message: improved robustness of Question and Answer content checks and tests

chatmastermind/main.py

@@ -7,10 +7,11 @@ import sys
 import argcomplete
 import argparse
 import pathlib
-from .utils import terminal_width, print_tag_args, print_chat_hist, display_source_code, print_tags_frequency, ChatType
-from .storage import save_answers, create_chat_hist, get_tags, get_tags_unique, read_file, dump_data
+from .utils import terminal_width, print_tag_args, print_chat_hist, display_source_code, ChatType
+from .storage import save_answers, create_chat_hist, get_tags_unique, read_file, dump_data
 from .api_client import ai, openai_api_key, print_models
 from .configuration import Config
+from .chat import ChatDB
 from itertools import zip_longest
 from typing import Any
 
@@ -61,8 +62,12 @@ def tag_cmd(args: argparse.Namespace, config: Config) -> None:
     """
     Handler for the 'tag' command.
     """
+    chat = ChatDB.from_dir(cache_path=pathlib.Path('.'),
+                           db_path=pathlib.Path(config.db))
     if args.list:
-        print_tags_frequency(get_tags(config, None))
+        tags_freq = chat.tags_frequency(args.prefix, args.contain)
+        for tag, freq in tags_freq.items():
+            print(f"- {tag}: {freq}")
 
 
 def config_cmd(args: argparse.Namespace, config: Config) -> None:
@@ -195,6 +200,8 @@ def create_parser() -> argparse.ArgumentParser:
     tag_group = tag_cmd_parser.add_mutually_exclusive_group(required=True)
     tag_group.add_argument('-l', '--list', help="List all tags and their frequency",
                            action='store_true')
+    tag_cmd_parser.add_argument('-p', '--prefix', help="Filter tags by prefix")
+    tag_cmd_parser.add_argument('-c', '--contain', help="Filter tags by contained substring")
 
     # 'config' command parser
     config_cmd_parser = cmdparser.add_parser('config',

chatmastermind/message.py

@@ -128,29 +128,29 @@ class ModelLine(str):
         return cls(' '.join([cls.prefix, model]))
 
 
-class Question(str):
+class Answer(str):
     """
-    A single question with a defined header.
+    A single answer with a defined header.
     """
-    tokens: int = 0   # tokens used by this question
-    txt_header: ClassVar[str] = '=== QUESTION ==='
-    yaml_key: ClassVar[str] = 'question'
+    tokens: int = 0  # tokens used by this answer
+    txt_header: ClassVar[str] = '=== ANSWER ==='
+    yaml_key: ClassVar[str] = 'answer'
 
-    def __new__(cls: Type[QuestionInst], string: str) -> QuestionInst:
+    def __new__(cls: Type[AnswerInst], string: str) -> AnswerInst:
         """
-        Make sure the question string does not contain the header.
+        Make sure the answer string does not contain the header as a whole line.
         """
-        if cls.txt_header in string:
-            raise MessageError(f"Question '{string}' contains the header '{cls.txt_header}'")
+        if cls.txt_header in string.split('\n'):
+            raise MessageError(f"Answer '{string}' contains the header '{cls.txt_header}'")
         instance = super().__new__(cls, string)
         return instance
 
     @classmethod
-    def from_list(cls: Type[QuestionInst], strings: list[str]) -> QuestionInst:
+    def from_list(cls: Type[AnswerInst], strings: list[str]) -> AnswerInst:
         """
         Build Question from a list of strings. Make sure strings do not contain the header.
         """
-        if any(cls.txt_header in string for string in strings):
+        if cls.txt_header in strings:
             raise MessageError(f"Question contains the header '{cls.txt_header}'")
         instance = super().__new__(cls, '\n'.join(strings).strip())
         return instance
@@ -162,29 +162,33 @@ class Question(str):
         return source_code(self, include_delims)
 
 
-class Answer(str):
+class Question(str):
     """
-    A single answer with a defined header.
+    A single question with a defined header.
     """
-    tokens: int = 0  # tokens used by this answer
-    txt_header: ClassVar[str] = '=== ANSWER ==='
-    yaml_key: ClassVar[str] = 'answer'
+    tokens: int = 0   # tokens used by this question
+    txt_header: ClassVar[str] = '=== QUESTION ==='
+    yaml_key: ClassVar[str] = 'question'
 
-    def __new__(cls: Type[AnswerInst], string: str) -> AnswerInst:
+    def __new__(cls: Type[QuestionInst], string: str) -> QuestionInst:
         """
-        Make sure the answer string does not contain the header.
+        Make sure the question string does not contain the header as a whole line
+        (also not that from 'Answer', so it's always clear where the answer starts).
         """
-        if cls.txt_header in string:
-            raise MessageError(f"Answer '{string}' contains the header '{cls.txt_header}'")
+        string_lines = string.split('\n')
+        if cls.txt_header in string_lines:
+            raise MessageError(f"Question '{string}' contains the header '{cls.txt_header}'")
+        if Answer.txt_header in string_lines:
+            raise MessageError(f"Question '{string}' contains the header '{Answer.txt_header}'")
         instance = super().__new__(cls, string)
         return instance
 
     @classmethod
-    def from_list(cls: Type[AnswerInst], strings: list[str]) -> AnswerInst:
+    def from_list(cls: Type[QuestionInst], strings: list[str]) -> QuestionInst:
         """
         Build Question from a list of strings. Make sure strings do not contain the header.
         """
-        if any(cls.txt_header in string for string in strings):
+        if cls.txt_header in strings:
             raise MessageError(f"Question contains the header '{cls.txt_header}'")
         instance = super().__new__(cls, '\n'.join(strings).strip())
         return instance

chatmastermind/utils.py

@@ -78,8 +78,3 @@ def print_chat_hist(chat: ChatType, dump: bool = False, source_code: bool = Fals
             print(message['content'])
         else:
             print(f"{message['role'].upper()}: {message['content']}")
-
-
-def print_tags_frequency(tags: list[str]) -> None:
-    for tag in sorted(set(tags)):
-        print(f"- {tag}: {tags.count(tag)}")

tests/test_message.py

@@ -61,22 +61,39 @@ class SourceCodeTestCase(CmmTestCase):
 
 
 class QuestionTestCase(CmmTestCase):
-    def test_question_with_prefix(self) -> None:
+    def test_question_with_header(self) -> None:
         with self.assertRaises(MessageError):
-            Question("=== QUESTION === What is your name?")
+            Question(f"{Question.txt_header}\nWhat is your name?")
 
-    def test_question_without_prefix(self) -> None:
+    def test_question_with_answer_header(self) -> None:
+        with self.assertRaises(MessageError):
+            Question(f"{Answer.txt_header}\nBob")
+
+    def test_question_with_legal_header(self) -> None:
+        """
+        If the header is just a part of a line, it's fine.
+        """
+        question = Question(f"This is a line contaning '{Question.txt_header}'\nWhat does that mean?")
+        self.assertIsInstance(question, Question)
+        self.assertEqual(question, f"This is a line contaning '{Question.txt_header}'\nWhat does that mean?")
+
+    def test_question_without_header(self) -> None:
         question = Question("What is your favorite color?")
         self.assertIsInstance(question, Question)
         self.assertEqual(question, "What is your favorite color?")
 
 
 class AnswerTestCase(CmmTestCase):
-    def test_answer_with_prefix(self) -> None:
+    def test_answer_with_header(self) -> None:
         with self.assertRaises(MessageError):
-            Answer("=== ANSWER === Yes")
+            Answer(f"{Answer.txt_header}\nno")
 
-    def test_answer_without_prefix(self) -> None:
+    def test_answer_with_legal_header(self) -> None:
+        answer = Answer(f"This is a line contaning '{Answer.txt_header}'\nIt is what it is.")
+        self.assertIsInstance(answer, Answer)
+        self.assertEqual(answer, f"This is a line contaning '{Answer.txt_header}'\nIt is what it is.")
+
+    def test_answer_without_header(self) -> None:
         answer = Answer("No")
         self.assertIsInstance(answer, Answer)
         self.assertEqual(answer, "No")