oarepo-runtime 1.5.36__tar.gz → 1.5.38__tar.gz

{oarepo_runtime-1.5.36/oarepo_runtime.egg-info → oarepo_runtime-1.5.38}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: oarepo-runtime
-Version: 1.5.36
+Version: 1.5.38
 Summary: A set of runtime extensions of Invenio repository
 Description-Content-Type: text/markdown
 License-File: LICENSE

oarepo_runtime-1.5.38/oarepo_runtime/services/files/components.py

@@ -0,0 +1,55 @@
+import mimetypes
+import os
+from invenio_records_resources.services.files.components import FileServiceComponent
+from marshmallow.exceptions import ValidationError
+
+class AllowedFileTypesComponent(FileServiceComponent):
+
+    def guess_content_type(self, filename: str | None) -> str | None:
+        if filename:
+            return mimetypes.guess_type(filename)[0] or "application/octet-stream"
+        return None
+
+    @property
+    def allowed_mimetypes(self):
+        """Returns files attribute (field) key."""
+        return getattr(self.service.config, "allowed_mimetypes", [])
+
+    def guess_extension(self, file, mimetype):
+        """File extension."""
+        # The ``ext`` property is used to in search to aggregate file types, and we want e.g. both ``.jpeg`` and
+        # ``.jpg`` to be aggregated under ``.jpg``
+        ext_guessed = mimetypes.guess_extension(mimetype)
+
+        # Check if a valid extension is guessed and it's not the default mimetype
+        if (
+                ext_guessed is not None
+                and mimetype != "application/octet-stream"
+        ):
+            return ext_guessed[1:]
+
+        # Support non-standard file extensions that cannot be guessed
+        _, ext = os.path.splitext(file)
+        if ext and len(ext) <= 5:
+            return ext[1:].lower()
+
+        if ext_guessed:
+            return ext_guessed[1:]
+
+    @property
+    def allowed_extensions(self):
+        """Returns files attribute (field) key."""
+        return getattr(self.service.config, "allowed_extensions", [])
+
+    def init_files(self, identity, id_, data, uow=None):
+        """Initialize the file upload for the record."""
+        list_files = list(data.files.entries)
+
+        for file in list_files:
+            allowed_type = self.guess_content_type(file)
+            allowed_ext = self.guess_extension(file, allowed_type)
+            if len(self.allowed_mimetypes) > 0 and allowed_type not in self.allowed_mimetypes:
+                raise ValidationError(f"Mimetype not supported, supported mimetypes: {self.allowed_mimetypes}")
+            elif len(self.allowed_extensions) > 0 and allowed_ext not in self.allowed_extensions:
+                raise ValidationError(f"Extension not supported, supported extensions: {self.allowed_extensions}")
+

{oarepo_runtime-1.5.36 → oarepo_runtime-1.5.38}/oarepo_runtime/services/search.py

@@ -12,6 +12,8 @@ from invenio_records_resources.services.records.params import (
     SortParam,
 )
 from invenio_records_resources.services.records.queryparser import SuggestQueryParser
+from invenio_search.engine import dsl
+
 
 # TODO: integrate this to invenio_records_resources.services.records and remove SearchOptions class
 from oarepo_runtime.i18n import lazy_gettext as _
@@ -25,6 +27,27 @@ except ImportError:
     from invenio_i18n.babel import get_locale
 
 
+class FuzzySuggestQueryParser(SuggestQueryParser):
+    def __init__(self, identity=None, extra_params=None, **kwargs):
+        """Constructor."""
+        super().__init__(identity=identity, extra_params=extra_params)
+        self.fields = self.extra_params.get("fields", [])
+        self.extra_params.setdefault("type", "bool_prefix")
+
+    def parse(self, query_str):
+        """Parse the query."""
+        # default behavior
+        multi_match_with_bool_prefix = dsl.Q(
+            "multi_match", query=query_str, **self.extra_params
+        )
+        # fuzziness does not seem to work with bool_prefix multimatch query, so we turn this into
+        # should multi match query with two clauses
+        multi_match_fuzzy = dsl.Q(
+            "multi_match", query=query_str, fields=self.fields, fuzziness="AUTO"
+        )
+        return dsl.Q("bool", should=[multi_match_with_bool_prefix, multi_match_fuzzy])
+
+
 class SearchOptions(InvenioSearchOptions):
     params_interpreters_cls = [
         QueryStrParam,
@@ -114,9 +137,7 @@ class ICUSuggestParser:
                     f"{fld.field}._index_prefix^{fld.boost * fld.boost_prefix}"
                 )
 
-        return SuggestQueryParser.factory(
-            fields=fields,
-        )
+        return FuzzySuggestQueryParser.factory(fields=fields)
 
 
 @dataclasses.dataclass

{oarepo_runtime-1.5.36 → oarepo_runtime-1.5.38/oarepo_runtime.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: oarepo-runtime
-Version: 1.5.36
+Version: 1.5.38
 Summary: A set of runtime extensions of Invenio repository
 Description-Content-Type: text/markdown
 License-File: LICENSE

{oarepo_runtime-1.5.36 → oarepo_runtime-1.5.38}/oarepo_runtime.egg-info/SOURCES.txt

@@ -101,6 +101,7 @@ oarepo_runtime/services/facets/max_facet.py
 oarepo_runtime/services/facets/nested_facet.py
 oarepo_runtime/services/facets/params.py
 oarepo_runtime/services/files/__init__.py
+oarepo_runtime/services/files/components.py
 oarepo_runtime/services/files/service.py
 oarepo_runtime/services/relations/__init__.py
 oarepo_runtime/services/relations/components.py

{oarepo_runtime-1.5.36 → oarepo_runtime-1.5.38}/setup.cfg

@@ -1,6 +1,6 @@
 [metadata]
 name = oarepo-runtime
-version = 1.5.36
+version = 1.5.38
 description = A set of runtime extensions of Invenio repository
 authors = Alzbeta Pokorna
 readme = README.md