pysema addition and also dcat catalog addition to the metadata.ttl

vliz-be-opsci · Nov 20, 2024 · bcfe032 · bcfe032
1 parent f892381
commit bcfe032
Show file tree

Hide file tree

Showing 5 changed files with 105 additions and 123 deletions.
diff --git a/pysubyt/commands.sh b/pysubyt/commands.sh
@@ -5,7 +5,7 @@ rm -rf ./outputs/*
 
 #then run the pysubyt command
 echo "running the pysubyt commands"
-python -m pysubyt -t ./templates/  \
+sema-subyt -t ./templates/  \
        -s contact ../src/data/contacts.json \
        -s main ../src/data/main_data.json \
        -s publications ../src/data/publications.json \

diff --git a/pysubyt/outputs/metadata.ttl b/pysubyt/outputs/metadata.ttl
@@ -3,6 +3,7 @@
 @prefix foaf: <http://xmlns.com/foaf/0.1/> .
 @prefix dc: <http://purl.org/dc/elements/1.1/> .
 @prefix da: <https://open-vocabularies.vliz.be/dmbonassistant#> .
+@prefix dcat: <http://www.w3.org/ns/dcat#> .
 @prefix schema: <http://schema.org/> .
 
 <https://orcid.org/0000-0003-0264-9631>
@@ -36,132 +37,125 @@
     .
 
 
-<base_uri> rdf:type foaf:Project ;
+
+<base_uri> rdf:type foaf:Project dcat:Catalog ;
     dc:title "Autonomous Reef Monitoring System" ;
     foaf:name "ARMS" ;
     dc:description "The European ARMS programme (ARMS-MBON) is a network of Autonomous Reef Monitoring Structures (ARMS) placed in the vicinity of marine stations, ports, marinas, and Long-Term Ecological Research (LTER) sites distributed over Europe and polar regions. The aim of ARMS-MBON is to assess the status of, and changes in, hard-bottom communities of near-coast environments, using genetic methods supplemented with image analysis and visual inspection methods. ARMS-MBON is part of GEO BONs Marine Biodiversity Observation Network (MBON)." ;
-    foaf:logo <https://www.researchobject.org/ro-crate/assets/img/ro-crate-w-text.svg>;
-    schema:mainEntityOfPage <base_uri>;
-    schema:hasPart [
-        rdf:type schema:ItemList;
-        schema:partOf <base_uri>;
-        schema:name "Profiles in Autonomous Reef Monitoring System" ;
-        schema:itemListElement <http://www.google.com>;
-        schema:itemListElement <http://www.google.com>;
-        schema:itemListElement <http://www.google.com>;
-        schema:itemListElement <http://www.google.com>;
-        ],
-    [
-        rdf:type schema:ItemList;
-        schema:partOf <base_uri>;
-        schema:name "Crates in Autonomous Reef Monitoring System" ;
-        schema:itemListElement <http://www.google.com>;
-        schema:itemListElement <base_uri/Data/QualityControlledData>;
-        schema:itemListElement <base_uri/Data/QualityControlledData/FromPlutoF>;
-        schema:itemListElement <http://www.google.com>;
-        schema:itemListElement <http://www.google.com>;
-        schema:itemListElement <http://www.google.com>;
-        schema:itemListElement <http://www.google.com>;
-        schema:itemListElement <http://www.google.com>;
-        ];
-
-    schema:hasPart <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley>;
-
-    schema:hasPart <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley>;
-
-    schema:hasPart <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley>;
-
-    schema:hasPart <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley>;
-
-    .
-
+    foaf:logo <https://www.researchobject.org/ro-crate/assets/img/ro-crate-w-text.svg> ;
+    schema:mainEntityOfPage <base_uri> ;
+    foas:homepage <base_uri> ;
+    dcat:dataset 
+    <http://www.google.com>,
+    <http://www.google.com>,
+    <http://www.google.com>,
+    <http://www.google.com>,
+
+    <base_urihttps://w3id.org/ro/crate/1/lod>,
+    <base_uri/Data/QualityControlledData>,
+    <base_uri/Data/QualityControlledData/FromPlutoF>,
+    <base_urihttps://w3id.org/ro/crate/1/lod>,
+    <base_uri>,
+    <base_urihttps://w3id.org/ro/crate/1/lod>,
+    <base_urihttps://w3id.org/ro/crate/1/lod>,
+    <base_urihttps://w3id.org/ro/crate/1/lod>
+    # 
+    # schema:hasPart <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley>;
+    # 
+    # schema:hasPart <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley>;
+    # 
+    # schema:hasPart <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley>;
+    # 
+    # schema:hasPart <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley>;
+    # 
 <https://www.researchobject.org/ro-crate/assets/img/ro-crate-w-text.svg> rdf:type foaf:Image 
     .
 
 
-<http://www.google.com> rdf:type schema:ListItem, schema:CreativeWork ;
+<http://www.google.com> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:title "Lorem ipsum" ;
     dc:description "Lorem ipsum dolor sit amet placerat facilisis felis mi in tempus eleifend pellentesque natoque etiam." ;
     schema:item <http://www.google.com> ;
     .
 
-<http://www.google.com> rdf:type schema:ListItem, schema:CreativeWork ;
+<http://www.google.com> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:title "Lorem ipsum" ;
     dc:description "Lorem ipsum dolor sit amet placerat facilisis felis mi in tempus eleifend pellentesque natoque etiam." ;
     schema:item <http://www.google.com> ;
     .
 
-<http://www.google.com> rdf:type schema:ListItem, schema:CreativeWork ;
+<http://www.google.com> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:title "Lorem ipsum" ;
     dc:description "Lorem ipsum dolor sit amet placerat facilisis felis mi in tempus eleifend pellentesque natoque etiam." ;
     schema:item <http://www.google.com> ;
     .
 
-<http://www.google.com> rdf:type schema:ListItem, schema:CreativeWork ;
+<http://www.google.com> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:title "Lorem ipsum" ;
     dc:description "Lorem ipsum dolor sit amet placerat facilisis felis mi in tempus eleifend pellentesque natoque etiam." ;
     schema:item <http://www.google.com> ;
     .
 
-<http://www.google.com> rdf:type schema:ListItem, schema:CreativeWork ;
+<https://w3id.org/ro/crate/1/lod> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:title "Lorem ipsum dolor" ;
     dc:description "Lorem ipsum dolor sit amet, consectetur adipiscing elit." ;
     schema:item <http://www.google.com> ;
     .
-<base_uri/Data/QualityControlledData> rdf:type schema:ListItem, schema:CreativeWork ;
+<base_uri/QualityControlledData> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:description "This is the repository for data taken from different sources, which are then quality controlled and combined" ;
-    schema:item <base_uri/Data/QualityControlledData>;
+    schema:item <base_uri/QualityControlledData>;
     .
-<base_uri/Data/QualityControlledData/FromPlutoF> rdf:type schema:ListItem, schema:CreativeWork ;
+<base_uri/QualityControlledData/FromPlutoF> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:description "The metadata taken from PlutoF, which are quality controlled" ;
-    schema:item <base_uri/Data/QualityControlledData/FromPlutoF>;
+    schema:item <base_uri/QualityControlledData/FromPlutoF>;
     .
-<http://www.google.com> rdf:type schema:ListItem, schema:CreativeWork ;
+<https://w3id.org/ro/crate/1/lod> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:title "Consectetur adipiscing" ;
     dc:description "Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis sed dapibus leo nec ornare diam sedasd commodo nibh ante facilisis bibendum dolor feugiat at." ;
     schema:item <http://www.google.com> ;
     .
-<http://www.google.com> rdf:type schema:ListItem, schema:CreativeWork ;
+<> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:title "Lorem ipsum dolor" ;
     dc:description "Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis sed dapibus leo nec ornare diam sedasd commodo nibh ante facilisis bibendum dolor feugiat at." ;
     schema:item <http://www.google.com> ;
     .
-<http://www.google.com> rdf:type schema:ListItem, schema:CreativeWork ;
+<https://w3id.org/ro/crate/1/lod> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:title "Consectetur adipiscing" ;
     dc:description "Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis sed dapibus leo nec ornare diam sedasd commodo nibh ante facilisis bibendum dolor feugiat at." ;
     schema:item <http://www.google.com> ;
     .
-<http://www.google.com> rdf:type schema:ListItem, schema:CreativeWork ;
+<https://w3id.org/ro/crate/1/lod> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:title "Lorem ipsum dolor" ;
     dc:description "Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis sed dapibus leo nec ornare diam sedasd commodo nibh ante facilisis bibendum dolor feugiat at." ;
     schema:item <http://www.google.com> ;
     .
-<http://www.google.com> rdf:type schema:ListItem, schema:CreativeWork ;
+<https://w3id.org/ro/crate/1/lod> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:title "Consectetur adipiscing" ;
     dc:description "Lorem ipsum dolor sit amet, consectetur adipiscing elit. Duis sed dapibus leo nec ornare diam sedasd commodo nibh ante facilisis bibendum dolor feugiat at." ;
     schema:item <http://www.google.com> ;
     .
-
-<https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> rdf:type schema:CreativeWork ;
-    dc:title "Something Something interesting" ;
-    dc:description "Text about something interesting here A" ;
-    schema:url <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> ;
-    .
-
-<https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> rdf:type schema:CreativeWork ;
-    dc:title "The cat in the hat" ;
-    dc:description "The cat in the hat sat on the mat. One hopes he didn't leave a fat cat track in the form of a shattered mat." ;
-    schema:url <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> ;
-    .
-
-<https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> rdf:type schema:CreativeWork ;
-    dc:title "The bone in the bone" ;
-    dc:description "Spooky scary skeletons send shivers down your spine. Shrieking skulls will shock your soul, seal your doom tonight." ;
-    schema:url <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> ;
-    .
-
-<https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> rdf:type schema:CreativeWork ;
-    dc:title "The eel in the seal caught the whale in the mail" ;
-    dc:description "The eel in the seal caught the whale in the mail. The whale in the mail was a very big whale." ;
-    schema:url <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> ;
-    .
+# 
+# <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> rdf:type schema:CreativeWork ;
+#     dc:title "Something Something interesting" ;
+#     dc:description "Text about something interesting here A" ;
+#     schema:url <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> ;
+#     .
+# 
+# <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> rdf:type schema:CreativeWork ;
+#     dc:title "The cat in the hat" ;
+#     dc:description "The cat in the hat sat on the mat. One hopes he didn't leave a fat cat track in the form of a shattered mat." ;
+#     schema:url <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> ;
+#     .
+# 
+# <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> rdf:type schema:CreativeWork ;
+#     dc:title "The bone in the bone" ;
+#     dc:description "Spooky scary skeletons send shivers down your spine. Shrieking skulls will shock your soul, seal your doom tonight." ;
+#     schema:url <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> ;
+#     .
+# 
+# <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> rdf:type schema:CreativeWork ;
+#     dc:title "The eel in the seal caught the whale in the mail" ;
+#     dc:description "The eel in the seal caught the whale in the mail. The whale in the mail was a very big whale." ;
+#     schema:url <https://www.youtube.com/watch?v=dQw4w9WgXcQ&ab_channel=RickAstley> ;
+#     .
+# 
diff --git a/pysubyt/sema.log b/pysubyt/sema.log
@@ -0,0 +1,10 @@
+2024-11-20 14:49:39 @clitools.py              [DEBUG ] sema-subyt parsed args ns=Namespace(logconf=None, name='metadata.ttl', set=[['contact', '../src/data/contacts.json'], ['main', '../src/data/main_data.json'], ['publications', '../src/data/publications.json'], ['project_crate', '../src/data/project_crates.json'], ['project_profile', '../src/data/project_profiles.json']], var=[['base_uri', 'base_uri']], templates='./templates/', input=None, output='outputs/metadata.ttl', force=False, mode=None, allow_repeated_sink_paths=False, conditional=False)
+2024-11-20 14:49:39 @sources.py               [DEBUG ] mapping ext 'json' to mime 'None'
+2024-11-20 14:49:40 @sources.py               [DEBUG ] mapping ext 'json' to mime 'None'
+2024-11-20 14:49:40 @sources.py               [DEBUG ] mapping ext 'json' to mime 'None'
+2024-11-20 14:49:40 @sources.py               [DEBUG ] mapping ext 'json' to mime 'None'
+2024-11-20 14:49:40 @sources.py               [DEBUG ] mapping ext 'json' to mime 'None'
+2024-11-20 14:49:40 @subyt.py                 [DEBUG ] Subyt initialized with {'template_name': 'metadata.ttl', '_inputs': {'contact': JsonFileSource('C:\Users\cedricd\Documents\GitHub\space-to-pages\src\data\contacts.json'), 'main': JsonFileSource('C:\Users\cedricd\Documents\GitHub\space-to-pages\src\data\main_data.json'), 'publications': JsonFileSource('C:\Users\cedricd\Documents\GitHub\space-to-pages\src\data\publications.json'), 'project_crate': JsonFileSource('C:\Users\cedricd\Documents\GitHub\space-to-pages\src\data\project_crates.json'), 'project_profile': JsonFileSource('C:\Users\cedricd\Documents\GitHub\space-to-pages\src\data\project_profiles.json')}, '_conditional': False, '_variables': {'base_uri': 'base_uri'}, '_generator_settings': GeneratorSettings('ignorecase,flatten,iteration'), '_sink': SingleFileSink('C:\Users\cedricd\Documents\GitHub\space-to-pages\pysubyt\outputs\metadata.ttl', False)}
+2024-11-20 14:49:40 @api.py                   [DEBUG ] processing item _ = None
+2024-11-20 14:49:40 @sinks.py                 [INFO  ] Creating outputs/metadata.ttl
+2024-11-20 14:49:40 @__main__.py              [DEBUG ] processing done
diff --git a/pysubyt/templates/metadata.ttl b/pysubyt/templates/metadata.ttl
@@ -3,6 +3,7 @@
 @prefix foaf: <http://xmlns.com/foaf/0.1/> .
 @prefix dc: <http://purl.org/dc/elements/1.1/> .
 @prefix da: <https://open-vocabularies.vliz.be/dmbonassistant#> .
+@prefix dcat: <http://www.w3.org/ns/dcat#> .
 @prefix schema: <http://schema.org/> .
 
 {% for contacte in sets['contact'] -%}
@@ -15,62 +16,39 @@
     .
 
 {% endfor %}
+
 {% for maindata in sets['main'] -%}
-<{{base_uri}}> rdf:type foaf:Project ;
+<{{base_uri}}> rdf:type foaf:Project dcat:Catalog ;
     dc:title "{{maindata.long_name}}" ;
     foaf:name "{{maindata.name_site}}" ;
     dc:description "{{maindata.description}}" ;
-    foaf:logo <{{maindata.logo}}>;
-    schema:mainEntityOfPage <{{base_uri}}>;
-    schema:hasPart [
-        rdf:type schema:ItemList;
-        schema:partOf <{{base_uri}}>;
-        schema:name "Profiles in {{maindata.long_name}}" ;
-        {% for profile in sets['project_profile'] -%}
-        schema:itemListElement <{{profile.url}}>;
-        {% endfor -%}
-    ],
-    [
-        rdf:type schema:ItemList;
-        schema:partOf <{{base_uri}}>;
-        schema:name "Crates in {{maindata.long_name}}" ;
-        {% for crate in sets['project_crate'] -%}
-        {% if crate.index -%}
-        {% for index in crate.index -%}
-        schema:itemListElement <{{base_uri}}{{crate.url}}{{index.path}}>;
-        {% endfor -%}
-        {% else -%}
-        schema:itemListElement <{{base_uri}}{{crate.lodurl}}>;
-        {% endif -%}
-        {% endfor -%}
-    ],
-    [
-        rdf:type schema:ItemList;
-        schema:partOf <{{base_uri}}>;
-        schema:name "Vocabularies in {{maindata.long_name}}" ;
-        {% for vocab in sets['vocabulary'] -%}
-        schema:itemListElement <{{vocab}}>;
-        {% endfor -%}
-    ],
-    [
-        rdf:type schema:ItemList;
-        schema:partOf <{{base_uri}}>;
-        schema:name "Ontologies in {{maindata.long_name}}" ;
-        {% for vocab in sets['ontology'] -%}
-        schema:itemListElement <{{vocab}}>;
-        {% endfor -%}
-    ];
+    foaf:logo <{{maindata.logo}}> ;
+    schema:mainEntityOfPage <{{base_uri}}> ;
+    foas:homepage <{{base_uri}}> ;
+    dcat:dataset 
+    {% for profile in sets['project_profile'] -%}
+    <{{profile.url}}>,
+    {% endfor %} 
+    {% for crate in sets['project_crate'] -%}
+    {% if crate.index -%}
+    {% for index in crate.index -%}
+    <{{base_uri}}{{crate.url}}{{index.path}}>,
+    {% endfor -%}
+    {% else -%}
+    <{{base_uri}}{{crate.lodurl}}>{% if not crate == sets['project_crate']|list|last %},{% endif %}
+    {% endif -%}
+    {% endfor -%}
+
+
     # {% for publication in sets['publications']%}
     # schema:hasPart <{{publication.link}}>;
     # {% endfor %}
-    .
-
 <{{maindata.logo}}> rdf:type foaf:Image 
     .
-
 {% endfor %}
+
 {% for profile in sets['project_profile'] -%}
-<{{profile.url}}> rdf:type schema:ListItem, schema:CreativeWork ;
+<{{profile.url}}> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:title "{{profile.title}}" ;
     dc:description "{{profile.text}}" ;
     schema:item <{{profile.url}}> ;
@@ -80,13 +58,13 @@
 {% for crate in sets['project_crate'] -%}
 {% if crate.index -%}
 {% for index in crate.index -%}
-<{{base_uri}}{{crate.lodurl}}{{index.path}}> rdf:type schema:ListItem, schema:CreativeWork ;
+<{{base_uri}}{{crate.lodurl}}{{index.path}}> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:description "{{index.label}}" ;
     schema:item <{{base_uri}}{{crate.lodurl}}{{index.path}}>;
     .
 {% endfor -%}
 {% else -%}
-<{{crate.lodurl}}> rdf:type schema:ListItem, schema:CreativeWork ;
+<{{crate.lodurl}}> rdf:type dcat:Dataset, schema:CreativeWork ;
     dc:title "{{crate.name}}" ;
     dc:description "{{crate.text}}" ;
     schema:item <{{crate.url}}> ;

diff --git a/requirements.txt b/requirements.txt
@@ -1 +1 @@
-git+https://github.com/vliz-be-opsci/pysembench.git@main
+git+https://github.com/vliz-be-opsci/py-sema.git@main
Original file line number	Diff line number	Diff line change
		@@ -1 +1 @@
		git+https://github.com/vliz-be-opsci/pysembench.git@main
		git+https://github.com/vliz-be-opsci/py-sema.git@main