Skip to content

Commit c58b789

Browse files
simontaurussimontaurus
committed
Merge remote-tracking branch 'origin/main' into oold-and-opensemantic-package-integration
2 parents ce384a1 + af50ff5 commit c58b789

10 files changed

Lines changed: 594 additions & 15 deletions

File tree

‎.github/workflows/ci.yml‎

Lines changed: 4 additions & 3 deletions
Original file line numberDiff line numberDiff line change
@@ -54,7 +54,7 @@ jobs:
5454
strategy:
5555
matrix:
5656
python:
57-
- "3.8" # oldest Python supported by PSF
57+
- "3.9" # oldest Python supported by PSF
5858
- "3.12" # newest Python that is stable
5959
platform:
6060
- ubuntu-latest
@@ -64,15 +64,16 @@ jobs:
6464
steps:
6565
- uses: actions/checkout@v4
6666
- uses: actions/setup-python@v5
67+
id: setup-python
6768
with:
6869
python-version: ${{ matrix.python }}
6970
- name: Retrieve pre-built distribution files
7071
uses: actions/download-artifact@v4
7172
with: {name: python-distribution-files, path: dist/}
7273
- name: Run tests
7374
run: >-
74-
pipx run tox -e test
75-
--installpkg '${{ needs.prepare.outputs.wheel-distribution }}'
75+
pipx run --python '${{ steps.setup-python.outputs.python-path }}'
76+
tox -e test --installpkg '${{ needs.prepare.outputs.wheel-distribution }}'
7677
-- -rFEx --durations 10 --color yes
7778
--wiki_domain "${{ secrets.INTEGR_TEST_WIKI_DOMAIN }}"
7879
--wiki_username "${{ secrets.INTEGR_TEST_WIKI_USERNAME }}"

‎examples/ontology_import_mwo.py‎

Lines changed: 250 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,250 @@
1+
import os
2+
3+
from osw.auth import CredentialManager
4+
from osw.core import model
5+
from osw.express import OswExpress
6+
from osw.ontology import ImportConfig, OntologyImporter, ParserSettings
7+
8+
# Use credentials from file. if none are found, the user will be prompted to enter them
9+
cm = CredentialManager(
10+
cred_filepath=os.path.join(
11+
os.path.dirname(os.path.abspath(__file__)), "accounts.pwd.yaml"
12+
)
13+
)
14+
# Create the OSW object
15+
# Set a local password in you user settings when using SSO login (e.g. ORCID)
16+
osw_obj = OswExpress(domain="demo.open-semantic-lab.org", cred_mngr=cm)
17+
# Load the required schemas / data classes
18+
DEPENDENCIES = {
19+
# "OwlClass": "Category:OSW725a3cf5458f4daea86615fcbd0029f8",
20+
"OwlIndividual": "Category:OSW6b9ef2784a934b8ab96523366e23e906",
21+
"OwlOntology": "Category:OSW662db0a2ad0946148422245f84e82f64",
22+
"EmmoClass": "Category:OSW57beed5e1294434ba77bb6516e461456",
23+
"Item": "Category:Item",
24+
"ObjectProperty": "Category:ObjectProperty",
25+
"DataProperty": "Category:DataProperty",
26+
"AnnotationProperty": "Category:AnnotationProperty",
27+
}
28+
osw_obj.install_dependencies(DEPENDENCIES, mode="append") # , policy="if-missing")
29+
30+
ontology_name = "MWO"
31+
32+
# Define ontology metadata
33+
mwo = model.OwlOntology(
34+
name="MWO",
35+
label=[model.Label(text="MWO", lang="en")],
36+
iri="http://purls.helmholtz-metadaten.de/mwo",
37+
prefix="http://purls.helmholtz-metadaten.de/mwo/",
38+
prefix_name="mwo",
39+
see_also=["https://ise-fizkarlsruhe.github.io/mwo"],
40+
)
41+
pmdco = model.OwlOntology(
42+
name="PMDco",
43+
label=[model.Label(text="PMDco", lang="en")],
44+
iri="https://w3id.org/pmd/co",
45+
prefix="https://w3id.org/pmd/co/",
46+
prefix_name="pmdco",
47+
see_also=["https://w3id.org/pmd/co"],
48+
)
49+
nfdi = model.OwlOntology(
50+
name="NFDI",
51+
label=[model.Label(text="NFDI Ontology", lang="en")],
52+
iri="https://nfdi.fiz-karlsruhe.de/ontology",
53+
prefix="https://nfdi.fiz-karlsruhe.de/ontology/",
54+
prefix_name="nfdicore",
55+
see_also=["https://nfdi.fiz-karlsruhe.de/ontology"],
56+
)
57+
owl = model.OwlOntology(
58+
name="OWL",
59+
label=[model.Label(text="Web Ontology Language", lang="en")],
60+
iri="http://www.w3.org/2002/07/owl",
61+
prefix="http://www.w3.org/2002/07/owl#",
62+
prefix_name="owl",
63+
see_also=["http://www.w3.org/2002/07/owl"],
64+
)
65+
schema = model.OwlOntology(
66+
name="Schema.org",
67+
label=[model.Label(text="Schema.org", lang="en")],
68+
iri="http://schema.org/",
69+
prefix="http://schema.org/",
70+
prefix_name="schema",
71+
see_also=["http://schema.org/"],
72+
)
73+
skos = model.OwlOntology(
74+
name="SKOS",
75+
label=[model.Label(text="Simple Knowledge Organization System", lang="en")],
76+
iri="http://www.w3.org/2004/02/skos/core",
77+
prefix="http://www.w3.org/2004/02/skos/core#",
78+
prefix_name="skos",
79+
see_also=["http://www.w3.org/2004/02/skos/core"],
80+
)
81+
foaf = model.OwlOntology(
82+
name="FOAF",
83+
label=[model.Label(text="Friend of a Friend", lang="en")],
84+
iri="http://xmlns.com/foaf/0.1/",
85+
prefix="http://xmlns.com/foaf/0.1/",
86+
prefix_name="foaf",
87+
see_also=["http://xmlns.com/foaf/0.1/"],
88+
)
89+
dc = model.OwlOntology(
90+
name="DC",
91+
label=[model.Label(text="Dublin Core", lang="en")],
92+
iri="http://purl.org/dc/elements/1.1",
93+
prefix="http://purl.org/dc/elements/1.1/",
94+
prefix_name="dc",
95+
see_also=["http://purl.org/dc"],
96+
)
97+
bibo = model.OwlOntology(
98+
name="BIBO",
99+
label=[model.Label(text="The Bibliographic Ontology", lang="en")],
100+
iri="http://purl.org/ontology/bibo/",
101+
prefix="http://purl.org/ontology/bibo/",
102+
prefix_name="bibo",
103+
see_also=["http://purl.org/ontology/bibo/"],
104+
)
105+
vann = model.OwlOntology(
106+
name="VANN",
107+
label=[model.Label(text="Vocabulary Annotation", lang="en")],
108+
iri="http://purl.org/vocab/vann/",
109+
prefix="http://purl.org/vocab/vann/",
110+
prefix_name="vann",
111+
see_also=["http://purl.org/vocab/vann/"],
112+
)
113+
adms = model.OwlOntology(
114+
name="ADMS",
115+
label=[model.Label(text="ADMS Ontology", lang="en")],
116+
iri="http://www.w3.org/ns/adms",
117+
prefix="http://www.w3.org/ns/adms#",
118+
prefix_name="adms",
119+
see_also=["http://www.w3.org/ns/adms#"],
120+
)
121+
dcat = model.OwlOntology(
122+
name="DCAT",
123+
label=[model.Label(text="Data Catalog Vocabulary", lang="en")],
124+
iri="http://www.w3.org/ns/dcat",
125+
prefix="http://www.w3.org/ns/dcat#",
126+
prefix_name="dcat",
127+
see_also=["http://www.w3.org/ns/dcat#"],
128+
)
129+
dcterms = model.OwlOntology(
130+
name="DCTerms",
131+
label=[model.Label(text="Dublin Core Terms", lang="en")],
132+
iri="http://purl.org/dc/terms",
133+
prefix="http://purl.org/dc/terms/",
134+
prefix_name="dct",
135+
see_also=["http://purl.org/dc/terms"],
136+
)
137+
doap = model.OwlOntology(
138+
name="DOAP",
139+
label=[model.Label(text="Description of a Project", lang="en")],
140+
iri="http://usefulinc.com/ns/doap",
141+
prefix="http://usefulinc.com/ns/doap#",
142+
prefix_name="doap",
143+
see_also=["http://usefulinc.com/ns/doap"],
144+
)
145+
edam = model.OwlOntology(
146+
name="EDAM",
147+
label=[model.Label(text="EDAM Ontology", lang="en")],
148+
iri="http://edamontology.org",
149+
prefix="http://edamontology.org/",
150+
prefix_name="edam",
151+
see_also=["http://edamontology.org"],
152+
)
153+
edam_obo = model.OwlOntology(
154+
name="EDAM-OBO",
155+
label=[model.Label(text="EDAM Ontology (OBO)", lang="en")],
156+
iri="http://purl.obolibrary.org/obo/edam",
157+
prefix="http://purl.obolibrary.org/obo/edam#",
158+
prefix_name="edamobo",
159+
see_also=["http://edamontology.org"],
160+
)
161+
bfo = model.OwlOntology(
162+
name="BFO",
163+
label=[model.Label(text="Basic Formal Ontology", lang="en")],
164+
iri="http://purl.obolibrary.org/obo/",
165+
prefix="http://purl.obolibrary.org/obo/BFO_",
166+
prefix_name="bfo",
167+
see_also=["http://purl.obolibrary.org/obo/bfo"],
168+
)
169+
iao = model.OwlOntology(
170+
name="IAO",
171+
label=[model.Label(text="Information Artifact Ontology", lang="en")],
172+
iri="http://purl.obolibrary.org/obo/",
173+
prefix="http://purl.obolibrary.org/obo/IAO_",
174+
prefix_name="iao",
175+
see_also=["http://purl.obolibrary.org/obo/iao"],
176+
)
177+
omo = model.OwlOntology(
178+
name="OMO",
179+
label=[model.Label(text="Ontology for Media Objects", lang="en")],
180+
iri="http://purl.obolibrary.org/obo/",
181+
prefix="http://purl.obolibrary.org/obo/OMO_",
182+
prefix_name="omo",
183+
see_also=["http://purl.obolibrary.org/obo/omo"],
184+
)
185+
swo = model.OwlOntology(
186+
name="SWO",
187+
label=[model.Label(text="Software Ontology", lang="en")],
188+
iri="http://ebi.ac.uk/swo",
189+
prefix="http://ebi.ac.uk/swo/",
190+
prefix_name="swo",
191+
see_also=["http://ebi.ac.uk/swo/"],
192+
)
193+
obo_in_owl = model.OwlOntology(
194+
name="OBOInOWL",
195+
label=[model.Label(text="OBO In OWL", lang="en")],
196+
iri="http://www.geneontology.org/formats/oboInOwl",
197+
prefix="http://www.geneontology.org/formats/oboInOwl#",
198+
prefix_name="obo_in_owl",
199+
see_also=["http://www.geneontology.org/formats/oboInOwl"],
200+
)
201+
rdfs = model.OwlOntology(
202+
name="RDFS",
203+
label=[model.Label(text="RDF Schema", lang="en")],
204+
iri="http://www.w3.org/2000/01/rdf-schema",
205+
prefix="http://www.w3.org/2000/01/rdf-schema#",
206+
prefix_name="rdfs",
207+
see_also=["http://www.w3.org/2000/01/rdf-schema"],
208+
)
209+
210+
import_config = ImportConfig(
211+
ontology_name=ontology_name,
212+
ontologies=[
213+
mwo,
214+
pmdco,
215+
nfdi,
216+
owl,
217+
schema,
218+
skos,
219+
foaf,
220+
dc,
221+
bibo,
222+
vann,
223+
adms,
224+
dcat,
225+
dcterms,
226+
doap,
227+
edam,
228+
edam_obo,
229+
bfo,
230+
obo_in_owl,
231+
rdfs,
232+
iao,
233+
omo,
234+
swo,
235+
],
236+
file="https://raw.githubusercontent.com/ISE-FIZKarlsruhe/mwo/refs/heads/main/mwo.ttl",
237+
base_class=model.OwlClass,
238+
base_class_title="Category:OSW725a3cf5458f4daea86615fcbd0029f8", # OwlClass
239+
dump_files=True,
240+
dump_path=os.path.dirname(os.path.abspath(__file__)),
241+
dry_run=False,
242+
change_id="84b563ec-4fd4-41da-8038-9cc4891fde85",
243+
)
244+
parser_settings = ParserSettings()
245+
parser_settings.ensure_array.append("elucidation")
246+
parser_settings.ensure_multilang.append("elucidation")
247+
248+
# Import ontologies
249+
importer = OntologyImporter(osw=osw_obj, parser_settings=parser_settings)
250+
prefix_dict = importer.import_ontology(import_config)

‎setup.cfg‎

Lines changed: 2 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -41,7 +41,7 @@ package_dir =
4141
=src
4242

4343
# Require a min/specific Python version (comma-separated conditions)
44-
# python_requires = >=3.8
44+
python_requires = >=3.9
4545

4646
# Add here dependencies of your project (line-separated), e.g. requests>=2.2,<3.0.
4747
# Version specifiers like >=2.2,<3.0 avoid problems due to API changes in
@@ -53,6 +53,7 @@ install_requires =
5353
opensemantic.core
5454
pydantic>=1.10.17
5555
datamodel-code-generator>=0.51.0
56+
ruamel.yaml<0.19.0
5657
mwclient>=0.11.0
5758
rdflib
5859
PyLD

‎src/osw/utils/templates.py‎

Lines changed: 42 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -65,6 +65,47 @@ def helper_join(this, options, context, separator=None, intro=None, outro=None):
6565
return intro + separator.join(items) + outro
6666

6767

68+
def helper_replace(this, options, find, replace, flags=None):
69+
"""
70+
Replace text in block content with plain string or regex
71+
72+
Plain string replacement:
73+
{{#replace "old" "new"}}text{{/replace}}
74+
75+
Regex replacement (with flags):
76+
{{#replace "(\\d{3})-(\\d{3})-(\\d{4})" "(\\1) \\2-\\3" "g"}}555-123-4567{{/replace}}
77+
78+
Note: Use \\1, \\2 for capture groups in Python templates
79+
(or $1, $2 will be auto-converted from JavaScript style)
80+
"""
81+
82+
# Get the block content
83+
string = "".join(options["fn"](this))
84+
85+
# If flags provided, use regex mode
86+
if flags:
87+
# Convert JavaScript flags to Python re flags
88+
re_flags = 0
89+
if "i" in flags:
90+
re_flags |= re.IGNORECASE
91+
if "m" in flags:
92+
re_flags |= re.MULTILINE
93+
if "s" in flags:
94+
re_flags |= re.DOTALL
95+
# Note: 'g' flag is implicit in Python's re.sub (replaces all by default)
96+
97+
# Convert JavaScript-style backreferences ($1, $2) to Python style (\1, \2)
98+
replace_python = re.sub(r"\$(\d+)", r"\\\1", replace)
99+
replace_python = replace_python.replace("$&", r"\g<0>") # Full match
100+
101+
# Use re.sub
102+
result = re.sub(find, replace_python, string, flags=re_flags)
103+
return result
104+
105+
# Default: plain string replacement (replace all occurrences)
106+
return string.replace(find, replace)
107+
108+
68109
def eval_compiled_handlebars_template(
69110
compiled_template, data, helpers=None, partials=None, add_self_as_partial=True
70111
):
@@ -90,6 +131,7 @@ def eval_compiled_handlebars_template(
90131

91132
default_helpers = {
92133
"join": helper_join,
134+
"replace": helper_replace,
93135
}
94136
if helpers is None:
95137
helpers = {}

‎src/osw/wiki_tools.py‎

Lines changed: 12 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -134,8 +134,9 @@ class SearchParam(OswBaseModel):
134134

135135
query: Union[str, List[str]]
136136
parallel: Optional[bool] = None # is set to true if query is a list longer than 5
137-
debug: Optional[bool] = True
137+
debug: Optional[bool] = False
138138
limit: Optional[int] = 1000
139+
return_json: Optional[bool] = False
139140

140141
def __init__(self, **data):
141142
super().__init__(**data)
@@ -208,7 +209,7 @@ def prefix_search_(single_text):
208209

209210
def semantic_search(
210211
site: mwclient.client.Site, query: Union[str, List[str], SearchParam]
211-
) -> List[str]:
212+
) -> Union[List[str], List[dict]]:
212213
"""Semantic query
213214
214215
Parameters
@@ -240,13 +241,16 @@ def semantic_search_(single_query):
240241
single_query, len(result["query"]["results"])
241242
)
242243
)
244+
if query.return_json:
245+
return result
246+
243247
for page in result["query"]["results"].values():
244-
# why do we do the following?
245-
if "printouts" in page:
246-
title = page["fulltext"]
247-
if "#" not in title and query.debug:
248-
print(title)
249-
# original position of "page_list.append(title)" line
248+
title = page["fulltext"]
249+
exists = page["exists"]
250+
if "#" not in title and query.debug:
251+
print(title)
252+
# original position of "page_list.append(title)" line
253+
if exists == "1":
250254
page_list.append(title)
251255
return page_list
252256

0 commit comments

Comments
 (0)