[llm] Add support for --help flag

CHANGELOG.md

@@ -2,7 +2,7 @@
 ## TBD
 
 ### Features
-
+* Support `--help` in the `\llm`and `\llm+` command. ([#214](https://github.com/dbcli/litecli/pull/214))
 * Make the history file location configurable. ([#206](https://github.com/dbcli/litecli/issues/206))
 
 ### Internal

litecli/packages/special/llm.py

@@ -236,6 +236,10 @@ def handle_llm(text, cur) -> Tuple[str, Optional[str]]:
     elif parts[0] in LLM_CLI_COMMANDS:
         capture_output = False
         use_context = False
+    # If the user wants to use --help option to see each command and it's description
+    elif "--help" == parts[0]:
+        capture_output = False
+        use_context = False
     # If the parts doesn't have any known LLM_CLI_COMMANDS then the user is
     # invoking a question. eg: \llm -m ollama "Most visited urls?"
     elif not set(parts).intersection(LLM_CLI_COMMANDS):

tests/test_llm_special.py

@@ -87,6 +87,23 @@ def test_llm_command_known_subcommand(mock_run_cmd, mock_llm, executor):
     # And the function should raise FinishIteration(None)
     assert exc_info.value.args[0] is None
 
+@patch("litecli.packages.special.llm.llm")
+@patch("litecli.packages.special.llm.run_external_cmd")
+def test_llm_command_with_help_flag(mock_run_cmd, mock_llm, executor):
+    """
+    If the parts[0] is --help, we do NOT capture output, we just call run_external_cmd
+    and then raise FinishIteration.
+    """
+    # Let's assume 'models' is in LLM_CLI_COMMANDS
+    test_text = r"\llm --help"
+
+    with pytest.raises(FinishIteration) as exc_info:
+        handle_llm(test_text, executor)
+
+    # We check that run_external_cmd was called with these arguments:
+    mock_run_cmd.assert_called_once_with("llm", "--help", restart_cli=False)
+    # And the function should raise FinishIteration(None)
+    assert exc_info.value.args[0] is None
 
 @patch("litecli.packages.special.llm.llm")
 @patch("litecli.packages.special.llm.run_external_cmd")