Support additional ontology pre-release annotations

README.md

@@ -269,21 +269,64 @@ type: `http://spdx.invalid./AbstractClass`, but this is not preferred.
 
 ### Pre-Release models
 
-`shacl2code` has a custom annotation that can be used to mark an entire
-ontology as "pre-release" meaning it is still subject to break changes. In the
-python bindings, this will cause a `FutureWarning` to be emitted. In order to
-do this, you must declare your ontology, and then use the custom annotation to
-mark it as pre-release:
+`shacl2code` can detect if an ontology is a "pre-release" version (still
+subject to breaking changes). For language bindings that support it (such as
+Python), importing a pre-release ontology binding will emit a warning
+(e.g. `FutureWarning`).
 
-```ttl
+Pre-release status can be specified explicitly via command-line options,
+or inferred automatically from various ontology annotations. For example,
 
+```ttl
 <http://example.org/my-ontology> a ow:Ontology ;
-    sh-to-code:isPreRelease true
-    .
+   sh-to-code:isPreRelease true
+   .
 ```
 
+In the event of conflicting annotations, `shacl2code` evaluates pre-release
+status using the following order of precedence (1 = the highest priority):
+
+1. **`--pre-release` or `--no-pre-release` command-line options**:
+   Explicitly marks the generated bindings as pre-release or stable,
+   overriding any annotations in the input ontology.
+2. **`sh-to-code:isPreRelease`**:
+   The `shacl2code` custom boolean annotation
+   (`sh-to-code:isPreRelease true` or `false`).
+3. **`adms:status` (EU SEMIC Vocabulary)**:
+   If the status is
+   `<http://publications.europa.eu/resource/authority/dataset-status/DEVELOP>`,
+   it is considered a pre-release.
+   Other values in the dataset-status vocabulary space (e.g. `COMPLETED`)
+   indicate a stable release.
+4. **`adms:status` (Original ADMS Vocabulary)**:
+   If the status is `<http://purl.org/adms/status/UnderDevelopment>`,
+   it is considered a pre-release.
+   Other values in the ADMS status namespace (e.g. `Completed`)
+   indicate a stable release.
+5. **`bibo:status` (Bibliographic Ontology)**:
+   If set to `<http://purl.org/ontology/bibo/status/draft>`, it is
+   considered a pre-release.
+   Other values in the BIBO status namespace (e.g. `published`, `legal`)
+   indicate a stable release.
+6. **`schema:creativeWorkStatus`**:
+   If set to `"Draft"` or `"Incomplete"`, it is considered a pre-release.
+   Other values (e.g. `"Published"`) indicate a stable release.
+7. **`vs:term_status`**:
+   If set to `"unstable"` or `"testing"`, it is considered a pre-release.
+   Other values (e.g. `"stable"`) indicate a stable release.
+8. **`owl:versionInfo` (pre-release extension)**:
+   If the version string contains a pre-release extension suffix
+   (e.g., `-alpha`, `-beta`, `-dev`, `-rc`, `-SNAPSHOT`, `.alpha`, etc.).
+9. **`owl:versionInfo` (major version zero)**:
+   If the version string corresponds to a major version zero in
+   [Semantic Versioning][semver] (e.g., `0.7.1`).
+10. **Default Fallback**:
+    If none of the above are present, the ontology is assumed to be a stable
+    release (`false`).
+
 Note that the IRI of the ontology must be the prefix of all IRIs that belong to
 that ontology.
 
 [pytest]: https://www.pytest.org
 [pytest-cov]: https://pytest-cov.readthedocs.io/en/latest/
+[semver]: https://semver.org/

pyproject.toml

@@ -39,7 +39,7 @@ dev = [
     "mypy >= 1.19.1",  # latest version for Python 3.9 is 1.19.x
     "pyrefly >= 0.55.0",
     "pyright >= 1.1.403",
-    "pyshacl >= 0.25.0",
+    "pyshacl >= 0.25.0, < 0.40.0",  # 0.40.0 uses Python 3.10+ type union
     "pytest >= 7.4",
     "pytest-cov >= 4.1",
     "pytest-xdist >= 3.5",

src/shacl2code/main.py

@@ -51,7 +51,7 @@ def handle_generate(parser, args):
                     data = json.load(f)
             contexts.append(ContextData(data, url))
 
-        m = Model(graph, UrlContext(contexts))
+        m = Model(graph, UrlContext(contexts), is_prerelease=args.pre_release)
 
         render = args.lang(args)
         render.output(m)
@@ -120,6 +120,11 @@ def handle_version(parser, args):
         help="SPDX License Identifier to use for generated source code. Default is %(default)s",
         default="0BSD",
     )
+    generate_parser.add_argument(
+        "--pre-release",
+        action=argparse.BooleanOptionalAction,
+        help="Mark the generated binding as pre-release. Overrides any ontology annotations",
+    )
     generate_parser.set_defaults(func=handle_generate)
 
     lang_subparser = generate_parser.add_subparsers(

src/shacl2code/model.py

@@ -5,6 +5,7 @@
 # SPDX-License-Identifier: MIT
 """SHACL model parsing and data class definitions"""
 
+import re
 from dataclasses import dataclass, field
 from typing import List, Optional
 
@@ -19,6 +20,8 @@
     XSD,
 )
 
+from .util import convert_version_string
+
 PATTERN_DATATYPES = [
     str(XSD.string),
     str(XSD.dateTime),
@@ -114,7 +117,7 @@ class Class:
 
 
 class Model(object):
-    def __init__(self, graph, context=None):
+    def __init__(self, graph, context=None, is_prerelease=None):
         self.model = graph
         self.context = context
         self.compact_ids = {}
@@ -197,6 +200,109 @@ def is_abstract(s):
 
             return False
 
+        def is_semver_prerelease(version_str):
+            # Only treat a hyphen as a semver pre-release marker when it
+            # directly follows a dotted numeric core (e.g. "1.2.3-beta"),
+            # not e.g. a date-like version such as "2024-01-15".
+            if re.match(r"^\d+\.\d+(?:\.\d+)?-[0-9A-Za-z]", version_str):
+                return True
+            if re.search(
+                r"\b(alpha|beta|dev|pre|rc|snapshot|test)\b", version_str, re.IGNORECASE
+            ):
+                return True
+            return False
+
+        def get_is_prerelease(onto_iri):
+            # 1) --pre-release command line option
+            if is_prerelease is not None:
+                return is_prerelease
+
+            # 2) sh-to-code:isPreRelease
+            val = self.model.value(onto_iri, SHACL2CODE.isPreRelease)
+            if val is not None:
+                return bool(val)
+
+            adms_statuses = list(
+                self.model.objects(onto_iri, URIRef("http://www.w3.org/ns/adms#status"))
+            )
+            if adms_statuses:
+                semic = [
+                    str(s)
+                    for s in adms_statuses
+                    if str(s).startswith(
+                        "http://publications.europa.eu/resource/authority/dataset-status/"
+                    )
+                ]
+                # 3) adms:status (EU SEMIC vocab)
+                if semic:
+                    return any(
+                        s
+                        == "http://publications.europa.eu/resource/authority/dataset-status/DEVELOP"
+                        for s in semic
+                    )
+                original = [
+                    str(s)
+                    for s in adms_statuses
+                    if str(s).startswith("http://purl.org/adms/status/")
+                ]
+                # 4) adms:status (Original ADMS vocab)
+                if original:
+                    return any(
+                        s == "http://purl.org/adms/status/UnderDevelopment"
+                        for s in original
+                    )
+
+            # 5) bibo:status (Bibliographic Ontology)
+            bibo_statuses = list(
+                self.model.objects(
+                    onto_iri, URIRef("http://purl.org/ontology/bibo/status")
+                )
+            )
+            if bibo_statuses:
+                return any(
+                    str(s) == "http://purl.org/ontology/bibo/status/draft"
+                    for s in bibo_statuses
+                )
+
+            # 6) schema:creativeWorkStatus
+            schema_statuses = list(
+                self.model.objects(
+                    onto_iri, URIRef("http://schema.org/creativeWorkStatus")
+                )
+            ) or list(
+                self.model.objects(
+                    onto_iri, URIRef("https://schema.org/creativeWorkStatus")
+                )
+            )
+            if schema_statuses:
+                return any(str(s) in ("Draft", "Incomplete") for s in schema_statuses)
+
+            # 7) vs:term_status
+            vs_statuses = list(
+                self.model.objects(
+                    onto_iri,
+                    URIRef("http://www.w3.org/2003/06/sw-vocab-status/ns#term_status"),
+                )
+            )
+            if vs_statuses:
+                return any(str(s) in ("unstable", "testing") for s in vs_statuses)
+
+            # 8) & 9) owl:versionInfo
+            versions = list(self.model.objects(onto_iri, OWL.versionInfo))
+            if versions:
+                for version in versions:
+                    version_str = str(version)
+                    # 8) owl:versionInfo (pre-release extension e.g., "-beta", "-alpha", "-rc" etc)
+                    if is_semver_prerelease(version_str):
+                        return True
+                    # 9) owl:versionInfo (major version zero)
+                    parts = convert_version_string(version_str)
+                    if parts and parts[0] == 0:
+                        return True
+                return False
+
+            return False
+
         for onto_iri in self.model.subjects(RDF.type, OWL.Ontology):
             label = str(self.model.value(onto_iri, RDFS.label, default=""))
             o = Ontology(
@@ -205,9 +311,7 @@ def is_abstract(s):
                 label=label,
                 comment=str(self.model.value(onto_iri, RDFS.comment, default="")),
                 version=str(self.model.value(onto_iri, OWL.versionInfo, default="")),
-                is_prerelease=bool(
-                    self.model.value(onto_iri, SHACL2CODE.isPreRelease, default=False)
-                ),
+                is_prerelease=get_is_prerelease(onto_iri),
             )
             self.ontologies.append(o)