FIX: Allowing Umlaut in cfood's $expression

CHANGELOG.md

@@ -21,6 +21,8 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed ###
 
+* [93](https://gitlab.com/linkahead/linkahead-crawler/-/issues/93) cfood.yaml does not allow umlaut in $expression
+
 ### Security ###
 
 ### Documentation ###

src/caoscrawler/converters.py

@@ -53,7 +53,7 @@ from .utils import has_parent
 # by the converters:
 SPECIAL_PROPERTIES = ("description", "name", "id", "path",
                       "file", "checksum", "size")
-SINGLE_VAR_RE = re.compile(r"^\$(\{)?(?P<varname>[0-9a-zA-Z_]+)(\})?$")
+SINGLE_VAR_RE = re.compile(r"^\$(\{)?(?P<varname>\w+)(\})?$")
 logger = logging.getLogger(__name__)
 
 

unittests/test_issues.py

@@ -22,13 +22,11 @@
 
 from pytest import mark
 
-import caosdb as db
-
+from caoscrawler.converters import replace_variables
 from caoscrawler.crawl import Crawler
-from caoscrawler.identifiable import Identifiable
-from caoscrawler.identifiable_adapters import CaosDBIdentifiableAdapter
-from caoscrawler.structure_elements import DictElement
 
+from caoscrawler.structure_elements import DictElement
+from caoscrawler.stores import GeneralStore
 from caoscrawler.scanner import create_converter_registry, scan_structure_elements
 
 
@@ -110,3 +108,19 @@ def test_list_datatypes():
     assert isinstance(records[0].get_property("Subject").value, list)
     assert records[0].get_property("Subject").datatype is not None
     assert records[0].get_property("Subject").datatype.startswith("LIST")
+
+
+def test_issue_93():
+    """https://gitlab.com/linkahead/linkahead-crawler/-/issues/93
+
+    cfood.yaml does not allow umlaut in $expression"""
+    values = GeneralStore()
+    expressions = [
+        "1",
+        "Ä",
+        "ųøîµ",
+    ]
+    for exp in expressions:
+        values[exp] = f"This is {exp}"
+    for exp in expressions:
+        assert replace_variables(f"${exp}", values) == f"This is {exp}"