translation: added test module for glossary

chatmastermind/ais/openai.py

@@ -68,6 +68,7 @@ class OpenAI(AI):
         self.ID = config.ID
         self.name = config.name
         self.config = config
+        openai.api_key = self.config.api_key
 
     def request(self,
                 question: Message,
@@ -80,7 +81,6 @@ class OpenAI(AI):
         nr. of messages in the 'AIResponse'.
         """
         self.encoding = tiktoken.encoding_for_model(self.config.model)
-        openai.api_key = self.config.api_key
         oai_chat, prompt_tokens = self.openai_chat(chat, self.config.system, question)
         tokens: Tokens = Tokens(prompt_tokens, 0, prompt_tokens)
         response = openai.ChatCompletion.create(

chatmastermind/commands/common.py

@@ -0,0 +1,69 @@
+"""
+Contains shared functions for the various CMM subcommands.
+"""
+
+import argparse
+from pathlib import Path
+from ..message import Message, MessageError, source_code
+
+
+def read_text_file(file: Path) -> str:
+    with open(file) as r:
+        content = r.read().strip()
+    return content
+
+
+def add_file_as_text(question_parts: list[str], file: str) -> None:
+    """
+    Add the given file as plain text to the question part list.
+    If the file is a Message, add the answer.
+    """
+    file_path = Path(file)
+    content: str
+    try:
+        message = Message.from_file(file_path)
+        if message and message.answer:
+            content = message.answer
+    except MessageError:
+        content = read_text_file(Path(file))
+    if len(content) > 0:
+        question_parts.append(content)
+
+
+def add_file_as_code(question_parts: list[str], file: str) -> None:
+    """
+    Add all source code from the given file. If no code segments can be extracted,
+    the whole content is added as source code segment. If the file is a Message,
+    extract the source code from the answer.
+    """
+    file_path = Path(file)
+    content: str
+    try:
+        message = Message.from_file(file_path)
+        if message and message.answer:
+            content = message.answer
+    except MessageError:
+        with open(file) as r:
+            content = r.read().strip()
+    # extract and add source code
+    code_parts = source_code(content, include_delims=True)
+    if len(code_parts) > 0:
+        question_parts += code_parts
+    else:
+        question_parts.append(f"```\n{content}\n```")
+
+
+def invert_input_tag_args(args: argparse.Namespace) -> None:
+    """
+    Changes the semantics of the INPUT tags for this command:
+    * not tags specified on the CLI -> no tags are selected
+    * empty tags specified on the CLI -> all tags are selected
+    """
+    if args.or_tags is None:
+        args.or_tags = set()
+    elif len(args.or_tags) == 0:
+        args.or_tags = None
+    if args.and_tags is None:
+        args.and_tags = set()
+    elif len(args.and_tags) == 0:
+        args.and_tags = None

chatmastermind/commands/question.py

@@ -3,9 +3,10 @@ import argparse
 from pathlib import Path
 from itertools import zip_longest
 from copy import deepcopy
+from .common import invert_input_tag_args, add_file_as_code, add_file_as_text
 from ..configuration import Config
 from ..chat import ChatDB, msg_location
-from ..message import Message, MessageFilter, MessageError, Question, source_code
+from ..message import Message, MessageFilter, Question
 from ..ai_factory import create_ai
 from ..ai import AI, AIResponse
 
@@ -14,47 +15,6 @@ class QuestionCmdError(Exception):
     pass
 
 
-def add_file_as_text(question_parts: list[str], file: str) -> None:
-    """
-    Add the given file as plain text to the question part list.
-    If the file is a Message, add the answer.
-    """
-    file_path = Path(file)
-    content: str
-    try:
-        message = Message.from_file(file_path)
-        if message and message.answer:
-            content = message.answer
-    except MessageError:
-        with open(file) as r:
-            content = r.read().strip()
-    if len(content) > 0:
-        question_parts.append(content)
-
-
-def add_file_as_code(question_parts: list[str], file: str) -> None:
-    """
-    Add all source code from the given file. If no code segments can be extracted,
-    the whole content is added as source code segment. If the file is a Message,
-    extract the source code from the answer.
-    """
-    file_path = Path(file)
-    content: str
-    try:
-        message = Message.from_file(file_path)
-        if message and message.answer:
-            content = message.answer
-    except MessageError:
-        with open(file) as r:
-            content = r.read().strip()
-    # extract and add source code
-    code_parts = source_code(content, include_delims=True)
-    if len(code_parts) > 0:
-        question_parts += code_parts
-    else:
-        question_parts.append(f"```\n{content}\n```")
-
-
 def create_msg_args(msg: Message, args: argparse.Namespace) -> argparse.Namespace:
     """
     Takes an existing message and CLI arguments, and returns modified args based
@@ -163,22 +123,6 @@ def repeat_messages(messages: list[Message], chat: ChatDB, args: argparse.Namesp
             make_request(ai, chat, message, msg_args)
 
 
-def invert_input_tag_args(args: argparse.Namespace) -> None:
-    """
-    Changes the semantics of the INPUT tags for this command:
-    * not tags specified on the CLI -> no tags are selected
-    * empty tags specified on the CLI -> all tags are selected
-    """
-    if args.or_tags is None:
-        args.or_tags = set()
-    elif len(args.or_tags) == 0:
-        args.or_tags = None
-    if args.and_tags is None:
-        args.and_tags = set()
-    elif len(args.and_tags) == 0:
-        args.and_tags = None
-
-
 def question_cmd(args: argparse.Namespace, config: Config) -> None:
     """
     Handler for the 'question' command.

chatmastermind/commands/translation.py

@@ -0,0 +1,105 @@
+import argparse
+import mimetypes
+from pathlib import Path
+from .common import invert_input_tag_args, read_text_file
+from ..configuration import Config
+from ..message import MessageFilter, Message, Question
+from ..chat import ChatDB, msg_location
+
+
+class TranslationCmdError(Exception):
+    pass
+
+
+text_separator: str = 'TEXT:'
+
+
+def assert_document_type_supported_openai(document_file: Path) -> None:
+    doctype = mimetypes.guess_type(document_file)
+    if doctype != 'text/plain':
+        raise TranslationCmdError("AI 'OpenAI' only supports document type 'text/plain''")
+
+
+def translation_prompt_openai(source_lang: str, target_lang: str) -> str:
+    """
+    Return the prompt for GPT that tells it to do the translation.
+    """
+    return f"Translate the text below the line {text_separator} from {source_lang} to {target_lang}."
+
+
+def create_message_openai(chat: ChatDB, args: argparse.Namespace) -> Message:
+    """
+    Create a new message from the given arguments and write it to the cache directory.
+
+    Message format
+    1. Translation prompt (tells GPT to do a translation)
+    2. Glossary (if specified as an argument)
+    3. User provided prompt enhancements
+    4. Translation separator
+    5. User provided text to be translated
+
+    The text to be translated is determined as a follows:
+    - if a document is provided in the arguments, translate its content
+    - if no document is provided, translate the last text argument
+
+    The other text arguments will be put into the "header" and can be used
+    to improve the translation prompt.
+    """
+    text_args: list[str] = []
+    if args.create is not None:
+        text_args = args.create
+    elif args.ask is not None:
+        text_args = args.ask
+    else:
+        raise TranslationCmdError("No input text found")
+
+    # extract user prompt and user text to be translated
+    user_text: str
+    user_prompt: str
+    if args.input_document is not None:
+        assert_document_type_supported_openai(Path(args.input_document))
+        user_text = read_text_file(Path(args.input_document))
+        user_prompt = '\n\n'.join([str(s) for s in text_args])
+    else:
+        user_text = text_args[-1]
+        user_prompt = '\n\n'.join([str(s) for s in text_args[:-1]])
+
+    # build full question string
+    # FIXME: add glossaries if given
+    question_text: str = '\n\n'.join([translation_prompt_openai(args.source_lang, args.target_lang),
+                                      user_prompt,
+                                      text_separator,
+                                      user_text])
+    # create and write the message
+    message = Message(question=Question(question_text),
+                      tags=args.output_tags,
+                      ai=args.AI,
+                      model=args.model)
+    # only write the new message to the cache,
+    # don't add it to the internal list
+    chat.cache_write([message])
+    return message
+
+
+def translation_cmd(args: argparse.Namespace, config: Config) -> None:
+    """
+    Handler for the 'translation' command. Creates and executes translation
+    requests based on the input and selected AI. Depending on the AI, the
+    whole process may be significantly different (e.g. DeepL vs OpenAI).
+    """
+    invert_input_tag_args(args)
+    mfilter = MessageFilter(tags_or=args.or_tags,
+                            tags_and=args.and_tags,
+                            tags_not=args.exclude_tags)
+    chat = ChatDB.from_dir(cache_path=Path(config.cache),
+                           db_path=Path(config.db),
+                           mfilter=mfilter,
+                           glob=args.glob,
+                           loc=msg_location(args.location))
+    # if it's a new translation, create and store it immediately
+    # FIXME: check AI type
+    if args.ask or args.create:
+        # message = create_message(chat, args)
+        create_message_openai(chat, args)
+    if args.create:
+        return

chatmastermind/glossary.py

@@ -0,0 +1,116 @@
+"""
+Module implementing glossaries for translations.
+"""
+import yaml
+import tempfile
+import shutil
+import csv
+from pathlib import Path
+from dataclasses import dataclass, field
+from typing import Type, TypeVar
+
+GlossaryInst = TypeVar('GlossaryInst', bound='Glossary')
+
+
+class GlossaryError(Exception):
+    pass
+
+
+@dataclass
+class Glossary:
+    """
+    A glossary consists of the following parameters:
+        - Name (freely selectable)
+        - Path (full file path)
+        - Source language
+        - Target language
+        - Entries (pairs of source lang and target lang terms)
+        - ID (automatically generated / modified, required by DeepL)
+    """
+
+    name: str
+    source_lang: str
+    target_lang: str
+    entries: dict[str, str] = field(default_factory=lambda: dict())
+    file_path: Path | None = None
+    ID: str | None = None
+
+    @classmethod
+    def from_file(cls: Type[GlossaryInst], file_path: Path) -> GlossaryInst:
+        """
+        Create a glossary from the given file.
+        """
+        with open(file_path, "r") as fd:
+            try:
+                data = yaml.load(fd, Loader=yaml.FullLoader)
+                return cls(name=data['Name'],
+                           source_lang=data['SourceLang'],
+                           target_lang=data['TargetLang'],
+                           entries=data['Entries'],
+                           file_path=file_path,
+                           ID=data['ID'] if data['ID'] != 'None' else None)
+            except Exception:
+                raise GlossaryError(f"'{file_path}' does not contain a valid glossary")
+
+    def to_file(self, file_path: Path | None = None) -> None:
+        """
+        Write glossary to given file.
+        """
+        if file_path:
+            self.file_path = file_path
+        if not self.file_path:
+            raise GlossaryError("Got no valid path to write glossary")
+        # write YAML
+        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,
+                    'ID': str(self.ID),
+                    'SourceLang': self.source_lang,
+                    'TargetLang': self.target_lang,
+                    'Entries': self.entries}
+            yaml.dump(data, temp_fd, sort_keys=False)
+        shutil.move(temp_file_path, self.file_path)
+
+    def export_csv(self, dictionary: dict[str, str], file_path: Path) -> None:
+        """
+        Export the 'entries' of this glossary to a file in CSV format (compatible with DeepL).
+        """
+        with open(file_path, 'w', newline='', encoding='utf-8') as csvfile:
+            writer = csv.writer(csvfile, delimiter=',', quotechar='"', quoting=csv.QUOTE_ALL)
+            for source_entry, target_entry in self.entries.items():
+                writer.writerow([source_entry, target_entry])
+
+    def export_tsv(self, entries: dict[str, str], file_path: Path) -> None:
+        """
+        Export the 'entries' of this glossary to a file in TSV format (compatible with DeepL).
+        """
+        with open(file_path, 'w', encoding='utf-8') as file:
+            for source_entry, target_entry in self.entries.items():
+                file.write(f"{source_entry}\t{target_entry}\n")
+
+    def import_csv(self, file_path: Path) -> None:
+        """
+        Import the entries from the given CSV file to those of the current glossary.
+        Existing entries are overwritten.
+        """
+        try:
+            with open(file_path, mode='r', encoding='utf-8') as csvfile:
+                reader = csv.reader(csvfile, delimiter=',', quotechar='"')
+                self.entries = {rows[0]: rows[1] for rows in reader if len(rows) >= 2}
+        except Exception as e:
+            raise GlossaryError(f"Error importing CSV: {e}")
+
+    def import_tsv(self, file_path: Path) -> None:
+        """
+        Import the entries from the given CSV file to those of the current glossary.
+        Existing entries are overwritten.
+        """
+        try:
+            with open(file_path, mode='r', encoding='utf-8') as tsvfile:
+                self.entries = {}
+                for line in tsvfile:
+                    parts = line.strip().split('\t')
+                    if len(parts) == 2:
+                        self.entries[parts[0]] = parts[1]
+        except Exception as e:
+            raise GlossaryError(f"Error importing TSV: {e}")

chatmastermind/main.py

@@ -14,6 +14,7 @@ from .commands.tags import tags_cmd
 from .commands.config import config_cmd
 from .commands.hist import hist_cmd
 from .commands.print import print_cmd
+from .commands.translation import translation_cmd
 from .chat import msg_location
 
 
@@ -102,7 +103,7 @@ def create_parser() -> argparse.ArgumentParser:
     # 'tags' command parser
     tags_cmd_parser = cmdparser.add_parser('tags',
                                            help="Manage tags.",
-                                           aliases=['t'])
+                                           aliases=['T'])
     tags_cmd_parser.set_defaults(func=tags_cmd)
     tags_group = tags_cmd_parser.add_mutually_exclusive_group(required=True)
     tags_group.add_argument('-l', '--list', help="List all tags and their frequency",
@@ -136,6 +137,21 @@ def create_parser() -> argparse.ArgumentParser:
     print_cmd_modes.add_argument('-a', '--answer', help='Only print the answer', action='store_true')
     print_cmd_modes.add_argument('-S', '--only-source-code', help='Only print embedded source code', action='store_true')
 
+    # 'translation' command parser
+    translation_cmd_parser = cmdparser.add_parser('translation', parents=[ai_parser, tag_parser],
+                                                  help="ask, create and repeat translations.",
+                                                  aliases=['t'])
+    translation_cmd_parser.set_defaults(func=translation_cmd)
+    translation_group = translation_cmd_parser.add_mutually_exclusive_group(required=True)
+    translation_group.add_argument('-a', '--ask', nargs='+', help='Ask to translate the given text', metavar='TEXT')
+    translation_group.add_argument('-c', '--create', nargs='+', help='Create a translation', metavar='TEXT')
+    translation_group.add_argument('-r', '--repeat', nargs='*', help='Repeat a translation', metavar='MESSAGE')
+    translation_cmd_parser.add_argument('-S', '--source-lang', help="Source language", metavar="LANGUAGE", required=True)
+    translation_cmd_parser.add_argument('-T', '--target-lang', help="Target language", metavar="LANGUAGE", required=True)
+    translation_cmd_parser.add_argument('-G', '--glossaries', nargs='+', help="List of glossaries", metavar="GLOSSARY")
+    translation_cmd_parser.add_argument('-d', '--input-document', help="Document to translate", metavar="FILE")
+    translation_cmd_parser.add_argument('-D', '--output-document', help="Path for the translated document", metavar="FILE")
+
     argcomplete.autocomplete(parser)
     return parser
 

tests/test_glossary.py

@@ -0,0 +1,73 @@
+import unittest
+import tempfile
+from pathlib import Path
+from chatmastermind.glossary import Glossary
+
+
+class TestGlossary(unittest.TestCase):
+
+    def test_from_file_valid_yaml(self) -> None:
+        # Prepare a temporary YAML file with valid content
+        with tempfile.NamedTemporaryFile('w', delete=False) as yaml_file:
+            yaml_file.write("Name: Sample\n"
+                            "ID: '123'\n"
+                            "SourceLang: en\n"
+                            "TargetLang: es\n"
+                            "Entries:\n"
+                            "  hello: hola\n"
+                            "  goodbye: adiós\n")
+            yaml_file_path = Path(yaml_file.name)
+
+        glossary = Glossary.from_file(yaml_file_path)
+        self.assertEqual(glossary.name, "Sample")
+        self.assertEqual(glossary.source_lang, "en")
+        self.assertEqual(glossary.target_lang, "es")
+        self.assertEqual(glossary.entries, {"hello": "hola", "goodbye": "adiós"})
+        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"})
+
+        # Use a temporary file
+        with tempfile.NamedTemporaryFile('w', delete=False) as tmp_file:
+            file_path = Path(tmp_file.name)
+            glossary.to_file(file_path)
+
+        with open(file_path, 'r') as file:
+            content = file.read()
+
+        self.assertIn("Name: Test", content)
+        self.assertIn("SourceLang: en", content)
+        self.assertIn("TargetLang: fr", content)
+        self.assertIn("Entries", content)
+        self.assertIn("yes: oui", content)
+        file_path.unlink()  # Remove the temporary file
+
+    def test_import_export_csv(self) -> None:
+        glossary = Glossary(name="Test", source_lang="en", target_lang="fr", entries={})
+
+        # First export to CSV
+        with tempfile.NamedTemporaryFile('w', delete=False) as csvfile:
+            csv_file_path = Path(csvfile.name)
+        glossary.entries = {"hello": "salut", "goodbye": "au revoir"}
+        glossary.export_csv(glossary.entries, csv_file_path)
+
+        # Now import CSV
+        glossary.import_csv(csv_file_path)
+        self.assertEqual(glossary.entries, {"hello": "salut", "goodbye": "au revoir"})
+        csv_file_path.unlink()  # Remove the temporary file
+
+    def test_import_export_tsv(self) -> None:
+        glossary = Glossary(name="Test", source_lang="en", target_lang="fr", entries={})
+
+        # First export to TSV
+        with tempfile.NamedTemporaryFile('w', delete=False) as tsvfile:
+            tsv_file_path = Path(tsvfile.name)
+        glossary.entries = {"hello": "salut", "goodbye": "au revoir"}
+        glossary.export_tsv(glossary.entries, tsv_file_path)
+
+        # Now import TSV
+        glossary.import_tsv(tsv_file_path)
+        self.assertEqual(glossary.entries, {"hello": "salut", "goodbye": "au revoir"})
+        tsv_file_path.unlink()  # Remove the temporary file