What Microsoft APIs can export from Purview and load into OneLake
The short answer
As of 2026-11-21, we found no Microsoft tool that moves a Purview glossary to the OneLake catalog. A move runs on Microsoft's APIs instead. Glossary terms can be exported, and Fabric's API can write tags. A tag holds a name only, so each term's definition, owner and parent need another home.
Why count the APIs
Microsoft announced on November 12, 2026 that Purview Data Governance is retiring: “Support for Purview Data Governance will end Nov. 12, 2027” (Microsoft Security blog, read 2026-11-21). The announcement describes no migration tooling. As of 2026-11-21, we found no Microsoft tool that moves Purview governance content to the OneLake catalog. Microsoft's transition guide (read 2026-11-21) gives a table instead: each Purview outcome, the Fabric target that covers it today, and the guidance for the move.
Without a tool, a move runs on the APIs: one API to read each kind of object out of Purview, and one API to write the object into the target Microsoft names. Microsoft publishes the specification of every one of those APIs. So this guide reads the specifications themselves and counts, for each kind of Purview object, the operations that can export it and the operations that can load its target in Fabric.
The result
| Purview object type | Purview API | Export operations | Target the transition guide names | Fabric API tags | Write operations |
|---|---|---|---|---|---|
| Business domains | Unified Catalog | 2 | Fabric domains and subdomains, workspace assignment, domain admins, owners | Domains | 11 |
| Glossary terms | Unified Catalog | 7 | Tags, descriptions, endorsement | Tags | 5 |
| Critical data elements | Unified Catalog | 6 | Tags, descriptions, endorsement | Tags | 5 |
| Data products | Unified Catalog | 6 | Domains, endorsement, sharing, discoverable semantic models | Domains | 11 |
| Objectives and key results | Unified Catalog | 7 | - | - | |
| Data access policies | Unified Catalog | 1 | Fabric Policies (preview); workspace, item, OneLake, and source permissions | PolicySets; PolicyRules | 6 |
| Data assets | Unified Catalog | 4 | OneLake catalog Explore, item details, filters, Catalog Search API | Items | 242 |
| Data columns | Unified Catalog | 3 | OneLake catalog Explore, item details, filters, Catalog Search API | - | - |
| Classic glossary | Data Map | 13 | Tags, descriptions, endorsement | Tags | 5 |
| Lineage | Data Map | 3 | Fabric lineage and impact analysis for supported items | - | - |
| Classification rules | Scanning | 3 | Purview sensitivity labels and DLP on supported Fabric items | - | - |
| Scan rule sets and scans | Scanning | 4 | Shortcuts, mirroring, external sharing, scanner/admin APIs | - | - |
The script sorted 1,124 operations in 66 specification files. All 12 Purview object types have an export operation. For 7 object types, the Fabric target that Microsoft's transition guide names has write operations in Fabric's API. For the other 5 object types (objectives and key results, data columns, lineage, classification rules, and scan rule sets and scans), our table names no Fabric API object that takes them as input. A script can read those 5 object types out of Purview but has no Fabric object to write them into.
- Write operations exist means the Fabric target Microsoft names can be written by API. A tag holds a name only, so a glossary term that becomes a tag loses its definition, owner and parent. The count says a load can be scripted; the count does not say the target holds everything the Purview object held. A tag name holds at most 40 characters (Fabric tag specification, read 2026-11-21), so longer term names need shortening. For data assets, the 242 Items operations create and change Fabric items. Those operations do not load a Purview record of a source outside Fabric.
- No Fabric API object to load into means the transition guide names no Fabric target for that object type, or names a target that does not take the Purview object as input. For scan rule sets and scans, the guide names shortcuts, mirroring and scanner APIs. Fabric's API can create shortcuts and start mirroring, but no operation takes a Purview scan rule set, so the table counts none. For lineage, none of the Fabric specifications has an operation whose path or name mentions lineage. For data columns, the guide names the same catalog target as data assets, but none of the Fabric specifications has an operation whose path or name mentions columns. For classification rules, the guide's answer is to keep classification in Purview.
- All 12 object types in the table have at least one export operation. An export operation does not mean every field reads out. The open question is where each object type lands.
Download landing.csv (the table above) and operations.csv (all 1,124 operations, each marked read or write).
Method
One Python script, count_api_operations.py, does every step:
- Read the specifications, pinned. The Purview Unified Catalog (version 2026-03-20-preview), Data Map and Scanning (both 2023-09-01) specifications come from Azure/azure-rest-api-specs at commit be93aca. Every Fabric specification comes from microsoft/fabric-rest-api-specs at commit ec85f05. Both repositories are MIT licensed. The script prints each file's size and SHA-256, so a reader can tell whether a re-run read the same files.
- Sort every operation. A GET reads. A POST reads when its name starts with List, Get, Query, Search, Count, Facets, Find, Read or Export, for example
Terms_Query. Every other operation writes. - Match object types to targets. The table in the script,
LANDING, gives each Purview object type, the API tags that hold it, the row of the transition guide it belongs to, the target that row names and the Fabric API tags that would write that target. That table is a judgment, written out so a reader can disagree with it; the rows are shown below. - Check the work. Every API tag in the table must exist in its specification, and the read and write counts must add up to every operation read. Either failure stops the script with SELF-CHECK FAILED, before any CSV is written, so a failed run never replaces earlier results.
| Purview object type | Transition-guide row | Fabric API tags counted |
|---|---|---|
| Business domains | Collections, domains, ownership | Domains |
| Glossary terms | Glossary, critical data elements, custom context | Tags |
| Critical data elements | Glossary, critical data elements, custom context | Tags |
| Data products | Data products and publishing | Domains |
| Objectives and key results | (not in the table) | - |
| Data access policies | Access policy and enforcement | PolicySets; PolicyRules |
| Data assets | Catalog search and browse | Items |
| Data columns | Catalog search and browse | - |
| Classic glossary | Glossary, critical data elements, custom context | Tags |
| Lineage | Lineage and impact analysis | - |
| Classification rules | Classification and sensitive-data protection | - |
| Scan rule sets and scans | External metadata scanning | - |
Run 1. The measurement
python3 count_api_operations.py
Python 3.14.8 duckdb 1.5.6 Read 66 specification files, 4,159,637 bytes: Purview Unified Catalog 272,123 bytes sha256 70d32c954a8ce3b8 CatalogApiService.json Purview Data Map 257,767 bytes sha256 adb679263fe3d0c5 purviewdatamap.json Purview Scanning 272,678 bytes sha256 1dc732ab31fc03b6 scanningService.json Fabric (63 files) 3,357,069 bytes sha256 76480bd793f4a367 (of the file hashes) Operations by API Fabric 901 operations 390 read 511 write Purview Data Map 96 operations 44 read 52 write Purview Scanning 58 operations 30 read 28 write Purview Unified Catalog 69 operations 36 read 33 write Fabric operations whose path or name mentions lineage: 0 Each Purview object type, and where Microsoft's transition guide says it lands Object type Export ops Fabric API tags Load ops Result Business domains 2 Domains 11 export by API; the named Fabric target has write operations Glossary terms 7 Tags 5 export by API; the named Fabric target has write operations Critical data elements 6 Tags 5 export by API; the named Fabric target has write operations Data products 6 Domains 11 export by API; the named Fabric target has write operations Objectives and key results 7 - 0 export only: no Fabric API object to load into Data access policies 1 PolicySets; PolicyRules 6 export by API; the named Fabric target has write operations Data assets 4 Items 242 export by API; the named Fabric target has write operations Data columns 3 - 0 export only: no Fabric API object to load into Classic glossary 13 Tags 5 export by API; the named Fabric target has write operations Lineage 3 - 0 export only: no Fabric API object to load into Classification rules 3 - 0 export only: no Fabric API object to load into Scan rule sets and scans 4 - 0 export only: no Fabric API object to load into Summary 7 object types: export by API; the named Fabric target has write operations 5 object types: export only: no Fabric API object to load into Wrote operations.csv (1124 rows) and landing.csv (12 rows) to . Self-checks passed: every tag in the landing table exists in its specification, and the counts add up.
Run 2. A planted fault the self-check must catch
Run 2 adds one row to the table for an object type, glossary term templates, under an API tag that does not exist in the Unified Catalog specification. A table key that matches nothing would quietly count zero, so the self-check refuses it.
python3 count_api_operations.py --plant-fault
Python 3.14.8 duckdb 1.5.6 Read 66 specification files, 4,159,637 bytes: Purview Unified Catalog 272,123 bytes sha256 70d32c954a8ce3b8 CatalogApiService.json Purview Data Map 257,767 bytes sha256 adb679263fe3d0c5 purviewdatamap.json Purview Scanning 272,678 bytes sha256 1dc732ab31fc03b6 scanningService.json Fabric (63 files) 3,357,069 bytes sha256 76480bd793f4a367 (of the file hashes) Operations by API Fabric 901 operations 390 read 511 write Purview Data Map 96 operations 44 read 52 write Purview Scanning 58 operations 30 read 28 write Purview Unified Catalog 69 operations 36 read 33 write Fabric operations whose path or name mentions lineage: 0 Each Purview object type, and where Microsoft's transition guide says it lands Object type Export ops Fabric API tags Load ops Result Business domains 2 Domains 11 export by API; the named Fabric target has write operations Glossary terms 7 Tags 5 export by API; the named Fabric target has write operations Critical data elements 6 Tags 5 export by API; the named Fabric target has write operations Data products 6 Domains 11 export by API; the named Fabric target has write operations Objectives and key results 7 - 0 export only: no Fabric API object to load into Data access policies 1 PolicySets; PolicyRules 6 export by API; the named Fabric target has write operations Data assets 4 Items 242 export by API; the named Fabric target has write operations Data columns 3 - 0 export only: no Fabric API object to load into Classic glossary 13 Tags 5 export by API; the named Fabric target has write operations Lineage 3 - 0 export only: no Fabric API object to load into Classification rules 3 - 0 export only: no Fabric API object to load into Scan rule sets and scans 4 - 0 export only: no Fabric API object to load into Glossary term templates 0 Tags 5 no export operation Summary 7 object types: export by API; the named Fabric target has write operations 5 object types: export only: no Fabric API object to load into 1 object types: no export operation SELF-CHECK FAILED (no CSV written, so earlier results are kept): Glossary term templates: no tag 'TermTemplate' in the Purview Unified Catalog specification
Run the count yourself
Download the script, install DuckDB (pip install duckdb) and run python3 count_api_operations.py. The first run downloads about 4 MB of specifications into spec-cache beside the script; later runs read the cache. To count a newer version, change the commit in AZURE or FABRIC at the top of the script, and edit LANDING if Microsoft adds or renames an API tag.
Get the full guide as a PDF
We'll email you a link to it. Please use your work email.
Thanks, . We're sending the link to now. If it isn't there in a few minutes, check your spam or junk folder, or email info@datameaning.com.
Limits
- A specification says what an API offers, not what the API does in your tenant. Counting an operation is not running the operation.
- The read or write sort uses each operation's name and method. A few operations could be sorted the other way; operations.csv lists every one, so a reader can check.
- The Fabric write counts include every operation under a tag. The Items tag covers every kind of Fabric item, so most of its write operations have nothing to do with governance.
- The Unified Catalog API is in preview, and Fabric policy sets are in beta. Microsoft can add, rename or remove operations; the counts are for the commits named above.
- This is a measurement of Microsoft's public specifications, not a client result. Dates are as of 2026-11-21. This guide is about Purview Data Governance, whose support ends on November 12, 2027; after that date the guide describes history.
Tested on
Tested on
- Platform
- DuckDB 1.5.6, run from Python 3.14.8
- Edition
- Open-source DuckDB library (
pip install duckdb) - Date
- 2026-11-21
- Data
- Microsoft's REST API specifications: Azure/azure-rest-api-specs at commit be93aca and microsoft/fabric-rest-api-specs at commit ec85f05, both MIT licensed
- Code
- count_api_operations.py
- Results
- landing.csv, operations.csv
When this is not enough
The count tells you which moves can be scripted at all. Each object type still needs a decision: whether the object moves to the OneLake catalog, to another catalog, to rules in code or to an archive, and what the move loses on the way.
Deciding where each kind of Purview governance content should land?
Purview to OneLake Migration is in early access. The offering starts with the four-day Catalog Transition Assessment: one fixed fee of $12,000, and no commitment to anything else. You get an inventory of every object in your Purview governance catalog and a decision for each object. AI does the first pass, and your stewards sign every domain.