refactor: use iri util consistently (DEV-2951) (#642)

src/dsp_tools/commands/xmlupload/stash/construct_and_analyze_graph.py

@@ -7,6 +7,7 @@
 from lxml import etree
 
 from dsp_tools.commands.xmlupload.stash.graph_models import Cost, Edge, ResptrLink, XMLLink
+from dsp_tools.utils.iri_util import is_resource_iri
 
 
 def create_info_from_xml_for_graph(
@@ -54,7 +55,7 @@ def _create_resptr_link_objects(subject_id: str, resptr_prop: etree._Element) ->
     resptr_links = []
     for resptr in resptr_prop.getchildren():
         resptr.text = cast(str, resptr.text)
-        if not regex.search(r"https?://rdfh.ch/[a-fA-F0-9]{4}/[\w-]{22}", resptr.text):
+        if not is_resource_iri(resptr.text):
             link_object = ResptrLink(subject_id, resptr.text)
             # this UUID is so that the links that were stashed can be identified in the XML data file
             resptr.attrib["linkUUID"] = link_object.link_uuid

src/dsp_tools/models/helpers.py

@@ -9,6 +9,7 @@
 import regex
 
 from dsp_tools.models.exceptions import BaseError
+from dsp_tools.utils.iri_util import is_iri
 
 #
 # here we do some data typing that should help
@@ -32,15 +33,6 @@ class OntoIri:
 ContextType = dict[str, OntoIri]
 
 
-class IriTest:  # pylint: disable=too-few-public-methods
-    __iri_regexp = regex.compile("^(http)s?://([\\w\\.\\-~]+)?(:\\d{,6})?(/[\\w\\-~]+)*(#[\\w\\-~]*)?")
-
-    @classmethod
-    def test(cls, val: str) -> bool:
-        m = cls.__iri_regexp.match(val)
-        return m.span()[1] == len(val) if m else False
-
-
 @unique
 class Actions(Enum):
     Create = 1
@@ -225,7 +217,7 @@ def iri_from_prefix(self, prefix: str) -> Optional[str]:
         :return: The full IRI without trailing "#"
         """
         # if self.__is_iri(prefix):
-        if IriTest.test(prefix):
+        if is_iri(prefix):
             return prefix
         if self._context.get(prefix) is not None:
             return self._context.get(prefix).iri
@@ -243,7 +235,7 @@ def prefix_from_iri(self, iri: str) -> Optional[str]:
         :return: the prefix of this context element, or None, if not found
         """
         # if not self.__is_iri(iri):
-        if not IriTest.test(iri):
+        if not is_iri(iri):
             raise BaseError("String does not conform to IRI patter: " + iri)
         if iri.endswith("#"):
             iri = iri[:-1]
@@ -278,7 +270,7 @@ def get_qualified_iri(self, val: Optional[str]) -> Optional[str]:
         """
         if not val:
             return None
-        if IriTest.test(val):
+        if is_iri(val):
             return val
         tmp = val.split(":")
         if len(tmp) < 2:
@@ -313,7 +305,7 @@ def get_prefixed_iri(self, iri: Optional[str]) -> Optional[str]:
         if m and m.span()[1] == len(iri):
             return iri
 
-        if not IriTest.test(iri):
+        if not is_iri(iri):
             raise BaseError(f"The IRI '{iri}' does not conform to the IRI pattern.")
 
         split_point = iri.find("#")
@@ -356,7 +348,7 @@ def reduce_iri(self, iri_str: str, onto_name: Optional[str] = None) -> str:
         knora_api = self.prefix_from_iri("http://api.knora.org/ontology/knora-api/v2#")
         salsah_gui = self.prefix_from_iri("http://api.knora.org/ontology/salsah-gui/v2#")
 
-        if IriTest.test(iri_str):
+        if is_iri(iri_str):
             if self.get_prefixed_iri(iri_str):
                 iri_str = self.get_prefixed_iri(iri_str)
         tmp = iri_str.split(":")

src/dsp_tools/utils/iri_util.py

@@ -1,8 +1,14 @@
 import regex
 
+_iri_pattern = r"^(http)s?://[\w\.\-~]*(:\d{,6})?(/[\w\-~]+)*(#[\w\-~]*)?"
 _resource_iri_pattern = r"https?://rdfh.ch/[a-fA-F0-9]{4}/[\w-]{22}"
 
 
+def is_iri(s: str) -> bool:
+    """Checks whether a string is a valid IRI."""
+    return regex.fullmatch(_iri_pattern, s) is not None
+
+
 def is_resource_iri(s: str) -> bool:
     """Checks whether a string is a valid resource IRI."""
     return regex.fullmatch(_resource_iri_pattern, s) is not None