Add all content for experiment

.gitignore

@@ -0,0 +1,4 @@
+venv/
+__pycache__
+personinfo_busted.py
+personinfo_workaround.py

README.md

@@ -1,2 +1,65 @@
-# linkml-experiments
-LinkML experiments
+# linkml-multivalued-default
+
+An experiment in generating an empty list default for multivalued slots in a
+LinkML model.
+
+## Background
+
+It is difficult or impossible in LinkML to specify that you want an empty list
+as the default value for a slot that is both `required` and `multivalued`. This
+repository contains some materials to help demonstrate this, as well as a
+workaround.
+
+For more context, see the following issues:
+- https://github.com/dandi/dandi-archive/issues/2150
+- https://github.com/dandi/dandi-schema/issues/286
+
+## Setup
+
+1. If you are using Nix flakes, run `nix develop` to install dependencies.
+   Otherwise, ensure you have a modern Python installed (e.g., Python 3.10).
+
+2. Create a virtual environment with `python -m venv venv` and activate it with
+   `. ./venv/bin/activate`.
+
+3. Install the Python dependencies with `pip install -r requirements.txt`.
+
+## Demonstration of non-optional list types with default value.
+
+1. Generate a Pydantic model by running `gen-pydantic personinfo_busted.yaml >personinfo_busted.py`.
+   Note that setting both `multivalued=true` and `required=true` correctly
+   infers the type of the `aliases3` field as `List[str]` (rather than
+   `Optional[List[str]]` as for `aliases2`, which is not required), but that it
+   is not possible to set a default value of `[]` directly (the Pydantic
+   generator does not have a way to encode list values in the `ifabsent`
+   attribute; attempts to do so generate the string `"[]"`).
+
+2. Generate another Pydantic model by running `gen-pydantic
+   personinfo_workaround.yaml >personinfo_workaround.py`. Note that this model
+   sets a globally unique and incorrectly typed default on `aliases3`.
+
+3. Run `gen-pydantic personinfo_workaround.yaml | sed s/\"aliases3dummy\"/[]/ >personinfo_workaround.py`.
+   Note that now the definition of `aliases3` correctly has an empty list
+   default value.
+
+4. Fire up Python and run this script:
+
+   ```python
+   from personinfo_workaround import Person
+   p = Person(name2='Ada Byron', aliases2=['Lady Byron'])
+   p
+   ```
+
+   Note that `p` contains the expected values for all fields, including a
+   default `[]` for `aliases3`.
+
+The need to post-process like this is unfortunate, but seems to be necessary
+due to unresolved issues of defaults for multivalued slots in LinkML
+([[1]](https://github.com/orgs/linkml/discussions/1975),
+[[2]](https://github.com/orgs/linkml/discussions/2314)).
+
+## Conclusion
+
+It *is* possible to generate correct Pydantic models from LinkML schemata
+featuring **required**, **multivalued** attributes **with a list-valued
+default**, but it currently requires a transformation of the generated code.

flake.nix

@@ -0,0 +1,18 @@
+{
+  description = "Nix dependencies for running LinkML experiments";
+
+  inputs = {
+    nixpkgs.url = "github:nixos/nixpkgs?ref=nixos-unstable";
+  };
+
+  outputs = { self, nixpkgs }: {
+    devShell.x86_64-linux =
+      let
+        pkgs = nixpkgs.legacyPackages.x86_64-linux;
+      in pkgs.mkShell {
+        buildInputs = with pkgs; [
+          python310Packages.python
+        ];
+      };
+  };
+}

personinfo_busted.yaml

@@ -0,0 +1,45 @@
+id: https://w3id.org/linkml/examples/personinfo
+name: personinfo
+prefixes:
+  linkml: https://w3id.org/linkml/
+  personinfo: https://w3id.org/linkml/examples/personinfo
+imports:
+  - linkml:types
+default_range: string
+default_prefix: personinfo
+
+classes:
+  Person:
+    attributes:
+      # Not required, implicit default value is None.
+      name1:
+
+      # Required, but no default specified.
+      name2:
+        required: true
+
+      # Required, with a default.
+      name3:
+        required: true
+        ifabsent: string(Alan Turing)
+
+      # Not required, implicit default value is None.
+      aliases1:
+        multivalued: true
+
+      # Required, but no default specified.
+      aliases2:
+        multivalued: true
+        required: true
+
+      # Required, with a default (ends up as `"[]"`, which is wrong).
+      aliases3:
+        multivalued: true
+        required: true
+        ifabsent: []
+
+      # Required, with a dummy default value to be replaced later.
+      aliases4:
+        multivalued: true
+        required: true
+        ifabsent: string(aliases4dummy)

personinfo_workaround.yaml

@@ -0,0 +1,39 @@
+id: https://w3id.org/linkml/examples/personinfo
+name: personinfo
+prefixes:
+  linkml: https://w3id.org/linkml/
+  personinfo: https://w3id.org/linkml/examples/personinfo
+imports:
+  - linkml:types
+default_range: string
+default_prefix: personinfo
+
+classes:
+  Person:
+    attributes:
+      # Not required, implicit default value is None.
+      name1:
+
+      # Required, but no default specified.
+      name2:
+        required: true
+
+      # Required, with a default.
+      name3:
+        required: true
+        ifabsent: string(Alan Turing)
+
+      # Not required, implicit default value is None.
+      aliases1:
+        multivalued: true
+
+      # Required, but no default specified.
+      aliases2:
+        multivalued: true
+        required: true
+
+      # Required, with a dummy default value to be replaced later.
+      aliases3:
+        multivalued: true
+        required: true
+        ifabsent: string(aliases3dummy)

requirements.txt

@@ -0,0 +1,76 @@
+annotated-types==0.7.0
+antlr4-python3-runtime==4.9.3
+arrow==1.3.0
+attrs==25.1.0
+certifi==2025.1.31
+CFGraph==0.2.1
+chardet==5.2.0
+charset-normalizer==3.4.1
+click==8.1.8
+curies==0.10.6
+Deprecated==1.2.18
+et_xmlfile==2.0.0
+exceptiongroup==1.2.2
+fqdn==1.5.1
+graphviz==0.20.3
+greenlet==3.1.1
+hbreader==0.9.1
+idna==3.10
+iniconfig==2.0.0
+isodate==0.7.2
+isoduration==20.11.0
+Jinja2==3.1.5
+json-flattener==0.1.9
+jsonasobj==1.3.1
+jsonasobj2==1.0.4
+jsonpatch==1.33
+jsonpath-ng==1.7.0
+jsonpointer==3.0.0
+jsonschema==4.23.0
+jsonschema-specifications==2024.10.1
+linkml==1.8.7
+linkml-dataops==0.1.0
+linkml-runtime==1.8.3
+MarkupSafe==3.0.2
+openpyxl==3.1.5
+packaging==24.2
+parse==1.20.2
+pluggy==1.5.0
+ply==3.11
+prefixcommons==0.1.12
+prefixmaps==0.2.6
+pydantic==2.10.6
+pydantic_core==2.27.2
+PyJSG==0.11.10
+pyparsing==3.2.1
+PyShEx==0.8.1
+PyShExC==0.9.1
+pytest==8.3.4
+pytest-logging==2015.11.4
+python-dateutil==2.9.0.post0
+PyTrie==0.4.0
+PyYAML==6.0.2
+rdflib==7.1.3
+rdflib-jsonld==0.6.1
+rdflib-shim==1.0.3
+referencing==0.36.2
+requests==2.32.3
+rfc3339-validator==0.1.4
+rfc3987==1.3.8
+rpds-py==0.23.1
+ruamel.yaml==0.18.10
+ruamel.yaml.clib==0.2.12
+ShExJSG==0.8.2
+six==1.17.0
+sortedcontainers==2.4.0
+sparqlslurper==0.5.1
+SPARQLWrapper==2.0.0
+SQLAlchemy==2.0.38
+tomli==2.2.1
+types-python-dateutil==2.9.0.20241206
+typing_extensions==4.12.2
+uri-template==1.3.0
+urllib3==2.3.0
+watchdog==6.0.0
+webcolors==24.11.1
+wrapt==1.17.2