fix: date validation forbids legal values (DEV-3059) (#668)

dasch-swiss · Dec 5, 2023 · 13131ff · 13131ff
1 parent 5b333c2
commit 13131ff
Show file tree

Hide file tree

Showing 5 changed files with 27 additions and 17 deletions.
diff --git a/src/dsp_tools/commands/excel2xml/excel2xml_lib.py b/src/dsp_tools/commands/excel2xml/excel2xml_lib.py
@@ -18,6 +18,7 @@
 from dsp_tools.commands.excel2xml.propertyelement import PropertyElement
 from dsp_tools.models.exceptions import BaseError
 from dsp_tools.models.helpers import DateTimeStamp
+from dsp_tools.utils.date_util import is_full_date
 from dsp_tools.utils.shared import check_notna, simplify_name, validate_xml_against_schema
 from dsp_tools.utils.uri_util import is_uri
 
@@ -660,13 +661,8 @@ def make_date_prop(
     values = prepare_value(value)
 
     # check value type
-    validation_regex = (
-        r"^(GREGORIAN:|JULIAN:)?(CE:|BCE:)?"
-        + r"(\d{4})(-\d{1,2})?(-\d{1,2})?"
-        + r"((:CE|:BCE)?(:\d{4})(-\d{1,2})?(-\d{1,2})?)?$"
-    )
     for val in values:
-        if not regex.search(validation_regex, str(val.value).strip()):
+        if not is_full_date(str(val.value).strip()):
             raise BaseError(
                 f"Failed validation in resource '{calling_resource}', property '{name}': "
                 f"'{val.value}' is not a valid DSP date."

diff --git a/src/dsp_tools/resources/schema/data.xsd b/src/dsp_tools/resources/schema/data.xsd
@@ -31,7 +31,7 @@
     <xs:simpleType name="knoradate_type">
         <xs:restriction base="xs:token">
             <xs:pattern
-                value="(GREGORIAN:|JULIAN:)?(CE:|BCE:)?(\d{4})(-\d{1,2})?(-\d{1,2})?((:CE|:BCE)?(:\d{4})(-\d{1,2})?(-\d{1,2})?)?"/>
+                value="(GREGORIAN:|JULIAN:|ISLAMIC:)?(CE:|BCE:|AD:|BC:)?(\d{1,4})(-\d{1,2})?(-\d{1,2})?((:CE|:BCE|:AD|:BC)?(:\d{1,4})(-\d{1,2})?(-\d{1,2})?)?"/>
         </xs:restriction>
     </xs:simpleType>
 

diff --git a/src/dsp_tools/utils/date_util.py b/src/dsp_tools/utils/date_util.py
@@ -8,16 +8,21 @@
 
 from dsp_tools.models.exceptions import BaseError
 
-_full_date_pattern = r"""
+_calendar = r"GREGORIAN|JULIAN|ISLAMIC"
+_era = r"CE|BCE|BC|AD"
+_year = r"\d{1,4}"
+_month = r"\d{1,2}"
+_day = r"\d{1,2}"
+_full_date_pattern = rf"""
 ^
-(?:(GREGORIAN|JULIAN|ISLAMIC):)?        # optional calendar
-(?:(CE|BCE|BC|AD):)?                    # optional era
-(\d{1,4}(?:-\d{1,2})?(?:-\d{1,2})?)       # date
-(?::(CE|BCE|BC|AD))?                    # optional era
-(?::(\d{4}(?:-\d{1,2})?(?:-\d{1,2})?))? # optional date
+(?:({_calendar}):)?                         # optional calendar
+(?:({_era}):)?                              # optional era
+({_year}(?:-{_month})?(?:-{_day})?)         # date
+(?::({_era}))?                              # optional era
+(?::({_year}(?:-{_month})?(?:-{_day})?))?   # optional date
 $
 """
-_single_date_pattern = r"^(\d{1,4})(?:-(\d{1,2}))?(?:-(\d{1,2}))?$"
+_single_date_pattern = rf"^({_year})(?:-({_month}))?(?:-({_day}))?$"
 
 
 class Calendar(Enum):
@@ -114,6 +119,13 @@ def parse_date_string(s: str) -> Date:
     return Date(calendar_enum, start, end)
 
 
+def is_full_date(s: str) -> bool:
+    """
+    Check if a string is a full DSP date string of the scheme calendar:epoch:yyyy-mm-dd:epoch:yyyy-mm-dd.
+    """
+    return bool(regex.search(_full_date_pattern, s, flags=regex.VERBOSE))
+
+
 def _split_date_string(s: str) -> tuple[str | None, str | None, str, str | None, str | None]:
     date_match = regex.search(_full_date_pattern, s, flags=regex.VERBOSE)
     if not date_match:

diff --git a/test/unittests/commands/excel2xml/test_excel2xml_lib.py b/test/unittests/commands/excel2xml/test_excel2xml_lib.py
@@ -396,8 +396,9 @@ def test_make_date_prop(self) -> None:
             "2022",
             "GREGORIAN:CE:0476-09-04:CE:0476-09-04",
             "GREGORIAN:CE:2014-01-31",
+            "JULIAN:BC:1:AD:200",
         ]
-        invalid_values = ["GREGORIAN:CE:0476-09-04:CE:09-04", "GREGORIAN:CE:0476-09-010:CE:0476-09-04"]
+        invalid_values = ["GREGORIAN:CE:0476-09-010:CE:0476-09-04"]
         run_test(self, prop, method, different_values, invalid_values)
 
     def test_make_decimal_prop(self) -> None:

diff --git a/testdata/xml-data/test-data-systematic.xml b/testdata/xml-data/test-data-systematic.xml
@@ -110,12 +110,13 @@
             <boolean permissions="prop-default">true</boolean>
         </boolean-prop>
         <date-prop name=":hasDate">
-            <date permissions="prop-default">JULIAN:BCE:0700:BCE:0600</date>
+            <date permissions="prop-default">JULIAN:BCE:70:CE:1</date>
             <date permissions="prop-default">CE:1849:CE:1850</date>
             <date permissions="prop-default">GREGORIAN:1848-01:1849-02</date>
             <date permissions="prop-default">2022</date>
-            <date permissions="prop-default">GREGORIAN:CE:0476-09-04:CE:0476-09-04</date>
+            <date permissions="prop-default">GREGORIAN:AD:0476-09-04:AD:0476-09-04</date>
             <date permissions="prop-default">GREGORIAN:CE:2014-01-31</date>
+            <date permissions="prop-default">ISLAMIC:1:100</date>
         </date-prop>
         <integer-prop name=":hasInteger">
             <integer permissions="prop-default">4711</integer>