Merge pull request #20 from acdh-oeaw/17-upgrade-extract_begin_end-fu…

…nction

feature: hardend `extract_begin_end` #17

README.md

@@ -299,7 +299,7 @@ print(normalize_string(string))
 
 ### extract date attributes (begin, end)
 
-expects typical TEI date attributes like `@when, @when-iso, @notBefore, @notAfter` and returns a tuple containg start- and enddate values. If only `@when or @when-iso` or only `@notBefore or @notAfter` are provided, the returned values are the same
+expects typical TEI date attributes like `@when, @when-iso, @notBefore, @notAfter, @from, @to, ...` and returns a tuple containg start- and enddate values. If only `@when or @when-iso` or only `@notBefore or @notAfter` are provided, the returned values are the same, unless the default parameter `fill_missing` is set to false. 
 
 ```python
 from lxml.etree import Element
@@ -313,6 +313,14 @@ print(extract_begin_end(date_object))
 # returns
 # ('1900-12-12', '1900-12-12')
 
+date_string = "1900-12-12"
+date_object = Element("{http://www.tei-c.org/ns/1.0}tei")
+date_object.attrib["when-iso"] = date_string
+print(extract_begin_end(date_object, fill_missing=False))
+
+# returns
+# ('1900-12-12', None)
+
 date_object = Element("{http://www.tei-c.org/ns/1.0}tei")
 date_object.attrib["notAfter"] = "1900-12-12"
 date_object.attrib["notBefore"] = "1800"

acdh_cidoc_pyutils/__init__.py

@@ -5,7 +5,7 @@
 from rdflib import Graph, Literal, URIRef, XSD, RDF, RDFS, OWL
 from slugify import slugify
 from acdh_tei_pyutils.utils import make_entity_label
-from acdh_cidoc_pyutils.namespaces import CIDOC, NSMAP
+from acdh_cidoc_pyutils.namespaces import CIDOC, NSMAP, DATE_ATTRIBUTE_DICT
 
 
 def normalize_string(string: str) -> str:
@@ -45,19 +45,41 @@ def coordinates_to_p168(
     return g
 
 
-def extract_begin_end(date_object: Union[Element, dict]) -> tuple[str, str]:
-    begin, end = "", ""
-    if date_object.get("when-iso", "") != "":
-        return (date_object.get("when-iso"), date_object.get("when-iso"))
-    elif date_object.get("when", "") != "":
-        return (date_object.get("when"), date_object.get("when"))
-    begin = date_object.get("notBefore", "")
-    end = date_object.get("notAfter", "")
-    if begin != "" and end == "":
-        end = begin
-    if end != "" and begin == "":
-        begin = end
-    return (begin, end)
+def extract_begin_end(
+    date_object: Union[Element, dict],
+    fill_missing=True,
+    attribute_map=DATE_ATTRIBUTE_DICT,
+) -> tuple[Union[str, bool], Union[str, bool]]:
+    final_start, final_end = None, None
+    start, end, when = None, None, None
+    for key, value in attribute_map.items():
+        date_value = date_object.get(key)
+        if date_value and value == "start":
+            start = date_value
+        if date_value and value == "end":
+            end = date_value
+        if date_value and value == "when":
+            when = date_value
+    if fill_missing:
+        if start or end or when:
+            if start and end:
+                final_start, final_end = start, end
+            elif start and not end and not when:
+                final_start, final_end = start, start
+            elif end and not start and not when:
+                final_start, final_end = end, end
+            elif when and not start and not end:
+                final_start, final_end = when, when
+    else:
+        if start and end:
+            final_start, final_end = start, end
+        elif start and not end and not when:
+            final_start, final_end = start, None
+        elif end and not start and not when:
+            final_start, final_end = None, end
+        elif when and not start and not end:
+            final_start, final_end = when, when
+    return final_start, final_end
 
 
 def date_to_literal(date_str: str) -> Literal:

acdh_cidoc_pyutils/namespaces.py

@@ -7,3 +7,16 @@
     "tei": "http://www.tei-c.org/ns/1.0",
     "xml": "http://www.w3.org/XML/1998/namespace",
 }
+
+DATE_ATTRIBUTE_DICT = {
+    "notBefore": "start",
+    "notBefore-iso": "start",
+    "from": "start",
+    "from-iso": "start",
+    "notAfter": "end",
+    "notAfter-iso": "end",
+    "to": "end",
+    "to-iso": "end",
+    "when": "when",
+    "when-iso": "when"
+}

tests/test_cidoc_pyutils.py

@@ -144,37 +144,61 @@ def test_006_begin_end(self):
         date_object = Element("hansi")
         date_object.attrib["when-iso"] = date_string
         begin, end = extract_begin_end(date_object)
-        self.assertTrue(begin, date_string)
-        self.assertTrue(end, date_string)
+        self.assertEqual(begin, date_string)
+        self.assertEqual(end, date_string)
+        begin, end = extract_begin_end(date_object, fill_missing=False)
+        self.assertEqual(begin, date_string)
+        self.assertEqual(end, date_string)
+
+        date_string = "1900-12-12"
+        date_object = Element("hansi")
+        date_object.attrib["from-iso"] = date_string
+        begin, end = extract_begin_end(date_object, fill_missing=False)
+        self.assertEqual(begin, date_string)
+        self.assertEqual(end, None)
 
         date_string = "1900-12-12"
         date_object = Element("hansi")
         date_object.attrib["when"] = date_string
         begin, end = extract_begin_end(date_object)
-        self.assertTrue(begin, date_string)
-        self.assertTrue(end, date_string)
+        self.assertEqual(begin, date_string)
+        self.assertEqual(end, date_string)
 
         date_string = "1900-12-12"
         date_object = Element("hansi")
         date_object.attrib["notAfter"] = date_string
         begin, end = extract_begin_end(date_object)
-        self.assertTrue(begin, date_string)
-        self.assertTrue(end, date_string)
+        self.assertEqual(begin, date_string)
+        self.assertEqual(end, date_string)
 
         date_string = "1900-12-12"
         date_object = Element("hansi")
         date_object.attrib["notBefore"] = date_string
         begin, end = extract_begin_end(date_object)
-        self.assertTrue(begin, date_string)
-        self.assertTrue(end, date_string)
+        self.assertEqual(begin, date_string)
+        self.assertEqual(end, date_string)
 
         date_string = "1900-12-12"
         date_object = Element("hansi")
         date_object.attrib["notAfter"] = date_string
         date_object.attrib["notBefore"] = "1800"
         begin, end = extract_begin_end(date_object)
-        self.assertTrue(begin, "1800")
-        self.assertTrue(end, date_string)
+        self.assertEqual(begin, "1800")
+        self.assertEqual(end, date_string)
+
+        date_string = "1900-12-12"
+        date_object = Element("hansi")
+        date_object.attrib["notAfter"] = date_string
+        date_object.attrib["notBefore"] = "1800"
+        begin, end = extract_begin_end(date_object, fill_missing=False)
+        self.assertEqual(begin, "1800")
+        self.assertEqual(end, date_string)
+        date_string = "1900-12-12"
+        date_object = Element("hansi")
+        date_object.attrib["to"] = date_string
+        begin, end = extract_begin_end(date_object, fill_missing=False)
+        self.assertEqual(begin, None)
+        self.assertEqual(end, date_string)
 
     def test_007_make_appelations(self):
         g = Graph()