From f6109949c816d350244d396a3d13ea302f8c38b8 Mon Sep 17 00:00:00 2001
From: juk0de <git@kadidlo.de>
Date: Thu, 14 Sep 2023 16:05:18 +0200
Subject: [PATCH] chat: ChatDB now correctly ignores files that contain no
 valid messages

---
 chatmastermind/chat.py    |  2 +-
 chatmastermind/message.py | 11 +++++++----
 tests/test_chat.py        | 14 +++++++++++++-
 3 files changed, 21 insertions(+), 6 deletions(-)

diff --git a/chatmastermind/chat.py b/chatmastermind/chat.py
index 8d64f86..c1464c3 100644
--- a/chatmastermind/chat.py
+++ b/chatmastermind/chat.py
@@ -57,7 +57,7 @@ def read_dir(dir_path: Path,
                 if message:
                     messages.append(message)
             except MessageError as e:
-                print(f"Error processing message in '{file_path}': {str(e)}")
+                print(f"WARNING: Skipping message in '{file_path}': {str(e)}")
     return messages
 
 
diff --git a/chatmastermind/message.py b/chatmastermind/message.py
index 64929a3..402a6d1 100644
--- a/chatmastermind/message.py
+++ b/chatmastermind/message.py
@@ -370,7 +370,7 @@ class Message():
             try:
                 question_idx = text.index(Question.txt_header) + 1
             except ValueError:
-                raise MessageError(f"Question header '{Question.txt_header}' not found in '{file_path}'")
+                raise MessageError(f"'{file_path}' does not contain a valid message")
             try:
                 answer_idx = text.index(Answer.txt_header)
                 question = Question.from_list(text[question_idx:answer_idx])
@@ -390,9 +390,12 @@ class Message():
             * Message.model_yaml_key: str [Optional]
         """
         with open(file_path, "r") as fd:
-            data = yaml.load(fd, Loader=yaml.FullLoader)
-            data[cls.file_yaml_key] = file_path
-            return cls.from_dict(data)
+            try:
+                data = yaml.load(fd, Loader=yaml.FullLoader)
+                data[cls.file_yaml_key] = file_path
+                return cls.from_dict(data)
+            except Exception:
+                raise MessageError(f"'{file_path}' does not contain a valid message")
 
     def to_str(self, with_tags: bool = False, with_file: bool = False, source_code_only: bool = False) -> str:
         """
diff --git a/tests/test_chat.py b/tests/test_chat.py
index ff44cda..bca3a9f 100644
--- a/tests/test_chat.py
+++ b/tests/test_chat.py
@@ -2,6 +2,7 @@ import unittest
 import pathlib
 import tempfile
 import time
+import yaml
 from io import StringIO
 from unittest.mock import patch
 from chatmastermind.tags import TagLine
@@ -156,13 +157,24 @@ class TestChatDB(unittest.TestCase):
         next_fname = pathlib.Path(self.db_path.name) / '.next'
         with open(next_fname, 'w') as f:
             f.write('4')
+        # add some "trash" in order to test if it's correctly handled / ignored
+        self.trash_files = ['.config.yaml', 'foo.yaml', 'bla.txt']
+        for file in self.trash_files:
+            with open(pathlib.Path(self.db_path.name) / file, 'w') as f:
+                f.write('test trash')
+        # also create a file with actual yaml content
+        with open(pathlib.Path(self.db_path.name) / 'content.yaml', 'w') as f:
+            yaml.dump({'key': 'value'}, f)
+        self.trash_files.append('content.yaml')
+
+        self.maxDiff = None
 
     def message_list(self, tmp_dir: tempfile.TemporaryDirectory) -> list[pathlib.Path]:
         """
         List all Message files in the given TemporaryDirectory.
         """
         # exclude '.next'
-        return list(pathlib.Path(tmp_dir.name).glob('*.[ty]*'))
+        return [f for f in pathlib.Path(tmp_dir.name).glob('*.[ty]*') if f.name not in self.trash_files]
 
     def tearDown(self) -> None:
         self.db_path.cleanup()