main: missing directories are now created if user agrees

chatmastermind/configuration.py

@@ -118,6 +118,7 @@ class Config:
     # a default configuration
     cache: str = '.'
     db: str = './db/'
+    glossaries: str | None = './glossaries/'
     ais: dict[str, AIConfig] = field(default_factory=create_default_ai_configs)
 
     @classmethod
@@ -135,7 +136,8 @@ class Config:
         return cls(
             cache=str(source['cache']) if 'cache' in source else '.',
             db=str(source['db']),
-            ais=ais
+            ais=ais,
+            glossaries=str(source['glossaries']) if 'glossaries' in source else None
         )
 
     @classmethod

chatmastermind/glossary.py

@@ -30,9 +30,10 @@ class Glossary:
     """
     A glossary consists of the following parameters:
         - Name (freely selectable)
-        - Path (full file path)
+        - Path (full file path, suffix is automatically generated)
         - Source language
         - Target language
+        - Description (optional)
         - Entries (pairs of source lang and target lang terms)
         - ID (automatically generated / modified, required by DeepL)
     """
@@ -40,8 +41,9 @@ class Glossary:
     name: str
     source_lang: str
     target_lang: str
-    entries: dict[str, str] = field(default_factory=lambda: dict())
     file_path: Path | None = None
+    desc: str | None = None
+    entries: dict[str, str] = field(default_factory=lambda: dict())
     ID: str | None = None
     file_suffix: ClassVar[str] = '.glo'
 
@@ -56,15 +58,18 @@ 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)
-                # remove any quotes from the entries that YAML may have added while dumping
-                # (e. g. for special keywords like 'yes')
-                clean_entries = {key.strip('\"\' '): value for key, value in data['Entries'].items()}
+                # 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'],
                            target_lang=data['TargetLang'],
-                           entries=clean_entries,
                            file_path=file_path,
+                           desc=data['Description'],
+                           entries=clean_entries,
                            ID=data['ID'] if data['ID'] != 'None' else None)
             except Exception:
                 raise GlossaryError(f"'{file_path}' does not contain a valid glossary")
@@ -86,6 +91,7 @@ class Glossary:
         with tempfile.NamedTemporaryFile(dir=self.file_path.parent, prefix=self.file_path.name, mode="w", delete=False) as temp_fd:
             temp_file_path = Path(temp_fd.name)
             data = {'Name': self.name,
+                    'Description': self.desc,
                     'ID': str(self.ID),
                     'SourceLang': self.source_lang,
                     'TargetLang': self.target_lang,
@@ -136,3 +142,20 @@ class Glossary:
                         self.entries[parts[0]] = parts[1]
         except Exception as e:
             raise GlossaryError(f"Error importing TSV: {e}")
+
+    def to_str(self, with_entries: bool = False) -> str:
+        """
+        Return the current glossary as a string.
+        """
+        output: list[str] = []
+        output.append(f'{self.name} (ID: {self.ID}):')
+        if self.desc:
+            output.append('- ' + self.desc)
+        output.append(f'- Languages: {self.source_lang} -> {self.target_lang}')
+        if with_entries:
+            output.append('- Entries:')
+            for source, target in self.entries.items():
+                output.append(f'  {source} : {target}')
+        else:
+            output.append(f'- Entries: {len(self.entries)}')
+        return '\n'.join(output)

chatmastermind/main.py

@@ -3,6 +3,7 @@
 # vim: set fileencoding=utf-8 :
 
 import sys
+import os
 import argcomplete
 import argparse
 from pathlib import Path
@@ -156,6 +157,46 @@ def create_parser() -> argparse.ArgumentParser:
     return parser
 
 
+def create_directories(config: Config) -> None:  # noqa: 11
+    """
+    Create the directories in the given configuration if they don't exist.
+    """
+    def make_dir(path: Path) -> None:
+        try:
+            os.makedirs(path.absolute())
+        except Exception as e:
+            print(f"Creating directory '{path.absolute()}' failed with: {e}")
+            sys.exit(1)
+    # Cache
+    cache_path = Path(config.cache)
+    if not cache_path.exists():
+        answer = input(f"Cache directory '{cache_path}' does not exist. Create it? [y/n]")
+        if answer.lower() in ['y', 'yes']:
+            make_dir(cache_path.absolute())
+        else:
+            print("Can't continue without a valid cache directory!")
+            sys.exit(1)
+    # DB
+    db_path = Path(config.db)
+    if not db_path.exists():
+        answer = input(f"DB directory '{db_path}' does not exist. Create it? [y/n]")
+        if answer.lower() in ['y', 'yes']:
+            make_dir(db_path.absolute())
+        else:
+            print("Can't continue without a valid DB directory!")
+            sys.exit(1)
+    # Glossaries
+    if config.glossaries:
+        glossaries_path = Path(config.glossaries)
+        if not glossaries_path.exists():
+            answer = input(f"Glossaries directory '{glossaries_path}' does not exist. Create it? [y/n]")
+            if answer.lower() in ['y', 'yes']:
+                make_dir(glossaries_path.absolute())
+            else:
+                print("Can't continue without a valid glossaries directory. Create it or remove it from the configuration.")
+                sys.exit(1)
+
+
 def main() -> int:
     parser = create_parser()
     args = parser.parse_args()
@@ -165,6 +206,7 @@ def main() -> int:
         command.func(command)
     else:
         config = Config.from_file(args.config)
+        create_directories(config)
         command.func(command, config)
 
     return 0

tests/test_configuration.py

@@ -71,11 +71,13 @@ class TestConfig(unittest.TestCase):
                     'frequency_penalty': 0.7,
                     'presence_penalty': 0.2
                 }
-            }
+            },
+            'glossaries': './glossaries/'
         }
         config = Config.from_dict(source_dict)
         self.assertEqual(config.cache, '.')
         self.assertEqual(config.db, './test_db/')
+        self.assertEqual(config.glossaries, './glossaries/')
         self.assertEqual(len(config.ais), 1)
         self.assertEqual(config.ais['myopenai'].name, 'openai')
         self.assertEqual(cast(OpenAIConfig, config.ais['myopenai']).system, 'Custom system')
@@ -105,6 +107,7 @@ class TestConfig(unittest.TestCase):
                     'frequency_penalty': 0.7,
                     'presence_penalty': 0.2
                 }
+                # omit glossaries, since it's optional
             }
         }
         with open(self.test_file.name, 'w') as f:
@@ -113,6 +116,8 @@ class TestConfig(unittest.TestCase):
         self.assertIsInstance(config, Config)
         self.assertEqual(config.cache, './test_cache/')
         self.assertEqual(config.db, './test_db/')
+        # missing 'glossaries' should result in 'None'
+        self.assertEqual(config.glossaries, None)
         self.assertEqual(len(config.ais), 1)
         self.assertIsInstance(config.ais['default'], AIConfig)
         self.assertEqual(cast(OpenAIConfig, config.ais['default']).system, 'Custom system')

tests/test_glossary.py

@@ -9,29 +9,73 @@ 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"
                             "ID: '123'\n"
                             "SourceLang: en\n"
                             "TargetLang: es\n"
                             "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)
         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"})
+        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:
         # Create glossary instance
-        glossary = Glossary(name="Test", source_lang="en", target_lang="fr", entries={"yes": "oui"})
+        glossary = Glossary(name="Test",
+                            desc="Test description",
+                            ID="666",
+                            source_lang="en",
+                            target_lang="fr",
+                            entries={"yes": "oui"})
 
         with tempfile.NamedTemporaryFile('w', delete=False, suffix=glossary_suffix) as tmp_file:
             file_path = Path(tmp_file.name)
@@ -41,6 +85,8 @@ class TestGlossary(unittest.TestCase):
             content = file.read()
 
         self.assertIn("Name: Test", content)
+        self.assertIn("Description: Test description", content)
+        self.assertIn("ID: '666'", content)
         self.assertIn("SourceLang: en", content)
         self.assertIn("TargetLang: fr", content)
         self.assertIn("Entries", content)
@@ -115,3 +161,44 @@ class TestGlossary(unittest.TestCase):
             glossary.to_file(file_path)
         assert glossary.file_path is not None
         self.assertEqual(glossary.file_path.suffix, glossary_suffix)
+
+    def test_to_str_with_id(self) -> None:
+        # Create a Glossary instance with an ID
+        glossary_with_id = Glossary(name="TestGlossary", source_lang="en", target_lang="fr",
+                                    desc="A simple test glossary", ID="1001", entries={"one": "un"})
+        glossary_str = glossary_with_id.to_str()
+        self.assertIn("TestGlossary (ID: 1001):", glossary_str)
+        self.assertIn("- A simple test glossary", glossary_str)
+        self.assertIn("- Languages: en -> fr", glossary_str)
+        self.assertIn("- Entries: 1", glossary_str)
+
+    def test_to_str_with_id_and_entries(self) -> None:
+        # Create a Glossary instance with an ID and include entries
+        glossary_with_entries = Glossary(name="TestGlossaryWithEntries", source_lang="en", target_lang="fr",
+                                         desc="Another test glossary", ID="2002",
+                                         entries={"hello": "salut", "goodbye": "au revoir"})
+        glossary_str_with_entries = glossary_with_entries.to_str(with_entries=True)
+        self.assertIn("TestGlossaryWithEntries (ID: 2002):", glossary_str_with_entries)
+        self.assertIn("- Entries:", glossary_str_with_entries)
+        self.assertIn("  hello : salut", glossary_str_with_entries)
+        self.assertIn("  goodbye : au revoir", glossary_str_with_entries)
+
+    def test_to_str_without_id(self) -> None:
+        # Create a Glossary instance without an ID
+        glossary_without_id = Glossary(name="TestGlossaryNoID", source_lang="en", target_lang="fr",
+                                       desc="A test glossary without an ID", ID=None, entries={"yes": "oui"})
+        glossary_str_no_id = glossary_without_id.to_str()
+        self.assertIn("TestGlossaryNoID (ID: None):", glossary_str_no_id)
+        self.assertIn("- A test glossary without an ID", glossary_str_no_id)
+        self.assertIn("- Languages: en -> fr", glossary_str_no_id)
+        self.assertIn("- Entries: 1", glossary_str_no_id)
+
+    def test_to_str_without_id_and_no_entries(self) -> None:
+        # Create a Glossary instance without an ID and no entries
+        glossary_no_id_no_entries = Glossary(name="EmptyGlossary", source_lang="en", target_lang="fr",
+                                             desc="An empty test glossary", ID=None, entries={})
+        glossary_str_no_id_no_entries = glossary_no_id_no_entries.to_str()
+        self.assertIn("EmptyGlossary (ID: None):", glossary_str_no_id_no_entries)
+        self.assertIn("- An empty test glossary", glossary_str_no_id_no_entries)
+        self.assertIn("- Languages: en -> fr", glossary_str_no_id_no_entries)
+        self.assertIn("- Entries: 0", glossary_str_no_id_no_entries)