Conversation
Previously, we did not test that `AASXWrite.write_all_aas_objects` did indeed that: it includes all `Submodel`s inside the given `ObjectStore`. This adds a unittest to `test_aasx_utils` that checks for this expected behavior. Fixes #209
…ge (#436) adapter.aasx: Add support for loading and saving thumbnails in AASX package Previously, the `AASXReader` did not load the thumbnail in the `read_into` function, instead a separate function needed to be called. This behavior did also occur in `AASXWriter` with writing the thumbnail. This was unintuitive, as it required a separat call of the load/save function. This PR implements the loading/storing of the thumbnail directly in the loading/storing of other supplementary files. Fixes #435
…434) `adapter.aasx`: Add `rename_file` to `DictSupplementaryFileContainer` Before: There was no supported way to rename supplementary files in `DictSupplementaryFileContainer`. Workflow had to manipulate internals or create duplicates, making file handling error-prone. Now: A public rename operation preserves content de-duplication and resolves name conflicts, simplifying AASX read/write flows and reducing inconsistent mappings.
Previously, multilingual fields (`display_name`, `description`, `MultiLanguageProperty.value`) could be assigned invalid types, leading to inconsistent and sometimes invalid JSON/XML serialization. As a result, this behavior could cause interoperability issues. The AAS specification requires these fields to be of type `LangStringSet / MultiLanguageTextType`. This requirement is now enforced to ensure spec-compliant serialization. Fixes #416
It appears that we forgot to update some of the copyright notices in the module docstrings, where actual changes were done in 2026. This adapts these ocurrences.
Previously the script only output the first problem with the copyright issue, which is inefficient. This adapts the script to output all copyright issues. Fixes #451
Previously, the serialization of `adapter.xml` uses the general `object_to_xml_element()` method. In it, there was branch to a call of `lang_string_set_to_xml()` without setting the tag parameter, which would have caused an exception, if the code was ever run. We decided to remove the call of `lang_string_set_to_xml` completely as this specific case in `object_to_xml()` is considered dead code, as the XML tag for a `LangStringSet` varies based on its context (the tag is determined by the *parent's* attribute), it must always be called explicitly with a tag parameter by the *parent's serialization logic*. Since there is no scenario where a `LangStringSet` is serialized in isolation (It **cannot** be a standalone xml object), this branch is unreachable and should be removed to maintain code cleanliness. Fixes #397
This adds an additional unittest for the `adatper.aasx.AASXReader` class. While the reading of AASX was previously implicitly tested by first reading and then writing AASX when testing the `AASXWriter`, this adds dedicated test cases for the reading. Furthermore, we fix some years in the copyright notices of some files. Fixes #441
Just as described in #453, now that the hotfix fixed our CI pipeline on push to `main`, we need to make sure that `develop` stays up to date with `main`. Improve CI.yml definition (#453) Previously, we had some weird bugs with the CI pipeline sometimes failing (#400), but not always reproducible. Namely, sometimes the CI failed, due to `mypy` not finding the `sdk` types when running from the `compliance_tool` environment. Since currently, we are at a point where it is impossible to reproduce the failing CI (at least for me), I decided to clean up the job definitions a little bit and make some things more explicit. Namely, instead of calling scripts like `pip` or `mypy` from their PATH, we now explicitly call them via `python -m pip` and `python -m mypy`. This theoretically ensures, that it always uses the script we just installed with the dependencies and not something the VM already had in its path via `actions/setup-python@v5`. This should ensure that a script like `mypy` actually has all the necessary dependencies installed. Secondly, we had a `pip install -e ../sdk[dev]`, therefore installing the development dependencies of the `sdk` in the `compliance_tool` CI check. This is technically incorrect, since we use the `sdk` as external dependency and therefore shouldn't depend on the development dependencies. I therefore removed this. Lastly, the `sdk-readme-codeblocks` check uses `bash` syntax. In theory, the Ubuntu environment should use `bash` by default, but now it is made explicit. Fixes #400
Previously, the `AbstractObjectProvider` only worked with `Identifiables`, which made it incompatible with the new version of the AAS metamodel. This renames and restructures both the `AbstractObjectProvider` and the `AbstractObjectStore`. In all non-abstract subclasses where `Object` appears in the class or method name, it has been replaced with `Identifiable`. Old classes are still available with a deprecation warning. Moreover, `AbstractObjectProvider` and `AbstractObjectStore` are now generic to be able to handle more classes than just `Identifiables`. In order to handle the new `AASDescriptor`, a new class `HasIdentifier` has been added. It is intended to be an abstract superclass for all classes that have an identifier, but are not `Identifiables`. As of now, this PR is intended to serve as a basis for discussion. Therefore, the documentation has not yet been adapted. Fixes #428
Previously, the `_object_cache` of the `LocalFileIdentifiableStore` was initialised empty. Calling `discard()` immediately after store initialisation raised a `KeyError` if the Identifiable to delete existed on disk but had not yet been loaded into the in-memory cache. This adds explicit handling of missing cache entries in `discard()`, preventing errors when a key is not present in the in-memory cache. Moreover, a corresponding regression test is added. Finally, the interim empty-cache workaround in `sync()` is removed. Fixes #438
Previously, we lacked a tutorial demonstrating how to navigate a Submodel's hierarchy using IdShorts and IdShortPaths. This adds a simple tutorial for Submodel navigation and updates the documentation to reference it. The end user is shown how to navigate SubmodelElements, such as simple Properties, Property Collections, Property Lists and Collection Lists. Fixes #351
Previously, the `EXIT_CODE` of the copyright year check was hardcoded to the value 0, causing the CI to pass unconditionally. This fixes the bug by coupling the `EXIT_CODE` to the result of the copyright year check. Moreover, the check has been extended to additionally verify the copyright year of the `LICENSE` file. This also updates the relevant copyright years that were missed before. Fixes #458
Previously, the `server` was built only with the `repository` API endpoints in mind. Since we now plan to also support `registry` and `discovery`, we need to adapt the `./server` project structure. For this reason, we did the following refactoring: - This moves the `pyproject.toml` from the `app` directory into the server project's top-level directory (`./server/pyproject.toml`). This aligns the `server` project directory with the `sdk` and the `compliance-tool` project directories. See #466 - This also moves the files for Docker image definition into the `./server/docker` subdirectories, where `./server/docker/common` are shared files across different Docker images, and `./server/docker/repository` and others contain the `Dockerfile`, as well as the image specific configuration files. See #468 - This moves the Docker `compose.yaml` into the `./server/example_configurations` and its subdirectories. Each example configuration should have its own `README.md` specifying what the example contains and how to run it. See #468 - Lastly, this moves the repository specific `main.py` to `.server/app/services/run_repository.py` And this adapts (hopefully) all relevant paths in all the different files accordingly. Fixes #466 Fixes #468 * server: Fix review issues in server package restructure - Rename CI job server-package -> server-repository-docker to reflect that it specifically builds the repository Docker image - Narrow pycodestyle scope to 'app test' to match mypy, avoiding scanning docker/ and example_configurations/ directories - Fix broken #running-examples anchor and stale Dockerfile path in example_configurations/repository_standalone/README.md --------- Co-authored-by: zrgt <garmaev@gmx.de>
Previously we had OpenAPI files in the sdk tests, which were used to check the first iteration of our server implementation. Since these files are not used anymore, this removes them. Fixes #476
Previously, the `sdk`'s `util.traversal.walk_submodel()` method skipped `Entity` and `Operation` children, both `SubmodelElement`s that can contain further children `SubmodelElement`s. This fixes this bug and adds additional unittests to check for successful traversal of all kinds of `SubmodelElement`s. Fixes #423
Previously the LangStringSet checked language tags that were added to match the format of xx or xx-XX. This did not follow the documented behavior, which requires all IETF BCP 47 conform language tags to be accepted. These changes replaced the previous constraint check, with a check by regular expression, following the IETF BCP 47 standard. Fixes #157
Previously, there were lots of occurrences of the old `acplt.org` domain name used whenever we needed an URL name example. This replaces occurrences of this domain name with `example.org`, one of the official [IANA example domains]. Fixes #460 [IANA example domains]: https://www.iana.org/help/example-domains
Previously, the server only contained the AAS Repository API, more precisely the AAS Repository, Submodel Repository and Concept Description Repository APIs. This also adds the Registry and Discovery APIs. Because of this, we had to refactor the way we start the server, as now effectively, we have multiple different servers that could be executed in the `server` package.
Till now we had not implemented the additional primitive data types introduced in the [API specification]. Many of them are still not usable since we for example do not support asynchronous operations yet. But I still think it is better to have them implemented for later use. [API specification]: https://industrialdigitaltwin.io/aas-specifications/IDTA-01002/v3.1/specification/interfaces-payload.html#_primitive_data_types --------- Co-authored-by: s-heppner <iat@s-heppner.com>
* Update SDK to V3.0.1 of AAS Part 5 (AASX) * Adapt link to specs and adapt link to schema files --------- Co-authored-by: Moritz Sommer <m.sommer@iat.rwth-aachen.de>
Fixes #443 --------- Co-authored-by: Moritz Sommer <m.sommer@iat.rwth-aachen.de>
Previously, the flag to surpress output in the compliance-tool `cli.py` was erroneously: "`--quite`", when it should have been `--quiet`, of course. This changes all occurences of this flag to `--quiet`. # Fixes 505
This implements the changes of the metamodel for version 3.1.2 of the spec. Fixes #437 --------- Co-authored-by: Moritz Sommer <m.sommer@iat.rwth-aachen.de> Co-authored-by: Leon Huang <hleon04@icloud.com> Co-authored-by: Sercan Sahin <s.sahin@iat.rwth-aachen.de>
Add deprecated discovery route Previously, we decided not to implement any routes that are deprecated in the specification. However, as it turns out, for compatibility with the [basyx-aas-web-ui], the `/lookup/shells` route is required. This implements the missing route. Fixes #522 [basyx-aas-web-ui]: https://github.com/eclipse-basyx/basyx-aas-web-ui/ --------- Co-authored-by: s-heppner <iat@s-heppner.com>
Previously we had two implementations of the service specification: one in interfaces.base.py and the other one in model.service_specification.py. This removes the implementation in interfaces.base.py and adapt the correponding usages to the one in model.service_specification.py. Fixes #523 --------- Co-authored-by: s-heppner <iat@s-heppner.com>
…que_name (#513) `DictSupplementaryFileContainer_store_refcount` is designed to track how many `_name_map` entries reference the same content hash, so `delete_file()` can free the underlying bytes only when the last reference is removed. The refcount is never incremented, so files are never freed. Previously, because `_assign_unique_name()` never increments `_store_refcount`, the count stays at 0 after any `add_file()`. Every `delete_file()` decrements to -1 and the equality check `== 0` is never true, so `_store[hash]` and `_store_refcount[hash]` are never cleaned up. Every file ever added leaks indefinitely. This fixes this bug by incrementing `_store_refcount[sha] += 1` inside `_assign_unique_name()` when a new `_name_map` entry is created (the first branch of the `while True` loop). Also decrement it (and skip the increment) inside the second branch when a duplicate name already maps to the same hash. Fixes #495
…rned empty results (#512) When 2 or more `globalAssetId` query parameters were sent, a `len(global_asset_ids) <= 1` guard in the filter lambda in `repository.py` evaluated `False` for every shell, causing an empty HTTP 200 response with no error. The guard was likely intended to reject invalid input with a 400 error, not silently discard all results. The guard is replaced with proper input validation that raises `BadRequest` when multiple global asset IDs are provided. Fixes #500
The compliance tool still had code to check a json or xml file against the defined schema from admin-shell-io/aas-specs-metamodel. The option to run this schema check was removed earlier from cli.py with commit af73a4b. As no other code uses the implemented schema check, this function is now removed, including the schema files and the unittests.
Previously, `provider.py` imported `ServerAASFromJsonDecoder` directly from `app.adapter`, which caused a circular import. Reverted the import to use the `app.adapter` module reference instead, accessing `adapter.ServerAASFromJsonDecoder` at the call site. Co-authored-by: s-heppner <iat@s-heppner.com>
PR #370 removed `Referable.commit()` and all call sites in the `server` handlers without replacing the write-back mechanism. Since then, any mutation to an object retrieved from `LocalFileIdentifiableStore`, `LocalFileDescriptorStore`, or `CouchDBIdentifiableStore` was silently lost on cache eviction, visible only within the same in-process `WeakValueDictionary` cache entry. A different uWSGI worker, or any request after the cache entry expired, would re-read the stale on-disk or on-database state. There was also a compounding bug: `get_item()` / `get_identifiable_by_hash()` always re-read from storage even on a cache hit, then called `update_from()` on the cached object, discarding any in-memory mutations even within the same request. This change fixes both issues across all three backends: - `get_identifiable_by_hash()` / `get_identifiable_by_couchdb_id()`: return the cached instance on a hit instead of overwriting it with a freshly-deserialized copy. - `get_item()`: check the cache first and return immediately on a hit. - Add `commit()` to `LocalFileIdentifiableStore` (re-serializes to .json), `LocalFileDescriptorStore` (same), and `CouchDBIdentifiableStore` (PUT with stored `_rev`, updates revision on success). - `AbstractObjectStore.commit()` is added as a no-op default so in-memory stores (`DictIdentifiableStore`) require no changes. All mutating handlers in `server/app/interfaces/repository.py` and `registry.py` now call `self.object_store.commit()` after each mutation. - A regression test `test_mutation_persistence` is added to `sdk/test/backend/test_local_file.py`. Fixes #552
After refactoring the server package, the CI for building and releasing did not longer work. Additionally the registry and discovery services are not published. These changes fix the build in the release CI for the repository service and add the missing ones. Builds are now executed using QEMU for amd64, armv7 and arm64 to create a multi-platform image. Fixes #555
Previously, the `sync()` method in `model.provider.AbstractObjectStore` was not tested at all. This adds a unittest with full line coverage of this method.
Previously, tutorial_navigate_aas.py was not covered by Unit Tests. This adds the module to the unittests.
Previously `_parse_xsd_date()` and `_parse_xsd_datetime()` raised `ValueError("Negative Dates are not supported by Python")` when given an XSD value with a leading `-` (BCE year). This was technically false as input is not malformed, only not supported.
This change raises `NotImplementedError` instead, allowing callers to distinguish "invalid input" from "known SDK limitation.". Additionally pointing users to https://github.com/eclipse-basyx/basyx-python-sdk/issues to report the lack of implementation.
Fixes #565
The project root contained an empty `1.0.0` file since commit 4f15ce2 [1]. As no usage of this file is known or observable, nor did the commit author document any reason for adding the file, we concluded that this must be a mistake from refactoring a `pyproject.toml`. Therefore the file is now removed. [1]: 4f15ce2
Previously, JSON deserialization accepted both `ModelReference` and `ExternalReference` for the `dataSpecification` field of `EmbeddedDataSpecification`. As the spec (constraint AASc-3a-050) requires `EmbeddedDataSpecification.dataSpecification` to be an `ExternalReference`, JSON deserialization was tightened to only accept `ExternalReference`. It now matches the requirement and the current behavior of XML deserialization. Fixes #567
Currently, the JSON and XMOL deserializiers raise an error when parsing Blob or File submodels that do not include a contentType The contentType attribute has a cardinality of 0..1 for both classes and should therefore be optional. The implementation now aligns with this specification and defaults to None if the field is absent in the data. Fixes #561
Previously, the check_directory method was not covered by unit tests. Unit tests for all three directory state branches are added.
Previously compliance_tool tests relied on handcrafted examples.
Replaced examples by mocking the underlying sdk functionality to
just test the output parsing. Currently only for json and only for
'check_deserialization' and 'check_example'.
Applied the same refactoring to resulting equivalence check of json
implementation and the xml implementation.
Tests for the compliance check for aasx files were refactored so
they use mocking instead of actual files. Additional test cases
were added to ensure
1. alignment of core properties is checked
2. content of supplementary files is equal
Replace the old fixture-dependent subprocess tests with direct calls to
main() and parse_cli_arguments(), mocking compliance_check_* modules to
verify routing without touching the filesystem.
Previously the aasx file equivalence check would
1. still execute subsequent steps if the loading of one file fails.
This was caused by checking only if `state_manager.status is Status.FAILED`,
missing that `Status.NOT_EXECUTED > Status.FAILED`.
2. use blank assertions to ensure core_properties `created` attribute
is of type `datetime.datetime`.
This resulted in the compliance_tool failing with `AssertionError`
in cases that were caused by (1).
Status guards for fast-failing are now corrected to take
`Status.NOT_EXECUTED` into account.
The assertions are removed and replaced with `DataChecker` checks
to inform the user gracefully on problems. `cast(...)` is used
to inform the type-checker about the `isinstance(...)` result.
Previously the supplementary files of the aasx container were not
tested for presence or equality although the check_aas_example did
it. Now a two-way check, coparing presence, content-type and sha256
is implemented. Tests were adapted to expect the additional step.
The previous check on the supplementary file was broken, as it
compared the sha256 value of the TestFile to itself.
Introduced a new step which checks for presence and equality of
supplementary file `/TestFile.pdf` in the AASX container.
Refactored the core property checks to no longer use `assert` in
combination with `try ... except AssertionError`.
The `failed_checks` is a property that returns a fresh iterator
at each call. With `mock_data_checker.return_value.failed_checks = iter([])`
only one iterator is created that may be exhausted at subsequent calls.
The introduced `PropertyMock` fixes this.
As the new test structure uses mocking extensively, the risk for
missing error cases in these tests increases. To overcome this
simple end-to-end integration tests are integrated now, which do
not use mocking but operate on real temporary files.
For all three adapters a full cycle is implemented:
1. create -> check_example (both success and fail)
2. create x2 -> check_file_equivalence (both sucess and fail)
The help output of the compliance tool is cleaned. Old fragments
of the long ago removed schema-checking functionality were still
in place.
Fixes #486
Previously the API base path (currently `api/v3.1`) was hardcoded across interface modules, dockerfiles and tests. Bumping the version risks drift across the occurences. This change adds a unittest that extracts the version from each of the currently known files via regex and asserts the version matches. The risk of newly added occurrences not being tracked remains. Fixes #535
Previously, only the `DictIdentifiableStore` was tested in the unittests. This extends the unittests by adding a `_STORE_CLASSES` class variable and consolidates duplicated test methods into `subTest` loops, covering both `Dict` and `Set` store implementations.
…exhausted iterator (#517) * fix: OrderedNamespaceSet.__setitem__ slice uses exhausted iterator slice branch stored exhausted islice iterator back into _order instead of the materialized successful_new_items list, resulting in an empty slice assignment that discarded all newly added items. Fixes #494 * fix: OrderedNamespaceSet.__setitem__ int removes old item before adding new add-before-remove caused false AASConstraintViolation when new item shares an attribute (e.g. id_short) with the item being replaced. Fix removes old item first, rolls back if the subsequent add fails. Fixes #498
Previously, the compliance-tool package name still reflected the time we kept it separate from the basyx-python-sdk repository and published it ourselves. As we now prepare to officially publish it via the Eclipse PyPI account, we update the name to reflect its affiliation with Eclipse BaSyx via: `basyx-compliance-tool` in the compliance-tool's `pyproject.toml`.
Previously, UnsignedInt was mapped to the XSD type name xs:unsignedByte in XSD_TYPE_NAMES, and UnsignedByte had no entry. As a result, serialized UnsignedInt values were tagged with the wrong XSD type and UnsignedByte values could not be serialized. This change corrects the mapping of UnsignedInt to xs:unsignedInt and adds the missing UnsignedByte entry mapped to xs:unsignedByte. Fixes #560
Until now, each value of a MultiLanguageNameType was constrained to a maximum length of 64 via check_short_name_type. This never matched the specification: LangStringNameType/text has had a maximum length of 128 characters since AAS Part 1 Metamodel V3.0, identical to NameType. MultiLanguageNameType now uses check_name_type instead. The now-unused check_short_name_type and constrain_short_name_type functions are removed accordingly. Fixes #576
Metamodel v3.1.2 changed `RelationshipElement.first` and `second` to be optional. This was missed earlier and required by JSON and XML adapters. - `AnnotatedRelationshipElement.__init__` now accepts `first` and `second` as `Optional[Reference]`, matching the parent class. - The JSON and XML deserializers treat both fields as optional default to `None` when absent. - The JSON serializer omits the fields when `None` (XML serializer already handled this). - Added tests for this behaviour. Fixes #525
Previously, `_parse_xsd_datetime` and `_parse_xsd_time` rejected the
XSD end-of-day midnight notation (`24:00:00`), even though the XSD
spec allows it as an alternative representation of midnight. Error
messages on invalid input were also generic ("Value is not a valid
XSD datetime string"), giving no indication of which value failed.
This change accepts `hour=24` when minutes, seconds, and
microseconds are all zero, normalizing it to `00:00:00` of the
following day for `DateTime` (and to `00:00:00` for `Time`, which has
no notion of a following day). Any other non-zero component combined
with `hour=24` now raises a clear, value-specific error. Error
messages throughout both parsers now include the offending value.
Additionally, the `Date`, `DateTime`, and `Time` constructor calls in
these parsers were switched from positional to keyword arguments, so
each parsed field is clearly labeled instead of being a same-typed
`int(...)` in a long positional list.
Fixes #564
* compliance_tool: Fix packaging and CI SDK install Previously, `compliance_tool/pyproject.toml` was missing the `[project.scripts]` console entry point, so the `aas-compliance-check` command documented in the README did not exist and only `python -m aas_compliance_tool.cli` worked. It also declared its SDK dependency as a direct reference (`basyx-python-sdk @ file:../sdk`), which bakes a local path into the built wheel and is rejected by PyPI on upload, so the package could not be published. This restores the `aas-compliance-check = aas_compliance_tool.cli:main` console script and replaces the direct reference with a normal `basyx-python-sdk>=1.0.0` version constraint, matching the original `setup.py`. Switching to the loose `>=1.0.0` constraint broke the compliance-tool CI jobs, however: they checked out shallow (no tags), so `setuptools_scm` built the local sdk as `0.1.dev1`. That does not satisfy `>=1.0.0`, so `pip install .[dev]` uninstalled the local sdk and pulled `basyx-python-sdk` from PyPI instead, and the tests failed against a release that predates the local refactors. Both compliance tool jobs now check out with `fetch-depth: 0` so the local sdk builds as a `>=1.0.0` dev version and pip keeps it (see #592). * compliance_tool: Document SDK version caveat in README Previously, the README described what the compliance tool checks but not how the installed `basyx-python-sdk` determines which metamodel version is actually verified. Because the SDK dependency is declared loosely (`basyx-python-sdk>=1.0.0`), `pip` keeps an already-installed, older SDK instead of upgrading it, so the tool can silently check against an older metamodel than the README advertises, with no error to signal the mismatch. This adds an `Installation` section that splits the default PyPI install (`pip install basyx-compliance-tool`) from the developer install against the sibling `./sdk` source tree, and documents — in the developer flow, where it is most relevant — that the SDK version must be matched explicitly until the pinning is tightened (see #592).
Previously, `_construct_blob` in `json_deserialization.py` was decorated with `@classmethod` twice, producing a `classmethod(classmethod(func))` wrapper. On Python 3.10 and 3.12 (used in CI) this went unnoticed, because `classmethod` chained through the inner descriptor and still resolved to a callable bound method. Python 3.13 removed that descriptor chaining, so accessing `cls._construct_blob` now yields a raw, non-callable `classmethod` object. As a result, deserializing any `Blob` raised `TypeError: 'classmethod' object is not callable`, breaking JSON deserialization and the AASX round-trip tests on 3.13. This change removes the redundant decorator so `_construct_blob` behaves like all other constructor methods. This finding motivated #594.
Previously, `GYEAR_RE` and `GYEARMONTH_RE` only matched exactly four digits and did not allow a leading minus sign. This rejected valid XSD values such as years with more than four digits (e.g. "20000") and negative years (e.g. "-2001"), even though the XSD spec permits both. Additionally, `GMonthDay.from_date(...)` passed `date.year` instead of `date.day` to the constructor, and `into_date(...)` on `GYear`/ `GYearMonth` raised Python's cryptic "year -2001 is out of range" error for negative years, since `datetime.date` does not support them. This change widens the regular expressions to accept any number of digits and an optional leading minus sign, fixes the `GMonthDay. from_date(...)` argument order bug, and makes `into_date(...)` raise a clear `ValueError` when called on a G-Class with a negative year. Fixes #566
Did a general pass over the root, `sdk`, and `server` READMEs and brought them back in line with the current state of the code. Previously these files still described the pre-2.1.0 state in a few places and carried some long-standing typos and stale references, for example missing server interfaces, an outdated `API_BASE_PATH` default, a wrong package name and submodule count, and dead links. The changes correct these outdated sections so the documentation matches the actual packages, interfaces, and configuration again.
Member
Author
|
The |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
12 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Release Notes
Version 2.1.0 of the BaSyx-Python SDK is a feature release that brings the SDK up to date with the v3.1 line of the AAS specifications and significantly expands the server.
The metamodel now implements Part 1 v3.1.2, with matching updates to the IEC 61360 Data Specification (Part 3a v3.1.1) and the AASX Package File Format (Part 5 v3.1). On the server side, the HTTP API has been updated to Part 2 API v3.1, and the server now ships Registry and Discovery interfaces in addition to the Repository, each available as its own standalone Docker service.
As part of expanding the server, the
AbstractObjectProviderand its object stores were refactored to support the updated AAS metamodel through generic typing and naming aligned withIdentifiable. This is a refactoring only: the usability of the SDK remains unchanged and you do not have to change any existing code. The previous classes and methods (DictObjectStore,SetObjectStore,LocalFileObjectStore,CouchDBObjectStore, andAbstractObjectProvider.get_identifiable()) are now deprecated and emit warnings, but continue to work as before. Going forward, please use the newDictIdentifiableStore,SetIdentifiableStore,LocalFileIdentifiableStore,CouchDBIdentifiableStoreclasses and the genericget_item()method instead.This release implements the following versions of the AAS specification:
Changelog
Notable:
Improvements:
first/secondin relationship elements (See: Allow optionalfirst/secondin relationship elements #587)valueIdinValueReferencePairfor JSON and XML (See: adapter: allow optionalvalueIdinValueReferencePairfor JSON and XML #579)contentTypeinBlobandFile(See: adapter: Allow optional contentType in Blob and File #572)dataSpecificationreference type in JSON (See: adapter.json: TightendataSpecificationref type #581)rename_filetoDictSupplementaryFileContainer(See: adapter.aasx: Addrename_filetoDictSupplementaryFileContainer#434)AbstractObjectProvider(See: sdk: Refactor AbstractObjectProvider #430)paging_metadatain paginated list responses (See: Always include paging_metadata in paginated list responses #538)Bugfixes:
24:00:00(See: Fix: Handle XSD end-of-day midnight notation 24:00:00 #582)MultiLanguageNameTypevalue length to 128 (See: model.base: Fix MultiLanguageNameType value length to 128 #577)UnsignedIntXSD type name (See: FixUnsignedIntXSD type name #589)OrderedNamespaceSet.__setitem__int remove-before-add and exhausted slice iterator (See: Fix OrderedNamespaceSet.__setitem__: int remove-before-add and slice exhausted iterator #517)GET /shells?assetIdsreturning empty for multipleglobalAssetIdvalues (See: Fix GET /shells?assetIds: multiple globalAssetId values silently returned empty results #512)DictSupplementaryFileContainer._assign_unique_name(See: Fix DictSupplementaryFileContainer: refcount never incremented, delete_file never frees memory #513)DataSpecificationIEC61360.valueindependent ofvalue_format(See: Fix XML deserialization: DataSpecificationIEC61360.value dropped when value_format absent #510)LocalFileIdentifiableStoreto count and iterate only.jsonstore files (See: Fix LocalFileIdentifiableStore: count and iterate only .json store files #507)load_directorysilently dropping all descriptors (See: Fix load_directory silently dropping all descriptors #545)NamespaceSet.pop()to remove items from all backends (See: Fix NamespaceSet.pop() removes item from all backends #514)ServiceSpecificationProfileEnumwith IDTA-01002 v3.1.2 (See: fix: align ServiceSpecificationProfileEnum with IDTA-01002 v3.1.2 spec #541)paging_metadatakwarg correctly in AAS/Submodel-ref and ConceptDescription listings (See: fix: pass paging_metadata kwarg correctly in get_aas_submodel_refs and get_concept_description_all #540)cursorcalculation andlimitdefault (See: server: Correct paginationcursorcalculation andlimitdefault #526)BadRequest(400) on malformedassetIdsinstead of a 500 (See: Fix assetIds decoding: return 400 instead of 500 on malformed JSON #511)walk_submodel()skippingEntityandOperationchildren (See: sdk: Fix walk_submodel() skipping Entity and Operation children #465)LocalFileIdentifiableStore(See: sdk: Fix empty-cache handling in LocalFileIdentifiableStore #463)LangStringSet(See: sdk: fix langtag constraints in LangStringSet #478)DataSpecificationIEC61360.value_idJSON deserialization code (See: Remove dead DataSpecificationIEC61360 value_id JSON deserialization code #509)--quitetypo, renamed to--quietin the compliance tool CLI (See: Fix --quite typo: rename to --quiet in compliance_tool CLI #506)Testing & Internal:
ObjectStore/AbstractObjectStoretests (See: test sdk: Increase test coverage #483, Extend ObjectStore tests to both implementations #569, test_provider: add tests forAbstractObjectStore.sync()#557)check_directoryand the AASX reader (See: backend.local_file: Add unit tests for check_directory method #571, AASXReader additional Unittests #448)compliance-toolname inpyproject.toml(See: Update pyproject.toml name of compliance-tool #585)acplt.orgwithexample.orgin example data (See: fix: Replace acplt.org with example.org in example data (#460) #474)For more details, check out https://projects.eclipse.org/projects/dt.basyx/releases/2.1.0-python