pyobo 0.12.6__py3-none-any.whl → 0.12.7__py3-none-any.whl

pyobo/__init__.py

@@ -65,6 +65,7 @@ from .plugins import (
     run_nomenclature_plugin,
 )
 from .struct import (
+    Annotation,
     Obo,
     Reference,
     StanzaType,
@@ -80,6 +81,7 @@ from .utils.path import ensure_path
 from .version import get_version
 
 __all__ = [
+    "Annotation",
     "Obo",
     "Reference",
     "StanzaType",

pyobo/sources/iana_media_type.py

@@ -5,14 +5,18 @@
 
 from collections.abc import Iterable
 
-from pyobo import Obo, Reference, Term, default_reference
-from pyobo.struct.typedef import term_replaced_by
+import requests
+
+from pyobo import Obo, Reference, Term, TypeDef, default_reference
+from pyobo.struct import Annotation
+from pyobo.struct.typedef import has_source, term_replaced_by
 from pyobo.utils.path import ensure_df
 
 __all__ = ["IANAGetter"]
 
 PREFIX = "iana.mediatype"
-ROOT = Term.from_triple(prefix="dcterms", identifier="MediaType", name="media type")
+ROOT_MEDIA_TYPE = Term.from_triple(prefix="dcterms", identifier="MediaType", name="media type")
+ROOT_FILE_FORMAT = Term.from_triple(prefix="dcterms", identifier="FileFormat", name="file format")
 
 #: The top-level types listed on https://www.iana.org/assignments/media-types/media-types.xhtml
 MEDIA_TYPE_GROUPS = [
@@ -31,13 +35,37 @@ MEDIA_TYPE_GROUPS = [
 GROUP_TO_CSV = {
     media_type_group: (
         f"https://www.iana.org/assignments/media-types/{media_type_group}.csv",
-        Term(reference=default_reference(PREFIX, media_type_group, media_type_group)).append_parent(
-            ROOT
-        ),
+        Term(
+            reference=Reference(prefix=PREFIX, identifier=media_type_group, name=media_type_group)
+        ).append_parent(ROOT_MEDIA_TYPE),
     )
     for media_type_group in MEDIA_TYPE_GROUPS
 }
 
+MIMETYPE_IO_URL = (
+    "https://github.com/patrickmccallum/mimetype-io/raw/refs/heads/master/src/mimeData.json"
+)
+
+
+def _get_mimetypes():
+    records = requests.get(MIMETYPE_IO_URL, timeout=5).json()
+    rv = {}
+    for record in records:
+        name = record.pop("name")
+        rv[name] = record
+    return rv
+
+
+PREDICATE = TypeDef(
+    reference=default_reference(
+        prefix=PREFIX, identifier="extension", name="appears with file extension"
+    ),
+    domain=ROOT_MEDIA_TYPE.reference,
+    range=ROOT_FILE_FORMAT.reference,
+    definition="Connects a media type with a file format that has been observed to encode it",
+    is_metadata_tag=True,
+)
+
 
 class IANAGetter(Obo):
     """An ontology representation of IANA media types (i.e. MIME types)."""
@@ -45,9 +73,10 @@ class IANAGetter(Obo):
     ontology = bioregistry_key = PREFIX
     name = "IANA Media Types"
     dynamic_version = True
-    root_terms = [t.reference for _, (_, t) in sorted(GROUP_TO_CSV.items())]
+    root_terms = [ROOT_MEDIA_TYPE.reference, ROOT_FILE_FORMAT.reference]
     typedefs = [
         term_replaced_by,
+        PREDICATE,
     ]
 
     def iter_terms(self, force: bool = False) -> Iterable[Term]:
@@ -55,26 +84,53 @@ class IANAGetter(Obo):
         return get_terms()
 
 
-def get_terms() -> list[Term]:
+def get_terms() -> Iterable[Term]:
     """Get IANA Media Type terms."""
+    mimetypes_data = _get_mimetypes()
+
+    filetype_to_term = {}
+    for record in mimetypes_data.values():
+        for filetype_str in record.get("fileTypes", []):
+            filetype_id = filetype_str.removeprefix(".")
+            filetype_term = Term(
+                reference=default_reference(PREFIX, identifier=filetype_id, name=filetype_str),
+                type="Instance",
+            )
+            filetype_term.append_parent(ROOT_FILE_FORMAT)
+            filetype_term.annotate_string(has_source, MIMETYPE_IO_URL)
+            filetype_to_term[filetype_str] = filetype_term
+            yield filetype_term
+
     terms: dict[str, Term] = {}
     forwards: dict[Term, str] = {}
     for key, (url, parent) in GROUP_TO_CSV.items():
         df = ensure_df(PREFIX, url=url, sep=",")
         terms[key] = parent
         for name, identifier, references in df.values:
+            mimetypes_record = mimetypes_data.get(identifier, {})
+
             if "OBSOLE" in name or "DEPRECATED" in name:
                 is_obsolete = True
             else:
                 is_obsolete = None
+
             term = Term(
                 reference=Reference(prefix=PREFIX, identifier=identifier, name=name),
                 is_obsolete=is_obsolete,
+                # TODO how to add definition source?
+                definition=mimetypes_record.get("description"),
             ).append_parent(parent)
             for reference in _process_references(references):
                 term.append_see_also_uri(reference)
             terms[identifier.casefold()] = term
 
+            for filetype_str in mimetypes_record.get("fileTypes", []):
+                term.annotate_object(
+                    PREDICATE,
+                    filetype_to_term[filetype_str],
+                    annotations=[Annotation.uri(has_source, MIMETYPE_IO_URL)],
+                )
+
             if "in favor of" in name:
                 _, _, new = name.partition("in favor of ")
                 forwards[term] = new.casefold().strip().rstrip(")").strip()
@@ -84,7 +140,7 @@ def get_terms() -> list[Term]:
             new = "application/vnd.afpc.afplinedata"
         old.append_replaced_by(terms[new].reference)
 
-    return list(terms.values())
+    yield from terms.values()
 
 
 def _process_references(cell: str) -> list[str]:

pyobo/struct/functional/obo_to_functional.py

@@ -39,9 +39,15 @@ def get_ofn_from_obo(
     prefix = obo_ontology.ontology
     base = f"{_BASE}/{prefix}"
     if iri is None:
-        iri = f"{base}/{prefix}.ofn"
-    if version_iri is None and obo_ontology.data_version:
-        version_iri = f"{base}/{obo_ontology.data_version}/{prefix}.ofn"
+        if obo_ontology.ontology_iri:
+            iri = obo_ontology.ontology_iri
+        else:
+            iri = f"{base}/{prefix}.ofn"
+    if version_iri is None:
+        if obo_ontology.ontology_version_iri:
+            version_iri = obo_ontology.ontology_version_iri
+        elif obo_ontology.data_version:
+            version_iri = f"{base}/{obo_ontology.data_version}/{prefix}.ofn"
     ofn_ontology = Ontology(
         iri=iri,
         version_iri=version_iri,

pyobo/struct/obograph/export.py

@@ -70,8 +70,12 @@ def to_parsed_obograph(obo: Obo) -> og.StandardizedGraphDocument:
 
 
 def _to_parsed_graph(obo: Obo) -> og.StandardizedGraph:
+    if obo.ontology_iri:
+        ontology_iri = obo.ontology_iri
+    else:
+        ontology_iri = f"http://purl.obolibrary.org/obo/{obo.ontology}.owl"
     return og.StandardizedGraph(
-        id=f"http://purl.obolibrary.org/obo/{obo.ontology}.owl",
+        id=ontology_iri,
         meta=_get_meta(obo),
         nodes=_get_nodes(obo),
         edges=_get_edges(obo),
@@ -167,7 +171,9 @@ def _get_meta(obo: Obo) -> og.StandardizedMeta | None:
             )
         )
 
-    if obo.data_version:
+    if obo.ontology_version_iri:
+        version_iri = obo.ontology_version_iri
+    elif obo.data_version:
         version_iri = (
             f"http://purl.obolibrary.org/obo/{obo.ontology}/{obo.data_version}/{obo.ontology}.owl"
         )

pyobo/struct/struct.py

@@ -594,6 +594,10 @@ class Obo:
 
     imports: ClassVar[list[str] | None] = None
 
+    ontology_iri: ClassVar[str | None] = None
+
+    ontology_version_iri: ClassVar[str | None] = None
+
     def __post_init__(self):
         """Run post-init checks."""
         if self.ontology is None:
@@ -2319,6 +2323,8 @@ def build_ontology(
     mailing_list: str | None = None,
     logo: str | None = None,
     repository: str | None = None,
+    ontology_iri: str | None = None,
+    ontology_version_iri: str | None = None,
 ) -> Obo:
     """Build an ontology from parts."""
     if name is None:
@@ -2378,6 +2384,8 @@ def build_ontology(
         _subsetdefs=subsetdefs,
         _property_values=properties,
         _imports=imports,
+        _ontology_iri=ontology_iri,
+        _ontology_version_iri=ontology_version_iri,
         terms=terms,
     )
 
@@ -2395,6 +2403,8 @@ def make_ad_hoc_ontology(
     _subsetdefs: list[tuple[Reference, str]] | None = None,
     _property_values: list[Annotation] | None = None,
     _imports: list[str] | None = None,
+    _ontology_iri: str | None = None,
+    _ontology_version_iri: str | None = None,
     *,
     terms: list[Term] | None = None,
 ) -> Obo:
@@ -2413,6 +2423,8 @@ def make_ad_hoc_ontology(
         subsetdefs = _subsetdefs
         property_values = _property_values
         imports = _imports
+        ontology_iri = _ontology_iri
+        ontology_version_iri = _ontology_version_iri
 
         def __post_init__(self):
             self.date = _date

pyobo/struct/struct_utils.py

@@ -59,18 +59,32 @@ class Annotation(NamedTuple):
     value: Reference | OBOLiteral
 
     @classmethod
-    def float(cls, predicate: Reference, value: float) -> Self:
+    def float(cls, predicate: Reference | TypeDef, value: float) -> Self:
         """Return a literal property for a float."""
+        from .struct import TypeDef
+
+        if isinstance(predicate, TypeDef):
+            predicate = predicate.reference
         return cls(predicate, OBOLiteral.float(value))
 
     @classmethod
-    def uri(cls, predicate: Reference, uri: str) -> Self:
+    def uri(cls, predicate: Reference | TypeDef, uri: str) -> Self:
         """Return a literal property for a URI."""
+        from .struct import TypeDef
+
+        if isinstance(predicate, TypeDef):
+            predicate = predicate.reference
         return cls(predicate, OBOLiteral.uri(uri))
 
     @classmethod
-    def string(cls, predicate: Reference, value: str, *, language: str | None = None) -> Self:
+    def string(
+        cls, predicate: Reference | TypeDef, value: str, *, language: str | None = None
+    ) -> Self:
         """Return a literal property for a float."""
+        from .struct import TypeDef
+
+        if isinstance(predicate, TypeDef):
+            predicate = predicate.reference
         return cls(predicate, OBOLiteral.string(value, language=language))
 
     @staticmethod

pyobo/struct/typedef.py

@@ -135,6 +135,7 @@ rdf_type = TypeDef(reference=v.rdf_type)
 subproperty_of = TypeDef(reference=v.subproperty_of)
 see_also = TypeDef(reference=v.see_also, is_metadata_tag=True)
 comment = TypeDef(reference=v.comment, is_metadata_tag=True)
+label = TypeDef(reference=v.label, is_metadata_tag=True)
 has_member = TypeDef(
     reference=Reference(prefix=RO_PREFIX, identifier="0002351", name="has member"),
 )

pyobo/version.py

@@ -12,7 +12,7 @@ __all__ = [
     "get_version",
 ]
 
-VERSION = "0.12.6"
+VERSION = "0.12.7"
 
 
 def get_git_hash() -> str:

{pyobo-0.12.6.dist-info → pyobo-0.12.7.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: pyobo
-Version: 0.12.6
+Version: 0.12.7
 Summary: A python package for handling and generating OBO
 Keywords: snekpack,cookiecutter,ontologies,biomedical ontologies,life sciences,natural sciences,bioinformatics,cheminformatics,Open Biomedical Ontologies,OBO
 Author: Charles Tapley Hoyt

{pyobo-0.12.6.dist-info → pyobo-0.12.7.dist-info}/RECORD

@@ -1,5 +1,5 @@
 pyobo/.DS_Store,sha256=89c9b251ccb5e7628fa025768c73ba6621f3768619efd69855d3d7cdc1b33044,8196
-pyobo/__init__.py,sha256=570b3a632caf50b46fff8e64712a22f08892df6997c07cddd59372e2bbadb79c,3598
+pyobo/__init__.py,sha256=e86aa2d2fb934d7cc923d0588be45a493dd13e3465f48f19ed8b02332c160f75,3632
 pyobo/__main__.py,sha256=70e610d8e10bc45fbd4fa9ee67dc672d2bbd6cb1e4e9e8a7aec4a8cfb11c0fd3,83
 pyobo/api/__init__.py,sha256=6e194a5d230bf0297970d8b3267f5577130e536c440443bba286dee1dda3643e,3091
 pyobo/api/alts.py,sha256=478b3162574616f2827bd73107f80f5dfdaf5971eaf6c6813109109dba89e522,3032
@@ -95,7 +95,7 @@ pyobo/sources/gwascentral/gwascentral_study.py,sha256=7f30e246e594f38d3ce60725ce
 pyobo/sources/hgnc/__init__.py,sha256=d5084cec674febf03108ccda073da4dd244b9034ce912da15213cc2c44d21963,157
 pyobo/sources/hgnc/hgnc.py,sha256=7ac57bc377e25e6dac0f5ada71005f50a3bf0b82c1aff88ffe7072c04f38ced2,16699
 pyobo/sources/hgnc/hgncgenefamily.py,sha256=e3e8d0b750d59a7da3d0ac2b100e248e8dad51af54c045405a12704e36db20cd,3710
-pyobo/sources/iana_media_type.py,sha256=6c895261f8ca6175b62c94275dc30c9c597ba7dfc2049b5ec566f474ae23a81e,3038
+pyobo/sources/iana_media_type.py,sha256=2ad63b4f5d7caf651fb8f1b5a93b79abf7a23e3b175b0ee21862105441c59e8e,5019
 pyobo/sources/icd/__init__.py,sha256=0eb65597be8ff1eb2c278b6564f71ab8556b5d590821907ab09d1ce8a9ba4962,142
 pyobo/sources/icd/icd10.py,sha256=61ab55ce0af6dbd27f59e25c6b8150007a96af1edcecde1f1db85bd5c4354de5,2587
 pyobo/sources/icd/icd11.py,sha256=5b6e59f8028cfbd41f860151d1164514be2653a50078c0fc8748c20b815ee11b,4867
@@ -177,20 +177,20 @@ pyobo/struct/__init__.py,sha256=7efd602470a09f89f9820227baee10b9679b7a8422b2554e
 pyobo/struct/functional/__init__.py,sha256=d536edb7fe86a1aad19607df3dea62b49dd0b4bb2b30f31b8431a842700a499e,32
 pyobo/struct/functional/dsl.py,sha256=a1d0d7df79734a429c46372bb2d7ab68549cd34c0838f04726dc577bd73d4471,101591
 pyobo/struct/functional/macros.py,sha256=cc654f131fa685d4afc87d8c7b5e8565da1b789dd5d73045a77b3e033a8ced52,13785
-pyobo/struct/functional/obo_to_functional.py,sha256=785c46fdb00dc45f3903f67ad24ae4fcb6d187ce48208373fc9d42833a85f6b8,13273
+pyobo/struct/functional/obo_to_functional.py,sha256=71e8bc7d2c03dc7942cba56fadb6b78f585792420c63c9172deb902c308dd5cb,13493
 pyobo/struct/functional/ontology.py,sha256=f9af23c7eeb536c23776008111834ff957acc00cbcc16c4b06500414c26e7322,9342
 pyobo/struct/functional/utils.py,sha256=73e5c6d9227b0a162dcfb294dcad05f93ea870c8d87af7a2d16c6591f9249ad1,3487
 pyobo/struct/obo/__init__.py,sha256=c49005278cae796050bf706bb53b871c9ec0ccea8280eb8780a638450a1cb16d,157
 pyobo/struct/obo/reader.py,sha256=edef37b6592ddcd4dded06f3bd9ea9ffc5bbcd1cc11b247839923be078ff497e,50797
 pyobo/struct/obo/reader_utils.py,sha256=5324154d08576e167d7c28b3660dbaa394e67878acf7ea9930abd0170201460a,4423
 pyobo/struct/obograph/__init__.py,sha256=078acd3fd3824e73179bc629434adaaf5b309f8188c3d6f88100fe215baa1f2c,452
-pyobo/struct/obograph/export.py,sha256=7bf4b4bcd6c3c89e820fe542711dae70cb037ed4571d2901631fdc48430d2c0d,9752
+pyobo/struct/obograph/export.py,sha256=ca4dcc1ee94333f48ca217f088575e07adbe5c627664d8115191a31a38b2a276,9945
 pyobo/struct/obograph/reader.py,sha256=dbae3255e0fc6b78c6c7d11a194655c456d243fa7043fe9c915c3d4bcc2225f3,8525
 pyobo/struct/obograph/utils.py,sha256=8ded244a40b6e0c53da56447ab5fcaf89e635a18d611263a348dd3a59aaf67f4,1516
 pyobo/struct/reference.py,sha256=4fc7d3b665cd18c8dfbbb2b8b49de4bc30a8e101d14ff26bf5f88e9c562bc2d3,11735
-pyobo/struct/struct.py,sha256=e833c86ec9e302808afc3750a0d9b1cfc4237251fc10450e48bda8b181761a00,93297
-pyobo/struct/struct_utils.py,sha256=be63a06ade7ba29cf3c02410a245e715103884c5e886d4b23e49050bb8f7dd5a,39930
-pyobo/struct/typedef.py,sha256=eed11961bc01efa4f2fec3c7ae9ba8eff5dd6776d70bf0d4a9cc13b72c042e3f,13031
+pyobo/struct/struct.py,sha256=58fd6a184318b8bc7d44edb90f635e4277b83b454be8309996175fa86661af7f,93743
+pyobo/struct/struct_utils.py,sha256=fc3ebcf5b3651a0d7fafee2f410fa299510ec73abde5706a045c6a14c574efb5,40346
+pyobo/struct/typedef.py,sha256=88b9bcf83558d2ffdd2bf470a0d28278e17b9c8e9795a89e4b609a08778f4a52,13088
 pyobo/struct/utils.py,sha256=ce4a4e138d894087e4374adce6a34434ad375482a691fff9eed1e88cc24aef56,906
 pyobo/struct/vocabulary.py,sha256=2addf44eb64ab55f3ee06f87022c7112078f873c8436fc45d9a2ff7dfec9b53e,5243
 pyobo/utils/__init__.py,sha256=08035263cbc6abfeaff2b5a159121d9e4f96a392c0d91f568e0d67a9b5aa2cec,17
@@ -200,9 +200,9 @@ pyobo/utils/iter.py,sha256=ad845b6da1491f131a134c94fab419a026a0608aed7afd3db98d2
 pyobo/utils/misc.py,sha256=e3df697a77320ea5bdff19b41f40aad0584ec34328368abef485504ca87e79a1,7992
 pyobo/utils/ndex_utils.py,sha256=128902592d345ab93fe32f2575e42e53269a0bac8dcc18370da81497e2767336,2326
 pyobo/utils/path.py,sha256=129f254d15a0154f7f5c5e6fa895790771b83dc501825222b4c91a50dc63cefa,4091
-pyobo/version.py,sha256=a0f95e1dc5c6ae4d27ebf8475452242e445573a9b03e3602e3b3fd402309fd94,926
-pyobo-0.12.6.dist-info/licenses/LICENSE,sha256=41c80964a1b1956e41c013670812fc5592d5b51bd7b3cd4287d949450488a498,1076
-pyobo-0.12.6.dist-info/WHEEL,sha256=ec89e25a684ec3ad8fb2a9be806e47666a27f7dcdcf9b55bff3f1d3ed5e60e0a,78
-pyobo-0.12.6.dist-info/entry_points.txt,sha256=00d833beec05ffdff58a90a8c49b5b04ce80e22d8c92ddfd80c3f340eea1bc6b,42
-pyobo-0.12.6.dist-info/METADATA,sha256=9725b7bbe5fa4b43a5d11e33fc551f38bfec3c5e2a6066f44cf0ea94f6cc9686,21834
-pyobo-0.12.6.dist-info/RECORD,,
+pyobo/version.py,sha256=524be873cd1df560b36d00f90206d2db05da6d13b7332ae8ecae37d2fdbc49ac,926
+pyobo-0.12.7.dist-info/licenses/LICENSE,sha256=41c80964a1b1956e41c013670812fc5592d5b51bd7b3cd4287d949450488a498,1076
+pyobo-0.12.7.dist-info/WHEEL,sha256=ec89e25a684ec3ad8fb2a9be806e47666a27f7dcdcf9b55bff3f1d3ed5e60e0a,78
+pyobo-0.12.7.dist-info/entry_points.txt,sha256=00d833beec05ffdff58a90a8c49b5b04ce80e22d8c92ddfd80c3f340eea1bc6b,42
+pyobo-0.12.7.dist-info/METADATA,sha256=c0d2615e44897b0361e752bc2579a271dc2669d3b6987fb3b50e0399154cef9a,21834
+pyobo-0.12.7.dist-info/RECORD,,