small fixes; plus new function p89_falls_within

acdh-oeaw · Dec 13, 2024 · af6f38d · af6f38d
1 parent 8e852c9
commit af6f38d
Show file tree

Hide file tree

Showing 6 changed files with 128 additions and 33 deletions.
diff --git a/.flake8 b/.flake8
@@ -7,3 +7,4 @@ exclude =
     build
     dist
     env
+    venv
diff --git a/README.md b/README.md
@@ -369,6 +369,41 @@ g.serialize(format="ttl")
     rdfs:label "Tallinn"@und ;
     ns1:P2_has_type <http://hansi/4/ever/alt-label> .
 ```
+
+### connects to places (E53_Place) with P89_falls_within
+
+```python
+domain = "https://foo/bar/"
+subj = URIRef(f"{domain}place__237979")
+sample = """
+<TEI xmlns="http://www.tei-c.org/ns/1.0">
+    <place xml:id="place__237979">
+        <placeName>Lerchenfelder G&#252;rtel 48</placeName>
+        <desc type="entity_type">Wohngeb&#228;ude (K.WHS)</desc>
+        <desc type="entity_type_id">36</desc>
+        <location type="coords">
+            <geo>48,209035 16,339257</geo>
+        </location>
+        <location>
+            <placeName ref="place__50">Wien</placeName>
+            <geo>48,208333 16,373056</geo>
+        </location>
+    </place>
+</TEI>"""
+doc = ET.fromstring(sample)
+node = doc.xpath(".//tei:place[1]", namespaces=NSMAP)[0]
+g = p89_falls_within(
+    subj, node, domain, location_id_xpath="./tei:location/tei:placeName/@ref"
+)
+result = g.serialize(format="ttl")
+```
+returns
+```ttl
+@prefix ns1: <http://www.cidoc-crm.org/cidoc-crm/> .
+
+<https://foo/bar/place__237979> ns1:P89_falls_within <https://foo/bar/place__50> .
+```
+
 ### normalize_string
 
 ```python

diff --git a/acdh_cidoc_pyutils/__init__.py b/acdh_cidoc_pyutils/__init__.py
@@ -343,24 +343,13 @@ def make_e42_identifiers(
         type_domain = f"{type_domain}/"
     app_uri = URIRef(f"{subj}/identifier/{xml_id}")
     type_uri = URIRef(f"{type_domain}idno/xml-id")
-    approx_uri = URIRef(f"{type_domain}date/approx")
-    g.add((approx_uri, RDF.type, CIDOC["E55_Type"]))
-    g.add((approx_uri, RDFS.label, Literal("approx")))
     g.add((type_uri, RDF.type, CIDOC["E55_Type"]))
     g.add((subj, CIDOC["P1_is_identified_by"], app_uri))
     g.add((app_uri, RDF.type, CIDOC["E42_Identifier"]))
     g.add((app_uri, RDFS.label, Literal(label_value, lang=lang)))
     g.add((app_uri, RDF.value, Literal(normalize_string(xml_id))))
     g.add((app_uri, CIDOC["P2_has_type"], type_uri))
-    events_types = {}
-    for i, x in enumerate(node.xpath(".//tei:event[@type]", namespaces=NSMAP)):
-        events_types[x.attrib["type"]] = x.attrib["type"]
-    if events_types:
-        for i, x in enumerate(events_types.keys()):
-            event_type_uri = URIRef(f"{type_domain}event/{x}")
-            g.add((event_type_uri, RDF.type, CIDOC["E55_Type"]))
-            g.add((event_type_uri, RDFS.label, Literal(x, lang=default_lang)))
-    for i, x in enumerate(node.xpath(".//tei:idno", namespaces=NSMAP)):
+    for i, x in enumerate(node.xpath("./tei:idno", namespaces=NSMAP)):
         idno_type_base_uri = f"{type_domain}idno"
         if x.text:
             idno_uri = URIRef(f"{subj}/identifier/idno/{i}")
@@ -398,7 +387,7 @@ def make_occupations(
     id_xpath=False,
     default_lang="de",
     not_known_value="undefined",
-    special_label=None
+    special_label=None,
 ):
     g = Graph()
     occ_uris = []
@@ -423,7 +412,9 @@ def make_occupations(
         occ_uris.append(occ_uri)
         g.add((occ_uri, RDF.type, FRBROO["F51_Pursuit"]))
         if special_label:
-            g.add((occ_uri, RDFS.label, Literal(f"{special_label}{occ_text}", lang=lang)))
+            g.add(
+                (occ_uri, RDFS.label, Literal(f"{special_label}{occ_text}", lang=lang))
+            )
         else:
             g.add((occ_uri, RDFS.label, Literal(occ_text, lang=lang)))
         g.add((subj, CIDOC["P14i_performed"], occ_uri))
@@ -450,9 +441,6 @@ def make_affiliations(
     lang="en",
 ):
     g = Graph()
-    xml_id = node.attrib["{http://www.w3.org/XML/1998/namespace}id"]
-    item_id = f"{domain}{xml_id}"
-    subj = URIRef(item_id)
     for i, x in enumerate(node.xpath(".//tei:affiliation", namespaces=NSMAP)):
         try:
             affiliation_id = x.xpath(org_id_xpath, namespaces=NSMAP)[0]
@@ -556,3 +544,31 @@ def make_birth_death_entities(
         place_uri = URIRef(f"{domain}{place_node}")
         g.add((event_uri, CIDOC["P7_took_place_at"], place_uri))
     return (g, event_uri, time_stamp_uri)
+
+
+def p89_falls_within(
+    subj: URIRef,
+    node: Element,
+    domain: URIRef,
+    location_id_xpath="./tei:location[@type='located_in_place']/tei:placeName/@key",
+) -> Graph:
+    """connects to places (E53_Place) with P89_falls_within
+
+    Args:
+        subj (URIRef): The Uri of the Place
+        node (Element): The tei:place Element
+        domain (URIRef): An URI used to create the ID of the domain object: {domain}{ID of target object}
+        location_id_xpath (str, optional): An XPath expression pointing to the parent's place ID.
+        Defaults to "./tei:location[@type='located_in_place']/tei:placeName/@key".
+
+    Returns:
+        Graph: A Graph object linking two places via P89_falls_within
+    """
+    g = Graph()
+    try:
+        range_id = node.xpath(location_id_xpath, namespaces=NSMAP)[0]
+    except IndexError:
+        return g
+    range_uri = URIRef(f"{domain}{range_id}")
+    g.add((subj, CIDOC["P89_falls_within"], range_uri))
+    return g
diff --git a/acdh_cidoc_pyutils/namespaces.py b/acdh_cidoc_pyutils/namespaces.py
@@ -1,7 +1,9 @@
 from rdflib import Namespace
 
 CIDOC = Namespace("http://www.cidoc-crm.org/cidoc-crm/")
-FRBROO = Namespace("https://cidoc-crm.org/frbroo/sites/default/files/FRBR2.4-draft.rdfs#")
+FRBROO = Namespace(
+    "https://cidoc-crm.org/frbroo/sites/default/files/FRBR2.4-draft.rdfs#"
+)
 INT = Namespace("https://w3id.org/lso/intro/beta202304#")
 SCHEMA = Namespace("https://schema.org/")
 
@@ -20,5 +22,5 @@
     "to": "end",
     "to-iso": "end",
     "when": "when",
-    "when-iso": "when"
+    "when-iso": "when",
 }
diff --git a/requirements_dev.txt b/requirements_dev.txt
@@ -1,8 +1,7 @@
 black
-coverage>=6.4.4,<7
-flake8>=5.0.4,<6
-pytest>=7.1.3,<8
-lxml
-acdh_tei_pyutils
-python-slugify>=8.0.1
+coverage
+flake8
+pytest
+acdh-tei-pyutils
+python-slugify
 rdflib
diff --git a/tests/test_cidoc_pyutils.py b/tests/test_cidoc_pyutils.py
@@ -16,6 +16,7 @@
     make_birth_death_entities,
     make_occupations,
     make_affiliations,
+    p89_falls_within,
 )
 from acdh_cidoc_pyutils.namespaces import NSMAP, CIDOC
 
@@ -141,12 +142,12 @@ def test_004_create_e52(self):
         e52 = create_e52(uri, end_of_end=begin_of_begin)
         self.assertTrue(begin_of_begin in graph_string)
         e52 = create_e52(uri, begin_of_begin=begin_of_begin, end_of_end=begin_of_begin)
-        e52.serialize('e52.ttl')
+        e52.serialize("e52.ttl")
         self.assertTrue('rdfs:label "1234-05-06"^^xsd:string' in f"{e52.serialize()}")
         e52 = create_e52(uri, begin_of_begin="1222", end_of_end=begin_of_begin)
-        e52.serialize('e52.ttl')
+        e52.serialize("e52.ttl")
         self.assertFalse('rdfs:label "1234-05-06"^^xsd:string' in f"{e52.serialize()}")
-        e52.serialize('e521.ttl')
+        e52.serialize("e521.ttl")
 
     def test_005_normalize_string(self):
         string = """\n\nhallo
@@ -228,7 +229,10 @@ def test_007_make_appellations(self):
             subj = URIRef(item_id)
             g.add((subj, RDF.type, CIDOC["hansi"]))
             g += make_appellations(
-                subj, x, type_domain="https://sk.acdh.oeaw.ac.at/types", default_lang="it"
+                subj,
+                x,
+                type_domain="https://sk.acdh.oeaw.ac.at/types",
+                default_lang="it",
             )
         data = g.serialize(format="turtle")
         g.serialize("appellation.ttl", format="turtle")
@@ -247,7 +251,10 @@ def test_008_make_e42_identifiers(self):
             subj = URIRef(item_id)
             g.add((subj, RDF.type, CIDOC["hansi"]))
             g += make_e42_identifiers(
-                subj, x, type_domain="https://sk.acdh.oeaw.ac.at/types", default_lang="it"
+                subj,
+                x,
+                type_domain="https://sk.acdh.oeaw.ac.at/types",
+                default_lang="it",
             )
             data = g.serialize(format="turtle")
         g = Graph()
@@ -283,7 +290,7 @@ def test_008_make_e42_identifiers(self):
                 default_lang="it",
                 set_lang=True,
                 same_as=False,
-                default_prefix=default_prefix
+                default_prefix=default_prefix,
             )
             data = g.serialize(format="turtle")
             self.assertTrue("@it" in data)
@@ -413,7 +420,9 @@ def test_011_occupation(self):
         g, uris = make_occupations(subj, x)
         self.assertFalse("occupation/hansi" in g.serialize(format="turtle"))
         g.serialize("occupations.ttl")
-        g1, uris = make_occupations(subj, x, id_xpath="@key", not_known_value="ronjaundhanna")
+        g1, uris = make_occupations(
+            subj, x, id_xpath="@key", not_known_value="ronjaundhanna"
+        )
         g1.serialize("occupations1.ttl")
         self.assertTrue("occupation/hansi" in g1.serialize(format="turtle"))
         self.assertTrue("ronjaundhanna" in g1.serialize(format="turtle"))
@@ -474,6 +483,39 @@ def test_012_affiliations(self):
             domain,
             person_label=person_label,
             org_id_xpath="./tei:orgName[1]/@key",
-            org_label_xpath="./tei:orgName[1]//text()"
+            org_label_xpath="./tei:orgName[1]//text()",
         )
         g.serialize("affiliations1.ttl")
+
+    def test_013_p89_falls_within(self):
+        domain = "https://foo/bar/"
+        subj = URIRef(f"{domain}place__237979")
+        sample = """
+<TEI xmlns="http://www.tei-c.org/ns/1.0">
+    <place xml:id="place__237979">
+        <placeName>Lerchenfelder G&#252;rtel 48</placeName>
+        <desc type="entity_type">Wohngeb&#228;ude (K.WHS)</desc>
+        <desc type="entity_type_id">36</desc>
+        <location type="coords">
+            <geo>48,209035 16,339257</geo>
+        </location>
+        <location>
+            <placeName ref="place__50">Wien</placeName>
+            <geo>48,208333 16,373056</geo>
+        </location>
+    </place>
+</TEI>"""
+        doc = ET.fromstring(sample)
+        node = doc.xpath(".//tei:place[1]", namespaces=NSMAP)[0]
+        g = p89_falls_within(
+            subj, node, domain, location_id_xpath="./tei:location/tei:placeName/@ref"
+        )
+        result = g.serialize(format="ttl")
+        self.assertTrue("https://foo/bar/place__50" in result)
+        g.serialize("p89.ttl")
+
+        g = p89_falls_within(
+            subj, node, domain, location_id_xpath="./tei:location/tei:placeName/@key"
+        )
+        result = g.serialize(format="ttl")
+        self.assertFalse("https://foo/bar/place__50" in result)
-Original file line number
+Diff line change
@@ Expand Up / @@ -7,3 +7,4 @@ exclude = @@
         build
         dist
         env
+        venv