Add sus updates

great_ai/tests/sus/test_clean.py

@@ -3,6 +3,7 @@ import unittest
 from src.great_ai.utilities.clean import clean
 
 
+
 class TestClean(unittest.TestCase):
     def test_xml_handling(self) -> None:
         xml = '<strong>Hi, </strong> my name<br/>is <span style="color: hotpink;"> András</span>! &lt;&#51; <> < ></><> &lt;&gt; <|'
@@ -59,6 +60,13 @@ class TestClean(unittest.TestCase):
 
         self.assertEqual(clean(text), cleaned)
 
+    def test_T_I_T_L_E_case(self) -> None:
+        text = "an a r t i  c l e is to F I N   D the purpose of a tree"
+
+        cleaned = "an article is to FIND the purpose of a tree"
+
+        self.assertEqual(clean(text), cleaned)
+
     def test_bracket_removal(self) -> None:
         text = "something [0], and [frefe, ferf]"
         cleaned = "something, and"

great_ai/tests/sus/test_get_sentences.py

@@ -3,6 +3,7 @@ import unittest
 from src.great_ai.utilities.get_sentences import get_sentences
 
 
+
 class TestGetSentences(unittest.TestCase):
     def test_default(self) -> None:
         text = "This is a complete sentence. So is this. However this is n"  # ot.

great_ai/tests/sus/test_lemmatize_text.py

@@ -57,7 +57,7 @@ class TestLemmatizeText(unittest.TestCase):
             "to_PART",
             "create_VERB",
             "a_DET",
-            "less_ADJ",
+            "less_ADV",
             "resource_NOUN",
             "-_PUNCT",
             "intensive_ADJ",
@@ -117,7 +117,7 @@ class TestLemmatizeText(unittest.TestCase):
             "to_PART",
             "create_VERB",
             "a_DET",
-            "less_ADJ",
+            "less_ADV",
             "resource_NOUN",
             "-_PUNCT",
             "intensive_ADJ",

great_ai/tests/sus/test_match_names.py

@@ -3,6 +3,7 @@ import unittest
 from src.great_ai.utilities.match_names.match_names import match_names
 
 
+
 class TestMatchNames(unittest.TestCase):
     def test_grid(self) -> None:
         names = [

great_ai/tests/sus/test_publication_tei.py

@@ -3,7 +3,17 @@ from pathlib import Path
 
 from src.great_ai.utilities.publication_tei import PublicationTEI
 
-from .data.parsed import authors, content, metadata, sentences
+
+from .data.parsed import (
+    abstract,
+    authors,
+    bookmarks,
+    conclusion,
+    content,
+    introduction,
+    metadata,
+    sentences,
+)
 
 DATA_PATH = Path(__file__).parent.resolve() / "data"
 
@@ -30,6 +40,20 @@ class TestPublicationTEI(unittest.TestCase):
     def test_sentences(self) -> None:
         self.assertEqual(PublicationTEI(self.test_xml).sentences, sentences)
 
+    def test_bookmarks(self) -> None:
+        self.assertEqual(PublicationTEI(self.test_xml).bookmarks, bookmarks)
+
+    def test_abstract(self) -> None:
+        self.assertEqual(PublicationTEI(self.test_xml).abstract_sentences, abstract)
+
+    def test_introduction(self) -> None:
+        self.assertEqual(
+            PublicationTEI(self.test_xml).introduction_sentences, introduction
+        )
+
+    def test_conclusion(self) -> None:
+        self.assertEqual(PublicationTEI(self.test_xml).conclusion_sentences, conclusion)
+
     def test_empty1(self) -> None:
         tei = PublicationTEI("<TEI/>")
         tei.publication_metadata, tei.publication_metadata, tei.authors, tei.content, tei.sentences

great_ai/tests/sus/test_unique.py

@@ -2,6 +2,7 @@ import unittest
 
 from src.great_ai.utilities.unique import unique
 
+
 original = [
     ("a", 1),
     ("b", 5),