Create Signposting from RO-Crate (#2)

* improvate jsonld parsing to be compatible with ro-crate

* refactoring

* example with RO-Crate

* bump version

---------

Co-authored-by: Daniel Bauer <daniel.bauer@senckenberg.de>

django_signposting/middleware.py

@@ -1,14 +1,12 @@
 from typing import Callable
 from django.http import HttpRequest, HttpResponse
-from signposting import Signpost, LinkRel
+from signposting import Signpost
 
 from bs4 import BeautifulSoup
-import re
 import json
 from django.utils.deprecation import MiddlewareMixin
 from django.conf import settings
-from rdflib import Graph
-from . import sparql
+from .utils import jsonld_to_signposts
 
 
 class SignpostingMiddleware:
@@ -45,68 +43,6 @@ class SignpostingMiddleware:
 
 
 class JsonLdSignpostingParserMiddleware(MiddlewareMixin):
-    def is_url(self, url: str) -> bool:
-        url_pattern = re.compile(
-            r"^(https?|ftp)://"  # protocol
-            r"(?:(?:[a-zA-Z0-9-_]+\.)?[a-zA-Z0-9-]+\.[a-zA-Z]{2,6})"  # domain
-            r"(?::\d{1,5})?"  # optional port
-            r"(?:/.*)?$"  # path
-        )
-        return bool(url_pattern.match(url))
-
-    def select_url(self, elements: tuple[str, ...]) -> str | None:
-        for elem in elements[::-1]:
-            if self.is_url(str(elem)):
-                return str(elem)
-        return None
-
-    def _jsonld_to_signposts(self, jsonld: dict) -> dict:
-        signposts = []
-        # TODO use jsonld context in query as prefix
-        g = Graph().parse(data=json.dumps(jsonld), format="json-ld")
-
-        rootElement = next(iter(sparql.root_element_query(g)), None)
-        if rootElement:
-            rootElement = rootElement[0]
-        else:
-            print("No root element found")
-            return {}
-
-        types = sparql.type_query(g, rootElement)
-        for type in types:
-            signposts.append(Signpost(LinkRel.type, str(type[0])))
-
-        authors = sparql.author_query(g, rootElement)
-        for author in authors:
-            author = self.select_url(author)
-            if author:
-                signposts.append(Signpost(LinkRel.author, author))
-
-        license = next(iter(sparql.license_query(g, rootElement)), [])
-        license = self.select_url(license)
-        if license:
-            signposts.append(Signpost(LinkRel.license, license))
-
-        citations = sparql.cite_query(g, rootElement)
-        for citation in citations:
-            citation = self.select_url(citation)
-            if citation:
-                signposts.append(Signpost(LinkRel.cite_as, citation))
-
-        sameas = sparql.sameas_query(g, rootElement)
-        for sa in sameas:
-            sa_media_type = sa[-1]
-            sa = self.select_url(sa[:-1])
-            if sa:
-                signposts.append(Signpost(LinkRel.describedby, sa, sa_media_type))
-
-        items = sparql.item_query(g, rootElement)
-        for item in items:
-            item_media_type = item[-1]
-            item = self.select_url(item[:-1])
-            if item:
-                signposts.append(Signpost(LinkRel.item, item, item_media_type))
-        return signposts
 
     def process_response(
         self, request: HttpRequest, response: HttpResponse
@@ -119,7 +55,7 @@ class JsonLdSignpostingParserMiddleware(MiddlewareMixin):
             for script in soup.find_all("script", type="application/ld+json"):
                 try:
                     jsonld = json.loads(script.string)
-                    signposts = self._jsonld_to_signposts(jsonld)
+                    signposts = jsonld_to_signposts(jsonld)
 
                     response._signposts = signposts
                 except json.JSONDecodeError as e:

django_signposting/sparql.py

@@ -14,7 +14,21 @@ WHERE {
   (schema:license | schema:author | schema:creator ) ?value .
 
   # No incoming edges for ?rootElement
-  FILTER NOT EXISTS { ?s ?p ?rootElement }
+  FILTER NOT EXISTS {
+    ?incoming ?p ?rootElement .
+
+    # also allow incoming edges that link entities BACK to the dataset
+    FILTER(?p != schema:isPartOf)
+    FILTER(?p != schema:mainEntityOfPage)
+    FILTER(?p != schema:recordedIn)
+    FILTER(?p != schema:exampleOfWork)
+    FILTER(?p != schema:includedInDataCatalogue)
+    FILTER(?p != schema:subjectOf)
+    FILTER(?p != schema:dataset)
+
+    # allow incoming edges from ro-crate-metadata.json
+    FILTER(CONTAINS(?incoming, "ro-crate-metadata.json")) .
+  }
 }
 LIMIT 1
 """)

django_signposting/utils.py

@@ -1,5 +1,10 @@
+import json
 from django.http import HttpResponse
-from signposting import Signpost
+from rdflib import Graph
+from signposting import Signpost, LinkRel
+import re
+
+from django_signposting import sparql
 
 
 def add_signposts(response: HttpResponse, *args: Signpost):
@@ -15,3 +20,67 @@ def add_signposts(response: HttpResponse, *args: Signpost):
     for signpost in args:
         if signpost not in response._signposts:
             response._signposts.append(signpost)
+
+
+def select_url(elements: tuple[str, ...]) -> str | None:
+    def is_url(url: str) -> bool:
+        url_pattern = re.compile(
+            r"^(https?|ftp)://"  # protocol
+            r"(?:(?:[a-zA-Z0-9-_]+\.)?[a-zA-Z0-9-]+\.[a-zA-Z]{2,6})"  # domain
+            r"(?::\d{1,5})?"  # optional port
+            r"(?:/.*)?$"  # path
+        )
+        return bool(url_pattern.match(url))
+
+    for elem in elements[::-1]:
+        if is_url(str(elem)):
+            return str(elem)
+    return None
+
+
+def jsonld_to_signposts(jsonld: dict) -> []:
+    signposts = []
+    # TODO use jsonld context in query as prefix
+    g = Graph().parse(data=json.dumps(jsonld), format="json-ld")
+
+    rootElement = next(iter(sparql.root_element_query(g)), None)
+    if rootElement:
+        rootElement = rootElement[0]
+    else:
+        print("No root element found")
+        return {}
+    types = sparql.type_query(g, rootElement)
+    for type in types:
+        signposts.append(Signpost(LinkRel.type, str(type[0])))
+
+    authors = sparql.author_query(g, rootElement)
+    for author in authors:
+        author = select_url(author)
+        if author:
+            signposts.append(Signpost(LinkRel.author, author))
+
+    license = next(iter(sparql.license_query(g, rootElement)), [])
+    license = select_url(license)
+    if license:
+        signposts.append(Signpost(LinkRel.license, license))
+
+    citations = sparql.cite_query(g, rootElement)
+    for citation in citations:
+        citation = select_url(citation)
+        if citation:
+            signposts.append(Signpost(LinkRel.cite_as, citation))
+
+    sameas = sparql.sameas_query(g, rootElement)
+    for sa in sameas:
+        sa_media_type = sa[-1]
+        sa = select_url(sa[:-1])
+        if sa:
+            signposts.append(Signpost(LinkRel.describedby, sa, sa_media_type))
+
+    items = sparql.item_query(g, rootElement)
+    for item in items:
+        item_media_type = item[-1]
+        item = select_url(item[:-1])
+        if item:
+            signposts.append(Signpost(LinkRel.item, item, item_media_type))
+    return signposts