gptel: Turn API parameters into defcustoms

* gptel.el (gptel--request-data, gptel--system-message-alist,
gptel--model, gptel--temperature, gptel--max-tokens): Rename API
parameters and turn them into customizable variables. They are
still buffer-local.

Rename:
`gptel--system-message-alist' to `gptel-directives'
`gptel--max-tokens' to `gptel-max-tokens'
`gptel--model' to `gptel-model'
`gptel--temperature' to `gptel-temperature'

* gptel-transient.el (gptel-system-prompt,
gptel--infix-max-tokens, gptel--infix-model,
gptel--infix-temperature): Accommodating changes when setting the
renamed parameters.

gptel-transient.el

@@ -57,8 +57,7 @@ You are a helpful assistant. Answer as concisely as possible.
 Reply only with shell commands and no prose.
 You are a poet. Reply only in verse.
 
-Customize `gptel--system-message-alist' for task-specific
-prompts."
+Customize `gptel-directives' for task-specific prompts."
   [:description
    (lambda () (format "Directive: %s"
                  (truncate-string-to-width gptel--system-message (max (- (window-width) 14) 20) nil nil t)))
@@ -68,22 +67,22 @@ prompts."
    ("p" "Programming"
     (lambda () (interactive)
       (setq gptel--system-message
-            (alist-get 'programming gptel--system-message-alist)))
+            (alist-get 'programming gptel-directives)))
     :transient t)
    ("d" "Default"
     (lambda () (interactive)
       (setq gptel--system-message
-            (alist-get 'default gptel--system-message-alist)))
+            (alist-get 'default gptel-directives)))
     :transient t)
    ("w" "Writing"
     (lambda () (interactive)
       (setq gptel--system-message
-            (alist-get 'writing gptel--system-message-alist)))
+            (alist-get 'writing gptel-directives)))
     :transient t)
    ("c" "Chat"
     (lambda () (interactive)
       (setq gptel--system-message
-            (alist-get 'chat gptel--system-message-alist)))
+            (alist-get 'chat gptel-directives)))
     :transient t)])
 
 ;; TODO: Switch to dynamic Transient menus (below) once there's a new Transient release
@@ -99,7 +98,7 @@ prompts."
 ;;   "Set up suffixes for system prompt."
 ;;   (transient-parse-suffixes
 ;;    'gptel-system-prompt
-;;    (cl-loop for (type . prompt) in gptel--system-message-alist
+;;    (cl-loop for (type . prompt) in gptel-directives
 ;;        for name = (symbol-name type)
 ;;        for key = (substring name 0 1)
 ;;        collect (list (key-description key) (capitalize name)
@@ -137,7 +136,7 @@ count of the conversation so far in each message, so messages
 will get progressively longer!"
   :description "Response length (tokens)"
   :class 'transient-lisp-variable
-  :variable 'gptel--max-tokens
+  :variable 'gptel-max-tokens
   :key "<"
   :prompt "Response length in tokens (leave empty: default, 80-200: short, 200-500: long): "
   :reader 'transient-read-number-N+)
@@ -146,7 +145,7 @@ will get progressively longer!"
   "AI Model for Chat."
   :description "GPT Model: "
   :class 'transient-lisp-variable
-  :variable 'gptel--model
+  :variable 'gptel-model
   :key "m"
   :choices '("gpt-3.5-turbo-0301" "gpt-3.5-turbo" "gpt-4")
   :reader (lambda (prompt &rest _)
@@ -158,11 +157,11 @@ will get progressively longer!"
   "Temperature of request."
   :description "Randomness (0 - 2.0)"
   :class 'transient-lisp-variable
-  :variable 'gptel--temperature
+  :variable 'gptel-temperature
   :key "t"
   :reader (lambda (&rest _)
             (read-from-minibuffer "Set temperature (0.0-2.0, leave empty for default): "
-                                  (number-to-string gptel--temperature))))
+                                  (number-to-string gptel-temperature))))
 
 (transient-define-suffix gptel--suffix-send-existing ()
   "Send query in existing chat session."

gptel.el

@@ -124,17 +124,76 @@ is only inserted in dedicated gptel buffers."
 ;; Model and interaction parameters
 (defvar-local gptel--system-message
   "You are a large language model living in Emacs and a helpful assistant. Respond concisely.")
-(defvar gptel--system-message-alist
+
+(defcustom gptel-directives
   `((default . ,gptel--system-message)
     (programming . "You are a large language model and a careful programmer. Provide code and only code as output without any additional text, prompt or note.")
     (writing . "You are a large language model and a writing assistant. Respond concisely.")
     (chat . "You are a large language model and a conversation partner. Respond concisely."))
-  "Prompt templates (directives).")
-(defvar gptel--debug nil)
-(defvar-local gptel--max-tokens nil)
-(defvar-local gptel--model "gpt-3.5-turbo")
-(defvar-local gptel--temperature 1.0)
+  "System prompts (directives) for ChatGPT.
+
+These are system instructions sent at the beginning of each
+request to ChatGPT.
+
+Each entry in this alist maps a symbol naming the directive to
+the string that is sent. To set the directive for a chat session
+interactively call `gptel-send' with a prefix argument.
+
+Note: Currently the names (default, programming, writing and
+chat) are hard-coded and only their values may be customized.
+This will be fixed in an upcoming release."
+  :group 'gptel
+  :type '(alist :key-type symbol :value-type string))
+
+(defcustom gptel-max-tokens nil
+  "Max tokens per response.
+
+This is roughly the number of words in the response. 100-300 is a
+reasonable range for short answers, 400 or more for longer
+responses.
+
+To set the target token count for a chat session interactively
+call `gptel-send' with a prefix argument.
+
+If left unset, ChatGPT will target about 40% of the total token
+count of the conversation so far in each message, so messages
+will get progressively longer!"
+  :local t
+  :group 'gptel
+  :type '(choice (integer :tag "Specify Token count")
+                 (const :tag "Default" nil)))
+
+(defcustom gptel-model "gpt-3.5-turbo"
+  "GPT Model for chat.
+
+The current options are
+- \"gpt-3.5-turbo\"
+- \"gpt-3.5-turbo-0301\"
+- \"gpt-4\" (experimental)
+
+To set the model for a chat session interactively call
+`gptel-send' with a prefix argument."
+  :local t
+  :group 'gptel
+  :type '(choice
+          (const :tag "GPT 3.5 turbo" "gpt-3.5-turbo")
+          (const :tag "GPT 3.5 turbo 0301" "gpt-3.5-turbo-0301")
+          (const :tag "GPT 4 (experimental)" "gpt-4")))
+
+(defcustom gptel-temperature 1.0
+  "\"Temperature\" of ChatGPT response.
+
+This is a number between 0.0 and 2.0 that controls the randomness
+of the response, with 2.0 being the most random.
+
+To set the temperature for a chat session interactively call
+`gptel-send' with a prefix argument."
+  :local t
+  :group 'gptel
+  :type 'number)
+
 (defvar-local gptel--num-messages-to-send nil)
+(defvar gptel--debug nil)
 
 (defun gptel-api-key-from-auth-source (&optional host user)
   "Lookup api key in the auth source.
@@ -285,12 +344,12 @@ there."
 (defun gptel--request-data (prompts)
   "JSON encode PROMPTS for sending to ChatGPT."
   (let ((prompts-plist
-         `(:model ,gptel--model
+         `(:model ,gptel-model
            :messages [,@prompts])))
-    (when gptel--temperature
-      (plist-put prompts-plist :temperature (gptel--numberize gptel--temperature)))
-    (when gptel--max-tokens
-      (plist-put prompts-plist :max_tokens (gptel--numberize gptel--max-tokens)))
+    (when gptel-temperature
+      (plist-put prompts-plist :temperature (gptel--numberize gptel-temperature)))
+    (when gptel-max-tokens
+      (plist-put prompts-plist :max_tokens (gptel--numberize gptel-max-tokens)))
     prompts-plist))
 
 ;; TODO: Use `run-hook-wrapped' with an accumulator instead to handle