From 2d94f8a10838859e9bc0fc51da6b1f66f21c41da Mon Sep 17 00:00:00 2001
From: juk0de <git@kadidlo.de>
Date: Mon, 5 Feb 2024 08:26:27 +0100
Subject: [PATCH] glossary: now supports quoted and unquoted entries (incl.
 tests)

---
 chatmastermind/glossary.py |  6 +++++-
 tests/test_glossary.py     | 44 ++++++++++++++++++++++++++++++++++----
 2 files changed, 45 insertions(+), 5 deletions(-)

diff --git a/chatmastermind/glossary.py b/chatmastermind/glossary.py
index b412ecf..3737fbb 100644
--- a/chatmastermind/glossary.py
+++ b/chatmastermind/glossary.py
@@ -58,7 +58,11 @@ class Glossary:
             raise GlossaryError(f"File type '{file_path.suffix}' is not supported")
         with open(file_path, "r") as fd:
             try:
-                data = yaml.load(fd, Loader=yaml.FullLoader)
+                # use BaseLoader so every entry is read as a string
+                # - disables automatic conversions
+                # - makes it possible to omit quoting for YAML keywords in entries (e. g. 'yes')
+                # - also correctly reads quoted entries
+                data = yaml.load(fd, Loader=yaml.BaseLoader)
                 clean_entries = data['Entries']
                 return cls(name=data['Name'],
                            source_lang=data['SourceLang'],
diff --git a/tests/test_glossary.py b/tests/test_glossary.py
index c1ac1ed..3991774 100644
--- a/tests/test_glossary.py
+++ b/tests/test_glossary.py
@@ -9,8 +9,10 @@ glossary_suffix: str = Glossary.file_suffix
 
 class TestGlossary(unittest.TestCase):
 
-    def test_from_file_valid_yaml(self) -> None:
-        # Prepare a temporary YAML file with valid content
+    def test_from_file_yaml_unquoted(self) -> None:
+        """
+        Test glossary creatiom from YAML with unquoted entries.
+        """
         with tempfile.NamedTemporaryFile('w', delete=False, suffix=glossary_suffix) as yaml_file:
             yaml_file.write("Name: Sample\n"
                             "Description: A brief description\n"
@@ -20,7 +22,9 @@ class TestGlossary(unittest.TestCase):
                             "Entries:\n"
                             "  hello: hola\n"
                             "  goodbye: adiós\n"
-                            "  'yes': sí\n")  # 'yes' is a YAML keyword and therefore quoted
+                            # 'yes' is a YAML keyword and would normally be quoted
+                            "  yes: sí\n"
+                            "  I'm going home: me voy a casa\n")
             yaml_file_path = Path(yaml_file.name)
 
         glossary = Glossary.from_file(yaml_file_path)
@@ -29,7 +33,39 @@ class TestGlossary(unittest.TestCase):
         self.assertEqual(glossary.ID, "123")
         self.assertEqual(glossary.source_lang, "en")
         self.assertEqual(glossary.target_lang, "es")
-        self.assertEqual(glossary.entries, {"hello": "hola", "goodbye": "adiós", "yes": "sí"})
+        self.assertEqual(glossary.entries, {"hello": "hola",
+                                            "goodbye": "adiós",
+                                            "yes": "sí",
+                                            "I'm going home": "me voy a casa"})
+        yaml_file_path.unlink()  # Remove the temporary file
+
+    def test_from_file_yaml_quoted(self) -> None:
+        """
+        Test glossary creatiom from YAML with quoted entries.
+        """
+        with tempfile.NamedTemporaryFile('w', delete=False, suffix=glossary_suffix) as yaml_file:
+            yaml_file.write("Name: Sample\n"
+                            "Description: A brief description\n"
+                            "ID: '123'\n"
+                            "SourceLang: en\n"
+                            "TargetLang: es\n"
+                            "Entries:\n"
+                            "  'hello': 'hola'\n"
+                            "  'goodbye': 'adiós'\n"
+                            "  'yes': 'sí'\n"
+                            "  \"I'm going home\": 'me voy a casa'\n")
+            yaml_file_path = Path(yaml_file.name)
+
+        glossary = Glossary.from_file(yaml_file_path)
+        self.assertEqual(glossary.name, "Sample")
+        self.assertEqual(glossary.desc, "A brief description")
+        self.assertEqual(glossary.ID, "123")
+        self.assertEqual(glossary.source_lang, "en")
+        self.assertEqual(glossary.target_lang, "es")
+        self.assertEqual(glossary.entries, {"hello": "hola",
+                                            "goodbye": "adiós",
+                                            "yes": "sí",
+                                            "I'm going home": "me voy a casa"})
         yaml_file_path.unlink()  # Remove the temporary file
 
     def test_to_file_writes_yaml(self) -> None: