query by source

Query reverse-engineered tables, fields, procedures, and functions as dataset objects; use actual types for metamodel instances. DDM dictionaries and DDL produce table/field metadata, not attachment records. Harvested file-server resources are file objects. BI and custom reports share report queries, with possible relatedTable, path, and url fields.

See Metadata Management. Metadata, catalog details, and business data samples are separate; /entities/data/sample is not a default integration endpoint here.

Synchronize a data dictionary

1. Map external systems to authorized categoryId and modelId values.
2. Page through sources, schemas, and types; read details and fields.
3. Store IDs, semantics, technical properties, standards, and synchronization time.
4. Compare history and snapshots. Without a suitable incremental cursor, use controlled scans and differences.
5. Before treating a missing object as retired, exclude permission changes, failed harvesting, and incomplete pagination.

Validate customer objects and record pagination, permissions, and errors.

Create, schedule, and execute harvesting tasks

Source creation and scheduling are separate. Confirm account access, plugins, and schema scope before execution.

Method and complete path	Purpose	Inputs
POST /metadata/jobs/createJob	Create harvest/sync task	DatablauJobDescriptor: name, typeName, boundResourceId, boundResourceType, runCycle, schedule, parameters, disabled, @class
POST /metadata/jobs/startJob	Start now	Long query name described below; String task parameter body
POST /metadata/jobs/stopJob	Stop	Deployed ID and parameter format
POST /metadata/jobs/enableJob / disableJob	Enable or disable schedule	Task ID
POST /metadata/jobs/rescheduleJob	Change schedule	Same query name; schedule body
POST /metadata/jobs/query/jobs/byDto	Find tasks	Name, type, state, resource, ID, cycle, creator, time
POST /metadata/jobs/query/jobResults/byCriteria	Results	Task, instance, state, time
POST /metadata/jobs/query/jobEvents/byCriteria	Events	Instance, event type, time
POST /metadata/simplejobs/status	Simple job state	Long query jobId

The source definition's Chinese query name is \u4efb\u52a1ID in Unicode escape notation. Decode it to the literal parameter name if required; confirm gateway handling. Do not translate wire identifiers.

Example task structure:

{
  "name": "metadata-sync-demo",
  "typeName": "META_DATA_SYNC",
  "boundResourceId": 1001,
  "boundResourceType": "MODEL",
  "runCycle": "CRON",
  "schedule": "0 0 2 * * ?",
  "parameters": {},
  "disabled": false,
  "@class": "com.datablau.job.dto.DatablauJobDescriptor"
}

Reconfirm @class, typeName, schedule, and ID parameter names after upgrades or gateway transformations. After start timeout, check instances before retrying.

Compare models with DDM

Use harvested database metadata as the baseline and a DDM design model as the target: reverse-engineer, edit in DDM, compare, review, and apply approved changes.

Method and complete path	Purpose
GET /metadata/models/{modelId}/compareMappings	Available mappings
GET /metadata/models/{modelId}/{modelVersionId}/compareResult/{mappingId}/{ddmModelId}/{ddmModelVersionId}	Versioned differences
GET /metadata/models/{modelId}/{modelVersionId}/compareResultFile/{mappingId}/{ddmModelId}/{ddmModelVersionId}	Export differences
GET /metadata/models/{modelId}/{mappingId}/compareResultHistory	Comparison history
POST /metadata/models/{modelId}/{modelVersionId}/updateCompareResultFile	Apply reviewed comparison file
POST /metadata/models/model/{baseModelId}/compare/{targetModelId}	Compare models
POST /metadata/models/model/{baseModelId}/compare/{targetModelId}/download	Download report

Comparison does not replace DDM approval and publication. Retain revisions, mappings, reports, and rollback snapshots.

Controlled updates and bulk maintenance

Method and complete path	Contract	Guidance
PUT /metadata/entities/{dataObjectId}/summary	Long ID; DataObjectDto body explicitly requires the complete modified object	Not a partial PATCH with only logicalName
POST /metadata/models/{modelId}/exportAsync	Query tableIds, String tagIds, isSimple; DataObjectExportDto body including schemas	String response requires confirmed job/download handling
POST /metadata/models/{modelId}/updateMetadata/{ignoreError}	Long model ID, Boolean ignoreError, binary file	Use same-environment export template; editable columns follow template

Read and snapshot the complete object, modify only approved fields, submit, and read back. writable is an indicator; server authorization remains decisive. Re-read before concurrent updates rather than testing overwrite behavior with empty or omitted fields.

For batches, export, enrich, import, and inspect outcomes. Retain IDs, columns, and failed rows. Validate partial-success behavior for ignoreError; it does not guarantee atomicity.

The current definition declares binary upload under application/json and only String responses for import/export. Confirm actual media type, task tracking, and downloads with the service owner. Until then, use UI bulk maintenance.